Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * parser.c : an XML 1.0 parser, namespaces and validity support are mostly |
3 | | * implemented on top of the SAX interfaces |
4 | | * |
5 | | * References: |
6 | | * The XML specification: |
7 | | * http://www.w3.org/TR/REC-xml |
8 | | * Original 1.0 version: |
9 | | * http://www.w3.org/TR/1998/REC-xml-19980210 |
10 | | * XML second edition working draft |
11 | | * http://www.w3.org/TR/2000/WD-xml-2e-20000814 |
12 | | * |
13 | | * Okay this is a big file, the parser core is around 7000 lines, then it |
14 | | * is followed by the progressive parser top routines, then the various |
15 | | * high level APIs to call the parser and a few miscellaneous functions. |
16 | | * A number of helper functions and deprecated ones have been moved to |
17 | | * parserInternals.c to reduce this file size. |
18 | | * As much as possible the functions are associated with their relative |
19 | | * production in the XML specification. A few productions defining the |
20 | | * different ranges of character are actually implanted either in |
21 | | * parserInternals.h or parserInternals.c |
22 | | * The DOM tree build is realized from the default SAX callbacks in |
23 | | * the module SAX.c. |
24 | | * The routines doing the validation checks are in valid.c and called either |
25 | | * from the SAX callbacks or as standalone functions using a preparsed |
26 | | * document. |
27 | | * |
28 | | * See Copyright for the status of this software. |
29 | | * |
30 | | * daniel@veillard.com |
31 | | */ |
32 | | |
33 | | /* To avoid EBCDIC trouble when parsing on zOS */ |
34 | | #if defined(__MVS__) |
35 | | #pragma convert("ISO8859-1") |
36 | | #endif |
37 | | |
38 | | #define IN_LIBXML |
39 | | #include "libxml.h" |
40 | | |
41 | | #if defined(_WIN32) |
42 | | #define XML_DIR_SEP '\\' |
43 | | #else |
44 | | #define XML_DIR_SEP '/' |
45 | | #endif |
46 | | |
47 | | #include <stdlib.h> |
48 | | #include <limits.h> |
49 | | #include <string.h> |
50 | | #include <stdarg.h> |
51 | | #include <stddef.h> |
52 | | #include <ctype.h> |
53 | | #include <stdlib.h> |
54 | | #include <libxml/xmlmemory.h> |
55 | | #include <libxml/threads.h> |
56 | | #include <libxml/globals.h> |
57 | | #include <libxml/tree.h> |
58 | | #include <libxml/parser.h> |
59 | | #include <libxml/parserInternals.h> |
60 | | #include <libxml/HTMLparser.h> |
61 | | #include <libxml/valid.h> |
62 | | #include <libxml/entities.h> |
63 | | #include <libxml/xmlerror.h> |
64 | | #include <libxml/encoding.h> |
65 | | #include <libxml/xmlIO.h> |
66 | | #include <libxml/uri.h> |
67 | | #ifdef LIBXML_CATALOG_ENABLED |
68 | | #include <libxml/catalog.h> |
69 | | #endif |
70 | | #ifdef LIBXML_SCHEMAS_ENABLED |
71 | | #include <libxml/xmlschemastypes.h> |
72 | | #include <libxml/relaxng.h> |
73 | | #endif |
74 | | #if defined(LIBXML_XPATH_ENABLED) || defined(LIBXML_SCHEMAS_ENABLED) |
75 | | #include <libxml/xpath.h> |
76 | | #endif |
77 | | |
78 | | #include "private/buf.h" |
79 | | #include "private/enc.h" |
80 | | #include "private/error.h" |
81 | | #include "private/html.h" |
82 | | #include "private/io.h" |
83 | | #include "private/parser.h" |
84 | | #include "private/threads.h" |
85 | | |
86 | | struct _xmlStartTag { |
87 | | const xmlChar *prefix; |
88 | | const xmlChar *URI; |
89 | | int line; |
90 | | int nsNr; |
91 | | }; |
92 | | |
93 | | static void |
94 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info); |
95 | | |
96 | | static xmlParserCtxtPtr |
97 | | xmlCreateEntityParserCtxtInternal(xmlSAXHandlerPtr sax, void *userData, |
98 | | const xmlChar *URL, const xmlChar *ID, const xmlChar *base, |
99 | | xmlParserCtxtPtr pctx); |
100 | | |
101 | | static void xmlHaltParser(xmlParserCtxtPtr ctxt); |
102 | | |
103 | | static int |
104 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
105 | | |
106 | | static void |
107 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
108 | | |
109 | | /************************************************************************ |
110 | | * * |
111 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
112 | | * * |
113 | | ************************************************************************/ |
114 | | |
115 | 83.1M | #define XML_MAX_HUGE_LENGTH 1000000000 |
116 | | |
117 | 96.3k | #define XML_PARSER_BIG_ENTITY 1000 |
118 | | #define XML_PARSER_LOT_ENTITY 5000 |
119 | | |
120 | | /* |
121 | | * XML_PARSER_NON_LINEAR is the threshold where the ratio of parsed entity |
122 | | * replacement over the size in byte of the input indicates that you have |
123 | | * and exponential behaviour. A value of 10 correspond to at least 3 entity |
124 | | * replacement per byte of input. |
125 | | */ |
126 | 2.65M | #define XML_PARSER_NON_LINEAR 10 |
127 | | |
128 | | /* |
129 | | * xmlParserEntityCheck |
130 | | * |
131 | | * Function to check non-linear entity expansion behaviour |
132 | | * This is here to detect and stop exponential linear entity expansion |
133 | | * This is not a limitation of the parser but a safety |
134 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
135 | | * parser option. |
136 | | */ |
137 | | static int |
138 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, size_t size, |
139 | | xmlEntityPtr ent, size_t replacement) |
140 | 15.9M | { |
141 | 15.9M | size_t consumed = 0; |
142 | 15.9M | int i; |
143 | | |
144 | 15.9M | if ((ctxt == NULL) || (ctxt->options & XML_PARSE_HUGE)) |
145 | 11.8M | return (0); |
146 | 4.10M | if (ctxt->lastError.code == XML_ERR_ENTITY_LOOP) |
147 | 2.39k | return (1); |
148 | | |
149 | | /* |
150 | | * This may look absurd but is needed to detect |
151 | | * entities problems |
152 | | */ |
153 | 4.10M | if ((ent != NULL) && (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
154 | 4.10M | (ent->content != NULL) && (ent->checked == 0) && |
155 | 4.10M | (ctxt->errNo != XML_ERR_ENTITY_LOOP)) { |
156 | 481k | unsigned long oldnbent = ctxt->nbentities, diff; |
157 | 481k | xmlChar *rep; |
158 | | |
159 | 481k | ent->checked = 1; |
160 | | |
161 | 481k | ++ctxt->depth; |
162 | 481k | rep = xmlStringDecodeEntities(ctxt, ent->content, |
163 | 481k | XML_SUBSTITUTE_REF, 0, 0, 0); |
164 | 481k | --ctxt->depth; |
165 | 481k | if ((rep == NULL) || (ctxt->errNo == XML_ERR_ENTITY_LOOP)) { |
166 | 1.23k | ent->content[0] = 0; |
167 | 1.23k | } |
168 | | |
169 | 481k | diff = ctxt->nbentities - oldnbent + 1; |
170 | 481k | if (diff > INT_MAX / 2) |
171 | 0 | diff = INT_MAX / 2; |
172 | 481k | ent->checked = diff * 2; |
173 | 481k | if (rep != NULL) { |
174 | 480k | if (xmlStrchr(rep, '<')) |
175 | 8.94k | ent->checked |= 1; |
176 | 480k | xmlFree(rep); |
177 | 480k | rep = NULL; |
178 | 480k | } |
179 | 481k | } |
180 | | |
181 | | /* |
182 | | * Prevent entity exponential check, not just replacement while |
183 | | * parsing the DTD |
184 | | * The check is potentially costly so do that only once in a thousand |
185 | | */ |
186 | 4.10M | if ((ctxt->instate == XML_PARSER_DTD) && (ctxt->nbentities > 10000) && |
187 | 4.10M | (ctxt->nbentities % 1024 == 0)) { |
188 | 0 | for (i = 0;i < ctxt->inputNr;i++) { |
189 | 0 | consumed += ctxt->inputTab[i]->consumed + |
190 | 0 | (ctxt->inputTab[i]->cur - ctxt->inputTab[i]->base); |
191 | 0 | } |
192 | 0 | if (ctxt->nbentities > consumed * XML_PARSER_NON_LINEAR) { |
193 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
194 | 0 | ctxt->instate = XML_PARSER_EOF; |
195 | 0 | return (1); |
196 | 0 | } |
197 | 0 | consumed = 0; |
198 | 0 | } |
199 | | |
200 | | |
201 | | |
202 | 4.10M | if (replacement != 0) { |
203 | 36.7k | if (replacement < XML_MAX_TEXT_LENGTH) |
204 | 36.7k | return(0); |
205 | | |
206 | | /* |
207 | | * If the volume of entity copy reaches 10 times the |
208 | | * amount of parsed data and over the large text threshold |
209 | | * then that's very likely to be an abuse. |
210 | | */ |
211 | 0 | if (ctxt->input != NULL) { |
212 | 0 | consumed = ctxt->input->consumed + |
213 | 0 | (ctxt->input->cur - ctxt->input->base); |
214 | 0 | } |
215 | 0 | consumed += ctxt->sizeentities; |
216 | |
|
217 | 0 | if (replacement < XML_PARSER_NON_LINEAR * consumed) |
218 | 0 | return(0); |
219 | 4.07M | } else if (size != 0) { |
220 | | /* |
221 | | * Do the check based on the replacement size of the entity |
222 | | */ |
223 | 96.3k | if (size < XML_PARSER_BIG_ENTITY) |
224 | 91.7k | return(0); |
225 | | |
226 | | /* |
227 | | * A limit on the amount of text data reasonably used |
228 | | */ |
229 | 4.59k | if (ctxt->input != NULL) { |
230 | 4.59k | consumed = ctxt->input->consumed + |
231 | 4.59k | (ctxt->input->cur - ctxt->input->base); |
232 | 4.59k | } |
233 | 4.59k | consumed += ctxt->sizeentities; |
234 | | |
235 | 4.59k | if ((size < XML_PARSER_NON_LINEAR * consumed) && |
236 | 4.59k | (ctxt->nbentities * 3 < XML_PARSER_NON_LINEAR * consumed)) |
237 | 3.50k | return (0); |
238 | 3.97M | } else if (ent != NULL) { |
239 | | /* |
240 | | * use the number of parsed entities in the replacement |
241 | | */ |
242 | 2.65M | size = ent->checked / 2; |
243 | | |
244 | | /* |
245 | | * The amount of data parsed counting entities size only once |
246 | | */ |
247 | 2.65M | if (ctxt->input != NULL) { |
248 | 2.65M | consumed = ctxt->input->consumed + |
249 | 2.65M | (ctxt->input->cur - ctxt->input->base); |
250 | 2.65M | } |
251 | 2.65M | consumed += ctxt->sizeentities; |
252 | | |
253 | | /* |
254 | | * Check the density of entities for the amount of data |
255 | | * knowing an entity reference will take at least 3 bytes |
256 | | */ |
257 | 2.65M | if (size * 3 < consumed * XML_PARSER_NON_LINEAR) |
258 | 2.64M | return (0); |
259 | 2.65M | } else { |
260 | | /* |
261 | | * strange we got no data for checking |
262 | | */ |
263 | 1.32M | if (((ctxt->lastError.code != XML_ERR_UNDECLARED_ENTITY) && |
264 | 1.32M | (ctxt->lastError.code != XML_WAR_UNDECLARED_ENTITY)) || |
265 | 1.32M | (ctxt->nbentities <= 10000)) |
266 | 1.28M | return (0); |
267 | 1.32M | } |
268 | 41.2k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
269 | 41.2k | return (1); |
270 | 4.10M | } |
271 | | |
272 | | /** |
273 | | * xmlParserMaxDepth: |
274 | | * |
275 | | * arbitrary depth limit for the XML documents that we allow to |
276 | | * process. This is not a limitation of the parser but a safety |
277 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
278 | | * parser option. |
279 | | */ |
280 | | unsigned int xmlParserMaxDepth = 256; |
281 | | |
282 | | |
283 | | |
284 | | #define SAX2 1 |
285 | 1.56G | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
286 | 1.10G | #define XML_PARSER_BUFFER_SIZE 100 |
287 | 1.27M | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
288 | | |
289 | | /** |
290 | | * XML_PARSER_CHUNK_SIZE |
291 | | * |
292 | | * When calling GROW that's the minimal amount of data |
293 | | * the parser expected to have received. It is not a hard |
294 | | * limit but an optimization when reading strings like Names |
295 | | * It is not strictly needed as long as inputs available characters |
296 | | * are followed by 0, which should be provided by the I/O level |
297 | | */ |
298 | 177M | #define XML_PARSER_CHUNK_SIZE 100 |
299 | | |
300 | | /* |
301 | | * List of XML prefixed PI allowed by W3C specs |
302 | | */ |
303 | | |
304 | | static const char* const xmlW3CPIs[] = { |
305 | | "xml-stylesheet", |
306 | | "xml-model", |
307 | | NULL |
308 | | }; |
309 | | |
310 | | |
311 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
312 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
313 | | const xmlChar **str); |
314 | | |
315 | | static xmlParserErrors |
316 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
317 | | xmlSAXHandlerPtr sax, |
318 | | void *user_data, int depth, const xmlChar *URL, |
319 | | const xmlChar *ID, xmlNodePtr *list); |
320 | | |
321 | | static int |
322 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, |
323 | | const char *encoding); |
324 | | #ifdef LIBXML_LEGACY_ENABLED |
325 | | static void |
326 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
327 | | xmlNodePtr lastNode); |
328 | | #endif /* LIBXML_LEGACY_ENABLED */ |
329 | | |
330 | | static xmlParserErrors |
331 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
332 | | const xmlChar *string, void *user_data, xmlNodePtr *lst); |
333 | | |
334 | | static int |
335 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
336 | | |
337 | | /************************************************************************ |
338 | | * * |
339 | | * Some factorized error routines * |
340 | | * * |
341 | | ************************************************************************/ |
342 | | |
343 | | /** |
344 | | * xmlErrAttributeDup: |
345 | | * @ctxt: an XML parser context |
346 | | * @prefix: the attribute prefix |
347 | | * @localname: the attribute localname |
348 | | * |
349 | | * Handle a redefinition of attribute error |
350 | | */ |
351 | | static void |
352 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
353 | | const xmlChar * localname) |
354 | 119k | { |
355 | 119k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
356 | 119k | (ctxt->instate == XML_PARSER_EOF)) |
357 | 0 | return; |
358 | 119k | if (ctxt != NULL) |
359 | 119k | ctxt->errNo = XML_ERR_ATTRIBUTE_REDEFINED; |
360 | | |
361 | 119k | if (prefix == NULL) |
362 | 111k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
363 | 111k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
364 | 111k | (const char *) localname, NULL, NULL, 0, 0, |
365 | 111k | "Attribute %s redefined\n", localname); |
366 | 8.49k | else |
367 | 8.49k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
368 | 8.49k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
369 | 8.49k | (const char *) prefix, (const char *) localname, |
370 | 8.49k | NULL, 0, 0, "Attribute %s:%s redefined\n", prefix, |
371 | 8.49k | localname); |
372 | 119k | if (ctxt != NULL) { |
373 | 119k | ctxt->wellFormed = 0; |
374 | 119k | if (ctxt->recovery == 0) |
375 | 74.8k | ctxt->disableSAX = 1; |
376 | 119k | } |
377 | 119k | } |
378 | | |
379 | | /** |
380 | | * xmlFatalErr: |
381 | | * @ctxt: an XML parser context |
382 | | * @error: the error number |
383 | | * @extra: extra information string |
384 | | * |
385 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
386 | | */ |
387 | | static void |
388 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info) |
389 | 16.6M | { |
390 | 16.6M | const char *errmsg; |
391 | | |
392 | 16.6M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
393 | 16.6M | (ctxt->instate == XML_PARSER_EOF)) |
394 | 721k | return; |
395 | 15.9M | switch (error) { |
396 | 340k | case XML_ERR_INVALID_HEX_CHARREF: |
397 | 340k | errmsg = "CharRef: invalid hexadecimal value"; |
398 | 340k | break; |
399 | 728k | case XML_ERR_INVALID_DEC_CHARREF: |
400 | 728k | errmsg = "CharRef: invalid decimal value"; |
401 | 728k | break; |
402 | 0 | case XML_ERR_INVALID_CHARREF: |
403 | 0 | errmsg = "CharRef: invalid value"; |
404 | 0 | break; |
405 | 347k | case XML_ERR_INTERNAL_ERROR: |
406 | 347k | errmsg = "internal error"; |
407 | 347k | break; |
408 | 0 | case XML_ERR_PEREF_AT_EOF: |
409 | 0 | errmsg = "PEReference at end of document"; |
410 | 0 | break; |
411 | 0 | case XML_ERR_PEREF_IN_PROLOG: |
412 | 0 | errmsg = "PEReference in prolog"; |
413 | 0 | break; |
414 | 0 | case XML_ERR_PEREF_IN_EPILOG: |
415 | 0 | errmsg = "PEReference in epilog"; |
416 | 0 | break; |
417 | 0 | case XML_ERR_PEREF_NO_NAME: |
418 | 0 | errmsg = "PEReference: no name"; |
419 | 0 | break; |
420 | 19.5k | case XML_ERR_PEREF_SEMICOL_MISSING: |
421 | 19.5k | errmsg = "PEReference: expecting ';'"; |
422 | 19.5k | break; |
423 | 837k | case XML_ERR_ENTITY_LOOP: |
424 | 837k | errmsg = "Detected an entity reference loop"; |
425 | 837k | break; |
426 | 0 | case XML_ERR_ENTITY_NOT_STARTED: |
427 | 0 | errmsg = "EntityValue: \" or ' expected"; |
428 | 0 | break; |
429 | 1.60k | case XML_ERR_ENTITY_PE_INTERNAL: |
430 | 1.60k | errmsg = "PEReferences forbidden in internal subset"; |
431 | 1.60k | break; |
432 | 3.44k | case XML_ERR_ENTITY_NOT_FINISHED: |
433 | 3.44k | errmsg = "EntityValue: \" or ' expected"; |
434 | 3.44k | break; |
435 | 404k | case XML_ERR_ATTRIBUTE_NOT_STARTED: |
436 | 404k | errmsg = "AttValue: \" or ' expected"; |
437 | 404k | break; |
438 | 1.39M | case XML_ERR_LT_IN_ATTRIBUTE: |
439 | 1.39M | errmsg = "Unescaped '<' not allowed in attributes values"; |
440 | 1.39M | break; |
441 | 8.59k | case XML_ERR_LITERAL_NOT_STARTED: |
442 | 8.59k | errmsg = "SystemLiteral \" or ' expected"; |
443 | 8.59k | break; |
444 | 11.5k | case XML_ERR_LITERAL_NOT_FINISHED: |
445 | 11.5k | errmsg = "Unfinished System or Public ID \" or ' expected"; |
446 | 11.5k | break; |
447 | 1.05M | case XML_ERR_MISPLACED_CDATA_END: |
448 | 1.05M | errmsg = "Sequence ']]>' not allowed in content"; |
449 | 1.05M | break; |
450 | 7.58k | case XML_ERR_URI_REQUIRED: |
451 | 7.58k | errmsg = "SYSTEM or PUBLIC, the URI is missing"; |
452 | 7.58k | break; |
453 | 1.05k | case XML_ERR_PUBID_REQUIRED: |
454 | 1.05k | errmsg = "PUBLIC, the Public Identifier is missing"; |
455 | 1.05k | break; |
456 | 691k | case XML_ERR_HYPHEN_IN_COMMENT: |
457 | 691k | errmsg = "Comment must not contain '--' (double-hyphen)"; |
458 | 691k | break; |
459 | 266k | case XML_ERR_PI_NOT_STARTED: |
460 | 266k | errmsg = "xmlParsePI : no target name"; |
461 | 266k | break; |
462 | 7.69k | case XML_ERR_RESERVED_XML_NAME: |
463 | 7.69k | errmsg = "Invalid PI name"; |
464 | 7.69k | break; |
465 | 522 | case XML_ERR_NOTATION_NOT_STARTED: |
466 | 522 | errmsg = "NOTATION: Name expected here"; |
467 | 522 | break; |
468 | 2.36k | case XML_ERR_NOTATION_NOT_FINISHED: |
469 | 2.36k | errmsg = "'>' required to close NOTATION declaration"; |
470 | 2.36k | break; |
471 | 11.2k | case XML_ERR_VALUE_REQUIRED: |
472 | 11.2k | errmsg = "Entity value required"; |
473 | 11.2k | break; |
474 | 1.21k | case XML_ERR_URI_FRAGMENT: |
475 | 1.21k | errmsg = "Fragment not allowed"; |
476 | 1.21k | break; |
477 | 13.7k | case XML_ERR_ATTLIST_NOT_STARTED: |
478 | 13.7k | errmsg = "'(' required to start ATTLIST enumeration"; |
479 | 13.7k | break; |
480 | 577 | case XML_ERR_NMTOKEN_REQUIRED: |
481 | 577 | errmsg = "NmToken expected in ATTLIST enumeration"; |
482 | 577 | break; |
483 | 2.40k | case XML_ERR_ATTLIST_NOT_FINISHED: |
484 | 2.40k | errmsg = "')' required to finish ATTLIST enumeration"; |
485 | 2.40k | break; |
486 | 3.21k | case XML_ERR_MIXED_NOT_STARTED: |
487 | 3.21k | errmsg = "MixedContentDecl : '|' or ')*' expected"; |
488 | 3.21k | break; |
489 | 0 | case XML_ERR_PCDATA_REQUIRED: |
490 | 0 | errmsg = "MixedContentDecl : '#PCDATA' expected"; |
491 | 0 | break; |
492 | 73.7k | case XML_ERR_ELEMCONTENT_NOT_STARTED: |
493 | 73.7k | errmsg = "ContentDecl : Name or '(' expected"; |
494 | 73.7k | break; |
495 | 15.2k | case XML_ERR_ELEMCONTENT_NOT_FINISHED: |
496 | 15.2k | errmsg = "ContentDecl : ',' '|' or ')' expected"; |
497 | 15.2k | break; |
498 | 0 | case XML_ERR_PEREF_IN_INT_SUBSET: |
499 | 0 | errmsg = |
500 | 0 | "PEReference: forbidden within markup decl in internal subset"; |
501 | 0 | break; |
502 | 3.27M | case XML_ERR_GT_REQUIRED: |
503 | 3.27M | errmsg = "expected '>'"; |
504 | 3.27M | break; |
505 | 484 | case XML_ERR_CONDSEC_INVALID: |
506 | 484 | errmsg = "XML conditional section '[' expected"; |
507 | 484 | break; |
508 | 36.9k | case XML_ERR_EXT_SUBSET_NOT_FINISHED: |
509 | 36.9k | errmsg = "Content error in the external subset"; |
510 | 36.9k | break; |
511 | 2.38k | case XML_ERR_CONDSEC_INVALID_KEYWORD: |
512 | 2.38k | errmsg = |
513 | 2.38k | "conditional section INCLUDE or IGNORE keyword expected"; |
514 | 2.38k | break; |
515 | 2.88k | case XML_ERR_CONDSEC_NOT_FINISHED: |
516 | 2.88k | errmsg = "XML conditional section not closed"; |
517 | 2.88k | break; |
518 | 222 | case XML_ERR_XMLDECL_NOT_STARTED: |
519 | 222 | errmsg = "Text declaration '<?xml' required"; |
520 | 222 | break; |
521 | 164k | case XML_ERR_XMLDECL_NOT_FINISHED: |
522 | 164k | errmsg = "parsing XML declaration: '?>' expected"; |
523 | 164k | break; |
524 | 0 | case XML_ERR_EXT_ENTITY_STANDALONE: |
525 | 0 | errmsg = "external parsed entities cannot be standalone"; |
526 | 0 | break; |
527 | 2.32M | case XML_ERR_ENTITYREF_SEMICOL_MISSING: |
528 | 2.32M | errmsg = "EntityRef: expecting ';'"; |
529 | 2.32M | break; |
530 | 112k | case XML_ERR_DOCTYPE_NOT_FINISHED: |
531 | 112k | errmsg = "DOCTYPE improperly terminated"; |
532 | 112k | break; |
533 | 0 | case XML_ERR_LTSLASH_REQUIRED: |
534 | 0 | errmsg = "EndTag: '</' not found"; |
535 | 0 | break; |
536 | 8.98k | case XML_ERR_EQUAL_REQUIRED: |
537 | 8.98k | errmsg = "expected '='"; |
538 | 8.98k | break; |
539 | 34.3k | case XML_ERR_STRING_NOT_CLOSED: |
540 | 34.3k | errmsg = "String not closed expecting \" or '"; |
541 | 34.3k | break; |
542 | 8.08k | case XML_ERR_STRING_NOT_STARTED: |
543 | 8.08k | errmsg = "String not started expecting ' or \""; |
544 | 8.08k | break; |
545 | 1.07k | case XML_ERR_ENCODING_NAME: |
546 | 1.07k | errmsg = "Invalid XML encoding name"; |
547 | 1.07k | break; |
548 | 997 | case XML_ERR_STANDALONE_VALUE: |
549 | 997 | errmsg = "standalone accepts only 'yes' or 'no'"; |
550 | 997 | break; |
551 | 50.4k | case XML_ERR_DOCUMENT_EMPTY: |
552 | 50.4k | errmsg = "Document is empty"; |
553 | 50.4k | break; |
554 | 243k | case XML_ERR_DOCUMENT_END: |
555 | 243k | errmsg = "Extra content at the end of the document"; |
556 | 243k | break; |
557 | 3.24M | case XML_ERR_NOT_WELL_BALANCED: |
558 | 3.24M | errmsg = "chunk is not well balanced"; |
559 | 3.24M | break; |
560 | 0 | case XML_ERR_EXTRA_CONTENT: |
561 | 0 | errmsg = "extra content at the end of well balanced chunk"; |
562 | 0 | break; |
563 | 72.4k | case XML_ERR_VERSION_MISSING: |
564 | 72.4k | errmsg = "Malformed declaration expecting version"; |
565 | 72.4k | break; |
566 | 153 | case XML_ERR_NAME_TOO_LONG: |
567 | 153 | errmsg = "Name too long"; |
568 | 153 | break; |
569 | | #if 0 |
570 | | case: |
571 | | errmsg = ""; |
572 | | break; |
573 | | #endif |
574 | 83.1k | default: |
575 | 83.1k | errmsg = "Unregistered error message"; |
576 | 15.9M | } |
577 | 15.9M | if (ctxt != NULL) |
578 | 15.9M | ctxt->errNo = error; |
579 | 15.9M | if (info == NULL) { |
580 | 15.5M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
581 | 15.5M | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s\n", |
582 | 15.5M | errmsg); |
583 | 15.5M | } else { |
584 | 347k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
585 | 347k | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s: %s\n", |
586 | 347k | errmsg, info); |
587 | 347k | } |
588 | 15.9M | if (ctxt != NULL) { |
589 | 15.9M | ctxt->wellFormed = 0; |
590 | 15.9M | if (ctxt->recovery == 0) |
591 | 13.7M | ctxt->disableSAX = 1; |
592 | 15.9M | } |
593 | 15.9M | } |
594 | | |
595 | | /** |
596 | | * xmlFatalErrMsg: |
597 | | * @ctxt: an XML parser context |
598 | | * @error: the error number |
599 | | * @msg: the error message |
600 | | * |
601 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
602 | | */ |
603 | | static void LIBXML_ATTR_FORMAT(3,0) |
604 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
605 | | const char *msg) |
606 | 134M | { |
607 | 134M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
608 | 134M | (ctxt->instate == XML_PARSER_EOF)) |
609 | 0 | return; |
610 | 134M | if (ctxt != NULL) |
611 | 134M | ctxt->errNo = error; |
612 | 134M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
613 | 134M | XML_ERR_FATAL, NULL, 0, NULL, NULL, NULL, 0, 0, "%s", msg); |
614 | 134M | if (ctxt != NULL) { |
615 | 134M | ctxt->wellFormed = 0; |
616 | 134M | if (ctxt->recovery == 0) |
617 | 109M | ctxt->disableSAX = 1; |
618 | 134M | } |
619 | 134M | } |
620 | | |
621 | | /** |
622 | | * xmlWarningMsg: |
623 | | * @ctxt: an XML parser context |
624 | | * @error: the error number |
625 | | * @msg: the error message |
626 | | * @str1: extra data |
627 | | * @str2: extra data |
628 | | * |
629 | | * Handle a warning. |
630 | | */ |
631 | | static void LIBXML_ATTR_FORMAT(3,0) |
632 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
633 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
634 | 395k | { |
635 | 395k | xmlStructuredErrorFunc schannel = NULL; |
636 | | |
637 | 395k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
638 | 395k | (ctxt->instate == XML_PARSER_EOF)) |
639 | 0 | return; |
640 | 395k | if ((ctxt != NULL) && (ctxt->sax != NULL) && |
641 | 395k | (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
642 | 258k | schannel = ctxt->sax->serror; |
643 | 395k | if (ctxt != NULL) { |
644 | 395k | __xmlRaiseError(schannel, |
645 | 395k | (ctxt->sax) ? ctxt->sax->warning : NULL, |
646 | 395k | ctxt->userData, |
647 | 395k | ctxt, NULL, XML_FROM_PARSER, error, |
648 | 395k | XML_ERR_WARNING, NULL, 0, |
649 | 395k | (const char *) str1, (const char *) str2, NULL, 0, 0, |
650 | 395k | msg, (const char *) str1, (const char *) str2); |
651 | 395k | } else { |
652 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
653 | 0 | ctxt, NULL, XML_FROM_PARSER, error, |
654 | 0 | XML_ERR_WARNING, NULL, 0, |
655 | 0 | (const char *) str1, (const char *) str2, NULL, 0, 0, |
656 | 0 | msg, (const char *) str1, (const char *) str2); |
657 | 0 | } |
658 | 395k | } |
659 | | |
660 | | /** |
661 | | * xmlValidityError: |
662 | | * @ctxt: an XML parser context |
663 | | * @error: the error number |
664 | | * @msg: the error message |
665 | | * @str1: extra data |
666 | | * |
667 | | * Handle a validity error. |
668 | | */ |
669 | | static void LIBXML_ATTR_FORMAT(3,0) |
670 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
671 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
672 | 20.7k | { |
673 | 20.7k | xmlStructuredErrorFunc schannel = NULL; |
674 | | |
675 | 20.7k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
676 | 20.7k | (ctxt->instate == XML_PARSER_EOF)) |
677 | 0 | return; |
678 | 20.7k | if (ctxt != NULL) { |
679 | 20.7k | ctxt->errNo = error; |
680 | 20.7k | if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
681 | 13.8k | schannel = ctxt->sax->serror; |
682 | 20.7k | } |
683 | 20.7k | if (ctxt != NULL) { |
684 | 20.7k | __xmlRaiseError(schannel, |
685 | 20.7k | ctxt->vctxt.error, ctxt->vctxt.userData, |
686 | 20.7k | ctxt, NULL, XML_FROM_DTD, error, |
687 | 20.7k | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
688 | 20.7k | (const char *) str2, NULL, 0, 0, |
689 | 20.7k | msg, (const char *) str1, (const char *) str2); |
690 | 20.7k | ctxt->valid = 0; |
691 | 20.7k | } else { |
692 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
693 | 0 | ctxt, NULL, XML_FROM_DTD, error, |
694 | 0 | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
695 | 0 | (const char *) str2, NULL, 0, 0, |
696 | 0 | msg, (const char *) str1, (const char *) str2); |
697 | 0 | } |
698 | 20.7k | } |
699 | | |
700 | | /** |
701 | | * xmlFatalErrMsgInt: |
702 | | * @ctxt: an XML parser context |
703 | | * @error: the error number |
704 | | * @msg: the error message |
705 | | * @val: an integer value |
706 | | * |
707 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
708 | | */ |
709 | | static void LIBXML_ATTR_FORMAT(3,0) |
710 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
711 | | const char *msg, int val) |
712 | 122M | { |
713 | 122M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
714 | 122M | (ctxt->instate == XML_PARSER_EOF)) |
715 | 0 | return; |
716 | 122M | if (ctxt != NULL) |
717 | 122M | ctxt->errNo = error; |
718 | 122M | __xmlRaiseError(NULL, NULL, NULL, |
719 | 122M | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
720 | 122M | NULL, 0, NULL, NULL, NULL, val, 0, msg, val); |
721 | 122M | if (ctxt != NULL) { |
722 | 122M | ctxt->wellFormed = 0; |
723 | 122M | if (ctxt->recovery == 0) |
724 | 112M | ctxt->disableSAX = 1; |
725 | 122M | } |
726 | 122M | } |
727 | | |
728 | | /** |
729 | | * xmlFatalErrMsgStrIntStr: |
730 | | * @ctxt: an XML parser context |
731 | | * @error: the error number |
732 | | * @msg: the error message |
733 | | * @str1: an string info |
734 | | * @val: an integer value |
735 | | * @str2: an string info |
736 | | * |
737 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
738 | | */ |
739 | | static void LIBXML_ATTR_FORMAT(3,0) |
740 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
741 | | const char *msg, const xmlChar *str1, int val, |
742 | | const xmlChar *str2) |
743 | 21.6M | { |
744 | 21.6M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
745 | 21.6M | (ctxt->instate == XML_PARSER_EOF)) |
746 | 0 | return; |
747 | 21.6M | if (ctxt != NULL) |
748 | 21.6M | ctxt->errNo = error; |
749 | 21.6M | __xmlRaiseError(NULL, NULL, NULL, |
750 | 21.6M | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
751 | 21.6M | NULL, 0, (const char *) str1, (const char *) str2, |
752 | 21.6M | NULL, val, 0, msg, str1, val, str2); |
753 | 21.6M | if (ctxt != NULL) { |
754 | 21.6M | ctxt->wellFormed = 0; |
755 | 21.6M | if (ctxt->recovery == 0) |
756 | 19.1M | ctxt->disableSAX = 1; |
757 | 21.6M | } |
758 | 21.6M | } |
759 | | |
760 | | /** |
761 | | * xmlFatalErrMsgStr: |
762 | | * @ctxt: an XML parser context |
763 | | * @error: the error number |
764 | | * @msg: the error message |
765 | | * @val: a string value |
766 | | * |
767 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
768 | | */ |
769 | | static void LIBXML_ATTR_FORMAT(3,0) |
770 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
771 | | const char *msg, const xmlChar * val) |
772 | 18.8M | { |
773 | 18.8M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
774 | 18.8M | (ctxt->instate == XML_PARSER_EOF)) |
775 | 0 | return; |
776 | 18.8M | if (ctxt != NULL) |
777 | 18.8M | ctxt->errNo = error; |
778 | 18.8M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
779 | 18.8M | XML_FROM_PARSER, error, XML_ERR_FATAL, |
780 | 18.8M | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
781 | 18.8M | val); |
782 | 18.8M | if (ctxt != NULL) { |
783 | 18.8M | ctxt->wellFormed = 0; |
784 | 18.8M | if (ctxt->recovery == 0) |
785 | 15.0M | ctxt->disableSAX = 1; |
786 | 18.8M | } |
787 | 18.8M | } |
788 | | |
789 | | /** |
790 | | * xmlErrMsgStr: |
791 | | * @ctxt: an XML parser context |
792 | | * @error: the error number |
793 | | * @msg: the error message |
794 | | * @val: a string value |
795 | | * |
796 | | * Handle a non fatal parser error |
797 | | */ |
798 | | static void LIBXML_ATTR_FORMAT(3,0) |
799 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
800 | | const char *msg, const xmlChar * val) |
801 | 95.7k | { |
802 | 95.7k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
803 | 95.7k | (ctxt->instate == XML_PARSER_EOF)) |
804 | 0 | return; |
805 | 95.7k | if (ctxt != NULL) |
806 | 95.7k | ctxt->errNo = error; |
807 | 95.7k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
808 | 95.7k | XML_FROM_PARSER, error, XML_ERR_ERROR, |
809 | 95.7k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
810 | 95.7k | val); |
811 | 95.7k | } |
812 | | |
813 | | /** |
814 | | * xmlNsErr: |
815 | | * @ctxt: an XML parser context |
816 | | * @error: the error number |
817 | | * @msg: the message |
818 | | * @info1: extra information string |
819 | | * @info2: extra information string |
820 | | * |
821 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
822 | | */ |
823 | | static void LIBXML_ATTR_FORMAT(3,0) |
824 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
825 | | const char *msg, |
826 | | const xmlChar * info1, const xmlChar * info2, |
827 | | const xmlChar * info3) |
828 | 1.39M | { |
829 | 1.39M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
830 | 1.39M | (ctxt->instate == XML_PARSER_EOF)) |
831 | 0 | return; |
832 | 1.39M | if (ctxt != NULL) |
833 | 1.39M | ctxt->errNo = error; |
834 | 1.39M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
835 | 1.39M | XML_ERR_ERROR, NULL, 0, (const char *) info1, |
836 | 1.39M | (const char *) info2, (const char *) info3, 0, 0, msg, |
837 | 1.39M | info1, info2, info3); |
838 | 1.39M | if (ctxt != NULL) |
839 | 1.39M | ctxt->nsWellFormed = 0; |
840 | 1.39M | } |
841 | | |
842 | | /** |
843 | | * xmlNsWarn |
844 | | * @ctxt: an XML parser context |
845 | | * @error: the error number |
846 | | * @msg: the message |
847 | | * @info1: extra information string |
848 | | * @info2: extra information string |
849 | | * |
850 | | * Handle a namespace warning error |
851 | | */ |
852 | | static void LIBXML_ATTR_FORMAT(3,0) |
853 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
854 | | const char *msg, |
855 | | const xmlChar * info1, const xmlChar * info2, |
856 | | const xmlChar * info3) |
857 | 52.0k | { |
858 | 52.0k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
859 | 52.0k | (ctxt->instate == XML_PARSER_EOF)) |
860 | 0 | return; |
861 | 52.0k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
862 | 52.0k | XML_ERR_WARNING, NULL, 0, (const char *) info1, |
863 | 52.0k | (const char *) info2, (const char *) info3, 0, 0, msg, |
864 | 52.0k | info1, info2, info3); |
865 | 52.0k | } |
866 | | |
867 | | /************************************************************************ |
868 | | * * |
869 | | * Library wide options * |
870 | | * * |
871 | | ************************************************************************/ |
872 | | |
873 | | /** |
874 | | * xmlHasFeature: |
875 | | * @feature: the feature to be examined |
876 | | * |
877 | | * Examines if the library has been compiled with a given feature. |
878 | | * |
879 | | * Returns a non-zero value if the feature exist, otherwise zero. |
880 | | * Returns zero (0) if the feature does not exist or an unknown |
881 | | * unknown feature is requested, non-zero otherwise. |
882 | | */ |
883 | | int |
884 | | xmlHasFeature(xmlFeature feature) |
885 | 0 | { |
886 | 0 | switch (feature) { |
887 | 0 | case XML_WITH_THREAD: |
888 | 0 | #ifdef LIBXML_THREAD_ENABLED |
889 | 0 | return(1); |
890 | | #else |
891 | | return(0); |
892 | | #endif |
893 | 0 | case XML_WITH_TREE: |
894 | 0 | #ifdef LIBXML_TREE_ENABLED |
895 | 0 | return(1); |
896 | | #else |
897 | | return(0); |
898 | | #endif |
899 | 0 | case XML_WITH_OUTPUT: |
900 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
901 | 0 | return(1); |
902 | | #else |
903 | | return(0); |
904 | | #endif |
905 | 0 | case XML_WITH_PUSH: |
906 | 0 | #ifdef LIBXML_PUSH_ENABLED |
907 | 0 | return(1); |
908 | | #else |
909 | | return(0); |
910 | | #endif |
911 | 0 | case XML_WITH_READER: |
912 | 0 | #ifdef LIBXML_READER_ENABLED |
913 | 0 | return(1); |
914 | | #else |
915 | | return(0); |
916 | | #endif |
917 | 0 | case XML_WITH_PATTERN: |
918 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
919 | 0 | return(1); |
920 | | #else |
921 | | return(0); |
922 | | #endif |
923 | 0 | case XML_WITH_WRITER: |
924 | 0 | #ifdef LIBXML_WRITER_ENABLED |
925 | 0 | return(1); |
926 | | #else |
927 | | return(0); |
928 | | #endif |
929 | 0 | case XML_WITH_SAX1: |
930 | 0 | #ifdef LIBXML_SAX1_ENABLED |
931 | 0 | return(1); |
932 | | #else |
933 | | return(0); |
934 | | #endif |
935 | 0 | case XML_WITH_FTP: |
936 | | #ifdef LIBXML_FTP_ENABLED |
937 | | return(1); |
938 | | #else |
939 | 0 | return(0); |
940 | 0 | #endif |
941 | 0 | case XML_WITH_HTTP: |
942 | | #ifdef LIBXML_HTTP_ENABLED |
943 | | return(1); |
944 | | #else |
945 | 0 | return(0); |
946 | 0 | #endif |
947 | 0 | case XML_WITH_VALID: |
948 | 0 | #ifdef LIBXML_VALID_ENABLED |
949 | 0 | return(1); |
950 | | #else |
951 | | return(0); |
952 | | #endif |
953 | 0 | case XML_WITH_HTML: |
954 | 0 | #ifdef LIBXML_HTML_ENABLED |
955 | 0 | return(1); |
956 | | #else |
957 | | return(0); |
958 | | #endif |
959 | 0 | case XML_WITH_LEGACY: |
960 | | #ifdef LIBXML_LEGACY_ENABLED |
961 | | return(1); |
962 | | #else |
963 | 0 | return(0); |
964 | 0 | #endif |
965 | 0 | case XML_WITH_C14N: |
966 | 0 | #ifdef LIBXML_C14N_ENABLED |
967 | 0 | return(1); |
968 | | #else |
969 | | return(0); |
970 | | #endif |
971 | 0 | case XML_WITH_CATALOG: |
972 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
973 | 0 | return(1); |
974 | | #else |
975 | | return(0); |
976 | | #endif |
977 | 0 | case XML_WITH_XPATH: |
978 | 0 | #ifdef LIBXML_XPATH_ENABLED |
979 | 0 | return(1); |
980 | | #else |
981 | | return(0); |
982 | | #endif |
983 | 0 | case XML_WITH_XPTR: |
984 | 0 | #ifdef LIBXML_XPTR_ENABLED |
985 | 0 | return(1); |
986 | | #else |
987 | | return(0); |
988 | | #endif |
989 | 0 | case XML_WITH_XINCLUDE: |
990 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
991 | 0 | return(1); |
992 | | #else |
993 | | return(0); |
994 | | #endif |
995 | 0 | case XML_WITH_ICONV: |
996 | 0 | #ifdef LIBXML_ICONV_ENABLED |
997 | 0 | return(1); |
998 | | #else |
999 | | return(0); |
1000 | | #endif |
1001 | 0 | case XML_WITH_ISO8859X: |
1002 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
1003 | 0 | return(1); |
1004 | | #else |
1005 | | return(0); |
1006 | | #endif |
1007 | 0 | case XML_WITH_UNICODE: |
1008 | 0 | #ifdef LIBXML_UNICODE_ENABLED |
1009 | 0 | return(1); |
1010 | | #else |
1011 | | return(0); |
1012 | | #endif |
1013 | 0 | case XML_WITH_REGEXP: |
1014 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
1015 | 0 | return(1); |
1016 | | #else |
1017 | | return(0); |
1018 | | #endif |
1019 | 0 | case XML_WITH_AUTOMATA: |
1020 | 0 | #ifdef LIBXML_AUTOMATA_ENABLED |
1021 | 0 | return(1); |
1022 | | #else |
1023 | | return(0); |
1024 | | #endif |
1025 | 0 | case XML_WITH_EXPR: |
1026 | | #ifdef LIBXML_EXPR_ENABLED |
1027 | | return(1); |
1028 | | #else |
1029 | 0 | return(0); |
1030 | 0 | #endif |
1031 | 0 | case XML_WITH_SCHEMAS: |
1032 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
1033 | 0 | return(1); |
1034 | | #else |
1035 | | return(0); |
1036 | | #endif |
1037 | 0 | case XML_WITH_SCHEMATRON: |
1038 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
1039 | 0 | return(1); |
1040 | | #else |
1041 | | return(0); |
1042 | | #endif |
1043 | 0 | case XML_WITH_MODULES: |
1044 | 0 | #ifdef LIBXML_MODULES_ENABLED |
1045 | 0 | return(1); |
1046 | | #else |
1047 | | return(0); |
1048 | | #endif |
1049 | 0 | case XML_WITH_DEBUG: |
1050 | | #ifdef LIBXML_DEBUG_ENABLED |
1051 | | return(1); |
1052 | | #else |
1053 | 0 | return(0); |
1054 | 0 | #endif |
1055 | 0 | case XML_WITH_DEBUG_MEM: |
1056 | | #ifdef DEBUG_MEMORY_LOCATION |
1057 | | return(1); |
1058 | | #else |
1059 | 0 | return(0); |
1060 | 0 | #endif |
1061 | 0 | case XML_WITH_DEBUG_RUN: |
1062 | 0 | return(0); |
1063 | 0 | case XML_WITH_ZLIB: |
1064 | 0 | #ifdef LIBXML_ZLIB_ENABLED |
1065 | 0 | return(1); |
1066 | | #else |
1067 | | return(0); |
1068 | | #endif |
1069 | 0 | case XML_WITH_LZMA: |
1070 | 0 | #ifdef LIBXML_LZMA_ENABLED |
1071 | 0 | return(1); |
1072 | | #else |
1073 | | return(0); |
1074 | | #endif |
1075 | 0 | case XML_WITH_ICU: |
1076 | | #ifdef LIBXML_ICU_ENABLED |
1077 | | return(1); |
1078 | | #else |
1079 | 0 | return(0); |
1080 | 0 | #endif |
1081 | 0 | default: |
1082 | 0 | break; |
1083 | 0 | } |
1084 | 0 | return(0); |
1085 | 0 | } |
1086 | | |
1087 | | /************************************************************************ |
1088 | | * * |
1089 | | * SAX2 defaulted attributes handling * |
1090 | | * * |
1091 | | ************************************************************************/ |
1092 | | |
1093 | | /** |
1094 | | * xmlDetectSAX2: |
1095 | | * @ctxt: an XML parser context |
1096 | | * |
1097 | | * Do the SAX2 detection and specific initialization |
1098 | | */ |
1099 | | static void |
1100 | 6.39M | xmlDetectSAX2(xmlParserCtxtPtr ctxt) { |
1101 | 6.39M | xmlSAXHandlerPtr sax; |
1102 | | |
1103 | | /* Avoid unused variable warning if features are disabled. */ |
1104 | 6.39M | (void) sax; |
1105 | | |
1106 | 6.39M | if (ctxt == NULL) return; |
1107 | 6.39M | sax = ctxt->sax; |
1108 | 6.39M | #ifdef LIBXML_SAX1_ENABLED |
1109 | 6.39M | if ((sax) && (sax->initialized == XML_SAX2_MAGIC) && |
1110 | 6.39M | ((sax->startElementNs != NULL) || |
1111 | 1.64M | (sax->endElementNs != NULL) || |
1112 | 1.64M | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
1113 | 1.64M | ctxt->sax2 = 1; |
1114 | | #else |
1115 | | ctxt->sax2 = 1; |
1116 | | #endif /* LIBXML_SAX1_ENABLED */ |
1117 | | |
1118 | 6.39M | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
1119 | 6.39M | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
1120 | 6.39M | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
1121 | 6.39M | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
1122 | 6.39M | (ctxt->str_xml_ns == NULL)) { |
1123 | 0 | xmlErrMemory(ctxt, NULL); |
1124 | 0 | } |
1125 | 6.39M | } |
1126 | | |
1127 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
1128 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
1129 | | struct _xmlDefAttrs { |
1130 | | int nbAttrs; /* number of defaulted attributes on that element */ |
1131 | | int maxAttrs; /* the size of the array */ |
1132 | | #if __STDC_VERSION__ >= 199901L |
1133 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
1134 | | const xmlChar *values[]; /* array of localname/prefix/values/external */ |
1135 | | #else |
1136 | | const xmlChar *values[5]; |
1137 | | #endif |
1138 | | }; |
1139 | | |
1140 | | /** |
1141 | | * xmlAttrNormalizeSpace: |
1142 | | * @src: the source string |
1143 | | * @dst: the target string |
1144 | | * |
1145 | | * Normalize the space in non CDATA attribute values: |
1146 | | * If the attribute type is not CDATA, then the XML processor MUST further |
1147 | | * process the normalized attribute value by discarding any leading and |
1148 | | * trailing space (#x20) characters, and by replacing sequences of space |
1149 | | * (#x20) characters by a single space (#x20) character. |
1150 | | * Note that the size of dst need to be at least src, and if one doesn't need |
1151 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
1152 | | * passing src as dst is just fine. |
1153 | | * |
1154 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1155 | | * is needed. |
1156 | | */ |
1157 | | static xmlChar * |
1158 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
1159 | 137k | { |
1160 | 137k | if ((src == NULL) || (dst == NULL)) |
1161 | 0 | return(NULL); |
1162 | | |
1163 | 205k | while (*src == 0x20) src++; |
1164 | 4.00M | while (*src != 0) { |
1165 | 3.86M | if (*src == 0x20) { |
1166 | 394k | while (*src == 0x20) src++; |
1167 | 92.4k | if (*src != 0) |
1168 | 79.1k | *dst++ = 0x20; |
1169 | 3.77M | } else { |
1170 | 3.77M | *dst++ = *src++; |
1171 | 3.77M | } |
1172 | 3.86M | } |
1173 | 137k | *dst = 0; |
1174 | 137k | if (dst == src) |
1175 | 111k | return(NULL); |
1176 | 26.4k | return(dst); |
1177 | 137k | } |
1178 | | |
1179 | | /** |
1180 | | * xmlAttrNormalizeSpace2: |
1181 | | * @src: the source string |
1182 | | * |
1183 | | * Normalize the space in non CDATA attribute values, a slightly more complex |
1184 | | * front end to avoid allocation problems when running on attribute values |
1185 | | * coming from the input. |
1186 | | * |
1187 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1188 | | * is needed. |
1189 | | */ |
1190 | | static const xmlChar * |
1191 | | xmlAttrNormalizeSpace2(xmlParserCtxtPtr ctxt, xmlChar *src, int *len) |
1192 | 70.5k | { |
1193 | 70.5k | int i; |
1194 | 70.5k | int remove_head = 0; |
1195 | 70.5k | int need_realloc = 0; |
1196 | 70.5k | const xmlChar *cur; |
1197 | | |
1198 | 70.5k | if ((ctxt == NULL) || (src == NULL) || (len == NULL)) |
1199 | 0 | return(NULL); |
1200 | 70.5k | i = *len; |
1201 | 70.5k | if (i <= 0) |
1202 | 2.64k | return(NULL); |
1203 | | |
1204 | 67.8k | cur = src; |
1205 | 85.8k | while (*cur == 0x20) { |
1206 | 17.9k | cur++; |
1207 | 17.9k | remove_head++; |
1208 | 17.9k | } |
1209 | 1.52M | while (*cur != 0) { |
1210 | 1.47M | if (*cur == 0x20) { |
1211 | 59.4k | cur++; |
1212 | 59.4k | if ((*cur == 0x20) || (*cur == 0)) { |
1213 | 10.4k | need_realloc = 1; |
1214 | 10.4k | break; |
1215 | 10.4k | } |
1216 | 59.4k | } else |
1217 | 1.41M | cur++; |
1218 | 1.47M | } |
1219 | 67.8k | if (need_realloc) { |
1220 | 10.4k | xmlChar *ret; |
1221 | | |
1222 | 10.4k | ret = xmlStrndup(src + remove_head, i - remove_head + 1); |
1223 | 10.4k | if (ret == NULL) { |
1224 | 0 | xmlErrMemory(ctxt, NULL); |
1225 | 0 | return(NULL); |
1226 | 0 | } |
1227 | 10.4k | xmlAttrNormalizeSpace(ret, ret); |
1228 | 10.4k | *len = strlen((const char *)ret); |
1229 | 10.4k | return(ret); |
1230 | 57.3k | } else if (remove_head) { |
1231 | 1.39k | *len -= remove_head; |
1232 | 1.39k | memmove(src, src + remove_head, 1 + *len); |
1233 | 1.39k | return(src); |
1234 | 1.39k | } |
1235 | 55.9k | return(NULL); |
1236 | 67.8k | } |
1237 | | |
1238 | | /** |
1239 | | * xmlAddDefAttrs: |
1240 | | * @ctxt: an XML parser context |
1241 | | * @fullname: the element fullname |
1242 | | * @fullattr: the attribute fullname |
1243 | | * @value: the attribute value |
1244 | | * |
1245 | | * Add a defaulted attribute for an element |
1246 | | */ |
1247 | | static void |
1248 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1249 | | const xmlChar *fullname, |
1250 | | const xmlChar *fullattr, |
1251 | 401k | const xmlChar *value) { |
1252 | 401k | xmlDefAttrsPtr defaults; |
1253 | 401k | int len; |
1254 | 401k | const xmlChar *name; |
1255 | 401k | const xmlChar *prefix; |
1256 | | |
1257 | | /* |
1258 | | * Allows to detect attribute redefinitions |
1259 | | */ |
1260 | 401k | if (ctxt->attsSpecial != NULL) { |
1261 | 361k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1262 | 227k | return; |
1263 | 361k | } |
1264 | | |
1265 | 173k | if (ctxt->attsDefault == NULL) { |
1266 | 48.2k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1267 | 48.2k | if (ctxt->attsDefault == NULL) |
1268 | 0 | goto mem_error; |
1269 | 48.2k | } |
1270 | | |
1271 | | /* |
1272 | | * split the element name into prefix:localname , the string found |
1273 | | * are within the DTD and then not associated to namespace names. |
1274 | | */ |
1275 | 173k | name = xmlSplitQName3(fullname, &len); |
1276 | 173k | if (name == NULL) { |
1277 | 148k | name = xmlDictLookup(ctxt->dict, fullname, -1); |
1278 | 148k | prefix = NULL; |
1279 | 148k | } else { |
1280 | 25.3k | name = xmlDictLookup(ctxt->dict, name, -1); |
1281 | 25.3k | prefix = xmlDictLookup(ctxt->dict, fullname, len); |
1282 | 25.3k | } |
1283 | | |
1284 | | /* |
1285 | | * make sure there is some storage |
1286 | | */ |
1287 | 173k | defaults = xmlHashLookup2(ctxt->attsDefault, name, prefix); |
1288 | 173k | if (defaults == NULL) { |
1289 | 93.0k | defaults = (xmlDefAttrsPtr) xmlMalloc(sizeof(xmlDefAttrs) + |
1290 | 93.0k | (4 * 5) * sizeof(const xmlChar *)); |
1291 | 93.0k | if (defaults == NULL) |
1292 | 0 | goto mem_error; |
1293 | 93.0k | defaults->nbAttrs = 0; |
1294 | 93.0k | defaults->maxAttrs = 4; |
1295 | 93.0k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1296 | 93.0k | defaults, NULL) < 0) { |
1297 | 0 | xmlFree(defaults); |
1298 | 0 | goto mem_error; |
1299 | 0 | } |
1300 | 93.0k | } else if (defaults->nbAttrs >= defaults->maxAttrs) { |
1301 | 2.98k | xmlDefAttrsPtr temp; |
1302 | | |
1303 | 2.98k | temp = (xmlDefAttrsPtr) xmlRealloc(defaults, sizeof(xmlDefAttrs) + |
1304 | 2.98k | (2 * defaults->maxAttrs * 5) * sizeof(const xmlChar *)); |
1305 | 2.98k | if (temp == NULL) |
1306 | 0 | goto mem_error; |
1307 | 2.98k | defaults = temp; |
1308 | 2.98k | defaults->maxAttrs *= 2; |
1309 | 2.98k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1310 | 2.98k | defaults, NULL) < 0) { |
1311 | 0 | xmlFree(defaults); |
1312 | 0 | goto mem_error; |
1313 | 0 | } |
1314 | 2.98k | } |
1315 | | |
1316 | | /* |
1317 | | * Split the element name into prefix:localname , the string found |
1318 | | * are within the DTD and hen not associated to namespace names. |
1319 | | */ |
1320 | 173k | name = xmlSplitQName3(fullattr, &len); |
1321 | 173k | if (name == NULL) { |
1322 | 114k | name = xmlDictLookup(ctxt->dict, fullattr, -1); |
1323 | 114k | prefix = NULL; |
1324 | 114k | } else { |
1325 | 59.2k | name = xmlDictLookup(ctxt->dict, name, -1); |
1326 | 59.2k | prefix = xmlDictLookup(ctxt->dict, fullattr, len); |
1327 | 59.2k | } |
1328 | | |
1329 | 173k | defaults->values[5 * defaults->nbAttrs] = name; |
1330 | 173k | defaults->values[5 * defaults->nbAttrs + 1] = prefix; |
1331 | | /* intern the string and precompute the end */ |
1332 | 173k | len = xmlStrlen(value); |
1333 | 173k | value = xmlDictLookup(ctxt->dict, value, len); |
1334 | 173k | defaults->values[5 * defaults->nbAttrs + 2] = value; |
1335 | 173k | defaults->values[5 * defaults->nbAttrs + 3] = value + len; |
1336 | 173k | if (ctxt->external) |
1337 | 99.4k | defaults->values[5 * defaults->nbAttrs + 4] = BAD_CAST "external"; |
1338 | 74.0k | else |
1339 | 74.0k | defaults->values[5 * defaults->nbAttrs + 4] = NULL; |
1340 | 173k | defaults->nbAttrs++; |
1341 | | |
1342 | 173k | return; |
1343 | | |
1344 | 0 | mem_error: |
1345 | 0 | xmlErrMemory(ctxt, NULL); |
1346 | 0 | return; |
1347 | 173k | } |
1348 | | |
1349 | | /** |
1350 | | * xmlAddSpecialAttr: |
1351 | | * @ctxt: an XML parser context |
1352 | | * @fullname: the element fullname |
1353 | | * @fullattr: the attribute fullname |
1354 | | * @type: the attribute type |
1355 | | * |
1356 | | * Register this attribute type |
1357 | | */ |
1358 | | static void |
1359 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1360 | | const xmlChar *fullname, |
1361 | | const xmlChar *fullattr, |
1362 | | int type) |
1363 | 2.79M | { |
1364 | 2.79M | if (ctxt->attsSpecial == NULL) { |
1365 | 103k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1366 | 103k | if (ctxt->attsSpecial == NULL) |
1367 | 0 | goto mem_error; |
1368 | 103k | } |
1369 | | |
1370 | 2.79M | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1371 | 1.08M | return; |
1372 | | |
1373 | 1.70M | xmlHashAddEntry2(ctxt->attsSpecial, fullname, fullattr, |
1374 | 1.70M | (void *) (ptrdiff_t) type); |
1375 | 1.70M | return; |
1376 | | |
1377 | 0 | mem_error: |
1378 | 0 | xmlErrMemory(ctxt, NULL); |
1379 | 0 | return; |
1380 | 2.79M | } |
1381 | | |
1382 | | /** |
1383 | | * xmlCleanSpecialAttrCallback: |
1384 | | * |
1385 | | * Removes CDATA attributes from the special attribute table |
1386 | | */ |
1387 | | static void |
1388 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1389 | | const xmlChar *fullname, const xmlChar *fullattr, |
1390 | 1.70M | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1391 | 1.70M | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1392 | | |
1393 | 1.70M | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1394 | 507k | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1395 | 507k | } |
1396 | 1.70M | } |
1397 | | |
1398 | | /** |
1399 | | * xmlCleanSpecialAttr: |
1400 | | * @ctxt: an XML parser context |
1401 | | * |
1402 | | * Trim the list of attributes defined to remove all those of type |
1403 | | * CDATA as they are not special. This call should be done when finishing |
1404 | | * to parse the DTD and before starting to parse the document root. |
1405 | | */ |
1406 | | static void |
1407 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1408 | 438k | { |
1409 | 438k | if (ctxt->attsSpecial == NULL) |
1410 | 337k | return; |
1411 | | |
1412 | 101k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1413 | | |
1414 | 101k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1415 | 20.3k | xmlHashFree(ctxt->attsSpecial, NULL); |
1416 | 20.3k | ctxt->attsSpecial = NULL; |
1417 | 20.3k | } |
1418 | 101k | return; |
1419 | 438k | } |
1420 | | |
1421 | | /** |
1422 | | * xmlCheckLanguageID: |
1423 | | * @lang: pointer to the string value |
1424 | | * |
1425 | | * Checks that the value conforms to the LanguageID production: |
1426 | | * |
1427 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1428 | | * the XML Second edition. |
1429 | | * |
1430 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1431 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1432 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1433 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1434 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1435 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1436 | | * |
1437 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1438 | | * |
1439 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1440 | | * langtag = language |
1441 | | * ["-" script] |
1442 | | * ["-" region] |
1443 | | * *("-" variant) |
1444 | | * *("-" extension) |
1445 | | * ["-" privateuse] |
1446 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1447 | | * ["-" extlang] ; sometimes followed by |
1448 | | * ; extended language subtags |
1449 | | * / 4ALPHA ; or reserved for future use |
1450 | | * / 5*8ALPHA ; or registered language subtag |
1451 | | * |
1452 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1453 | | * *2("-" 3ALPHA) ; permanently reserved |
1454 | | * |
1455 | | * script = 4ALPHA ; ISO 15924 code |
1456 | | * |
1457 | | * region = 2ALPHA ; ISO 3166-1 code |
1458 | | * / 3DIGIT ; UN M.49 code |
1459 | | * |
1460 | | * variant = 5*8alphanum ; registered variants |
1461 | | * / (DIGIT 3alphanum) |
1462 | | * |
1463 | | * extension = singleton 1*("-" (2*8alphanum)) |
1464 | | * |
1465 | | * ; Single alphanumerics |
1466 | | * ; "x" reserved for private use |
1467 | | * singleton = DIGIT ; 0 - 9 |
1468 | | * / %x41-57 ; A - W |
1469 | | * / %x59-5A ; Y - Z |
1470 | | * / %x61-77 ; a - w |
1471 | | * / %x79-7A ; y - z |
1472 | | * |
1473 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1474 | | * The parser below doesn't try to cope with extension or privateuse |
1475 | | * that could be added but that's not interoperable anyway |
1476 | | * |
1477 | | * Returns 1 if correct 0 otherwise |
1478 | | **/ |
1479 | | int |
1480 | | xmlCheckLanguageID(const xmlChar * lang) |
1481 | 38.5k | { |
1482 | 38.5k | const xmlChar *cur = lang, *nxt; |
1483 | | |
1484 | 38.5k | if (cur == NULL) |
1485 | 1.21k | return (0); |
1486 | 37.2k | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1487 | 37.2k | ((cur[0] == 'I') && (cur[1] == '-')) || |
1488 | 37.2k | ((cur[0] == 'x') && (cur[1] == '-')) || |
1489 | 37.2k | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1490 | | /* |
1491 | | * Still allow IANA code and user code which were coming |
1492 | | * from the previous version of the XML-1.0 specification |
1493 | | * it's deprecated but we should not fail |
1494 | | */ |
1495 | 1.35k | cur += 2; |
1496 | 15.1k | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1497 | 15.1k | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1498 | 13.8k | cur++; |
1499 | 1.35k | return(cur[0] == 0); |
1500 | 1.35k | } |
1501 | 35.9k | nxt = cur; |
1502 | 135k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1503 | 135k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1504 | 99.3k | nxt++; |
1505 | 35.9k | if (nxt - cur >= 4) { |
1506 | | /* |
1507 | | * Reserved |
1508 | | */ |
1509 | 2.09k | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1510 | 1.78k | return(0); |
1511 | 303 | return(1); |
1512 | 2.09k | } |
1513 | 33.8k | if (nxt - cur < 2) |
1514 | 2.12k | return(0); |
1515 | | /* we got an ISO 639 code */ |
1516 | 31.7k | if (nxt[0] == 0) |
1517 | 20.2k | return(1); |
1518 | 11.5k | if (nxt[0] != '-') |
1519 | 1.26k | return(0); |
1520 | | |
1521 | 10.2k | nxt++; |
1522 | 10.2k | cur = nxt; |
1523 | | /* now we can have extlang or script or region or variant */ |
1524 | 10.2k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1525 | 1.38k | goto region_m49; |
1526 | | |
1527 | 46.4k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1528 | 46.4k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1529 | 37.5k | nxt++; |
1530 | 8.86k | if (nxt - cur == 4) |
1531 | 2.92k | goto script; |
1532 | 5.94k | if (nxt - cur == 2) |
1533 | 1.49k | goto region; |
1534 | 4.45k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1535 | 726 | goto variant; |
1536 | 3.72k | if (nxt - cur != 3) |
1537 | 863 | return(0); |
1538 | | /* we parsed an extlang */ |
1539 | 2.86k | if (nxt[0] == 0) |
1540 | 11 | return(1); |
1541 | 2.85k | if (nxt[0] != '-') |
1542 | 576 | return(0); |
1543 | | |
1544 | 2.27k | nxt++; |
1545 | 2.27k | cur = nxt; |
1546 | | /* now we can have script or region or variant */ |
1547 | 2.27k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1548 | 170 | goto region_m49; |
1549 | | |
1550 | 16.9k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1551 | 16.9k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1552 | 14.8k | nxt++; |
1553 | 2.10k | if (nxt - cur == 2) |
1554 | 297 | goto region; |
1555 | 1.80k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1556 | 582 | goto variant; |
1557 | 1.22k | if (nxt - cur != 4) |
1558 | 563 | return(0); |
1559 | | /* we parsed a script */ |
1560 | 3.58k | script: |
1561 | 3.58k | if (nxt[0] == 0) |
1562 | 74 | return(1); |
1563 | 3.51k | if (nxt[0] != '-') |
1564 | 693 | return(0); |
1565 | | |
1566 | 2.82k | nxt++; |
1567 | 2.82k | cur = nxt; |
1568 | | /* now we can have region or variant */ |
1569 | 2.82k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1570 | 355 | goto region_m49; |
1571 | | |
1572 | 18.1k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1573 | 18.1k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1574 | 15.6k | nxt++; |
1575 | | |
1576 | 2.46k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1577 | 407 | goto variant; |
1578 | 2.05k | if (nxt - cur != 2) |
1579 | 785 | return(0); |
1580 | | /* we parsed a region */ |
1581 | 4.07k | region: |
1582 | 4.07k | if (nxt[0] == 0) |
1583 | 584 | return(1); |
1584 | 3.49k | if (nxt[0] != '-') |
1585 | 2.00k | return(0); |
1586 | | |
1587 | 1.48k | nxt++; |
1588 | 1.48k | cur = nxt; |
1589 | | /* now we can just have a variant */ |
1590 | 89.3k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1591 | 89.3k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1592 | 87.8k | nxt++; |
1593 | | |
1594 | 1.48k | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1595 | 1.07k | return(0); |
1596 | | |
1597 | | /* we parsed a variant */ |
1598 | 2.12k | variant: |
1599 | 2.12k | if (nxt[0] == 0) |
1600 | 145 | return(1); |
1601 | 1.97k | if (nxt[0] != '-') |
1602 | 1.55k | return(0); |
1603 | | /* extensions and private use subtags not checked */ |
1604 | 428 | return (1); |
1605 | | |
1606 | 1.91k | region_m49: |
1607 | 1.91k | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1608 | 1.91k | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1609 | 1.01k | nxt += 3; |
1610 | 1.01k | goto region; |
1611 | 1.01k | } |
1612 | 902 | return(0); |
1613 | 1.91k | } |
1614 | | |
1615 | | /************************************************************************ |
1616 | | * * |
1617 | | * Parser stacks related functions and macros * |
1618 | | * * |
1619 | | ************************************************************************/ |
1620 | | |
1621 | | static xmlEntityPtr xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, |
1622 | | const xmlChar ** str); |
1623 | | |
1624 | | #ifdef SAX2 |
1625 | | /** |
1626 | | * nsPush: |
1627 | | * @ctxt: an XML parser context |
1628 | | * @prefix: the namespace prefix or NULL |
1629 | | * @URL: the namespace name |
1630 | | * |
1631 | | * Pushes a new parser namespace on top of the ns stack |
1632 | | * |
1633 | | * Returns -1 in case of error, -2 if the namespace should be discarded |
1634 | | * and the index in the stack otherwise. |
1635 | | */ |
1636 | | static int |
1637 | | nsPush(xmlParserCtxtPtr ctxt, const xmlChar *prefix, const xmlChar *URL) |
1638 | 1.29M | { |
1639 | 1.29M | if (ctxt->options & XML_PARSE_NSCLEAN) { |
1640 | 266k | int i; |
1641 | 385k | for (i = ctxt->nsNr - 2;i >= 0;i -= 2) { |
1642 | 208k | if (ctxt->nsTab[i] == prefix) { |
1643 | | /* in scope */ |
1644 | 89.4k | if (ctxt->nsTab[i + 1] == URL) |
1645 | 29.7k | return(-2); |
1646 | | /* out of scope keep it */ |
1647 | 59.6k | break; |
1648 | 89.4k | } |
1649 | 208k | } |
1650 | 266k | } |
1651 | 1.26M | if ((ctxt->nsMax == 0) || (ctxt->nsTab == NULL)) { |
1652 | 251k | ctxt->nsMax = 10; |
1653 | 251k | ctxt->nsNr = 0; |
1654 | 251k | ctxt->nsTab = (const xmlChar **) |
1655 | 251k | xmlMalloc(ctxt->nsMax * sizeof(xmlChar *)); |
1656 | 251k | if (ctxt->nsTab == NULL) { |
1657 | 0 | xmlErrMemory(ctxt, NULL); |
1658 | 0 | ctxt->nsMax = 0; |
1659 | 0 | return (-1); |
1660 | 0 | } |
1661 | 1.01M | } else if (ctxt->nsNr >= ctxt->nsMax) { |
1662 | 22.8k | const xmlChar ** tmp; |
1663 | 22.8k | ctxt->nsMax *= 2; |
1664 | 22.8k | tmp = (const xmlChar **) xmlRealloc((char *) ctxt->nsTab, |
1665 | 22.8k | ctxt->nsMax * sizeof(ctxt->nsTab[0])); |
1666 | 22.8k | if (tmp == NULL) { |
1667 | 0 | xmlErrMemory(ctxt, NULL); |
1668 | 0 | ctxt->nsMax /= 2; |
1669 | 0 | return (-1); |
1670 | 0 | } |
1671 | 22.8k | ctxt->nsTab = tmp; |
1672 | 22.8k | } |
1673 | 1.26M | ctxt->nsTab[ctxt->nsNr++] = prefix; |
1674 | 1.26M | ctxt->nsTab[ctxt->nsNr++] = URL; |
1675 | 1.26M | return (ctxt->nsNr); |
1676 | 1.26M | } |
1677 | | /** |
1678 | | * nsPop: |
1679 | | * @ctxt: an XML parser context |
1680 | | * @nr: the number to pop |
1681 | | * |
1682 | | * Pops the top @nr parser prefix/namespace from the ns stack |
1683 | | * |
1684 | | * Returns the number of namespaces removed |
1685 | | */ |
1686 | | static int |
1687 | | nsPop(xmlParserCtxtPtr ctxt, int nr) |
1688 | 229k | { |
1689 | 229k | int i; |
1690 | | |
1691 | 229k | if (ctxt->nsTab == NULL) return(0); |
1692 | 229k | if (ctxt->nsNr < nr) { |
1693 | 0 | xmlGenericError(xmlGenericErrorContext, "Pbm popping %d NS\n", nr); |
1694 | 0 | nr = ctxt->nsNr; |
1695 | 0 | } |
1696 | 229k | if (ctxt->nsNr <= 0) |
1697 | 0 | return (0); |
1698 | | |
1699 | 731k | for (i = 0;i < nr;i++) { |
1700 | 501k | ctxt->nsNr--; |
1701 | 501k | ctxt->nsTab[ctxt->nsNr] = NULL; |
1702 | 501k | } |
1703 | 229k | return(nr); |
1704 | 229k | } |
1705 | | #endif |
1706 | | |
1707 | | static int |
1708 | 344k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1709 | 344k | const xmlChar **atts; |
1710 | 344k | int *attallocs; |
1711 | 344k | int maxatts; |
1712 | | |
1713 | 344k | if (ctxt->atts == NULL) { |
1714 | 342k | maxatts = 55; /* allow for 10 attrs by default */ |
1715 | 342k | atts = (const xmlChar **) |
1716 | 342k | xmlMalloc(maxatts * sizeof(xmlChar *)); |
1717 | 342k | if (atts == NULL) goto mem_error; |
1718 | 342k | ctxt->atts = atts; |
1719 | 342k | attallocs = (int *) xmlMalloc((maxatts / 5) * sizeof(int)); |
1720 | 342k | if (attallocs == NULL) goto mem_error; |
1721 | 342k | ctxt->attallocs = attallocs; |
1722 | 342k | ctxt->maxatts = maxatts; |
1723 | 342k | } else if (nr + 5 > ctxt->maxatts) { |
1724 | 1.50k | maxatts = (nr + 5) * 2; |
1725 | 1.50k | atts = (const xmlChar **) xmlRealloc((void *) ctxt->atts, |
1726 | 1.50k | maxatts * sizeof(const xmlChar *)); |
1727 | 1.50k | if (atts == NULL) goto mem_error; |
1728 | 1.50k | ctxt->atts = atts; |
1729 | 1.50k | attallocs = (int *) xmlRealloc((void *) ctxt->attallocs, |
1730 | 1.50k | (maxatts / 5) * sizeof(int)); |
1731 | 1.50k | if (attallocs == NULL) goto mem_error; |
1732 | 1.50k | ctxt->attallocs = attallocs; |
1733 | 1.50k | ctxt->maxatts = maxatts; |
1734 | 1.50k | } |
1735 | 344k | return(ctxt->maxatts); |
1736 | 0 | mem_error: |
1737 | 0 | xmlErrMemory(ctxt, NULL); |
1738 | 0 | return(-1); |
1739 | 344k | } |
1740 | | |
1741 | | /** |
1742 | | * inputPush: |
1743 | | * @ctxt: an XML parser context |
1744 | | * @value: the parser input |
1745 | | * |
1746 | | * Pushes a new parser input on top of the input stack |
1747 | | * |
1748 | | * Returns -1 in case of error, the index in the stack otherwise |
1749 | | */ |
1750 | | int |
1751 | | inputPush(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1752 | 7.66M | { |
1753 | 7.66M | if ((ctxt == NULL) || (value == NULL)) |
1754 | 0 | return(-1); |
1755 | 7.66M | if (ctxt->inputNr >= ctxt->inputMax) { |
1756 | 7.30k | ctxt->inputMax *= 2; |
1757 | 7.30k | ctxt->inputTab = |
1758 | 7.30k | (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1759 | 7.30k | ctxt->inputMax * |
1760 | 7.30k | sizeof(ctxt->inputTab[0])); |
1761 | 7.30k | if (ctxt->inputTab == NULL) { |
1762 | 0 | xmlErrMemory(ctxt, NULL); |
1763 | 0 | ctxt->inputMax /= 2; |
1764 | 0 | return (-1); |
1765 | 0 | } |
1766 | 7.30k | } |
1767 | 7.66M | ctxt->inputTab[ctxt->inputNr] = value; |
1768 | 7.66M | ctxt->input = value; |
1769 | 7.66M | return (ctxt->inputNr++); |
1770 | 7.66M | } |
1771 | | /** |
1772 | | * inputPop: |
1773 | | * @ctxt: an XML parser context |
1774 | | * |
1775 | | * Pops the top parser input from the input stack |
1776 | | * |
1777 | | * Returns the input just removed |
1778 | | */ |
1779 | | xmlParserInputPtr |
1780 | | inputPop(xmlParserCtxtPtr ctxt) |
1781 | 19.8M | { |
1782 | 19.8M | xmlParserInputPtr ret; |
1783 | | |
1784 | 19.8M | if (ctxt == NULL) |
1785 | 0 | return(NULL); |
1786 | 19.8M | if (ctxt->inputNr <= 0) |
1787 | 12.2M | return (NULL); |
1788 | 7.58M | ctxt->inputNr--; |
1789 | 7.58M | if (ctxt->inputNr > 0) |
1790 | 1.76M | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1791 | 5.82M | else |
1792 | 5.82M | ctxt->input = NULL; |
1793 | 7.58M | ret = ctxt->inputTab[ctxt->inputNr]; |
1794 | 7.58M | ctxt->inputTab[ctxt->inputNr] = NULL; |
1795 | 7.58M | return (ret); |
1796 | 19.8M | } |
1797 | | /** |
1798 | | * nodePush: |
1799 | | * @ctxt: an XML parser context |
1800 | | * @value: the element node |
1801 | | * |
1802 | | * Pushes a new element node on top of the node stack |
1803 | | * |
1804 | | * Returns -1 in case of error, the index in the stack otherwise |
1805 | | */ |
1806 | | int |
1807 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
1808 | 31.4M | { |
1809 | 31.4M | if (ctxt == NULL) return(0); |
1810 | 31.4M | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1811 | 93.0k | xmlNodePtr *tmp; |
1812 | | |
1813 | 93.0k | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
1814 | 93.0k | ctxt->nodeMax * 2 * |
1815 | 93.0k | sizeof(ctxt->nodeTab[0])); |
1816 | 93.0k | if (tmp == NULL) { |
1817 | 0 | xmlErrMemory(ctxt, NULL); |
1818 | 0 | return (-1); |
1819 | 0 | } |
1820 | 93.0k | ctxt->nodeTab = tmp; |
1821 | 93.0k | ctxt->nodeMax *= 2; |
1822 | 93.0k | } |
1823 | 31.4M | if ((((unsigned int) ctxt->nodeNr) > xmlParserMaxDepth) && |
1824 | 31.4M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
1825 | 3 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
1826 | 3 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
1827 | 3 | xmlParserMaxDepth); |
1828 | 3 | xmlHaltParser(ctxt); |
1829 | 3 | return(-1); |
1830 | 3 | } |
1831 | 31.4M | ctxt->nodeTab[ctxt->nodeNr] = value; |
1832 | 31.4M | ctxt->node = value; |
1833 | 31.4M | return (ctxt->nodeNr++); |
1834 | 31.4M | } |
1835 | | |
1836 | | /** |
1837 | | * nodePop: |
1838 | | * @ctxt: an XML parser context |
1839 | | * |
1840 | | * Pops the top element node from the node stack |
1841 | | * |
1842 | | * Returns the node just removed |
1843 | | */ |
1844 | | xmlNodePtr |
1845 | | nodePop(xmlParserCtxtPtr ctxt) |
1846 | 30.0M | { |
1847 | 30.0M | xmlNodePtr ret; |
1848 | | |
1849 | 30.0M | if (ctxt == NULL) return(NULL); |
1850 | 30.0M | if (ctxt->nodeNr <= 0) |
1851 | 5.61M | return (NULL); |
1852 | 24.3M | ctxt->nodeNr--; |
1853 | 24.3M | if (ctxt->nodeNr > 0) |
1854 | 22.0M | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
1855 | 2.36M | else |
1856 | 2.36M | ctxt->node = NULL; |
1857 | 24.3M | ret = ctxt->nodeTab[ctxt->nodeNr]; |
1858 | 24.3M | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
1859 | 24.3M | return (ret); |
1860 | 30.0M | } |
1861 | | |
1862 | | /** |
1863 | | * nameNsPush: |
1864 | | * @ctxt: an XML parser context |
1865 | | * @value: the element name |
1866 | | * @prefix: the element prefix |
1867 | | * @URI: the element namespace name |
1868 | | * @line: the current line number for error messages |
1869 | | * @nsNr: the number of namespaces pushed on the namespace table |
1870 | | * |
1871 | | * Pushes a new element name/prefix/URL on top of the name stack |
1872 | | * |
1873 | | * Returns -1 in case of error, the index in the stack otherwise |
1874 | | */ |
1875 | | static int |
1876 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
1877 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
1878 | 102M | { |
1879 | 102M | xmlStartTag *tag; |
1880 | | |
1881 | 102M | if (ctxt->nameNr >= ctxt->nameMax) { |
1882 | 395k | const xmlChar * *tmp; |
1883 | 395k | xmlStartTag *tmp2; |
1884 | 395k | ctxt->nameMax *= 2; |
1885 | 395k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1886 | 395k | ctxt->nameMax * |
1887 | 395k | sizeof(ctxt->nameTab[0])); |
1888 | 395k | if (tmp == NULL) { |
1889 | 0 | ctxt->nameMax /= 2; |
1890 | 0 | goto mem_error; |
1891 | 0 | } |
1892 | 395k | ctxt->nameTab = tmp; |
1893 | 395k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
1894 | 395k | ctxt->nameMax * |
1895 | 395k | sizeof(ctxt->pushTab[0])); |
1896 | 395k | if (tmp2 == NULL) { |
1897 | 0 | ctxt->nameMax /= 2; |
1898 | 0 | goto mem_error; |
1899 | 0 | } |
1900 | 395k | ctxt->pushTab = tmp2; |
1901 | 101M | } else if (ctxt->pushTab == NULL) { |
1902 | 5.26M | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
1903 | 5.26M | sizeof(ctxt->pushTab[0])); |
1904 | 5.26M | if (ctxt->pushTab == NULL) |
1905 | 0 | goto mem_error; |
1906 | 5.26M | } |
1907 | 102M | ctxt->nameTab[ctxt->nameNr] = value; |
1908 | 102M | ctxt->name = value; |
1909 | 102M | tag = &ctxt->pushTab[ctxt->nameNr]; |
1910 | 102M | tag->prefix = prefix; |
1911 | 102M | tag->URI = URI; |
1912 | 102M | tag->line = line; |
1913 | 102M | tag->nsNr = nsNr; |
1914 | 102M | return (ctxt->nameNr++); |
1915 | 0 | mem_error: |
1916 | 0 | xmlErrMemory(ctxt, NULL); |
1917 | 0 | return (-1); |
1918 | 102M | } |
1919 | | #ifdef LIBXML_PUSH_ENABLED |
1920 | | /** |
1921 | | * nameNsPop: |
1922 | | * @ctxt: an XML parser context |
1923 | | * |
1924 | | * Pops the top element/prefix/URI name from the name stack |
1925 | | * |
1926 | | * Returns the name just removed |
1927 | | */ |
1928 | | static const xmlChar * |
1929 | | nameNsPop(xmlParserCtxtPtr ctxt) |
1930 | 1.07M | { |
1931 | 1.07M | const xmlChar *ret; |
1932 | | |
1933 | 1.07M | if (ctxt->nameNr <= 0) |
1934 | 0 | return (NULL); |
1935 | 1.07M | ctxt->nameNr--; |
1936 | 1.07M | if (ctxt->nameNr > 0) |
1937 | 1.03M | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1938 | 42.6k | else |
1939 | 42.6k | ctxt->name = NULL; |
1940 | 1.07M | ret = ctxt->nameTab[ctxt->nameNr]; |
1941 | 1.07M | ctxt->nameTab[ctxt->nameNr] = NULL; |
1942 | 1.07M | return (ret); |
1943 | 1.07M | } |
1944 | | #endif /* LIBXML_PUSH_ENABLED */ |
1945 | | |
1946 | | /** |
1947 | | * namePush: |
1948 | | * @ctxt: an XML parser context |
1949 | | * @value: the element name |
1950 | | * |
1951 | | * Pushes a new element name on top of the name stack |
1952 | | * |
1953 | | * Returns -1 in case of error, the index in the stack otherwise |
1954 | | */ |
1955 | | int |
1956 | | namePush(xmlParserCtxtPtr ctxt, const xmlChar * value) |
1957 | 0 | { |
1958 | 0 | if (ctxt == NULL) return (-1); |
1959 | | |
1960 | 0 | if (ctxt->nameNr >= ctxt->nameMax) { |
1961 | 0 | const xmlChar * *tmp; |
1962 | 0 | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1963 | 0 | ctxt->nameMax * 2 * |
1964 | 0 | sizeof(ctxt->nameTab[0])); |
1965 | 0 | if (tmp == NULL) { |
1966 | 0 | goto mem_error; |
1967 | 0 | } |
1968 | 0 | ctxt->nameTab = tmp; |
1969 | 0 | ctxt->nameMax *= 2; |
1970 | 0 | } |
1971 | 0 | ctxt->nameTab[ctxt->nameNr] = value; |
1972 | 0 | ctxt->name = value; |
1973 | 0 | return (ctxt->nameNr++); |
1974 | 0 | mem_error: |
1975 | 0 | xmlErrMemory(ctxt, NULL); |
1976 | 0 | return (-1); |
1977 | 0 | } |
1978 | | /** |
1979 | | * namePop: |
1980 | | * @ctxt: an XML parser context |
1981 | | * |
1982 | | * Pops the top element name from the name stack |
1983 | | * |
1984 | | * Returns the name just removed |
1985 | | */ |
1986 | | const xmlChar * |
1987 | | namePop(xmlParserCtxtPtr ctxt) |
1988 | 89.7M | { |
1989 | 89.7M | const xmlChar *ret; |
1990 | | |
1991 | 89.7M | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
1992 | 0 | return (NULL); |
1993 | 89.7M | ctxt->nameNr--; |
1994 | 89.7M | if (ctxt->nameNr > 0) |
1995 | 80.1M | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1996 | 9.57M | else |
1997 | 9.57M | ctxt->name = NULL; |
1998 | 89.7M | ret = ctxt->nameTab[ctxt->nameNr]; |
1999 | 89.7M | ctxt->nameTab[ctxt->nameNr] = NULL; |
2000 | 89.7M | return (ret); |
2001 | 89.7M | } |
2002 | | |
2003 | 174M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
2004 | 174M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
2005 | 415k | int *tmp; |
2006 | | |
2007 | 415k | ctxt->spaceMax *= 2; |
2008 | 415k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
2009 | 415k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
2010 | 415k | if (tmp == NULL) { |
2011 | 0 | xmlErrMemory(ctxt, NULL); |
2012 | 0 | ctxt->spaceMax /=2; |
2013 | 0 | return(-1); |
2014 | 0 | } |
2015 | 415k | ctxt->spaceTab = tmp; |
2016 | 415k | } |
2017 | 174M | ctxt->spaceTab[ctxt->spaceNr] = val; |
2018 | 174M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2019 | 174M | return(ctxt->spaceNr++); |
2020 | 174M | } |
2021 | | |
2022 | 163M | static int spacePop(xmlParserCtxtPtr ctxt) { |
2023 | 163M | int ret; |
2024 | 163M | if (ctxt->spaceNr <= 0) return(0); |
2025 | 163M | ctxt->spaceNr--; |
2026 | 163M | if (ctxt->spaceNr > 0) |
2027 | 163M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2028 | 108k | else |
2029 | 108k | ctxt->space = &ctxt->spaceTab[0]; |
2030 | 163M | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2031 | 163M | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2032 | 163M | return(ret); |
2033 | 163M | } |
2034 | | |
2035 | | /* |
2036 | | * Macros for accessing the content. Those should be used only by the parser, |
2037 | | * and not exported. |
2038 | | * |
2039 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2040 | | * use them |
2041 | | * |
2042 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2043 | | * To be used with extreme caution since operations consuming |
2044 | | * characters may move the input buffer to a different location ! |
2045 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2046 | | * This should be used internally by the parser |
2047 | | * only to compare to ASCII values otherwise it would break when |
2048 | | * running with UTF-8 encoding. |
2049 | | * RAW same as CUR but in the input buffer, bypass any token |
2050 | | * extraction that may have been done |
2051 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2052 | | * to compare on ASCII based substring. |
2053 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2054 | | * strings without newlines within the parser. |
2055 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2056 | | * defined char within the parser. |
2057 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2058 | | * |
2059 | | * NEXT Skip to the next character, this does the proper decoding |
2060 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2061 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2062 | | * CUR_CHAR(l) returns the current unicode character (int), set l |
2063 | | * to the number of xmlChars used for the encoding [0-5]. |
2064 | | * CUR_SCHAR same but operate on a string instead of the context |
2065 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2066 | | * the index |
2067 | | * GROW, SHRINK handling of input buffers |
2068 | | */ |
2069 | | |
2070 | 1.98G | #define RAW (*ctxt->input->cur) |
2071 | 264M | #define CUR (*ctxt->input->cur) |
2072 | 1.18G | #define NXT(val) ctxt->input->cur[(val)] |
2073 | 199M | #define CUR_PTR ctxt->input->cur |
2074 | 6.42M | #define BASE_PTR ctxt->input->base |
2075 | | |
2076 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2077 | 1.17G | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2078 | 586M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2079 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2080 | 1.15G | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2081 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2082 | 1.13G | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2083 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2084 | 1.12G | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2085 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2086 | 1.11G | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2087 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2088 | 554M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2089 | 554M | ((unsigned char *) s)[ 8 ] == c9 ) |
2090 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2091 | 272k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2092 | 272k | ((unsigned char *) s)[ 9 ] == c10 ) |
2093 | | |
2094 | 144M | #define SKIP(val) do { \ |
2095 | 144M | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2096 | 144M | if (*ctxt->input->cur == 0) \ |
2097 | 144M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2098 | 144M | } while (0) |
2099 | | |
2100 | 408k | #define SKIPL(val) do { \ |
2101 | 408k | int skipl; \ |
2102 | 18.0M | for(skipl=0; skipl<val; skipl++) { \ |
2103 | 17.6M | if (*(ctxt->input->cur) == '\n') { \ |
2104 | 231k | ctxt->input->line++; ctxt->input->col = 1; \ |
2105 | 17.4M | } else ctxt->input->col++; \ |
2106 | 17.6M | ctxt->input->cur++; \ |
2107 | 17.6M | } \ |
2108 | 408k | if (*ctxt->input->cur == 0) \ |
2109 | 408k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2110 | 408k | } while (0) |
2111 | | |
2112 | 1.26G | #define SHRINK if ((ctxt->progressive == 0) && \ |
2113 | 1.26G | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2114 | 1.26G | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2115 | 1.26G | xmlSHRINK (ctxt); |
2116 | | |
2117 | 1.74M | static void xmlSHRINK (xmlParserCtxtPtr ctxt) { |
2118 | 1.74M | xmlParserInputShrink(ctxt->input); |
2119 | 1.74M | if (*ctxt->input->cur == 0) |
2120 | 115k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2121 | 1.74M | } |
2122 | | |
2123 | 2.86G | #define GROW if ((ctxt->progressive == 0) && \ |
2124 | 2.86G | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2125 | 2.86G | xmlGROW (ctxt); |
2126 | | |
2127 | 168M | static void xmlGROW (xmlParserCtxtPtr ctxt) { |
2128 | 168M | ptrdiff_t curEnd = ctxt->input->end - ctxt->input->cur; |
2129 | 168M | ptrdiff_t curBase = ctxt->input->cur - ctxt->input->base; |
2130 | | |
2131 | 168M | if (((curEnd > XML_MAX_LOOKUP_LIMIT) || |
2132 | 168M | (curBase > XML_MAX_LOOKUP_LIMIT)) && |
2133 | 168M | ((ctxt->input->buf) && |
2134 | 0 | (ctxt->input->buf->readcallback != xmlInputReadCallbackNop)) && |
2135 | 168M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
2136 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
2137 | 0 | xmlHaltParser(ctxt); |
2138 | 0 | return; |
2139 | 0 | } |
2140 | 168M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2141 | 168M | if ((ctxt->input->cur > ctxt->input->end) || |
2142 | 168M | (ctxt->input->cur < ctxt->input->base)) { |
2143 | 0 | xmlHaltParser(ctxt); |
2144 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "cur index out of bound"); |
2145 | 0 | return; |
2146 | 0 | } |
2147 | 168M | if ((ctxt->input->cur != NULL) && (*ctxt->input->cur == 0)) |
2148 | 5.79M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2149 | 168M | } |
2150 | | |
2151 | 470M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2152 | | |
2153 | 464M | #define NEXT xmlNextChar(ctxt) |
2154 | | |
2155 | 255M | #define NEXT1 { \ |
2156 | 255M | ctxt->input->col++; \ |
2157 | 255M | ctxt->input->cur++; \ |
2158 | 255M | if (*ctxt->input->cur == 0) \ |
2159 | 255M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2160 | 255M | } |
2161 | | |
2162 | 3.31G | #define NEXTL(l) do { \ |
2163 | 3.31G | if (*(ctxt->input->cur) == '\n') { \ |
2164 | 51.5M | ctxt->input->line++; ctxt->input->col = 1; \ |
2165 | 3.25G | } else ctxt->input->col++; \ |
2166 | 3.31G | ctxt->input->cur += l; \ |
2167 | 3.31G | } while (0) |
2168 | | |
2169 | 3.45G | #define CUR_CHAR(l) xmlCurrentChar(ctxt, &l) |
2170 | 1.05G | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2171 | | |
2172 | | #define COPY_BUF(l,b,i,v) \ |
2173 | 4.07G | if (l == 1) b[i++] = v; \ |
2174 | 4.07G | else i += xmlCopyCharMultiByte(&b[i],v) |
2175 | | |
2176 | | #define CUR_CONSUMED \ |
2177 | 1.33G | (ctxt->input->consumed + (ctxt->input->cur - ctxt->input->base)) |
2178 | | |
2179 | | /** |
2180 | | * xmlSkipBlankChars: |
2181 | | * @ctxt: the XML parser context |
2182 | | * |
2183 | | * skip all blanks character found at that point in the input streams. |
2184 | | * It pops up finished entities in the process if allowable at that point. |
2185 | | * |
2186 | | * Returns the number of space chars skipped |
2187 | | */ |
2188 | | |
2189 | | int |
2190 | 470M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2191 | 470M | int res = 0; |
2192 | | |
2193 | | /* |
2194 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2195 | | * the ASCII range. |
2196 | | */ |
2197 | 470M | if (((ctxt->inputNr == 1) && (ctxt->instate != XML_PARSER_DTD)) || |
2198 | 470M | (ctxt->instate == XML_PARSER_START)) { |
2199 | 415M | const xmlChar *cur; |
2200 | | /* |
2201 | | * if we are in the document content, go really fast |
2202 | | */ |
2203 | 415M | cur = ctxt->input->cur; |
2204 | 415M | while (IS_BLANK_CH(*cur)) { |
2205 | 137M | if (*cur == '\n') { |
2206 | 5.13M | ctxt->input->line++; ctxt->input->col = 1; |
2207 | 131M | } else { |
2208 | 131M | ctxt->input->col++; |
2209 | 131M | } |
2210 | 137M | cur++; |
2211 | 137M | if (res < INT_MAX) |
2212 | 137M | res++; |
2213 | 137M | if (*cur == 0) { |
2214 | 280k | ctxt->input->cur = cur; |
2215 | 280k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2216 | 280k | cur = ctxt->input->cur; |
2217 | 280k | } |
2218 | 137M | } |
2219 | 415M | ctxt->input->cur = cur; |
2220 | 415M | } else { |
2221 | 55.1M | int expandPE = ((ctxt->external != 0) || (ctxt->inputNr != 1)); |
2222 | | |
2223 | 229M | while (1) { |
2224 | 229M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2225 | 168M | NEXT; |
2226 | 168M | } else if (CUR == '%') { |
2227 | | /* |
2228 | | * Need to handle support of entities branching here |
2229 | | */ |
2230 | 6.00M | if ((expandPE == 0) || (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2231 | 1.48M | break; |
2232 | 4.51M | xmlParsePEReference(ctxt); |
2233 | 55.1M | } else if (CUR == 0) { |
2234 | 1.49M | if (ctxt->inputNr <= 1) |
2235 | 49.9k | break; |
2236 | 1.44M | xmlPopInput(ctxt); |
2237 | 53.6M | } else { |
2238 | 53.6M | break; |
2239 | 53.6M | } |
2240 | | |
2241 | | /* |
2242 | | * Also increase the counter when entering or exiting a PERef. |
2243 | | * The spec says: "When a parameter-entity reference is recognized |
2244 | | * in the DTD and included, its replacement text MUST be enlarged |
2245 | | * by the attachment of one leading and one following space (#x20) |
2246 | | * character." |
2247 | | */ |
2248 | 174M | if (res < INT_MAX) |
2249 | 174M | res++; |
2250 | 174M | } |
2251 | 55.1M | } |
2252 | 470M | return(res); |
2253 | 470M | } |
2254 | | |
2255 | | /************************************************************************ |
2256 | | * * |
2257 | | * Commodity functions to handle entities * |
2258 | | * * |
2259 | | ************************************************************************/ |
2260 | | |
2261 | | /** |
2262 | | * xmlPopInput: |
2263 | | * @ctxt: an XML parser context |
2264 | | * |
2265 | | * xmlPopInput: the current input pointed by ctxt->input came to an end |
2266 | | * pop it and return the next char. |
2267 | | * |
2268 | | * Returns the current xmlChar in the parser context |
2269 | | */ |
2270 | | xmlChar |
2271 | 1.46M | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2272 | 1.46M | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2273 | 1.46M | if (xmlParserDebugEntities) |
2274 | 0 | xmlGenericError(xmlGenericErrorContext, |
2275 | 0 | "Popping input %d\n", ctxt->inputNr); |
2276 | 1.46M | if ((ctxt->inputNr > 1) && (ctxt->inSubset == 0) && |
2277 | 1.46M | (ctxt->instate != XML_PARSER_EOF)) |
2278 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
2279 | 0 | "Unfinished entity outside the DTD"); |
2280 | 1.46M | xmlFreeInputStream(inputPop(ctxt)); |
2281 | 1.46M | if (*ctxt->input->cur == 0) |
2282 | 360 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2283 | 1.46M | return(CUR); |
2284 | 1.46M | } |
2285 | | |
2286 | | /** |
2287 | | * xmlPushInput: |
2288 | | * @ctxt: an XML parser context |
2289 | | * @input: an XML parser input fragment (entity, XML fragment ...). |
2290 | | * |
2291 | | * xmlPushInput: switch to a new input stream which is stacked on top |
2292 | | * of the previous one(s). |
2293 | | * Returns -1 in case of error or the index in the input stack |
2294 | | */ |
2295 | | int |
2296 | 1.83M | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2297 | 1.83M | int ret; |
2298 | 1.83M | if (input == NULL) return(-1); |
2299 | | |
2300 | 1.83M | if (xmlParserDebugEntities) { |
2301 | 0 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
2302 | 0 | xmlGenericError(xmlGenericErrorContext, |
2303 | 0 | "%s(%d): ", ctxt->input->filename, |
2304 | 0 | ctxt->input->line); |
2305 | 0 | xmlGenericError(xmlGenericErrorContext, |
2306 | 0 | "Pushing input %d : %.30s\n", ctxt->inputNr+1, input->cur); |
2307 | 0 | } |
2308 | 1.83M | if (((ctxt->inputNr > 40) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2309 | 1.83M | (ctxt->inputNr > 1024)) { |
2310 | 1.63k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2311 | 298k | while (ctxt->inputNr > 1) |
2312 | 296k | xmlFreeInputStream(inputPop(ctxt)); |
2313 | 1.63k | return(-1); |
2314 | 1.63k | } |
2315 | 1.83M | ret = inputPush(ctxt, input); |
2316 | 1.83M | if (ctxt->instate == XML_PARSER_EOF) |
2317 | 0 | return(-1); |
2318 | 1.83M | GROW; |
2319 | 1.83M | return(ret); |
2320 | 1.83M | } |
2321 | | |
2322 | | /** |
2323 | | * xmlParseCharRef: |
2324 | | * @ctxt: an XML parser context |
2325 | | * |
2326 | | * DEPRECATED: Internal function, don't use. |
2327 | | * |
2328 | | * parse Reference declarations |
2329 | | * |
2330 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2331 | | * '&#x' [0-9a-fA-F]+ ';' |
2332 | | * |
2333 | | * [ WFC: Legal Character ] |
2334 | | * Characters referred to using character references must match the |
2335 | | * production for Char. |
2336 | | * |
2337 | | * Returns the value parsed (as an int), 0 in case of error |
2338 | | */ |
2339 | | int |
2340 | 11.8M | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2341 | 11.8M | int val = 0; |
2342 | 11.8M | int count = 0; |
2343 | | |
2344 | | /* |
2345 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2346 | | */ |
2347 | 11.8M | if ((RAW == '&') && (NXT(1) == '#') && |
2348 | 11.8M | (NXT(2) == 'x')) { |
2349 | 2.64M | SKIP(3); |
2350 | 2.64M | GROW; |
2351 | 6.61M | while (RAW != ';') { /* loop blocked by count */ |
2352 | 4.30M | if (count++ > 20) { |
2353 | 41.8k | count = 0; |
2354 | 41.8k | GROW; |
2355 | 41.8k | if (ctxt->instate == XML_PARSER_EOF) |
2356 | 0 | return(0); |
2357 | 41.8k | } |
2358 | 4.30M | if ((RAW >= '0') && (RAW <= '9')) |
2359 | 1.80M | val = val * 16 + (CUR - '0'); |
2360 | 2.50M | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2361 | 1.94M | val = val * 16 + (CUR - 'a') + 10; |
2362 | 558k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2363 | 220k | val = val * 16 + (CUR - 'A') + 10; |
2364 | 337k | else { |
2365 | 337k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2366 | 337k | val = 0; |
2367 | 337k | break; |
2368 | 337k | } |
2369 | 3.96M | if (val > 0x110000) |
2370 | 469k | val = 0x110000; |
2371 | | |
2372 | 3.96M | NEXT; |
2373 | 3.96M | count++; |
2374 | 3.96M | } |
2375 | 2.64M | if (RAW == ';') { |
2376 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2377 | 2.30M | ctxt->input->col++; |
2378 | 2.30M | ctxt->input->cur++; |
2379 | 2.30M | } |
2380 | 9.24M | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2381 | 9.24M | SKIP(2); |
2382 | 9.24M | GROW; |
2383 | 34.3M | while (RAW != ';') { /* loop blocked by count */ |
2384 | 25.8M | if (count++ > 20) { |
2385 | 79.7k | count = 0; |
2386 | 79.7k | GROW; |
2387 | 79.7k | if (ctxt->instate == XML_PARSER_EOF) |
2388 | 0 | return(0); |
2389 | 79.7k | } |
2390 | 25.8M | if ((RAW >= '0') && (RAW <= '9')) |
2391 | 25.1M | val = val * 10 + (CUR - '0'); |
2392 | 724k | else { |
2393 | 724k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2394 | 724k | val = 0; |
2395 | 724k | break; |
2396 | 724k | } |
2397 | 25.1M | if (val > 0x110000) |
2398 | 895k | val = 0x110000; |
2399 | | |
2400 | 25.1M | NEXT; |
2401 | 25.1M | count++; |
2402 | 25.1M | } |
2403 | 9.24M | if (RAW == ';') { |
2404 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2405 | 8.51M | ctxt->input->col++; |
2406 | 8.51M | ctxt->input->cur++; |
2407 | 8.51M | } |
2408 | 9.24M | } else { |
2409 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2410 | 0 | } |
2411 | | |
2412 | | /* |
2413 | | * [ WFC: Legal Character ] |
2414 | | * Characters referred to using character references must match the |
2415 | | * production for Char. |
2416 | | */ |
2417 | 11.8M | if (val >= 0x110000) { |
2418 | 8.23k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2419 | 8.23k | "xmlParseCharRef: character reference out of bounds\n", |
2420 | 8.23k | val); |
2421 | 11.8M | } else if (IS_CHAR(val)) { |
2422 | 10.8M | return(val); |
2423 | 10.8M | } else { |
2424 | 1.07M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2425 | 1.07M | "xmlParseCharRef: invalid xmlChar value %d\n", |
2426 | 1.07M | val); |
2427 | 1.07M | } |
2428 | 1.08M | return(0); |
2429 | 11.8M | } |
2430 | | |
2431 | | /** |
2432 | | * xmlParseStringCharRef: |
2433 | | * @ctxt: an XML parser context |
2434 | | * @str: a pointer to an index in the string |
2435 | | * |
2436 | | * parse Reference declarations, variant parsing from a string rather |
2437 | | * than an an input flow. |
2438 | | * |
2439 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2440 | | * '&#x' [0-9a-fA-F]+ ';' |
2441 | | * |
2442 | | * [ WFC: Legal Character ] |
2443 | | * Characters referred to using character references must match the |
2444 | | * production for Char. |
2445 | | * |
2446 | | * Returns the value parsed (as an int), 0 in case of error, str will be |
2447 | | * updated to the current value of the index |
2448 | | */ |
2449 | | static int |
2450 | 1.80M | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2451 | 1.80M | const xmlChar *ptr; |
2452 | 1.80M | xmlChar cur; |
2453 | 1.80M | int val = 0; |
2454 | | |
2455 | 1.80M | if ((str == NULL) || (*str == NULL)) return(0); |
2456 | 1.80M | ptr = *str; |
2457 | 1.80M | cur = *ptr; |
2458 | 1.80M | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2459 | 171k | ptr += 3; |
2460 | 171k | cur = *ptr; |
2461 | 469k | while (cur != ';') { /* Non input consuming loop */ |
2462 | 300k | if ((cur >= '0') && (cur <= '9')) |
2463 | 124k | val = val * 16 + (cur - '0'); |
2464 | 175k | else if ((cur >= 'a') && (cur <= 'f')) |
2465 | 88.1k | val = val * 16 + (cur - 'a') + 10; |
2466 | 87.1k | else if ((cur >= 'A') && (cur <= 'F')) |
2467 | 84.6k | val = val * 16 + (cur - 'A') + 10; |
2468 | 2.48k | else { |
2469 | 2.48k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2470 | 2.48k | val = 0; |
2471 | 2.48k | break; |
2472 | 2.48k | } |
2473 | 297k | if (val > 0x110000) |
2474 | 61.1k | val = 0x110000; |
2475 | | |
2476 | 297k | ptr++; |
2477 | 297k | cur = *ptr; |
2478 | 297k | } |
2479 | 171k | if (cur == ';') |
2480 | 168k | ptr++; |
2481 | 1.63M | } else if ((cur == '&') && (ptr[1] == '#')){ |
2482 | 1.63M | ptr += 2; |
2483 | 1.63M | cur = *ptr; |
2484 | 6.11M | while (cur != ';') { /* Non input consuming loops */ |
2485 | 4.47M | if ((cur >= '0') && (cur <= '9')) |
2486 | 4.47M | val = val * 10 + (cur - '0'); |
2487 | 3.71k | else { |
2488 | 3.71k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2489 | 3.71k | val = 0; |
2490 | 3.71k | break; |
2491 | 3.71k | } |
2492 | 4.47M | if (val > 0x110000) |
2493 | 1.17k | val = 0x110000; |
2494 | | |
2495 | 4.47M | ptr++; |
2496 | 4.47M | cur = *ptr; |
2497 | 4.47M | } |
2498 | 1.63M | if (cur == ';') |
2499 | 1.63M | ptr++; |
2500 | 1.63M | } else { |
2501 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2502 | 0 | return(0); |
2503 | 0 | } |
2504 | 1.80M | *str = ptr; |
2505 | | |
2506 | | /* |
2507 | | * [ WFC: Legal Character ] |
2508 | | * Characters referred to using character references must match the |
2509 | | * production for Char. |
2510 | | */ |
2511 | 1.80M | if (val >= 0x110000) { |
2512 | 295 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2513 | 295 | "xmlParseStringCharRef: character reference out of bounds\n", |
2514 | 295 | val); |
2515 | 1.80M | } else if (IS_CHAR(val)) { |
2516 | 1.80M | return(val); |
2517 | 1.80M | } else { |
2518 | 6.80k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2519 | 6.80k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2520 | 6.80k | val); |
2521 | 6.80k | } |
2522 | 7.10k | return(0); |
2523 | 1.80M | } |
2524 | | |
2525 | | /** |
2526 | | * xmlParserHandlePEReference: |
2527 | | * @ctxt: the parser context |
2528 | | * |
2529 | | * [69] PEReference ::= '%' Name ';' |
2530 | | * |
2531 | | * [ WFC: No Recursion ] |
2532 | | * A parsed entity must not contain a recursive |
2533 | | * reference to itself, either directly or indirectly. |
2534 | | * |
2535 | | * [ WFC: Entity Declared ] |
2536 | | * In a document without any DTD, a document with only an internal DTD |
2537 | | * subset which contains no parameter entity references, or a document |
2538 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2539 | | * entity must precede any reference to it... |
2540 | | * |
2541 | | * [ VC: Entity Declared ] |
2542 | | * In a document with an external subset or external parameter entities |
2543 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2544 | | * must precede any reference to it... |
2545 | | * |
2546 | | * [ WFC: In DTD ] |
2547 | | * Parameter-entity references may only appear in the DTD. |
2548 | | * NOTE: misleading but this is handled. |
2549 | | * |
2550 | | * A PEReference may have been detected in the current input stream |
2551 | | * the handling is done accordingly to |
2552 | | * http://www.w3.org/TR/REC-xml#entproc |
2553 | | * i.e. |
2554 | | * - Included in literal in entity values |
2555 | | * - Included as Parameter Entity reference within DTDs |
2556 | | */ |
2557 | | void |
2558 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2559 | 0 | switch(ctxt->instate) { |
2560 | 0 | case XML_PARSER_CDATA_SECTION: |
2561 | 0 | return; |
2562 | 0 | case XML_PARSER_COMMENT: |
2563 | 0 | return; |
2564 | 0 | case XML_PARSER_START_TAG: |
2565 | 0 | return; |
2566 | 0 | case XML_PARSER_END_TAG: |
2567 | 0 | return; |
2568 | 0 | case XML_PARSER_EOF: |
2569 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_AT_EOF, NULL); |
2570 | 0 | return; |
2571 | 0 | case XML_PARSER_PROLOG: |
2572 | 0 | case XML_PARSER_START: |
2573 | 0 | case XML_PARSER_MISC: |
2574 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_PROLOG, NULL); |
2575 | 0 | return; |
2576 | 0 | case XML_PARSER_ENTITY_DECL: |
2577 | 0 | case XML_PARSER_CONTENT: |
2578 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
2579 | 0 | case XML_PARSER_PI: |
2580 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
2581 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
2582 | | /* we just ignore it there */ |
2583 | 0 | return; |
2584 | 0 | case XML_PARSER_EPILOG: |
2585 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_EPILOG, NULL); |
2586 | 0 | return; |
2587 | 0 | case XML_PARSER_ENTITY_VALUE: |
2588 | | /* |
2589 | | * NOTE: in the case of entity values, we don't do the |
2590 | | * substitution here since we need the literal |
2591 | | * entity value to be able to save the internal |
2592 | | * subset of the document. |
2593 | | * This will be handled by xmlStringDecodeEntities |
2594 | | */ |
2595 | 0 | return; |
2596 | 0 | case XML_PARSER_DTD: |
2597 | | /* |
2598 | | * [WFC: Well-Formedness Constraint: PEs in Internal Subset] |
2599 | | * In the internal DTD subset, parameter-entity references |
2600 | | * can occur only where markup declarations can occur, not |
2601 | | * within markup declarations. |
2602 | | * In that case this is handled in xmlParseMarkupDecl |
2603 | | */ |
2604 | 0 | if ((ctxt->external == 0) && (ctxt->inputNr == 1)) |
2605 | 0 | return; |
2606 | 0 | if (IS_BLANK_CH(NXT(1)) || NXT(1) == 0) |
2607 | 0 | return; |
2608 | 0 | break; |
2609 | 0 | case XML_PARSER_IGNORE: |
2610 | 0 | return; |
2611 | 0 | } |
2612 | | |
2613 | 0 | xmlParsePEReference(ctxt); |
2614 | 0 | } |
2615 | | |
2616 | | /* |
2617 | | * Macro used to grow the current buffer. |
2618 | | * buffer##_size is expected to be a size_t |
2619 | | * mem_error: is expected to handle memory allocation failures |
2620 | | */ |
2621 | 1.57M | #define growBuffer(buffer, n) { \ |
2622 | 1.57M | xmlChar *tmp; \ |
2623 | 1.57M | size_t new_size = buffer##_size * 2 + n; \ |
2624 | 1.57M | if (new_size < buffer##_size) goto mem_error; \ |
2625 | 1.57M | tmp = (xmlChar *) xmlRealloc(buffer, new_size); \ |
2626 | 1.57M | if (tmp == NULL) goto mem_error; \ |
2627 | 1.57M | buffer = tmp; \ |
2628 | 1.57M | buffer##_size = new_size; \ |
2629 | 1.57M | } |
2630 | | |
2631 | | /** |
2632 | | * xmlStringLenDecodeEntities: |
2633 | | * @ctxt: the parser context |
2634 | | * @str: the input string |
2635 | | * @len: the string length |
2636 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2637 | | * @end: an end marker xmlChar, 0 if none |
2638 | | * @end2: an end marker xmlChar, 0 if none |
2639 | | * @end3: an end marker xmlChar, 0 if none |
2640 | | * |
2641 | | * Takes a entity string content and process to do the adequate substitutions. |
2642 | | * |
2643 | | * [67] Reference ::= EntityRef | CharRef |
2644 | | * |
2645 | | * [69] PEReference ::= '%' Name ';' |
2646 | | * |
2647 | | * Returns A newly allocated string with the substitution done. The caller |
2648 | | * must deallocate it ! |
2649 | | */ |
2650 | | xmlChar * |
2651 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2652 | 11.1M | int what, xmlChar end, xmlChar end2, xmlChar end3) { |
2653 | 11.1M | xmlChar *buffer = NULL; |
2654 | 11.1M | size_t buffer_size = 0; |
2655 | 11.1M | size_t nbchars = 0; |
2656 | | |
2657 | 11.1M | xmlChar *current = NULL; |
2658 | 11.1M | xmlChar *rep = NULL; |
2659 | 11.1M | const xmlChar *last; |
2660 | 11.1M | xmlEntityPtr ent; |
2661 | 11.1M | int c,l; |
2662 | | |
2663 | 11.1M | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2664 | 0 | return(NULL); |
2665 | 11.1M | last = str + len; |
2666 | | |
2667 | 11.1M | if (((ctxt->depth > 40) && |
2668 | 11.1M | ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2669 | 11.1M | (ctxt->depth > 1024)) { |
2670 | 13.0k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2671 | 13.0k | return(NULL); |
2672 | 13.0k | } |
2673 | | |
2674 | | /* |
2675 | | * allocate a translation buffer. |
2676 | | */ |
2677 | 11.0M | buffer_size = XML_PARSER_BIG_BUFFER_SIZE; |
2678 | 11.0M | buffer = (xmlChar *) xmlMallocAtomic(buffer_size); |
2679 | 11.0M | if (buffer == NULL) goto mem_error; |
2680 | | |
2681 | | /* |
2682 | | * OK loop until we reach one of the ending char or a size limit. |
2683 | | * we are operating on already parsed values. |
2684 | | */ |
2685 | 11.0M | if (str < last) |
2686 | 10.3M | c = CUR_SCHAR(str, l); |
2687 | 723k | else |
2688 | 723k | c = 0; |
2689 | 1.02G | while ((c != 0) && (c != end) && /* non input consuming loop */ |
2690 | 1.02G | (c != end2) && (c != end3) && |
2691 | 1.02G | (ctxt->instate != XML_PARSER_EOF)) { |
2692 | | |
2693 | 1.01G | if (c == 0) break; |
2694 | 1.01G | if ((c == '&') && (str[1] == '#')) { |
2695 | 1.80M | int val = xmlParseStringCharRef(ctxt, &str); |
2696 | 1.80M | if (val == 0) |
2697 | 7.10k | goto int_error; |
2698 | 1.80M | COPY_BUF(0,buffer,nbchars,val); |
2699 | 1.80M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2700 | 24.9k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2701 | 24.9k | } |
2702 | 1.01G | } else if ((c == '&') && (what & XML_SUBSTITUTE_REF)) { |
2703 | 921k | if (xmlParserDebugEntities) |
2704 | 0 | xmlGenericError(xmlGenericErrorContext, |
2705 | 0 | "String decoding Entity Reference: %.30s\n", |
2706 | 0 | str); |
2707 | 921k | ent = xmlParseStringEntityRef(ctxt, &str); |
2708 | 921k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2709 | 921k | if (ent != NULL) |
2710 | 786k | ctxt->nbentities += ent->checked / 2; |
2711 | 921k | if ((ent != NULL) && |
2712 | 921k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
2713 | 119k | if (ent->content != NULL) { |
2714 | 119k | COPY_BUF(0,buffer,nbchars,ent->content[0]); |
2715 | 119k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2716 | 15.5k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2717 | 15.5k | } |
2718 | 119k | } else { |
2719 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
2720 | 0 | "predefined entity has no content\n"); |
2721 | 0 | goto int_error; |
2722 | 0 | } |
2723 | 801k | } else if ((ent != NULL) && (ent->content != NULL)) { |
2724 | 649k | ctxt->depth++; |
2725 | 649k | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2726 | 649k | 0, 0, 0); |
2727 | 649k | ctxt->depth--; |
2728 | 649k | if (rep == NULL) { |
2729 | 467k | ent->content[0] = 0; |
2730 | 467k | goto int_error; |
2731 | 467k | } |
2732 | | |
2733 | 182k | current = rep; |
2734 | 11.5M | while (*current != 0) { /* non input consuming loop */ |
2735 | 11.3M | buffer[nbchars++] = *current++; |
2736 | 11.3M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2737 | 21.0k | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2738 | 20 | goto int_error; |
2739 | 63.1k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2740 | 63.1k | } |
2741 | 11.3M | } |
2742 | 182k | xmlFree(rep); |
2743 | 182k | rep = NULL; |
2744 | 182k | } else if (ent != NULL) { |
2745 | 17.3k | int i = xmlStrlen(ent->name); |
2746 | 17.3k | const xmlChar *cur = ent->name; |
2747 | | |
2748 | 17.3k | buffer[nbchars++] = '&'; |
2749 | 17.3k | if (nbchars + i + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2750 | 124 | growBuffer(buffer, i + XML_PARSER_BUFFER_SIZE); |
2751 | 124 | } |
2752 | 93.3k | for (;i > 0;i--) |
2753 | 76.0k | buffer[nbchars++] = *cur++; |
2754 | 17.3k | buffer[nbchars++] = ';'; |
2755 | 17.3k | } |
2756 | 1.01G | } else if (c == '%' && (what & XML_SUBSTITUTE_PEREF)) { |
2757 | 1.91M | if (xmlParserDebugEntities) |
2758 | 0 | xmlGenericError(xmlGenericErrorContext, |
2759 | 0 | "String decoding PE Reference: %.30s\n", str); |
2760 | 1.91M | ent = xmlParseStringPEReference(ctxt, &str); |
2761 | 1.91M | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2762 | 1.91M | if (ent != NULL) |
2763 | 1.28M | ctxt->nbentities += ent->checked / 2; |
2764 | 1.91M | if (ent != NULL) { |
2765 | 1.28M | if (ent->content == NULL) { |
2766 | | /* |
2767 | | * Note: external parsed entities will not be loaded, |
2768 | | * it is not required for a non-validating parser to |
2769 | | * complete external PEReferences coming from the |
2770 | | * internal subset |
2771 | | */ |
2772 | 38.3k | if (((ctxt->options & XML_PARSE_NOENT) != 0) || |
2773 | 38.3k | ((ctxt->options & XML_PARSE_DTDVALID) != 0) || |
2774 | 38.3k | (ctxt->validate != 0)) { |
2775 | 10.5k | xmlLoadEntityContent(ctxt, ent); |
2776 | 27.8k | } else { |
2777 | 27.8k | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
2778 | 27.8k | "not validating will not read content for PE entity %s\n", |
2779 | 27.8k | ent->name, NULL); |
2780 | 27.8k | } |
2781 | 38.3k | } |
2782 | 1.28M | ctxt->depth++; |
2783 | 1.28M | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2784 | 1.28M | 0, 0, 0); |
2785 | 1.28M | ctxt->depth--; |
2786 | 1.28M | if (rep == NULL) { |
2787 | 250k | if (ent->content != NULL) |
2788 | 217k | ent->content[0] = 0; |
2789 | 250k | goto int_error; |
2790 | 250k | } |
2791 | 1.03M | current = rep; |
2792 | 51.7M | while (*current != 0) { /* non input consuming loop */ |
2793 | 50.7M | buffer[nbchars++] = *current++; |
2794 | 50.7M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2795 | 112k | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2796 | 1.08k | goto int_error; |
2797 | 335k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2798 | 335k | } |
2799 | 50.7M | } |
2800 | 1.03M | xmlFree(rep); |
2801 | 1.03M | rep = NULL; |
2802 | 1.03M | } |
2803 | 1.01G | } else { |
2804 | 1.01G | COPY_BUF(l,buffer,nbchars,c); |
2805 | 1.01G | str += l; |
2806 | 1.01G | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2807 | 1.59M | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2808 | 1.59M | } |
2809 | 1.01G | } |
2810 | 1.01G | if (str < last) |
2811 | 1.00G | c = CUR_SCHAR(str, l); |
2812 | 9.64M | else |
2813 | 9.64M | c = 0; |
2814 | 1.01G | } |
2815 | 10.3M | buffer[nbchars] = 0; |
2816 | 10.3M | return(buffer); |
2817 | | |
2818 | 0 | mem_error: |
2819 | 0 | xmlErrMemory(ctxt, NULL); |
2820 | 726k | int_error: |
2821 | 726k | if (rep != NULL) |
2822 | 1.10k | xmlFree(rep); |
2823 | 726k | if (buffer != NULL) |
2824 | 726k | xmlFree(buffer); |
2825 | 726k | return(NULL); |
2826 | 0 | } |
2827 | | |
2828 | | /** |
2829 | | * xmlStringDecodeEntities: |
2830 | | * @ctxt: the parser context |
2831 | | * @str: the input string |
2832 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2833 | | * @end: an end marker xmlChar, 0 if none |
2834 | | * @end2: an end marker xmlChar, 0 if none |
2835 | | * @end3: an end marker xmlChar, 0 if none |
2836 | | * |
2837 | | * Takes a entity string content and process to do the adequate substitutions. |
2838 | | * |
2839 | | * [67] Reference ::= EntityRef | CharRef |
2840 | | * |
2841 | | * [69] PEReference ::= '%' Name ';' |
2842 | | * |
2843 | | * Returns A newly allocated string with the substitution done. The caller |
2844 | | * must deallocate it ! |
2845 | | */ |
2846 | | xmlChar * |
2847 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int what, |
2848 | 11.0M | xmlChar end, xmlChar end2, xmlChar end3) { |
2849 | 11.0M | if ((ctxt == NULL) || (str == NULL)) return(NULL); |
2850 | 11.0M | return(xmlStringLenDecodeEntities(ctxt, str, xmlStrlen(str), what, |
2851 | 11.0M | end, end2, end3)); |
2852 | 11.0M | } |
2853 | | |
2854 | | /************************************************************************ |
2855 | | * * |
2856 | | * Commodity functions, cleanup needed ? * |
2857 | | * * |
2858 | | ************************************************************************/ |
2859 | | |
2860 | | /** |
2861 | | * areBlanks: |
2862 | | * @ctxt: an XML parser context |
2863 | | * @str: a xmlChar * |
2864 | | * @len: the size of @str |
2865 | | * @blank_chars: we know the chars are blanks |
2866 | | * |
2867 | | * Is this a sequence of blank chars that one can ignore ? |
2868 | | * |
2869 | | * Returns 1 if ignorable 0 otherwise. |
2870 | | */ |
2871 | | |
2872 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2873 | 35.8M | int blank_chars) { |
2874 | 35.8M | int i, ret; |
2875 | 35.8M | xmlNodePtr lastChild; |
2876 | | |
2877 | | /* |
2878 | | * Don't spend time trying to differentiate them, the same callback is |
2879 | | * used ! |
2880 | | */ |
2881 | 35.8M | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2882 | 679k | return(0); |
2883 | | |
2884 | | /* |
2885 | | * Check for xml:space value. |
2886 | | */ |
2887 | 35.1M | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2888 | 35.1M | (*(ctxt->space) == -2)) |
2889 | 19.5M | return(0); |
2890 | | |
2891 | | /* |
2892 | | * Check that the string is made of blanks |
2893 | | */ |
2894 | 15.6M | if (blank_chars == 0) { |
2895 | 32.7M | for (i = 0;i < len;i++) |
2896 | 28.4M | if (!(IS_BLANK_CH(str[i]))) return(0); |
2897 | 7.21M | } |
2898 | | |
2899 | | /* |
2900 | | * Look if the element is mixed content in the DTD if available |
2901 | | */ |
2902 | 12.7M | if (ctxt->node == NULL) return(0); |
2903 | 10.6M | if (ctxt->myDoc != NULL) { |
2904 | 10.6M | ret = xmlIsMixedElement(ctxt->myDoc, ctxt->node->name); |
2905 | 10.6M | if (ret == 0) return(1); |
2906 | 9.60M | if (ret == 1) return(0); |
2907 | 9.60M | } |
2908 | | |
2909 | | /* |
2910 | | * Otherwise, heuristic :-\ |
2911 | | */ |
2912 | 9.58M | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2913 | 8.41M | if ((ctxt->node->children == NULL) && |
2914 | 8.41M | (RAW == '<') && (NXT(1) == '/')) return(0); |
2915 | | |
2916 | 8.40M | lastChild = xmlGetLastChild(ctxt->node); |
2917 | 8.40M | if (lastChild == NULL) { |
2918 | 1.97M | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2919 | 1.97M | (ctxt->node->content != NULL)) return(0); |
2920 | 6.43M | } else if (xmlNodeIsText(lastChild)) |
2921 | 2.30M | return(0); |
2922 | 4.12M | else if ((ctxt->node->children != NULL) && |
2923 | 4.12M | (xmlNodeIsText(ctxt->node->children))) |
2924 | 109k | return(0); |
2925 | 5.99M | return(1); |
2926 | 8.40M | } |
2927 | | |
2928 | | /************************************************************************ |
2929 | | * * |
2930 | | * Extra stuff for namespace support * |
2931 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2932 | | * * |
2933 | | ************************************************************************/ |
2934 | | |
2935 | | /** |
2936 | | * xmlSplitQName: |
2937 | | * @ctxt: an XML parser context |
2938 | | * @name: an XML parser context |
2939 | | * @prefix: a xmlChar ** |
2940 | | * |
2941 | | * parse an UTF8 encoded XML qualified name string |
2942 | | * |
2943 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2944 | | * |
2945 | | * [NS 6] Prefix ::= NCName |
2946 | | * |
2947 | | * [NS 7] LocalPart ::= NCName |
2948 | | * |
2949 | | * Returns the local part, and prefix is updated |
2950 | | * to get the Prefix if any. |
2951 | | */ |
2952 | | |
2953 | | xmlChar * |
2954 | 37.2M | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefix) { |
2955 | 37.2M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2956 | 37.2M | xmlChar *buffer = NULL; |
2957 | 37.2M | int len = 0; |
2958 | 37.2M | int max = XML_MAX_NAMELEN; |
2959 | 37.2M | xmlChar *ret = NULL; |
2960 | 37.2M | const xmlChar *cur = name; |
2961 | 37.2M | int c; |
2962 | | |
2963 | 37.2M | if (prefix == NULL) return(NULL); |
2964 | 37.2M | *prefix = NULL; |
2965 | | |
2966 | 37.2M | if (cur == NULL) return(NULL); |
2967 | | |
2968 | | #ifndef XML_XML_NAMESPACE |
2969 | | /* xml: prefix is not really a namespace */ |
2970 | | if ((cur[0] == 'x') && (cur[1] == 'm') && |
2971 | | (cur[2] == 'l') && (cur[3] == ':')) |
2972 | | return(xmlStrdup(name)); |
2973 | | #endif |
2974 | | |
2975 | | /* nasty but well=formed */ |
2976 | 37.2M | if (cur[0] == ':') |
2977 | 17.1k | return(xmlStrdup(name)); |
2978 | | |
2979 | 37.2M | c = *cur++; |
2980 | 160M | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
2981 | 123M | buf[len++] = c; |
2982 | 123M | c = *cur++; |
2983 | 123M | } |
2984 | 37.2M | if (len >= max) { |
2985 | | /* |
2986 | | * Okay someone managed to make a huge name, so he's ready to pay |
2987 | | * for the processing speed. |
2988 | | */ |
2989 | 18.0k | max = len * 2; |
2990 | | |
2991 | 18.0k | buffer = (xmlChar *) xmlMallocAtomic(max); |
2992 | 18.0k | if (buffer == NULL) { |
2993 | 0 | xmlErrMemory(ctxt, NULL); |
2994 | 0 | return(NULL); |
2995 | 0 | } |
2996 | 18.0k | memcpy(buffer, buf, len); |
2997 | 18.4M | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
2998 | 18.4M | if (len + 10 > max) { |
2999 | 31.9k | xmlChar *tmp; |
3000 | | |
3001 | 31.9k | max *= 2; |
3002 | 31.9k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3003 | 31.9k | if (tmp == NULL) { |
3004 | 0 | xmlFree(buffer); |
3005 | 0 | xmlErrMemory(ctxt, NULL); |
3006 | 0 | return(NULL); |
3007 | 0 | } |
3008 | 31.9k | buffer = tmp; |
3009 | 31.9k | } |
3010 | 18.4M | buffer[len++] = c; |
3011 | 18.4M | c = *cur++; |
3012 | 18.4M | } |
3013 | 18.0k | buffer[len] = 0; |
3014 | 18.0k | } |
3015 | | |
3016 | 37.2M | if ((c == ':') && (*cur == 0)) { |
3017 | 57.0k | if (buffer != NULL) |
3018 | 89 | xmlFree(buffer); |
3019 | 57.0k | *prefix = NULL; |
3020 | 57.0k | return(xmlStrdup(name)); |
3021 | 57.0k | } |
3022 | | |
3023 | 37.1M | if (buffer == NULL) |
3024 | 37.1M | ret = xmlStrndup(buf, len); |
3025 | 17.9k | else { |
3026 | 17.9k | ret = buffer; |
3027 | 17.9k | buffer = NULL; |
3028 | 17.9k | max = XML_MAX_NAMELEN; |
3029 | 17.9k | } |
3030 | | |
3031 | | |
3032 | 37.1M | if (c == ':') { |
3033 | 11.2M | c = *cur; |
3034 | 11.2M | *prefix = ret; |
3035 | 11.2M | if (c == 0) { |
3036 | 0 | return(xmlStrndup(BAD_CAST "", 0)); |
3037 | 0 | } |
3038 | 11.2M | len = 0; |
3039 | | |
3040 | | /* |
3041 | | * Check that the first character is proper to start |
3042 | | * a new name |
3043 | | */ |
3044 | 11.2M | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3045 | 11.2M | ((c >= 0x41) && (c <= 0x5A)) || |
3046 | 11.2M | (c == '_') || (c == ':'))) { |
3047 | 9.13k | int l; |
3048 | 9.13k | int first = CUR_SCHAR(cur, l); |
3049 | | |
3050 | 9.13k | if (!IS_LETTER(first) && (first != '_')) { |
3051 | 5.77k | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3052 | 5.77k | "Name %s is not XML Namespace compliant\n", |
3053 | 5.77k | name); |
3054 | 5.77k | } |
3055 | 9.13k | } |
3056 | 11.2M | cur++; |
3057 | | |
3058 | 46.2M | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3059 | 34.9M | buf[len++] = c; |
3060 | 34.9M | c = *cur++; |
3061 | 34.9M | } |
3062 | 11.2M | if (len >= max) { |
3063 | | /* |
3064 | | * Okay someone managed to make a huge name, so he's ready to pay |
3065 | | * for the processing speed. |
3066 | | */ |
3067 | 14.7k | max = len * 2; |
3068 | | |
3069 | 14.7k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3070 | 14.7k | if (buffer == NULL) { |
3071 | 0 | xmlErrMemory(ctxt, NULL); |
3072 | 0 | return(NULL); |
3073 | 0 | } |
3074 | 14.7k | memcpy(buffer, buf, len); |
3075 | 39.6M | while (c != 0) { /* tested bigname2.xml */ |
3076 | 39.6M | if (len + 10 > max) { |
3077 | 28.3k | xmlChar *tmp; |
3078 | | |
3079 | 28.3k | max *= 2; |
3080 | 28.3k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3081 | 28.3k | if (tmp == NULL) { |
3082 | 0 | xmlErrMemory(ctxt, NULL); |
3083 | 0 | xmlFree(buffer); |
3084 | 0 | return(NULL); |
3085 | 0 | } |
3086 | 28.3k | buffer = tmp; |
3087 | 28.3k | } |
3088 | 39.6M | buffer[len++] = c; |
3089 | 39.6M | c = *cur++; |
3090 | 39.6M | } |
3091 | 14.7k | buffer[len] = 0; |
3092 | 14.7k | } |
3093 | | |
3094 | 11.2M | if (buffer == NULL) |
3095 | 11.2M | ret = xmlStrndup(buf, len); |
3096 | 14.7k | else { |
3097 | 14.7k | ret = buffer; |
3098 | 14.7k | } |
3099 | 11.2M | } |
3100 | | |
3101 | 37.1M | return(ret); |
3102 | 37.1M | } |
3103 | | |
3104 | | /************************************************************************ |
3105 | | * * |
3106 | | * The parser itself * |
3107 | | * Relates to http://www.w3.org/TR/REC-xml * |
3108 | | * * |
3109 | | ************************************************************************/ |
3110 | | |
3111 | | /************************************************************************ |
3112 | | * * |
3113 | | * Routines to parse Name, NCName and NmToken * |
3114 | | * * |
3115 | | ************************************************************************/ |
3116 | | #ifdef DEBUG |
3117 | | static unsigned long nbParseName = 0; |
3118 | | static unsigned long nbParseNmToken = 0; |
3119 | | static unsigned long nbParseNCName = 0; |
3120 | | static unsigned long nbParseNCNameComplex = 0; |
3121 | | static unsigned long nbParseNameComplex = 0; |
3122 | | static unsigned long nbParseStringName = 0; |
3123 | | #endif |
3124 | | |
3125 | | /* |
3126 | | * The two following functions are related to the change of accepted |
3127 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3128 | | * They correspond to the modified production [4] and the new production [4a] |
3129 | | * changes in that revision. Also note that the macros used for the |
3130 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3131 | | * anymore. |
3132 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3133 | | * new XML_PARSE_OLD10 option is given to the parser. |
3134 | | */ |
3135 | | static int |
3136 | 10.1M | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3137 | 10.1M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3138 | | /* |
3139 | | * Use the new checks of production [4] [4a] amd [5] of the |
3140 | | * Update 5 of XML-1.0 |
3141 | | */ |
3142 | 6.18M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3143 | 6.18M | (((c >= 'a') && (c <= 'z')) || |
3144 | 6.08M | ((c >= 'A') && (c <= 'Z')) || |
3145 | 6.08M | (c == '_') || (c == ':') || |
3146 | 6.08M | ((c >= 0xC0) && (c <= 0xD6)) || |
3147 | 6.08M | ((c >= 0xD8) && (c <= 0xF6)) || |
3148 | 6.08M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3149 | 6.08M | ((c >= 0x370) && (c <= 0x37D)) || |
3150 | 6.08M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3151 | 6.08M | ((c >= 0x200C) && (c <= 0x200D)) || |
3152 | 6.08M | ((c >= 0x2070) && (c <= 0x218F)) || |
3153 | 6.08M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3154 | 6.08M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3155 | 6.08M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3156 | 6.08M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3157 | 6.08M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3158 | 2.45M | return(1); |
3159 | 6.18M | } else { |
3160 | 3.97M | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3161 | 1.37M | return(1); |
3162 | 3.97M | } |
3163 | 6.32M | return(0); |
3164 | 10.1M | } |
3165 | | |
3166 | | static int |
3167 | 71.0M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3168 | 71.0M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3169 | | /* |
3170 | | * Use the new checks of production [4] [4a] amd [5] of the |
3171 | | * Update 5 of XML-1.0 |
3172 | | */ |
3173 | 57.8M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3174 | 57.8M | (((c >= 'a') && (c <= 'z')) || |
3175 | 57.7M | ((c >= 'A') && (c <= 'Z')) || |
3176 | 57.7M | ((c >= '0') && (c <= '9')) || /* !start */ |
3177 | 57.7M | (c == '_') || (c == ':') || |
3178 | 57.7M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3179 | 57.7M | ((c >= 0xC0) && (c <= 0xD6)) || |
3180 | 57.7M | ((c >= 0xD8) && (c <= 0xF6)) || |
3181 | 57.7M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3182 | 57.7M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3183 | 57.7M | ((c >= 0x370) && (c <= 0x37D)) || |
3184 | 57.7M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3185 | 57.7M | ((c >= 0x200C) && (c <= 0x200D)) || |
3186 | 57.7M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3187 | 57.7M | ((c >= 0x2070) && (c <= 0x218F)) || |
3188 | 57.7M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3189 | 57.7M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3190 | 57.7M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3191 | 57.7M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3192 | 57.7M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3193 | 54.8M | return(1); |
3194 | 57.8M | } else { |
3195 | 13.1M | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3196 | 13.1M | (c == '.') || (c == '-') || |
3197 | 13.1M | (c == '_') || (c == ':') || |
3198 | 13.1M | (IS_COMBINING(c)) || |
3199 | 13.1M | (IS_EXTENDER(c))) |
3200 | 11.6M | return(1); |
3201 | 13.1M | } |
3202 | 4.48M | return(0); |
3203 | 71.0M | } |
3204 | | |
3205 | | static xmlChar * xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, |
3206 | | int *len, int *alloc, int normalize); |
3207 | | |
3208 | | static const xmlChar * |
3209 | 105M | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3210 | 105M | int len = 0, l; |
3211 | 105M | int c; |
3212 | 105M | int count = 0; |
3213 | 105M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3214 | 75.5M | XML_MAX_TEXT_LENGTH : |
3215 | 105M | XML_MAX_NAME_LENGTH; |
3216 | | |
3217 | | #ifdef DEBUG |
3218 | | nbParseNameComplex++; |
3219 | | #endif |
3220 | | |
3221 | | /* |
3222 | | * Handler for more complex cases |
3223 | | */ |
3224 | 105M | GROW; |
3225 | 105M | if (ctxt->instate == XML_PARSER_EOF) |
3226 | 0 | return(NULL); |
3227 | 105M | c = CUR_CHAR(l); |
3228 | 105M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3229 | | /* |
3230 | | * Use the new checks of production [4] [4a] amd [5] of the |
3231 | | * Update 5 of XML-1.0 |
3232 | | */ |
3233 | 58.2M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3234 | 58.2M | (!(((c >= 'a') && (c <= 'z')) || |
3235 | 57.7M | ((c >= 'A') && (c <= 'Z')) || |
3236 | 57.7M | (c == '_') || (c == ':') || |
3237 | 57.7M | ((c >= 0xC0) && (c <= 0xD6)) || |
3238 | 57.7M | ((c >= 0xD8) && (c <= 0xF6)) || |
3239 | 57.7M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3240 | 57.7M | ((c >= 0x370) && (c <= 0x37D)) || |
3241 | 57.7M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3242 | 57.7M | ((c >= 0x200C) && (c <= 0x200D)) || |
3243 | 57.7M | ((c >= 0x2070) && (c <= 0x218F)) || |
3244 | 57.7M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3245 | 57.7M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3246 | 57.7M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3247 | 57.7M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3248 | 57.7M | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3249 | 55.3M | return(NULL); |
3250 | 55.3M | } |
3251 | 2.88M | len += l; |
3252 | 2.88M | NEXTL(l); |
3253 | 2.88M | c = CUR_CHAR(l); |
3254 | 64.7M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3255 | 64.7M | (((c >= 'a') && (c <= 'z')) || |
3256 | 63.6M | ((c >= 'A') && (c <= 'Z')) || |
3257 | 63.6M | ((c >= '0') && (c <= '9')) || /* !start */ |
3258 | 63.6M | (c == '_') || (c == ':') || |
3259 | 63.6M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3260 | 63.6M | ((c >= 0xC0) && (c <= 0xD6)) || |
3261 | 63.6M | ((c >= 0xD8) && (c <= 0xF6)) || |
3262 | 63.6M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3263 | 63.6M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3264 | 63.6M | ((c >= 0x370) && (c <= 0x37D)) || |
3265 | 63.6M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3266 | 63.6M | ((c >= 0x200C) && (c <= 0x200D)) || |
3267 | 63.6M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3268 | 63.6M | ((c >= 0x2070) && (c <= 0x218F)) || |
3269 | 63.6M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3270 | 63.6M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3271 | 63.6M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3272 | 63.6M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3273 | 63.6M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3274 | 63.6M | )) { |
3275 | 61.8M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3276 | 467k | count = 0; |
3277 | 467k | GROW; |
3278 | 467k | if (ctxt->instate == XML_PARSER_EOF) |
3279 | 0 | return(NULL); |
3280 | 467k | } |
3281 | 61.8M | if (len <= INT_MAX - l) |
3282 | 61.8M | len += l; |
3283 | 61.8M | NEXTL(l); |
3284 | 61.8M | c = CUR_CHAR(l); |
3285 | 61.8M | } |
3286 | 47.5M | } else { |
3287 | 47.5M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3288 | 47.5M | (!IS_LETTER(c) && (c != '_') && |
3289 | 46.7M | (c != ':'))) { |
3290 | 45.4M | return(NULL); |
3291 | 45.4M | } |
3292 | 2.08M | len += l; |
3293 | 2.08M | NEXTL(l); |
3294 | 2.08M | c = CUR_CHAR(l); |
3295 | | |
3296 | 58.8M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3297 | 58.8M | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3298 | 58.0M | (c == '.') || (c == '-') || |
3299 | 58.0M | (c == '_') || (c == ':') || |
3300 | 58.0M | (IS_COMBINING(c)) || |
3301 | 58.0M | (IS_EXTENDER(c)))) { |
3302 | 56.7M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3303 | 412k | count = 0; |
3304 | 412k | GROW; |
3305 | 412k | if (ctxt->instate == XML_PARSER_EOF) |
3306 | 0 | return(NULL); |
3307 | 412k | } |
3308 | 56.7M | if (len <= INT_MAX - l) |
3309 | 56.7M | len += l; |
3310 | 56.7M | NEXTL(l); |
3311 | 56.7M | c = CUR_CHAR(l); |
3312 | 56.7M | } |
3313 | 2.08M | } |
3314 | 4.96M | if (len > maxLength) { |
3315 | 26 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3316 | 26 | return(NULL); |
3317 | 26 | } |
3318 | 4.96M | if (ctxt->input->cur - ctxt->input->base < len) { |
3319 | | /* |
3320 | | * There were a couple of bugs where PERefs lead to to a change |
3321 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3322 | | * pointer to xmlDictLookup. |
3323 | | */ |
3324 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3325 | 0 | "unexpected change of input buffer"); |
3326 | 0 | return (NULL); |
3327 | 0 | } |
3328 | 4.96M | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3329 | 14.7k | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len)); |
3330 | 4.94M | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len)); |
3331 | 4.96M | } |
3332 | | |
3333 | | /** |
3334 | | * xmlParseName: |
3335 | | * @ctxt: an XML parser context |
3336 | | * |
3337 | | * DEPRECATED: Internal function, don't use. |
3338 | | * |
3339 | | * parse an XML name. |
3340 | | * |
3341 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3342 | | * CombiningChar | Extender |
3343 | | * |
3344 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3345 | | * |
3346 | | * [6] Names ::= Name (#x20 Name)* |
3347 | | * |
3348 | | * Returns the Name parsed or NULL |
3349 | | */ |
3350 | | |
3351 | | const xmlChar * |
3352 | 311M | xmlParseName(xmlParserCtxtPtr ctxt) { |
3353 | 311M | const xmlChar *in; |
3354 | 311M | const xmlChar *ret; |
3355 | 311M | size_t count = 0; |
3356 | 311M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3357 | 240M | XML_MAX_TEXT_LENGTH : |
3358 | 311M | XML_MAX_NAME_LENGTH; |
3359 | | |
3360 | 311M | GROW; |
3361 | | |
3362 | | #ifdef DEBUG |
3363 | | nbParseName++; |
3364 | | #endif |
3365 | | |
3366 | | /* |
3367 | | * Accelerator for simple ASCII names |
3368 | | */ |
3369 | 311M | in = ctxt->input->cur; |
3370 | 311M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3371 | 311M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3372 | 311M | (*in == '_') || (*in == ':')) { |
3373 | 210M | in++; |
3374 | 1.12G | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3375 | 1.12G | ((*in >= 0x41) && (*in <= 0x5A)) || |
3376 | 1.12G | ((*in >= 0x30) && (*in <= 0x39)) || |
3377 | 1.12G | (*in == '_') || (*in == '-') || |
3378 | 1.12G | (*in == ':') || (*in == '.')) |
3379 | 910M | in++; |
3380 | 210M | if ((*in > 0) && (*in < 0x80)) { |
3381 | 206M | count = in - ctxt->input->cur; |
3382 | 206M | if (count > maxLength) { |
3383 | 14 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3384 | 14 | return(NULL); |
3385 | 14 | } |
3386 | 206M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3387 | 206M | ctxt->input->cur = in; |
3388 | 206M | ctxt->input->col += count; |
3389 | 206M | if (ret == NULL) |
3390 | 0 | xmlErrMemory(ctxt, NULL); |
3391 | 206M | return(ret); |
3392 | 206M | } |
3393 | 210M | } |
3394 | | /* accelerator for special cases */ |
3395 | 105M | return(xmlParseNameComplex(ctxt)); |
3396 | 311M | } |
3397 | | |
3398 | | static const xmlChar * |
3399 | 6.18M | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3400 | 6.18M | int len = 0, l; |
3401 | 6.18M | int c; |
3402 | 6.18M | int count = 0; |
3403 | 6.18M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3404 | 2.31M | XML_MAX_TEXT_LENGTH : |
3405 | 6.18M | XML_MAX_NAME_LENGTH; |
3406 | 6.18M | size_t startPosition = 0; |
3407 | | |
3408 | | #ifdef DEBUG |
3409 | | nbParseNCNameComplex++; |
3410 | | #endif |
3411 | | |
3412 | | /* |
3413 | | * Handler for more complex cases |
3414 | | */ |
3415 | 6.18M | GROW; |
3416 | 6.18M | startPosition = CUR_PTR - BASE_PTR; |
3417 | 6.18M | c = CUR_CHAR(l); |
3418 | 6.18M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3419 | 6.18M | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3420 | 5.94M | return(NULL); |
3421 | 5.94M | } |
3422 | | |
3423 | 10.9M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3424 | 10.9M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3425 | 10.6M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3426 | 89.0k | count = 0; |
3427 | 89.0k | GROW; |
3428 | 89.0k | if (ctxt->instate == XML_PARSER_EOF) |
3429 | 0 | return(NULL); |
3430 | 89.0k | } |
3431 | 10.6M | if (len <= INT_MAX - l) |
3432 | 10.6M | len += l; |
3433 | 10.6M | NEXTL(l); |
3434 | 10.6M | c = CUR_CHAR(l); |
3435 | 10.6M | if (c == 0) { |
3436 | 15.2k | count = 0; |
3437 | | /* |
3438 | | * when shrinking to extend the buffer we really need to preserve |
3439 | | * the part of the name we already parsed. Hence rolling back |
3440 | | * by current length. |
3441 | | */ |
3442 | 15.2k | ctxt->input->cur -= l; |
3443 | 15.2k | GROW; |
3444 | 15.2k | if (ctxt->instate == XML_PARSER_EOF) |
3445 | 0 | return(NULL); |
3446 | 15.2k | ctxt->input->cur += l; |
3447 | 15.2k | c = CUR_CHAR(l); |
3448 | 15.2k | } |
3449 | 10.6M | } |
3450 | 241k | if (len > maxLength) { |
3451 | 27 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3452 | 27 | return(NULL); |
3453 | 27 | } |
3454 | 241k | return(xmlDictLookup(ctxt->dict, (BASE_PTR + startPosition), len)); |
3455 | 241k | } |
3456 | | |
3457 | | /** |
3458 | | * xmlParseNCName: |
3459 | | * @ctxt: an XML parser context |
3460 | | * @len: length of the string parsed |
3461 | | * |
3462 | | * parse an XML name. |
3463 | | * |
3464 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3465 | | * CombiningChar | Extender |
3466 | | * |
3467 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3468 | | * |
3469 | | * Returns the Name parsed or NULL |
3470 | | */ |
3471 | | |
3472 | | static const xmlChar * |
3473 | 58.4M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3474 | 58.4M | const xmlChar *in, *e; |
3475 | 58.4M | const xmlChar *ret; |
3476 | 58.4M | size_t count = 0; |
3477 | 58.4M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3478 | 5.84M | XML_MAX_TEXT_LENGTH : |
3479 | 58.4M | XML_MAX_NAME_LENGTH; |
3480 | | |
3481 | | #ifdef DEBUG |
3482 | | nbParseNCName++; |
3483 | | #endif |
3484 | | |
3485 | | /* |
3486 | | * Accelerator for simple ASCII names |
3487 | | */ |
3488 | 58.4M | in = ctxt->input->cur; |
3489 | 58.4M | e = ctxt->input->end; |
3490 | 58.4M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3491 | 58.4M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3492 | 58.4M | (*in == '_')) && (in < e)) { |
3493 | 52.4M | in++; |
3494 | 160M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3495 | 160M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3496 | 160M | ((*in >= 0x30) && (*in <= 0x39)) || |
3497 | 160M | (*in == '_') || (*in == '-') || |
3498 | 160M | (*in == '.')) && (in < e)) |
3499 | 107M | in++; |
3500 | 52.4M | if (in >= e) |
3501 | 2.77k | goto complex; |
3502 | 52.4M | if ((*in > 0) && (*in < 0x80)) { |
3503 | 52.2M | count = in - ctxt->input->cur; |
3504 | 52.2M | if (count > maxLength) { |
3505 | 20 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3506 | 20 | return(NULL); |
3507 | 20 | } |
3508 | 52.2M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3509 | 52.2M | ctxt->input->cur = in; |
3510 | 52.2M | ctxt->input->col += count; |
3511 | 52.2M | if (ret == NULL) { |
3512 | 0 | xmlErrMemory(ctxt, NULL); |
3513 | 0 | } |
3514 | 52.2M | return(ret); |
3515 | 52.2M | } |
3516 | 52.4M | } |
3517 | 6.18M | complex: |
3518 | 6.18M | return(xmlParseNCNameComplex(ctxt)); |
3519 | 58.4M | } |
3520 | | |
3521 | | /** |
3522 | | * xmlParseNameAndCompare: |
3523 | | * @ctxt: an XML parser context |
3524 | | * |
3525 | | * parse an XML name and compares for match |
3526 | | * (specialized for endtag parsing) |
3527 | | * |
3528 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3529 | | * and the name for mismatch |
3530 | | */ |
3531 | | |
3532 | | static const xmlChar * |
3533 | 37.6M | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3534 | 37.6M | register const xmlChar *cmp = other; |
3535 | 37.6M | register const xmlChar *in; |
3536 | 37.6M | const xmlChar *ret; |
3537 | | |
3538 | 37.6M | GROW; |
3539 | 37.6M | if (ctxt->instate == XML_PARSER_EOF) |
3540 | 0 | return(NULL); |
3541 | | |
3542 | 37.6M | in = ctxt->input->cur; |
3543 | 198M | while (*in != 0 && *in == *cmp) { |
3544 | 160M | ++in; |
3545 | 160M | ++cmp; |
3546 | 160M | } |
3547 | 37.6M | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3548 | | /* success */ |
3549 | 28.6M | ctxt->input->col += in - ctxt->input->cur; |
3550 | 28.6M | ctxt->input->cur = in; |
3551 | 28.6M | return (const xmlChar*) 1; |
3552 | 28.6M | } |
3553 | | /* failure (or end of input buffer), check with full function */ |
3554 | 8.96M | ret = xmlParseName (ctxt); |
3555 | | /* strings coming from the dictionary direct compare possible */ |
3556 | 8.96M | if (ret == other) { |
3557 | 154k | return (const xmlChar*) 1; |
3558 | 154k | } |
3559 | 8.81M | return ret; |
3560 | 8.96M | } |
3561 | | |
3562 | | /** |
3563 | | * xmlParseStringName: |
3564 | | * @ctxt: an XML parser context |
3565 | | * @str: a pointer to the string pointer (IN/OUT) |
3566 | | * |
3567 | | * parse an XML name. |
3568 | | * |
3569 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3570 | | * CombiningChar | Extender |
3571 | | * |
3572 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3573 | | * |
3574 | | * [6] Names ::= Name (#x20 Name)* |
3575 | | * |
3576 | | * Returns the Name parsed or NULL. The @str pointer |
3577 | | * is updated to the current location in the string. |
3578 | | */ |
3579 | | |
3580 | | static xmlChar * |
3581 | 4.04M | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3582 | 4.04M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3583 | 4.04M | const xmlChar *cur = *str; |
3584 | 4.04M | int len = 0, l; |
3585 | 4.04M | int c; |
3586 | 4.04M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3587 | 1.81M | XML_MAX_TEXT_LENGTH : |
3588 | 4.04M | XML_MAX_NAME_LENGTH; |
3589 | | |
3590 | | #ifdef DEBUG |
3591 | | nbParseStringName++; |
3592 | | #endif |
3593 | | |
3594 | 4.04M | c = CUR_SCHAR(cur, l); |
3595 | 4.04M | if (!xmlIsNameStartChar(ctxt, c)) { |
3596 | 480k | return(NULL); |
3597 | 480k | } |
3598 | | |
3599 | 3.56M | COPY_BUF(l,buf,len,c); |
3600 | 3.56M | cur += l; |
3601 | 3.56M | c = CUR_SCHAR(cur, l); |
3602 | 32.0M | while (xmlIsNameChar(ctxt, c)) { |
3603 | 28.5M | COPY_BUF(l,buf,len,c); |
3604 | 28.5M | cur += l; |
3605 | 28.5M | c = CUR_SCHAR(cur, l); |
3606 | 28.5M | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3607 | | /* |
3608 | | * Okay someone managed to make a huge name, so he's ready to pay |
3609 | | * for the processing speed. |
3610 | | */ |
3611 | 5.50k | xmlChar *buffer; |
3612 | 5.50k | int max = len * 2; |
3613 | | |
3614 | 5.50k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3615 | 5.50k | if (buffer == NULL) { |
3616 | 0 | xmlErrMemory(ctxt, NULL); |
3617 | 0 | return(NULL); |
3618 | 0 | } |
3619 | 5.50k | memcpy(buffer, buf, len); |
3620 | 5.95M | while (xmlIsNameChar(ctxt, c)) { |
3621 | 5.94M | if (len + 10 > max) { |
3622 | 9.75k | xmlChar *tmp; |
3623 | | |
3624 | 9.75k | max *= 2; |
3625 | 9.75k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3626 | 9.75k | if (tmp == NULL) { |
3627 | 0 | xmlErrMemory(ctxt, NULL); |
3628 | 0 | xmlFree(buffer); |
3629 | 0 | return(NULL); |
3630 | 0 | } |
3631 | 9.75k | buffer = tmp; |
3632 | 9.75k | } |
3633 | 5.94M | COPY_BUF(l,buffer,len,c); |
3634 | 5.94M | cur += l; |
3635 | 5.94M | c = CUR_SCHAR(cur, l); |
3636 | 5.94M | if (len > maxLength) { |
3637 | 4 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3638 | 4 | xmlFree(buffer); |
3639 | 4 | return(NULL); |
3640 | 4 | } |
3641 | 5.94M | } |
3642 | 5.50k | buffer[len] = 0; |
3643 | 5.50k | *str = cur; |
3644 | 5.50k | return(buffer); |
3645 | 5.50k | } |
3646 | 28.5M | } |
3647 | 3.55M | if (len > maxLength) { |
3648 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3649 | 0 | return(NULL); |
3650 | 0 | } |
3651 | 3.55M | *str = cur; |
3652 | 3.55M | return(xmlStrndup(buf, len)); |
3653 | 3.55M | } |
3654 | | |
3655 | | /** |
3656 | | * xmlParseNmtoken: |
3657 | | * @ctxt: an XML parser context |
3658 | | * |
3659 | | * DEPRECATED: Internal function, don't use. |
3660 | | * |
3661 | | * parse an XML Nmtoken. |
3662 | | * |
3663 | | * [7] Nmtoken ::= (NameChar)+ |
3664 | | * |
3665 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3666 | | * |
3667 | | * Returns the Nmtoken parsed or NULL |
3668 | | */ |
3669 | | |
3670 | | xmlChar * |
3671 | 800k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3672 | 800k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3673 | 800k | int len = 0, l; |
3674 | 800k | int c; |
3675 | 800k | int count = 0; |
3676 | 800k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3677 | 322k | XML_MAX_TEXT_LENGTH : |
3678 | 800k | XML_MAX_NAME_LENGTH; |
3679 | | |
3680 | | #ifdef DEBUG |
3681 | | nbParseNmToken++; |
3682 | | #endif |
3683 | | |
3684 | 800k | GROW; |
3685 | 800k | if (ctxt->instate == XML_PARSER_EOF) |
3686 | 0 | return(NULL); |
3687 | 800k | c = CUR_CHAR(l); |
3688 | | |
3689 | 5.41M | while (xmlIsNameChar(ctxt, c)) { |
3690 | 4.61M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3691 | 0 | count = 0; |
3692 | 0 | GROW; |
3693 | 0 | } |
3694 | 4.61M | COPY_BUF(l,buf,len,c); |
3695 | 4.61M | NEXTL(l); |
3696 | 4.61M | c = CUR_CHAR(l); |
3697 | 4.61M | if (c == 0) { |
3698 | 856 | count = 0; |
3699 | 856 | GROW; |
3700 | 856 | if (ctxt->instate == XML_PARSER_EOF) |
3701 | 0 | return(NULL); |
3702 | 856 | c = CUR_CHAR(l); |
3703 | 856 | } |
3704 | 4.61M | if (len >= XML_MAX_NAMELEN) { |
3705 | | /* |
3706 | | * Okay someone managed to make a huge token, so he's ready to pay |
3707 | | * for the processing speed. |
3708 | | */ |
3709 | 4.19k | xmlChar *buffer; |
3710 | 4.19k | int max = len * 2; |
3711 | | |
3712 | 4.19k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3713 | 4.19k | if (buffer == NULL) { |
3714 | 0 | xmlErrMemory(ctxt, NULL); |
3715 | 0 | return(NULL); |
3716 | 0 | } |
3717 | 4.19k | memcpy(buffer, buf, len); |
3718 | 16.7M | while (xmlIsNameChar(ctxt, c)) { |
3719 | 16.7M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3720 | 166k | count = 0; |
3721 | 166k | GROW; |
3722 | 166k | if (ctxt->instate == XML_PARSER_EOF) { |
3723 | 0 | xmlFree(buffer); |
3724 | 0 | return(NULL); |
3725 | 0 | } |
3726 | 166k | } |
3727 | 16.7M | if (len + 10 > max) { |
3728 | 10.9k | xmlChar *tmp; |
3729 | | |
3730 | 10.9k | max *= 2; |
3731 | 10.9k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3732 | 10.9k | if (tmp == NULL) { |
3733 | 0 | xmlErrMemory(ctxt, NULL); |
3734 | 0 | xmlFree(buffer); |
3735 | 0 | return(NULL); |
3736 | 0 | } |
3737 | 10.9k | buffer = tmp; |
3738 | 10.9k | } |
3739 | 16.7M | COPY_BUF(l,buffer,len,c); |
3740 | 16.7M | NEXTL(l); |
3741 | 16.7M | c = CUR_CHAR(l); |
3742 | 16.7M | if (len > maxLength) { |
3743 | 17 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3744 | 17 | xmlFree(buffer); |
3745 | 17 | return(NULL); |
3746 | 17 | } |
3747 | 16.7M | } |
3748 | 4.18k | buffer[len] = 0; |
3749 | 4.18k | return(buffer); |
3750 | 4.19k | } |
3751 | 4.61M | } |
3752 | 795k | if (len == 0) |
3753 | 39.5k | return(NULL); |
3754 | 756k | if (len > maxLength) { |
3755 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3756 | 0 | return(NULL); |
3757 | 0 | } |
3758 | 756k | return(xmlStrndup(buf, len)); |
3759 | 756k | } |
3760 | | |
3761 | | /** |
3762 | | * xmlParseEntityValue: |
3763 | | * @ctxt: an XML parser context |
3764 | | * @orig: if non-NULL store a copy of the original entity value |
3765 | | * |
3766 | | * DEPRECATED: Internal function, don't use. |
3767 | | * |
3768 | | * parse a value for ENTITY declarations |
3769 | | * |
3770 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3771 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3772 | | * |
3773 | | * Returns the EntityValue parsed with reference substituted or NULL |
3774 | | */ |
3775 | | |
3776 | | xmlChar * |
3777 | 2.10M | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3778 | 2.10M | xmlChar *buf = NULL; |
3779 | 2.10M | int len = 0; |
3780 | 2.10M | int size = XML_PARSER_BUFFER_SIZE; |
3781 | 2.10M | int c, l; |
3782 | 2.10M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3783 | 643k | XML_MAX_HUGE_LENGTH : |
3784 | 2.10M | XML_MAX_TEXT_LENGTH; |
3785 | 2.10M | xmlChar stop; |
3786 | 2.10M | xmlChar *ret = NULL; |
3787 | 2.10M | const xmlChar *cur = NULL; |
3788 | 2.10M | xmlParserInputPtr input; |
3789 | | |
3790 | 2.10M | if (RAW == '"') stop = '"'; |
3791 | 535k | else if (RAW == '\'') stop = '\''; |
3792 | 0 | else { |
3793 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_STARTED, NULL); |
3794 | 0 | return(NULL); |
3795 | 0 | } |
3796 | 2.10M | buf = (xmlChar *) xmlMallocAtomic(size); |
3797 | 2.10M | if (buf == NULL) { |
3798 | 0 | xmlErrMemory(ctxt, NULL); |
3799 | 0 | return(NULL); |
3800 | 0 | } |
3801 | | |
3802 | | /* |
3803 | | * The content of the entity definition is copied in a buffer. |
3804 | | */ |
3805 | | |
3806 | 2.10M | ctxt->instate = XML_PARSER_ENTITY_VALUE; |
3807 | 2.10M | input = ctxt->input; |
3808 | 2.10M | GROW; |
3809 | 2.10M | if (ctxt->instate == XML_PARSER_EOF) |
3810 | 0 | goto error; |
3811 | 2.10M | NEXT; |
3812 | 2.10M | c = CUR_CHAR(l); |
3813 | | /* |
3814 | | * NOTE: 4.4.5 Included in Literal |
3815 | | * When a parameter entity reference appears in a literal entity |
3816 | | * value, ... a single or double quote character in the replacement |
3817 | | * text is always treated as a normal data character and will not |
3818 | | * terminate the literal. |
3819 | | * In practice it means we stop the loop only when back at parsing |
3820 | | * the initial entity and the quote is found |
3821 | | */ |
3822 | 120M | while (((IS_CHAR(c)) && ((c != stop) || /* checked */ |
3823 | 120M | (ctxt->input != input))) && (ctxt->instate != XML_PARSER_EOF)) { |
3824 | 117M | if (len + 5 >= size) { |
3825 | 434k | xmlChar *tmp; |
3826 | | |
3827 | 434k | size *= 2; |
3828 | 434k | tmp = (xmlChar *) xmlRealloc(buf, size); |
3829 | 434k | if (tmp == NULL) { |
3830 | 0 | xmlErrMemory(ctxt, NULL); |
3831 | 0 | goto error; |
3832 | 0 | } |
3833 | 434k | buf = tmp; |
3834 | 434k | } |
3835 | 117M | COPY_BUF(l,buf,len,c); |
3836 | 117M | NEXTL(l); |
3837 | | |
3838 | 117M | GROW; |
3839 | 117M | c = CUR_CHAR(l); |
3840 | 117M | if (c == 0) { |
3841 | 2.00k | GROW; |
3842 | 2.00k | c = CUR_CHAR(l); |
3843 | 2.00k | } |
3844 | | |
3845 | 117M | if (len > maxLength) { |
3846 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
3847 | 0 | "entity value too long\n"); |
3848 | 0 | goto error; |
3849 | 0 | } |
3850 | 117M | } |
3851 | 2.10M | buf[len] = 0; |
3852 | 2.10M | if (ctxt->instate == XML_PARSER_EOF) |
3853 | 0 | goto error; |
3854 | 2.10M | if (c != stop) { |
3855 | 3.44k | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3856 | 3.44k | goto error; |
3857 | 3.44k | } |
3858 | 2.10M | NEXT; |
3859 | | |
3860 | | /* |
3861 | | * Raise problem w.r.t. '&' and '%' being used in non-entities |
3862 | | * reference constructs. Note Charref will be handled in |
3863 | | * xmlStringDecodeEntities() |
3864 | | */ |
3865 | 2.10M | cur = buf; |
3866 | 95.2M | while (*cur != 0) { /* non input consuming */ |
3867 | 93.1M | if ((*cur == '%') || ((*cur == '&') && (cur[1] != '#'))) { |
3868 | 1.20M | xmlChar *name; |
3869 | 1.20M | xmlChar tmp = *cur; |
3870 | 1.20M | int nameOk = 0; |
3871 | | |
3872 | 1.20M | cur++; |
3873 | 1.20M | name = xmlParseStringName(ctxt, &cur); |
3874 | 1.20M | if (name != NULL) { |
3875 | 1.19M | nameOk = 1; |
3876 | 1.19M | xmlFree(name); |
3877 | 1.19M | } |
3878 | 1.20M | if ((nameOk == 0) || (*cur != ';')) { |
3879 | 18.7k | xmlFatalErrMsgInt(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3880 | 18.7k | "EntityValue: '%c' forbidden except for entities references\n", |
3881 | 18.7k | tmp); |
3882 | 18.7k | goto error; |
3883 | 18.7k | } |
3884 | 1.18M | if ((tmp == '%') && (ctxt->inSubset == 1) && |
3885 | 1.18M | (ctxt->inputNr == 1)) { |
3886 | 1.60k | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3887 | 1.60k | goto error; |
3888 | 1.60k | } |
3889 | 1.18M | if (*cur == 0) |
3890 | 0 | break; |
3891 | 1.18M | } |
3892 | 93.1M | cur++; |
3893 | 93.1M | } |
3894 | | |
3895 | | /* |
3896 | | * Then PEReference entities are substituted. |
3897 | | * |
3898 | | * NOTE: 4.4.7 Bypassed |
3899 | | * When a general entity reference appears in the EntityValue in |
3900 | | * an entity declaration, it is bypassed and left as is. |
3901 | | * so XML_SUBSTITUTE_REF is not set here. |
3902 | | */ |
3903 | 2.08M | ++ctxt->depth; |
3904 | 2.08M | ret = xmlStringDecodeEntities(ctxt, buf, XML_SUBSTITUTE_PEREF, |
3905 | 2.08M | 0, 0, 0); |
3906 | 2.08M | --ctxt->depth; |
3907 | 2.08M | if (orig != NULL) { |
3908 | 2.08M | *orig = buf; |
3909 | 2.08M | buf = NULL; |
3910 | 2.08M | } |
3911 | | |
3912 | 2.10M | error: |
3913 | 2.10M | if (buf != NULL) |
3914 | 23.8k | xmlFree(buf); |
3915 | 2.10M | return(ret); |
3916 | 2.08M | } |
3917 | | |
3918 | | /** |
3919 | | * xmlParseAttValueComplex: |
3920 | | * @ctxt: an XML parser context |
3921 | | * @len: the resulting attribute len |
3922 | | * @normalize: whether to apply the inner normalization |
3923 | | * |
3924 | | * parse a value for an attribute, this is the fallback function |
3925 | | * of xmlParseAttValue() when the attribute parsing requires handling |
3926 | | * of non-ASCII characters, or normalization compaction. |
3927 | | * |
3928 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
3929 | | */ |
3930 | | static xmlChar * |
3931 | 6.92M | xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) { |
3932 | 6.92M | xmlChar limit = 0; |
3933 | 6.92M | xmlChar *buf = NULL; |
3934 | 6.92M | xmlChar *rep = NULL; |
3935 | 6.92M | size_t len = 0; |
3936 | 6.92M | size_t buf_size = 0; |
3937 | 6.92M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3938 | 5.12M | XML_MAX_HUGE_LENGTH : |
3939 | 6.92M | XML_MAX_TEXT_LENGTH; |
3940 | 6.92M | int c, l, in_space = 0; |
3941 | 6.92M | xmlChar *current = NULL; |
3942 | 6.92M | xmlEntityPtr ent; |
3943 | | |
3944 | 6.92M | if (NXT(0) == '"') { |
3945 | 3.85M | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3946 | 3.85M | limit = '"'; |
3947 | 3.85M | NEXT; |
3948 | 3.85M | } else if (NXT(0) == '\'') { |
3949 | 3.06M | limit = '\''; |
3950 | 3.06M | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3951 | 3.06M | NEXT; |
3952 | 3.06M | } else { |
3953 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3954 | 0 | return(NULL); |
3955 | 0 | } |
3956 | | |
3957 | | /* |
3958 | | * allocate a translation buffer. |
3959 | | */ |
3960 | 6.92M | buf_size = XML_PARSER_BUFFER_SIZE; |
3961 | 6.92M | buf = (xmlChar *) xmlMallocAtomic(buf_size); |
3962 | 6.92M | if (buf == NULL) goto mem_error; |
3963 | | |
3964 | | /* |
3965 | | * OK loop until we reach one of the ending char or a size limit. |
3966 | | */ |
3967 | 6.92M | c = CUR_CHAR(l); |
3968 | 362M | while (((NXT(0) != limit) && /* checked */ |
3969 | 362M | (IS_CHAR(c)) && (c != '<')) && |
3970 | 362M | (ctxt->instate != XML_PARSER_EOF)) { |
3971 | 355M | if (c == '&') { |
3972 | 8.50M | in_space = 0; |
3973 | 8.50M | if (NXT(1) == '#') { |
3974 | 2.13M | int val = xmlParseCharRef(ctxt); |
3975 | | |
3976 | 2.13M | if (val == '&') { |
3977 | 170k | if (ctxt->replaceEntities) { |
3978 | 18.0k | if (len + 10 > buf_size) { |
3979 | 360 | growBuffer(buf, 10); |
3980 | 360 | } |
3981 | 18.0k | buf[len++] = '&'; |
3982 | 152k | } else { |
3983 | | /* |
3984 | | * The reparsing will be done in xmlStringGetNodeList() |
3985 | | * called by the attribute() function in SAX.c |
3986 | | */ |
3987 | 152k | if (len + 10 > buf_size) { |
3988 | 612 | growBuffer(buf, 10); |
3989 | 612 | } |
3990 | 152k | buf[len++] = '&'; |
3991 | 152k | buf[len++] = '#'; |
3992 | 152k | buf[len++] = '3'; |
3993 | 152k | buf[len++] = '8'; |
3994 | 152k | buf[len++] = ';'; |
3995 | 152k | } |
3996 | 1.96M | } else if (val != 0) { |
3997 | 1.68M | if (len + 10 > buf_size) { |
3998 | 7.96k | growBuffer(buf, 10); |
3999 | 7.96k | } |
4000 | 1.68M | len += xmlCopyChar(0, &buf[len], val); |
4001 | 1.68M | } |
4002 | 6.37M | } else { |
4003 | 6.37M | ent = xmlParseEntityRef(ctxt); |
4004 | 6.37M | ctxt->nbentities++; |
4005 | 6.37M | if (ent != NULL) |
4006 | 2.18M | ctxt->nbentities += ent->owner; |
4007 | 6.37M | if ((ent != NULL) && |
4008 | 6.37M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
4009 | 1.91M | if (len + 10 > buf_size) { |
4010 | 312 | growBuffer(buf, 10); |
4011 | 312 | } |
4012 | 1.91M | if ((ctxt->replaceEntities == 0) && |
4013 | 1.91M | (ent->content[0] == '&')) { |
4014 | 937k | buf[len++] = '&'; |
4015 | 937k | buf[len++] = '#'; |
4016 | 937k | buf[len++] = '3'; |
4017 | 937k | buf[len++] = '8'; |
4018 | 937k | buf[len++] = ';'; |
4019 | 982k | } else { |
4020 | 982k | buf[len++] = ent->content[0]; |
4021 | 982k | } |
4022 | 4.45M | } else if ((ent != NULL) && |
4023 | 4.45M | (ctxt->replaceEntities != 0)) { |
4024 | 72.9k | if (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) { |
4025 | 72.9k | ++ctxt->depth; |
4026 | 72.9k | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4027 | 72.9k | XML_SUBSTITUTE_REF, |
4028 | 72.9k | 0, 0, 0); |
4029 | 72.9k | --ctxt->depth; |
4030 | 72.9k | if (rep != NULL) { |
4031 | 69.3k | current = rep; |
4032 | 3.00M | while (*current != 0) { /* non input consuming */ |
4033 | 2.93M | if ((*current == 0xD) || (*current == 0xA) || |
4034 | 2.93M | (*current == 0x9)) { |
4035 | 67.5k | buf[len++] = 0x20; |
4036 | 67.5k | current++; |
4037 | 67.5k | } else |
4038 | 2.86M | buf[len++] = *current++; |
4039 | 2.93M | if (len + 10 > buf_size) { |
4040 | 6.84k | growBuffer(buf, 10); |
4041 | 6.84k | } |
4042 | 2.93M | } |
4043 | 69.3k | xmlFree(rep); |
4044 | 69.3k | rep = NULL; |
4045 | 69.3k | } |
4046 | 72.9k | } else { |
4047 | 0 | if (len + 10 > buf_size) { |
4048 | 0 | growBuffer(buf, 10); |
4049 | 0 | } |
4050 | 0 | if (ent->content != NULL) |
4051 | 0 | buf[len++] = ent->content[0]; |
4052 | 0 | } |
4053 | 4.38M | } else if (ent != NULL) { |
4054 | 191k | int i = xmlStrlen(ent->name); |
4055 | 191k | const xmlChar *cur = ent->name; |
4056 | | |
4057 | | /* |
4058 | | * This may look absurd but is needed to detect |
4059 | | * entities problems |
4060 | | */ |
4061 | 191k | if ((ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
4062 | 191k | (ent->content != NULL) && (ent->checked == 0)) { |
4063 | 23.1k | unsigned long oldnbent = ctxt->nbentities, diff; |
4064 | | |
4065 | 23.1k | ++ctxt->depth; |
4066 | 23.1k | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4067 | 23.1k | XML_SUBSTITUTE_REF, 0, 0, 0); |
4068 | 23.1k | --ctxt->depth; |
4069 | | |
4070 | 23.1k | diff = ctxt->nbentities - oldnbent + 1; |
4071 | 23.1k | if (diff > INT_MAX / 2) |
4072 | 0 | diff = INT_MAX / 2; |
4073 | 23.1k | ent->checked = diff * 2; |
4074 | 23.1k | if (rep != NULL) { |
4075 | 22.7k | if (xmlStrchr(rep, '<')) |
4076 | 1.30k | ent->checked |= 1; |
4077 | 22.7k | xmlFree(rep); |
4078 | 22.7k | rep = NULL; |
4079 | 22.7k | } else { |
4080 | 316 | ent->content[0] = 0; |
4081 | 316 | } |
4082 | 23.1k | } |
4083 | | |
4084 | | /* |
4085 | | * Just output the reference |
4086 | | */ |
4087 | 191k | buf[len++] = '&'; |
4088 | 193k | while (len + i + 10 > buf_size) { |
4089 | 3.80k | growBuffer(buf, i + 10); |
4090 | 3.80k | } |
4091 | 704k | for (;i > 0;i--) |
4092 | 513k | buf[len++] = *cur++; |
4093 | 191k | buf[len++] = ';'; |
4094 | 191k | } |
4095 | 6.37M | } |
4096 | 347M | } else { |
4097 | 347M | if ((c == 0x20) || (c == 0xD) || (c == 0xA) || (c == 0x9)) { |
4098 | 35.0M | if ((len != 0) || (!normalize)) { |
4099 | 34.8M | if ((!normalize) || (!in_space)) { |
4100 | 34.2M | COPY_BUF(l,buf,len,0x20); |
4101 | 34.3M | while (len + 10 > buf_size) { |
4102 | 176k | growBuffer(buf, 10); |
4103 | 176k | } |
4104 | 34.2M | } |
4105 | 34.8M | in_space = 1; |
4106 | 34.8M | } |
4107 | 312M | } else { |
4108 | 312M | in_space = 0; |
4109 | 312M | COPY_BUF(l,buf,len,c); |
4110 | 312M | if (len + 10 > buf_size) { |
4111 | 1.04M | growBuffer(buf, 10); |
4112 | 1.04M | } |
4113 | 312M | } |
4114 | 347M | NEXTL(l); |
4115 | 347M | } |
4116 | 355M | GROW; |
4117 | 355M | c = CUR_CHAR(l); |
4118 | 355M | if (len > maxLength) { |
4119 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4120 | 0 | "AttValue length too long\n"); |
4121 | 0 | goto mem_error; |
4122 | 0 | } |
4123 | 355M | } |
4124 | 6.92M | if (ctxt->instate == XML_PARSER_EOF) |
4125 | 0 | goto error; |
4126 | | |
4127 | 6.92M | if ((in_space) && (normalize)) { |
4128 | 67.0k | while ((len > 0) && (buf[len - 1] == 0x20)) len--; |
4129 | 32.9k | } |
4130 | 6.92M | buf[len] = 0; |
4131 | 6.92M | if (RAW == '<') { |
4132 | 1.39M | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4133 | 5.52M | } else if (RAW != limit) { |
4134 | 744k | if ((c != 0) && (!IS_CHAR(c))) { |
4135 | 513k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4136 | 513k | "invalid character in attribute value\n"); |
4137 | 513k | } else { |
4138 | 231k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4139 | 231k | "AttValue: ' expected\n"); |
4140 | 231k | } |
4141 | 744k | } else |
4142 | 4.78M | NEXT; |
4143 | | |
4144 | 6.92M | if (attlen != NULL) *attlen = len; |
4145 | 6.92M | return(buf); |
4146 | | |
4147 | 0 | mem_error: |
4148 | 0 | xmlErrMemory(ctxt, NULL); |
4149 | 0 | error: |
4150 | 0 | if (buf != NULL) |
4151 | 0 | xmlFree(buf); |
4152 | 0 | if (rep != NULL) |
4153 | 0 | xmlFree(rep); |
4154 | 0 | return(NULL); |
4155 | 0 | } |
4156 | | |
4157 | | /** |
4158 | | * xmlParseAttValue: |
4159 | | * @ctxt: an XML parser context |
4160 | | * |
4161 | | * DEPRECATED: Internal function, don't use. |
4162 | | * |
4163 | | * parse a value for an attribute |
4164 | | * Note: the parser won't do substitution of entities here, this |
4165 | | * will be handled later in xmlStringGetNodeList |
4166 | | * |
4167 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4168 | | * "'" ([^<&'] | Reference)* "'" |
4169 | | * |
4170 | | * 3.3.3 Attribute-Value Normalization: |
4171 | | * Before the value of an attribute is passed to the application or |
4172 | | * checked for validity, the XML processor must normalize it as follows: |
4173 | | * - a character reference is processed by appending the referenced |
4174 | | * character to the attribute value |
4175 | | * - an entity reference is processed by recursively processing the |
4176 | | * replacement text of the entity |
4177 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4178 | | * appending #x20 to the normalized value, except that only a single |
4179 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4180 | | * parsed entity or the literal entity value of an internal parsed entity |
4181 | | * - other characters are processed by appending them to the normalized value |
4182 | | * If the declared value is not CDATA, then the XML processor must further |
4183 | | * process the normalized attribute value by discarding any leading and |
4184 | | * trailing space (#x20) characters, and by replacing sequences of space |
4185 | | * (#x20) characters by a single space (#x20) character. |
4186 | | * All attributes for which no declaration has been read should be treated |
4187 | | * by a non-validating parser as if declared CDATA. |
4188 | | * |
4189 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4190 | | */ |
4191 | | |
4192 | | |
4193 | | xmlChar * |
4194 | 72.0M | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4195 | 72.0M | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4196 | 72.0M | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0)); |
4197 | 72.0M | } |
4198 | | |
4199 | | /** |
4200 | | * xmlParseSystemLiteral: |
4201 | | * @ctxt: an XML parser context |
4202 | | * |
4203 | | * DEPRECATED: Internal function, don't use. |
4204 | | * |
4205 | | * parse an XML Literal |
4206 | | * |
4207 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4208 | | * |
4209 | | * Returns the SystemLiteral parsed or NULL |
4210 | | */ |
4211 | | |
4212 | | xmlChar * |
4213 | 387k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4214 | 387k | xmlChar *buf = NULL; |
4215 | 387k | int len = 0; |
4216 | 387k | int size = XML_PARSER_BUFFER_SIZE; |
4217 | 387k | int cur, l; |
4218 | 387k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4219 | 130k | XML_MAX_TEXT_LENGTH : |
4220 | 387k | XML_MAX_NAME_LENGTH; |
4221 | 387k | xmlChar stop; |
4222 | 387k | int state = ctxt->instate; |
4223 | 387k | int count = 0; |
4224 | | |
4225 | 387k | SHRINK; |
4226 | 387k | if (RAW == '"') { |
4227 | 316k | NEXT; |
4228 | 316k | stop = '"'; |
4229 | 316k | } else if (RAW == '\'') { |
4230 | 63.3k | NEXT; |
4231 | 63.3k | stop = '\''; |
4232 | 63.3k | } else { |
4233 | 7.53k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4234 | 7.53k | return(NULL); |
4235 | 7.53k | } |
4236 | | |
4237 | 380k | buf = (xmlChar *) xmlMallocAtomic(size); |
4238 | 380k | if (buf == NULL) { |
4239 | 0 | xmlErrMemory(ctxt, NULL); |
4240 | 0 | return(NULL); |
4241 | 0 | } |
4242 | 380k | ctxt->instate = XML_PARSER_SYSTEM_LITERAL; |
4243 | 380k | cur = CUR_CHAR(l); |
4244 | 19.3M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4245 | 18.9M | if (len + 5 >= size) { |
4246 | 43.1k | xmlChar *tmp; |
4247 | | |
4248 | 43.1k | size *= 2; |
4249 | 43.1k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4250 | 43.1k | if (tmp == NULL) { |
4251 | 0 | xmlFree(buf); |
4252 | 0 | xmlErrMemory(ctxt, NULL); |
4253 | 0 | ctxt->instate = (xmlParserInputState) state; |
4254 | 0 | return(NULL); |
4255 | 0 | } |
4256 | 43.1k | buf = tmp; |
4257 | 43.1k | } |
4258 | 18.9M | count++; |
4259 | 18.9M | if (count > 50) { |
4260 | 274k | SHRINK; |
4261 | 274k | GROW; |
4262 | 274k | count = 0; |
4263 | 274k | if (ctxt->instate == XML_PARSER_EOF) { |
4264 | 0 | xmlFree(buf); |
4265 | 0 | return(NULL); |
4266 | 0 | } |
4267 | 274k | } |
4268 | 18.9M | COPY_BUF(l,buf,len,cur); |
4269 | 18.9M | NEXTL(l); |
4270 | 18.9M | cur = CUR_CHAR(l); |
4271 | 18.9M | if (cur == 0) { |
4272 | 5.57k | GROW; |
4273 | 5.57k | SHRINK; |
4274 | 5.57k | cur = CUR_CHAR(l); |
4275 | 5.57k | } |
4276 | 18.9M | if (len > maxLength) { |
4277 | 45 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4278 | 45 | xmlFree(buf); |
4279 | 45 | ctxt->instate = (xmlParserInputState) state; |
4280 | 45 | return(NULL); |
4281 | 45 | } |
4282 | 18.9M | } |
4283 | 380k | buf[len] = 0; |
4284 | 380k | ctxt->instate = (xmlParserInputState) state; |
4285 | 380k | if (!IS_CHAR(cur)) { |
4286 | 7.38k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4287 | 372k | } else { |
4288 | 372k | NEXT; |
4289 | 372k | } |
4290 | 380k | return(buf); |
4291 | 380k | } |
4292 | | |
4293 | | /** |
4294 | | * xmlParsePubidLiteral: |
4295 | | * @ctxt: an XML parser context |
4296 | | * |
4297 | | * DEPRECATED: Internal function, don't use. |
4298 | | * |
4299 | | * parse an XML public literal |
4300 | | * |
4301 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4302 | | * |
4303 | | * Returns the PubidLiteral parsed or NULL. |
4304 | | */ |
4305 | | |
4306 | | xmlChar * |
4307 | 134k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4308 | 134k | xmlChar *buf = NULL; |
4309 | 134k | int len = 0; |
4310 | 134k | int size = XML_PARSER_BUFFER_SIZE; |
4311 | 134k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4312 | 48.0k | XML_MAX_TEXT_LENGTH : |
4313 | 134k | XML_MAX_NAME_LENGTH; |
4314 | 134k | xmlChar cur; |
4315 | 134k | xmlChar stop; |
4316 | 134k | int count = 0; |
4317 | 134k | xmlParserInputState oldstate = ctxt->instate; |
4318 | | |
4319 | 134k | SHRINK; |
4320 | 134k | if (RAW == '"') { |
4321 | 82.0k | NEXT; |
4322 | 82.0k | stop = '"'; |
4323 | 82.0k | } else if (RAW == '\'') { |
4324 | 51.8k | NEXT; |
4325 | 51.8k | stop = '\''; |
4326 | 51.8k | } else { |
4327 | 1.05k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4328 | 1.05k | return(NULL); |
4329 | 1.05k | } |
4330 | 133k | buf = (xmlChar *) xmlMallocAtomic(size); |
4331 | 133k | if (buf == NULL) { |
4332 | 0 | xmlErrMemory(ctxt, NULL); |
4333 | 0 | return(NULL); |
4334 | 0 | } |
4335 | 133k | ctxt->instate = XML_PARSER_PUBLIC_LITERAL; |
4336 | 133k | cur = CUR; |
4337 | 4.97M | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop)) { /* checked */ |
4338 | 4.84M | if (len + 1 >= size) { |
4339 | 5.48k | xmlChar *tmp; |
4340 | | |
4341 | 5.48k | size *= 2; |
4342 | 5.48k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4343 | 5.48k | if (tmp == NULL) { |
4344 | 0 | xmlErrMemory(ctxt, NULL); |
4345 | 0 | xmlFree(buf); |
4346 | 0 | return(NULL); |
4347 | 0 | } |
4348 | 5.48k | buf = tmp; |
4349 | 5.48k | } |
4350 | 4.84M | buf[len++] = cur; |
4351 | 4.84M | count++; |
4352 | 4.84M | if (count > 50) { |
4353 | 45.9k | SHRINK; |
4354 | 45.9k | GROW; |
4355 | 45.9k | count = 0; |
4356 | 45.9k | if (ctxt->instate == XML_PARSER_EOF) { |
4357 | 0 | xmlFree(buf); |
4358 | 0 | return(NULL); |
4359 | 0 | } |
4360 | 45.9k | } |
4361 | 4.84M | NEXT; |
4362 | 4.84M | cur = CUR; |
4363 | 4.84M | if (cur == 0) { |
4364 | 863 | GROW; |
4365 | 863 | SHRINK; |
4366 | 863 | cur = CUR; |
4367 | 863 | } |
4368 | 4.84M | if (len > maxLength) { |
4369 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4370 | 0 | xmlFree(buf); |
4371 | 0 | return(NULL); |
4372 | 0 | } |
4373 | 4.84M | } |
4374 | 133k | buf[len] = 0; |
4375 | 133k | if (cur != stop) { |
4376 | 4.15k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4377 | 129k | } else { |
4378 | 129k | NEXT; |
4379 | 129k | } |
4380 | 133k | ctxt->instate = oldstate; |
4381 | 133k | return(buf); |
4382 | 133k | } |
4383 | | |
4384 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata); |
4385 | | |
4386 | | /* |
4387 | | * used for the test in the inner loop of the char data testing |
4388 | | */ |
4389 | | static const unsigned char test_char_data[256] = { |
4390 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4391 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4392 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4393 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4394 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4395 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4396 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4397 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4398 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4399 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4400 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4401 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4402 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4403 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4404 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4405 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4406 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4407 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4408 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4409 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4410 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4411 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4412 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4413 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4414 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4415 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4416 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4417 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4418 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4419 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4420 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4421 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4422 | | }; |
4423 | | |
4424 | | /** |
4425 | | * xmlParseCharData: |
4426 | | * @ctxt: an XML parser context |
4427 | | * @cdata: int indicating whether we are within a CDATA section |
4428 | | * |
4429 | | * DEPRECATED: Internal function, don't use. |
4430 | | * |
4431 | | * parse a CharData section. |
4432 | | * if we are within a CDATA section ']]>' marks an end of section. |
4433 | | * |
4434 | | * The right angle bracket (>) may be represented using the string ">", |
4435 | | * and must, for compatibility, be escaped using ">" or a character |
4436 | | * reference when it appears in the string "]]>" in content, when that |
4437 | | * string is not marking the end of a CDATA section. |
4438 | | * |
4439 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4440 | | */ |
4441 | | |
4442 | | void |
4443 | 302M | xmlParseCharData(xmlParserCtxtPtr ctxt, int cdata) { |
4444 | 302M | const xmlChar *in; |
4445 | 302M | int nbchar = 0; |
4446 | 302M | int line = ctxt->input->line; |
4447 | 302M | int col = ctxt->input->col; |
4448 | 302M | int ccol; |
4449 | | |
4450 | 302M | SHRINK; |
4451 | 302M | GROW; |
4452 | | /* |
4453 | | * Accelerated common case where input don't need to be |
4454 | | * modified before passing it to the handler. |
4455 | | */ |
4456 | 302M | if (!cdata) { |
4457 | 302M | in = ctxt->input->cur; |
4458 | 338M | do { |
4459 | 394M | get_more_space: |
4460 | 654M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4461 | 394M | if (*in == 0xA) { |
4462 | 78.0M | do { |
4463 | 78.0M | ctxt->input->line++; ctxt->input->col = 1; |
4464 | 78.0M | in++; |
4465 | 78.0M | } while (*in == 0xA); |
4466 | 56.7M | goto get_more_space; |
4467 | 56.7M | } |
4468 | 338M | if (*in == '<') { |
4469 | 46.5M | nbchar = in - ctxt->input->cur; |
4470 | 46.5M | if (nbchar > 0) { |
4471 | 46.4M | const xmlChar *tmp = ctxt->input->cur; |
4472 | 46.4M | ctxt->input->cur = in; |
4473 | | |
4474 | 46.4M | if ((ctxt->sax != NULL) && |
4475 | 46.4M | (ctxt->sax->ignorableWhitespace != |
4476 | 46.4M | ctxt->sax->characters)) { |
4477 | 18.9M | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4478 | 4.49M | if (ctxt->sax->ignorableWhitespace != NULL) |
4479 | 4.49M | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4480 | 4.49M | tmp, nbchar); |
4481 | 14.5M | } else { |
4482 | 14.5M | if (ctxt->sax->characters != NULL) |
4483 | 14.5M | ctxt->sax->characters(ctxt->userData, |
4484 | 14.5M | tmp, nbchar); |
4485 | 14.5M | if (*ctxt->space == -1) |
4486 | 3.94M | *ctxt->space = -2; |
4487 | 14.5M | } |
4488 | 27.4M | } else if ((ctxt->sax != NULL) && |
4489 | 27.4M | (ctxt->sax->characters != NULL)) { |
4490 | 27.4M | ctxt->sax->characters(ctxt->userData, |
4491 | 27.4M | tmp, nbchar); |
4492 | 27.4M | } |
4493 | 46.4M | } |
4494 | 46.5M | return; |
4495 | 46.5M | } |
4496 | | |
4497 | 351M | get_more: |
4498 | 351M | ccol = ctxt->input->col; |
4499 | 3.71G | while (test_char_data[*in]) { |
4500 | 3.36G | in++; |
4501 | 3.36G | ccol++; |
4502 | 3.36G | } |
4503 | 351M | ctxt->input->col = ccol; |
4504 | 351M | if (*in == 0xA) { |
4505 | 105M | do { |
4506 | 105M | ctxt->input->line++; ctxt->input->col = 1; |
4507 | 105M | in++; |
4508 | 105M | } while (*in == 0xA); |
4509 | 54.7M | goto get_more; |
4510 | 54.7M | } |
4511 | 297M | if (*in == ']') { |
4512 | 6.39M | if ((in[1] == ']') && (in[2] == '>')) { |
4513 | 767k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4514 | 767k | ctxt->input->cur = in + 1; |
4515 | 767k | return; |
4516 | 767k | } |
4517 | 5.62M | in++; |
4518 | 5.62M | ctxt->input->col++; |
4519 | 5.62M | goto get_more; |
4520 | 6.39M | } |
4521 | 290M | nbchar = in - ctxt->input->cur; |
4522 | 290M | if (nbchar > 0) { |
4523 | 137M | if ((ctxt->sax != NULL) && |
4524 | 137M | (ctxt->sax->ignorableWhitespace != |
4525 | 137M | ctxt->sax->characters) && |
4526 | 137M | (IS_BLANK_CH(*ctxt->input->cur))) { |
4527 | 15.2M | const xmlChar *tmp = ctxt->input->cur; |
4528 | 15.2M | ctxt->input->cur = in; |
4529 | | |
4530 | 15.2M | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4531 | 2.47M | if (ctxt->sax->ignorableWhitespace != NULL) |
4532 | 2.47M | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4533 | 2.47M | tmp, nbchar); |
4534 | 12.7M | } else { |
4535 | 12.7M | if (ctxt->sax->characters != NULL) |
4536 | 12.7M | ctxt->sax->characters(ctxt->userData, |
4537 | 12.7M | tmp, nbchar); |
4538 | 12.7M | if (*ctxt->space == -1) |
4539 | 4.40M | *ctxt->space = -2; |
4540 | 12.7M | } |
4541 | 15.2M | line = ctxt->input->line; |
4542 | 15.2M | col = ctxt->input->col; |
4543 | 121M | } else if (ctxt->sax != NULL) { |
4544 | 121M | if (ctxt->sax->characters != NULL) |
4545 | 121M | ctxt->sax->characters(ctxt->userData, |
4546 | 121M | ctxt->input->cur, nbchar); |
4547 | 121M | line = ctxt->input->line; |
4548 | 121M | col = ctxt->input->col; |
4549 | 121M | } |
4550 | | /* something really bad happened in the SAX callback */ |
4551 | 137M | if (ctxt->instate != XML_PARSER_CONTENT) |
4552 | 0 | return; |
4553 | 137M | } |
4554 | 290M | ctxt->input->cur = in; |
4555 | 290M | if (*in == 0xD) { |
4556 | 35.4M | in++; |
4557 | 35.4M | if (*in == 0xA) { |
4558 | 35.2M | ctxt->input->cur = in; |
4559 | 35.2M | in++; |
4560 | 35.2M | ctxt->input->line++; ctxt->input->col = 1; |
4561 | 35.2M | continue; /* while */ |
4562 | 35.2M | } |
4563 | 212k | in--; |
4564 | 212k | } |
4565 | 255M | if (*in == '<') { |
4566 | 105M | return; |
4567 | 105M | } |
4568 | 149M | if (*in == '&') { |
4569 | 18.9M | return; |
4570 | 18.9M | } |
4571 | 130M | SHRINK; |
4572 | 130M | GROW; |
4573 | 130M | if (ctxt->instate == XML_PARSER_EOF) |
4574 | 0 | return; |
4575 | 130M | in = ctxt->input->cur; |
4576 | 166M | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
4577 | 131M | nbchar = 0; |
4578 | 131M | } |
4579 | 131M | ctxt->input->line = line; |
4580 | 131M | ctxt->input->col = col; |
4581 | 131M | xmlParseCharDataComplex(ctxt, cdata); |
4582 | 131M | } |
4583 | | |
4584 | | /** |
4585 | | * xmlParseCharDataComplex: |
4586 | | * @ctxt: an XML parser context |
4587 | | * @cdata: int indicating whether we are within a CDATA section |
4588 | | * |
4589 | | * parse a CharData section.this is the fallback function |
4590 | | * of xmlParseCharData() when the parsing requires handling |
4591 | | * of non-ASCII characters. |
4592 | | */ |
4593 | | static void |
4594 | 131M | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata) { |
4595 | 131M | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4596 | 131M | int nbchar = 0; |
4597 | 131M | int cur, l; |
4598 | 131M | int count = 0; |
4599 | | |
4600 | 131M | SHRINK; |
4601 | 131M | GROW; |
4602 | 131M | cur = CUR_CHAR(l); |
4603 | 1.66G | while ((cur != '<') && /* checked */ |
4604 | 1.66G | (cur != '&') && |
4605 | 1.66G | (IS_CHAR(cur))) /* test also done in xmlCurrentChar() */ { |
4606 | 1.53G | if ((cur == ']') && (NXT(1) == ']') && |
4607 | 1.53G | (NXT(2) == '>')) { |
4608 | 282k | if (cdata) break; |
4609 | 282k | else { |
4610 | 282k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4611 | 282k | } |
4612 | 282k | } |
4613 | 1.53G | COPY_BUF(l,buf,nbchar,cur); |
4614 | | /* move current position before possible calling of ctxt->sax->characters */ |
4615 | 1.53G | NEXTL(l); |
4616 | 1.53G | cur = CUR_CHAR(l); |
4617 | 1.53G | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4618 | 3.75M | buf[nbchar] = 0; |
4619 | | |
4620 | | /* |
4621 | | * OK the segment is to be consumed as chars. |
4622 | | */ |
4623 | 3.75M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4624 | 190k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4625 | 3.51k | if (ctxt->sax->ignorableWhitespace != NULL) |
4626 | 3.51k | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4627 | 3.51k | buf, nbchar); |
4628 | 186k | } else { |
4629 | 186k | if (ctxt->sax->characters != NULL) |
4630 | 186k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4631 | 186k | if ((ctxt->sax->characters != |
4632 | 186k | ctxt->sax->ignorableWhitespace) && |
4633 | 186k | (*ctxt->space == -1)) |
4634 | 6.58k | *ctxt->space = -2; |
4635 | 186k | } |
4636 | 190k | } |
4637 | 3.75M | nbchar = 0; |
4638 | | /* something really bad happened in the SAX callback */ |
4639 | 3.75M | if (ctxt->instate != XML_PARSER_CONTENT) |
4640 | 0 | return; |
4641 | 3.75M | } |
4642 | 1.53G | count++; |
4643 | 1.53G | if (count > 50) { |
4644 | 26.8M | SHRINK; |
4645 | 26.8M | GROW; |
4646 | 26.8M | count = 0; |
4647 | 26.8M | if (ctxt->instate == XML_PARSER_EOF) |
4648 | 0 | return; |
4649 | 26.8M | } |
4650 | 1.53G | } |
4651 | 131M | if (nbchar != 0) { |
4652 | 9.99M | buf[nbchar] = 0; |
4653 | | /* |
4654 | | * OK the segment is to be consumed as chars. |
4655 | | */ |
4656 | 9.99M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4657 | 1.46M | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4658 | 20.7k | if (ctxt->sax->ignorableWhitespace != NULL) |
4659 | 20.7k | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4660 | 1.44M | } else { |
4661 | 1.44M | if (ctxt->sax->characters != NULL) |
4662 | 1.44M | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4663 | 1.44M | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4664 | 1.44M | (*ctxt->space == -1)) |
4665 | 306k | *ctxt->space = -2; |
4666 | 1.44M | } |
4667 | 1.46M | } |
4668 | 9.99M | } |
4669 | 131M | if ((cur != 0) && (!IS_CHAR(cur))) { |
4670 | | /* Generate the error and skip the offending character */ |
4671 | 120M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4672 | 120M | "PCDATA invalid Char value %d\n", |
4673 | 120M | cur); |
4674 | 120M | NEXTL(l); |
4675 | 120M | } |
4676 | 131M | } |
4677 | | |
4678 | | /** |
4679 | | * xmlParseExternalID: |
4680 | | * @ctxt: an XML parser context |
4681 | | * @publicID: a xmlChar** receiving PubidLiteral |
4682 | | * @strict: indicate whether we should restrict parsing to only |
4683 | | * production [75], see NOTE below |
4684 | | * |
4685 | | * DEPRECATED: Internal function, don't use. |
4686 | | * |
4687 | | * Parse an External ID or a Public ID |
4688 | | * |
4689 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4690 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
4691 | | * |
4692 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4693 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4694 | | * |
4695 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4696 | | * |
4697 | | * Returns the function returns SystemLiteral and in the second |
4698 | | * case publicID receives PubidLiteral, is strict is off |
4699 | | * it is possible to return NULL and have publicID set. |
4700 | | */ |
4701 | | |
4702 | | xmlChar * |
4703 | 739k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
4704 | 739k | xmlChar *URI = NULL; |
4705 | | |
4706 | 739k | SHRINK; |
4707 | | |
4708 | 739k | *publicID = NULL; |
4709 | 739k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4710 | 257k | SKIP(6); |
4711 | 257k | if (SKIP_BLANKS == 0) { |
4712 | 709 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4713 | 709 | "Space required after 'SYSTEM'\n"); |
4714 | 709 | } |
4715 | 257k | URI = xmlParseSystemLiteral(ctxt); |
4716 | 257k | if (URI == NULL) { |
4717 | 1.00k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4718 | 1.00k | } |
4719 | 482k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4720 | 134k | SKIP(6); |
4721 | 134k | if (SKIP_BLANKS == 0) { |
4722 | 1.07k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4723 | 1.07k | "Space required after 'PUBLIC'\n"); |
4724 | 1.07k | } |
4725 | 134k | *publicID = xmlParsePubidLiteral(ctxt); |
4726 | 134k | if (*publicID == NULL) { |
4727 | 1.05k | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4728 | 1.05k | } |
4729 | 134k | if (strict) { |
4730 | | /* |
4731 | | * We don't handle [83] so "S SystemLiteral" is required. |
4732 | | */ |
4733 | 129k | if (SKIP_BLANKS == 0) { |
4734 | 5.94k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4735 | 5.94k | "Space required after the Public Identifier\n"); |
4736 | 5.94k | } |
4737 | 129k | } else { |
4738 | | /* |
4739 | | * We handle [83] so we return immediately, if |
4740 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4741 | | * system literal was found, but this is harmless since we must |
4742 | | * be at the end of a NotationDecl. |
4743 | | */ |
4744 | 5.14k | if (SKIP_BLANKS == 0) return(NULL); |
4745 | 572 | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4746 | 572 | } |
4747 | 130k | URI = xmlParseSystemLiteral(ctxt); |
4748 | 130k | if (URI == NULL) { |
4749 | 6.57k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4750 | 6.57k | } |
4751 | 130k | } |
4752 | 734k | return(URI); |
4753 | 739k | } |
4754 | | |
4755 | | /** |
4756 | | * xmlParseCommentComplex: |
4757 | | * @ctxt: an XML parser context |
4758 | | * @buf: the already parsed part of the buffer |
4759 | | * @len: number of bytes in the buffer |
4760 | | * @size: allocated size of the buffer |
4761 | | * |
4762 | | * Skip an XML (SGML) comment <!-- .... --> |
4763 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4764 | | * must not occur within comments. " |
4765 | | * This is the slow routine in case the accelerator for ascii didn't work |
4766 | | * |
4767 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4768 | | */ |
4769 | | static void |
4770 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4771 | 1.57M | size_t len, size_t size) { |
4772 | 1.57M | int q, ql; |
4773 | 1.57M | int r, rl; |
4774 | 1.57M | int cur, l; |
4775 | 1.57M | size_t count = 0; |
4776 | 1.57M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4777 | 1.35M | XML_MAX_HUGE_LENGTH : |
4778 | 1.57M | XML_MAX_TEXT_LENGTH; |
4779 | 1.57M | int inputid; |
4780 | | |
4781 | 1.57M | inputid = ctxt->input->id; |
4782 | | |
4783 | 1.57M | if (buf == NULL) { |
4784 | 47.7k | len = 0; |
4785 | 47.7k | size = XML_PARSER_BUFFER_SIZE; |
4786 | 47.7k | buf = (xmlChar *) xmlMallocAtomic(size); |
4787 | 47.7k | if (buf == NULL) { |
4788 | 0 | xmlErrMemory(ctxt, NULL); |
4789 | 0 | return; |
4790 | 0 | } |
4791 | 47.7k | } |
4792 | 1.57M | GROW; /* Assure there's enough input data */ |
4793 | 1.57M | q = CUR_CHAR(ql); |
4794 | 1.57M | if (q == 0) |
4795 | 146k | goto not_terminated; |
4796 | 1.43M | if (!IS_CHAR(q)) { |
4797 | 376k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4798 | 376k | "xmlParseComment: invalid xmlChar value %d\n", |
4799 | 376k | q); |
4800 | 376k | xmlFree (buf); |
4801 | 376k | return; |
4802 | 376k | } |
4803 | 1.05M | NEXTL(ql); |
4804 | 1.05M | r = CUR_CHAR(rl); |
4805 | 1.05M | if (r == 0) |
4806 | 62.6k | goto not_terminated; |
4807 | 993k | if (!IS_CHAR(r)) { |
4808 | 24.3k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4809 | 24.3k | "xmlParseComment: invalid xmlChar value %d\n", |
4810 | 24.3k | q); |
4811 | 24.3k | xmlFree (buf); |
4812 | 24.3k | return; |
4813 | 24.3k | } |
4814 | 968k | NEXTL(rl); |
4815 | 968k | cur = CUR_CHAR(l); |
4816 | 968k | if (cur == 0) |
4817 | 39.7k | goto not_terminated; |
4818 | 396M | while (IS_CHAR(cur) && /* checked */ |
4819 | 396M | ((cur != '>') || |
4820 | 395M | (r != '-') || (q != '-'))) { |
4821 | 395M | if ((r == '-') && (q == '-')) { |
4822 | 691k | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
4823 | 691k | } |
4824 | 395M | if (len + 5 >= size) { |
4825 | 542k | xmlChar *new_buf; |
4826 | 542k | size_t new_size; |
4827 | | |
4828 | 542k | new_size = size * 2; |
4829 | 542k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
4830 | 542k | if (new_buf == NULL) { |
4831 | 0 | xmlFree (buf); |
4832 | 0 | xmlErrMemory(ctxt, NULL); |
4833 | 0 | return; |
4834 | 0 | } |
4835 | 542k | buf = new_buf; |
4836 | 542k | size = new_size; |
4837 | 542k | } |
4838 | 395M | COPY_BUF(ql,buf,len,q); |
4839 | 395M | q = r; |
4840 | 395M | ql = rl; |
4841 | 395M | r = cur; |
4842 | 395M | rl = l; |
4843 | | |
4844 | 395M | count++; |
4845 | 395M | if (count > 50) { |
4846 | 7.43M | SHRINK; |
4847 | 7.43M | GROW; |
4848 | 7.43M | count = 0; |
4849 | 7.43M | if (ctxt->instate == XML_PARSER_EOF) { |
4850 | 0 | xmlFree(buf); |
4851 | 0 | return; |
4852 | 0 | } |
4853 | 7.43M | } |
4854 | 395M | NEXTL(l); |
4855 | 395M | cur = CUR_CHAR(l); |
4856 | 395M | if (cur == 0) { |
4857 | 125k | SHRINK; |
4858 | 125k | GROW; |
4859 | 125k | cur = CUR_CHAR(l); |
4860 | 125k | } |
4861 | | |
4862 | 395M | if (len > maxLength) { |
4863 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4864 | 0 | "Comment too big found", NULL); |
4865 | 0 | xmlFree (buf); |
4866 | 0 | return; |
4867 | 0 | } |
4868 | 395M | } |
4869 | 928k | buf[len] = 0; |
4870 | 928k | if (cur == 0) { |
4871 | 125k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4872 | 125k | "Comment not terminated \n<!--%.50s\n", buf); |
4873 | 803k | } else if (!IS_CHAR(cur)) { |
4874 | 218k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4875 | 218k | "xmlParseComment: invalid xmlChar value %d\n", |
4876 | 218k | cur); |
4877 | 584k | } else { |
4878 | 584k | if (inputid != ctxt->input->id) { |
4879 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
4880 | 0 | "Comment doesn't start and stop in the same" |
4881 | 0 | " entity\n"); |
4882 | 0 | } |
4883 | 584k | NEXT; |
4884 | 584k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
4885 | 584k | (!ctxt->disableSAX)) |
4886 | 51.2k | ctxt->sax->comment(ctxt->userData, buf); |
4887 | 584k | } |
4888 | 928k | xmlFree(buf); |
4889 | 928k | return; |
4890 | 248k | not_terminated: |
4891 | 248k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4892 | 248k | "Comment not terminated\n", NULL); |
4893 | 248k | xmlFree(buf); |
4894 | 248k | return; |
4895 | 928k | } |
4896 | | |
4897 | | /** |
4898 | | * xmlParseComment: |
4899 | | * @ctxt: an XML parser context |
4900 | | * |
4901 | | * DEPRECATED: Internal function, don't use. |
4902 | | * |
4903 | | * Skip an XML (SGML) comment <!-- .... --> |
4904 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4905 | | * must not occur within comments. " |
4906 | | * |
4907 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4908 | | */ |
4909 | | void |
4910 | 13.0M | xmlParseComment(xmlParserCtxtPtr ctxt) { |
4911 | 13.0M | xmlChar *buf = NULL; |
4912 | 13.0M | size_t size = XML_PARSER_BUFFER_SIZE; |
4913 | 13.0M | size_t len = 0; |
4914 | 13.0M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4915 | 10.2M | XML_MAX_HUGE_LENGTH : |
4916 | 13.0M | XML_MAX_TEXT_LENGTH; |
4917 | 13.0M | xmlParserInputState state; |
4918 | 13.0M | const xmlChar *in; |
4919 | 13.0M | size_t nbchar = 0; |
4920 | 13.0M | int ccol; |
4921 | 13.0M | int inputid; |
4922 | | |
4923 | | /* |
4924 | | * Check that there is a comment right here. |
4925 | | */ |
4926 | 13.0M | if ((RAW != '<') || (NXT(1) != '!') || |
4927 | 13.0M | (NXT(2) != '-') || (NXT(3) != '-')) return; |
4928 | 12.9M | state = ctxt->instate; |
4929 | 12.9M | ctxt->instate = XML_PARSER_COMMENT; |
4930 | 12.9M | inputid = ctxt->input->id; |
4931 | 12.9M | SKIP(4); |
4932 | 12.9M | SHRINK; |
4933 | 12.9M | GROW; |
4934 | | |
4935 | | /* |
4936 | | * Accelerated common case where input don't need to be |
4937 | | * modified before passing it to the handler. |
4938 | | */ |
4939 | 12.9M | in = ctxt->input->cur; |
4940 | 12.9M | do { |
4941 | 12.9M | if (*in == 0xA) { |
4942 | 787k | do { |
4943 | 787k | ctxt->input->line++; ctxt->input->col = 1; |
4944 | 787k | in++; |
4945 | 787k | } while (*in == 0xA); |
4946 | 747k | } |
4947 | 45.1M | get_more: |
4948 | 45.1M | ccol = ctxt->input->col; |
4949 | 1.12G | while (((*in > '-') && (*in <= 0x7F)) || |
4950 | 1.12G | ((*in >= 0x20) && (*in < '-')) || |
4951 | 1.12G | (*in == 0x09)) { |
4952 | 1.08G | in++; |
4953 | 1.08G | ccol++; |
4954 | 1.08G | } |
4955 | 45.1M | ctxt->input->col = ccol; |
4956 | 45.1M | if (*in == 0xA) { |
4957 | 11.8M | do { |
4958 | 11.8M | ctxt->input->line++; ctxt->input->col = 1; |
4959 | 11.8M | in++; |
4960 | 11.8M | } while (*in == 0xA); |
4961 | 11.0M | goto get_more; |
4962 | 11.0M | } |
4963 | 34.0M | nbchar = in - ctxt->input->cur; |
4964 | | /* |
4965 | | * save current set of data |
4966 | | */ |
4967 | 34.0M | if (nbchar > 0) { |
4968 | 33.9M | if ((ctxt->sax != NULL) && |
4969 | 33.9M | (ctxt->sax->comment != NULL)) { |
4970 | 33.9M | if (buf == NULL) { |
4971 | 12.9M | if ((*in == '-') && (in[1] == '-')) |
4972 | 9.09M | size = nbchar + 1; |
4973 | 3.85M | else |
4974 | 3.85M | size = XML_PARSER_BUFFER_SIZE + nbchar; |
4975 | 12.9M | buf = (xmlChar *) xmlMallocAtomic(size); |
4976 | 12.9M | if (buf == NULL) { |
4977 | 0 | xmlErrMemory(ctxt, NULL); |
4978 | 0 | ctxt->instate = state; |
4979 | 0 | return; |
4980 | 0 | } |
4981 | 12.9M | len = 0; |
4982 | 20.9M | } else if (len + nbchar + 1 >= size) { |
4983 | 1.40M | xmlChar *new_buf; |
4984 | 1.40M | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
4985 | 1.40M | new_buf = (xmlChar *) xmlRealloc(buf, size); |
4986 | 1.40M | if (new_buf == NULL) { |
4987 | 0 | xmlFree (buf); |
4988 | 0 | xmlErrMemory(ctxt, NULL); |
4989 | 0 | ctxt->instate = state; |
4990 | 0 | return; |
4991 | 0 | } |
4992 | 1.40M | buf = new_buf; |
4993 | 1.40M | } |
4994 | 33.9M | memcpy(&buf[len], ctxt->input->cur, nbchar); |
4995 | 33.9M | len += nbchar; |
4996 | 33.9M | buf[len] = 0; |
4997 | 33.9M | } |
4998 | 33.9M | } |
4999 | 34.0M | if (len > maxLength) { |
5000 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5001 | 0 | "Comment too big found", NULL); |
5002 | 0 | xmlFree (buf); |
5003 | 0 | return; |
5004 | 0 | } |
5005 | 34.0M | ctxt->input->cur = in; |
5006 | 34.0M | if (*in == 0xA) { |
5007 | 0 | in++; |
5008 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5009 | 0 | } |
5010 | 34.0M | if (*in == 0xD) { |
5011 | 9.71M | in++; |
5012 | 9.71M | if (*in == 0xA) { |
5013 | 9.66M | ctxt->input->cur = in; |
5014 | 9.66M | in++; |
5015 | 9.66M | ctxt->input->line++; ctxt->input->col = 1; |
5016 | 9.66M | goto get_more; |
5017 | 9.66M | } |
5018 | 45.9k | in--; |
5019 | 45.9k | } |
5020 | 24.3M | SHRINK; |
5021 | 24.3M | GROW; |
5022 | 24.3M | if (ctxt->instate == XML_PARSER_EOF) { |
5023 | 0 | xmlFree(buf); |
5024 | 0 | return; |
5025 | 0 | } |
5026 | 24.3M | in = ctxt->input->cur; |
5027 | 24.3M | if (*in == '-') { |
5028 | 22.8M | if (in[1] == '-') { |
5029 | 15.1M | if (in[2] == '>') { |
5030 | 11.4M | if (ctxt->input->id != inputid) { |
5031 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5032 | 0 | "comment doesn't start and stop in the" |
5033 | 0 | " same entity\n"); |
5034 | 0 | } |
5035 | 11.4M | SKIP(3); |
5036 | 11.4M | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5037 | 11.4M | (!ctxt->disableSAX)) { |
5038 | 2.36M | if (buf != NULL) |
5039 | 2.35M | ctxt->sax->comment(ctxt->userData, buf); |
5040 | 6.48k | else |
5041 | 6.48k | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5042 | 2.36M | } |
5043 | 11.4M | if (buf != NULL) |
5044 | 11.4M | xmlFree(buf); |
5045 | 11.4M | if (ctxt->instate != XML_PARSER_EOF) |
5046 | 11.4M | ctxt->instate = state; |
5047 | 11.4M | return; |
5048 | 11.4M | } |
5049 | 3.68M | if (buf != NULL) { |
5050 | 3.68M | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5051 | 3.68M | "Double hyphen within comment: " |
5052 | 3.68M | "<!--%.50s\n", |
5053 | 3.68M | buf); |
5054 | 3.68M | } else |
5055 | 6.68k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5056 | 6.68k | "Double hyphen within comment\n", NULL); |
5057 | 3.68M | if (ctxt->instate == XML_PARSER_EOF) { |
5058 | 0 | xmlFree(buf); |
5059 | 0 | return; |
5060 | 0 | } |
5061 | 3.68M | in++; |
5062 | 3.68M | ctxt->input->col++; |
5063 | 3.68M | } |
5064 | 11.3M | in++; |
5065 | 11.3M | ctxt->input->col++; |
5066 | 11.3M | goto get_more; |
5067 | 22.8M | } |
5068 | 24.3M | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5069 | 1.57M | xmlParseCommentComplex(ctxt, buf, len, size); |
5070 | 1.57M | ctxt->instate = state; |
5071 | 1.57M | return; |
5072 | 12.9M | } |
5073 | | |
5074 | | |
5075 | | /** |
5076 | | * xmlParsePITarget: |
5077 | | * @ctxt: an XML parser context |
5078 | | * |
5079 | | * DEPRECATED: Internal function, don't use. |
5080 | | * |
5081 | | * parse the name of a PI |
5082 | | * |
5083 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5084 | | * |
5085 | | * Returns the PITarget name or NULL |
5086 | | */ |
5087 | | |
5088 | | const xmlChar * |
5089 | 3.50M | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5090 | 3.50M | const xmlChar *name; |
5091 | | |
5092 | 3.50M | name = xmlParseName(ctxt); |
5093 | 3.50M | if ((name != NULL) && |
5094 | 3.50M | ((name[0] == 'x') || (name[0] == 'X')) && |
5095 | 3.50M | ((name[1] == 'm') || (name[1] == 'M')) && |
5096 | 3.50M | ((name[2] == 'l') || (name[2] == 'L'))) { |
5097 | 1.04M | int i; |
5098 | 1.04M | if ((name[0] == 'x') && (name[1] == 'm') && |
5099 | 1.04M | (name[2] == 'l') && (name[3] == 0)) { |
5100 | 739k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5101 | 739k | "XML declaration allowed only at the start of the document\n"); |
5102 | 739k | return(name); |
5103 | 739k | } else if (name[3] == 0) { |
5104 | 7.69k | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5105 | 7.69k | return(name); |
5106 | 7.69k | } |
5107 | 452k | for (i = 0;;i++) { |
5108 | 452k | if (xmlW3CPIs[i] == NULL) break; |
5109 | 375k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5110 | 221k | return(name); |
5111 | 375k | } |
5112 | 77.2k | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5113 | 77.2k | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5114 | 77.2k | NULL, NULL); |
5115 | 77.2k | } |
5116 | 2.53M | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5117 | 26.6k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5118 | 26.6k | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5119 | 26.6k | } |
5120 | 2.53M | return(name); |
5121 | 3.50M | } |
5122 | | |
5123 | | #ifdef LIBXML_CATALOG_ENABLED |
5124 | | /** |
5125 | | * xmlParseCatalogPI: |
5126 | | * @ctxt: an XML parser context |
5127 | | * @catalog: the PI value string |
5128 | | * |
5129 | | * parse an XML Catalog Processing Instruction. |
5130 | | * |
5131 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5132 | | * |
5133 | | * Occurs only if allowed by the user and if happening in the Misc |
5134 | | * part of the document before any doctype information |
5135 | | * This will add the given catalog to the parsing context in order |
5136 | | * to be used if there is a resolution need further down in the document |
5137 | | */ |
5138 | | |
5139 | | static void |
5140 | 0 | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5141 | 0 | xmlChar *URL = NULL; |
5142 | 0 | const xmlChar *tmp, *base; |
5143 | 0 | xmlChar marker; |
5144 | |
|
5145 | 0 | tmp = catalog; |
5146 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5147 | 0 | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5148 | 0 | goto error; |
5149 | 0 | tmp += 7; |
5150 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5151 | 0 | if (*tmp != '=') { |
5152 | 0 | return; |
5153 | 0 | } |
5154 | 0 | tmp++; |
5155 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5156 | 0 | marker = *tmp; |
5157 | 0 | if ((marker != '\'') && (marker != '"')) |
5158 | 0 | goto error; |
5159 | 0 | tmp++; |
5160 | 0 | base = tmp; |
5161 | 0 | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5162 | 0 | if (*tmp == 0) |
5163 | 0 | goto error; |
5164 | 0 | URL = xmlStrndup(base, tmp - base); |
5165 | 0 | tmp++; |
5166 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5167 | 0 | if (*tmp != 0) |
5168 | 0 | goto error; |
5169 | | |
5170 | 0 | if (URL != NULL) { |
5171 | 0 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5172 | 0 | xmlFree(URL); |
5173 | 0 | } |
5174 | 0 | return; |
5175 | | |
5176 | 0 | error: |
5177 | 0 | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5178 | 0 | "Catalog PI syntax error: %s\n", |
5179 | 0 | catalog, NULL); |
5180 | 0 | if (URL != NULL) |
5181 | 0 | xmlFree(URL); |
5182 | 0 | } |
5183 | | #endif |
5184 | | |
5185 | | /** |
5186 | | * xmlParsePI: |
5187 | | * @ctxt: an XML parser context |
5188 | | * |
5189 | | * DEPRECATED: Internal function, don't use. |
5190 | | * |
5191 | | * parse an XML Processing Instruction. |
5192 | | * |
5193 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5194 | | * |
5195 | | * The processing is transferred to SAX once parsed. |
5196 | | */ |
5197 | | |
5198 | | void |
5199 | 3.50M | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5200 | 3.50M | xmlChar *buf = NULL; |
5201 | 3.50M | size_t len = 0; |
5202 | 3.50M | size_t size = XML_PARSER_BUFFER_SIZE; |
5203 | 3.50M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5204 | 2.97M | XML_MAX_HUGE_LENGTH : |
5205 | 3.50M | XML_MAX_TEXT_LENGTH; |
5206 | 3.50M | int cur, l; |
5207 | 3.50M | const xmlChar *target; |
5208 | 3.50M | xmlParserInputState state; |
5209 | 3.50M | int count = 0; |
5210 | | |
5211 | 3.50M | if ((RAW == '<') && (NXT(1) == '?')) { |
5212 | 3.50M | int inputid = ctxt->input->id; |
5213 | 3.50M | state = ctxt->instate; |
5214 | 3.50M | ctxt->instate = XML_PARSER_PI; |
5215 | | /* |
5216 | | * this is a Processing Instruction. |
5217 | | */ |
5218 | 3.50M | SKIP(2); |
5219 | 3.50M | SHRINK; |
5220 | | |
5221 | | /* |
5222 | | * Parse the target name and check for special support like |
5223 | | * namespace. |
5224 | | */ |
5225 | 3.50M | target = xmlParsePITarget(ctxt); |
5226 | 3.50M | if (target != NULL) { |
5227 | 3.23M | if ((RAW == '?') && (NXT(1) == '>')) { |
5228 | 470k | if (inputid != ctxt->input->id) { |
5229 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5230 | 0 | "PI declaration doesn't start and stop in" |
5231 | 0 | " the same entity\n"); |
5232 | 0 | } |
5233 | 470k | SKIP(2); |
5234 | | |
5235 | | /* |
5236 | | * SAX: PI detected. |
5237 | | */ |
5238 | 470k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5239 | 470k | (ctxt->sax->processingInstruction != NULL)) |
5240 | 123k | ctxt->sax->processingInstruction(ctxt->userData, |
5241 | 123k | target, NULL); |
5242 | 470k | if (ctxt->instate != XML_PARSER_EOF) |
5243 | 470k | ctxt->instate = state; |
5244 | 470k | return; |
5245 | 470k | } |
5246 | 2.76M | buf = (xmlChar *) xmlMallocAtomic(size); |
5247 | 2.76M | if (buf == NULL) { |
5248 | 0 | xmlErrMemory(ctxt, NULL); |
5249 | 0 | ctxt->instate = state; |
5250 | 0 | return; |
5251 | 0 | } |
5252 | 2.76M | if (SKIP_BLANKS == 0) { |
5253 | 636k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5254 | 636k | "ParsePI: PI %s space expected\n", target); |
5255 | 636k | } |
5256 | 2.76M | cur = CUR_CHAR(l); |
5257 | 374M | while (IS_CHAR(cur) && /* checked */ |
5258 | 374M | ((cur != '?') || (NXT(1) != '>'))) { |
5259 | 371M | if (len + 5 >= size) { |
5260 | 1.07M | xmlChar *tmp; |
5261 | 1.07M | size_t new_size = size * 2; |
5262 | 1.07M | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5263 | 1.07M | if (tmp == NULL) { |
5264 | 0 | xmlErrMemory(ctxt, NULL); |
5265 | 0 | xmlFree(buf); |
5266 | 0 | ctxt->instate = state; |
5267 | 0 | return; |
5268 | 0 | } |
5269 | 1.07M | buf = tmp; |
5270 | 1.07M | size = new_size; |
5271 | 1.07M | } |
5272 | 371M | count++; |
5273 | 371M | if (count > 50) { |
5274 | 6.36M | SHRINK; |
5275 | 6.36M | GROW; |
5276 | 6.36M | if (ctxt->instate == XML_PARSER_EOF) { |
5277 | 0 | xmlFree(buf); |
5278 | 0 | return; |
5279 | 0 | } |
5280 | 6.36M | count = 0; |
5281 | 6.36M | } |
5282 | 371M | COPY_BUF(l,buf,len,cur); |
5283 | 371M | NEXTL(l); |
5284 | 371M | cur = CUR_CHAR(l); |
5285 | 371M | if (cur == 0) { |
5286 | 171k | SHRINK; |
5287 | 171k | GROW; |
5288 | 171k | cur = CUR_CHAR(l); |
5289 | 171k | } |
5290 | 371M | if (len > maxLength) { |
5291 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5292 | 0 | "PI %s too big found", target); |
5293 | 0 | xmlFree(buf); |
5294 | 0 | ctxt->instate = state; |
5295 | 0 | return; |
5296 | 0 | } |
5297 | 371M | } |
5298 | 2.76M | buf[len] = 0; |
5299 | 2.76M | if (cur != '?') { |
5300 | 402k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5301 | 402k | "ParsePI: PI %s never end ...\n", target); |
5302 | 2.36M | } else { |
5303 | 2.36M | if (inputid != ctxt->input->id) { |
5304 | 63 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5305 | 63 | "PI declaration doesn't start and stop in" |
5306 | 63 | " the same entity\n"); |
5307 | 63 | } |
5308 | 2.36M | SKIP(2); |
5309 | | |
5310 | 2.36M | #ifdef LIBXML_CATALOG_ENABLED |
5311 | 2.36M | if (((state == XML_PARSER_MISC) || |
5312 | 2.36M | (state == XML_PARSER_START)) && |
5313 | 2.36M | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5314 | 0 | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5315 | 0 | if ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5316 | 0 | (allow == XML_CATA_ALLOW_ALL)) |
5317 | 0 | xmlParseCatalogPI(ctxt, buf); |
5318 | 0 | } |
5319 | 2.36M | #endif |
5320 | | |
5321 | | |
5322 | | /* |
5323 | | * SAX: PI detected. |
5324 | | */ |
5325 | 2.36M | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5326 | 2.36M | (ctxt->sax->processingInstruction != NULL)) |
5327 | 229k | ctxt->sax->processingInstruction(ctxt->userData, |
5328 | 229k | target, buf); |
5329 | 2.36M | } |
5330 | 2.76M | xmlFree(buf); |
5331 | 2.76M | } else { |
5332 | 266k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5333 | 266k | } |
5334 | 3.03M | if (ctxt->instate != XML_PARSER_EOF) |
5335 | 3.03M | ctxt->instate = state; |
5336 | 3.03M | } |
5337 | 3.50M | } |
5338 | | |
5339 | | /** |
5340 | | * xmlParseNotationDecl: |
5341 | | * @ctxt: an XML parser context |
5342 | | * |
5343 | | * DEPRECATED: Internal function, don't use. |
5344 | | * |
5345 | | * parse a notation declaration |
5346 | | * |
5347 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5348 | | * |
5349 | | * Hence there is actually 3 choices: |
5350 | | * 'PUBLIC' S PubidLiteral |
5351 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5352 | | * and 'SYSTEM' S SystemLiteral |
5353 | | * |
5354 | | * See the NOTE on xmlParseExternalID(). |
5355 | | */ |
5356 | | |
5357 | | void |
5358 | 14.8k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5359 | 14.8k | const xmlChar *name; |
5360 | 14.8k | xmlChar *Pubid; |
5361 | 14.8k | xmlChar *Systemid; |
5362 | | |
5363 | 14.8k | if (CMP10(CUR_PTR, '<', '!', 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5364 | 11.6k | int inputid = ctxt->input->id; |
5365 | 11.6k | SHRINK; |
5366 | 11.6k | SKIP(10); |
5367 | 11.6k | if (SKIP_BLANKS == 0) { |
5368 | 415 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5369 | 415 | "Space required after '<!NOTATION'\n"); |
5370 | 415 | return; |
5371 | 415 | } |
5372 | | |
5373 | 11.2k | name = xmlParseName(ctxt); |
5374 | 11.2k | if (name == NULL) { |
5375 | 251 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5376 | 251 | return; |
5377 | 251 | } |
5378 | 11.0k | if (xmlStrchr(name, ':') != NULL) { |
5379 | 472 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5380 | 472 | "colons are forbidden from notation names '%s'\n", |
5381 | 472 | name, NULL, NULL); |
5382 | 472 | } |
5383 | 11.0k | if (SKIP_BLANKS == 0) { |
5384 | 360 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5385 | 360 | "Space required after the NOTATION name'\n"); |
5386 | 360 | return; |
5387 | 360 | } |
5388 | | |
5389 | | /* |
5390 | | * Parse the IDs. |
5391 | | */ |
5392 | 10.6k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5393 | 10.6k | SKIP_BLANKS; |
5394 | | |
5395 | 10.6k | if (RAW == '>') { |
5396 | 8.84k | if (inputid != ctxt->input->id) { |
5397 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5398 | 3 | "Notation declaration doesn't start and stop" |
5399 | 3 | " in the same entity\n"); |
5400 | 3 | } |
5401 | 8.84k | NEXT; |
5402 | 8.84k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5403 | 8.84k | (ctxt->sax->notationDecl != NULL)) |
5404 | 7.70k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5405 | 8.84k | } else { |
5406 | 1.80k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5407 | 1.80k | } |
5408 | 10.6k | if (Systemid != NULL) xmlFree(Systemid); |
5409 | 10.6k | if (Pubid != NULL) xmlFree(Pubid); |
5410 | 10.6k | } |
5411 | 14.8k | } |
5412 | | |
5413 | | /** |
5414 | | * xmlParseEntityDecl: |
5415 | | * @ctxt: an XML parser context |
5416 | | * |
5417 | | * DEPRECATED: Internal function, don't use. |
5418 | | * |
5419 | | * parse <!ENTITY declarations |
5420 | | * |
5421 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5422 | | * |
5423 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5424 | | * |
5425 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5426 | | * |
5427 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5428 | | * |
5429 | | * [74] PEDef ::= EntityValue | ExternalID |
5430 | | * |
5431 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5432 | | * |
5433 | | * [ VC: Notation Declared ] |
5434 | | * The Name must match the declared name of a notation. |
5435 | | */ |
5436 | | |
5437 | | void |
5438 | 2.42M | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5439 | 2.42M | const xmlChar *name = NULL; |
5440 | 2.42M | xmlChar *value = NULL; |
5441 | 2.42M | xmlChar *URI = NULL, *literal = NULL; |
5442 | 2.42M | const xmlChar *ndata = NULL; |
5443 | 2.42M | int isParameter = 0; |
5444 | 2.42M | xmlChar *orig = NULL; |
5445 | | |
5446 | | /* GROW; done in the caller */ |
5447 | 2.42M | if (CMP8(CUR_PTR, '<', '!', 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5448 | 2.41M | int inputid = ctxt->input->id; |
5449 | 2.41M | SHRINK; |
5450 | 2.41M | SKIP(8); |
5451 | 2.41M | if (SKIP_BLANKS == 0) { |
5452 | 18.4k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5453 | 18.4k | "Space required after '<!ENTITY'\n"); |
5454 | 18.4k | } |
5455 | | |
5456 | 2.41M | if (RAW == '%') { |
5457 | 1.43M | NEXT; |
5458 | 1.43M | if (SKIP_BLANKS == 0) { |
5459 | 624 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5460 | 624 | "Space required after '%%'\n"); |
5461 | 624 | } |
5462 | 1.43M | isParameter = 1; |
5463 | 1.43M | } |
5464 | | |
5465 | 2.41M | name = xmlParseName(ctxt); |
5466 | 2.41M | if (name == NULL) { |
5467 | 94.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5468 | 94.2k | "xmlParseEntityDecl: no name\n"); |
5469 | 94.2k | return; |
5470 | 94.2k | } |
5471 | 2.32M | if (xmlStrchr(name, ':') != NULL) { |
5472 | 778 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5473 | 778 | "colons are forbidden from entities names '%s'\n", |
5474 | 778 | name, NULL, NULL); |
5475 | 778 | } |
5476 | 2.32M | if (SKIP_BLANKS == 0) { |
5477 | 11.1k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5478 | 11.1k | "Space required after the entity name\n"); |
5479 | 11.1k | } |
5480 | | |
5481 | 2.32M | ctxt->instate = XML_PARSER_ENTITY_DECL; |
5482 | | /* |
5483 | | * handle the various case of definitions... |
5484 | | */ |
5485 | 2.32M | if (isParameter) { |
5486 | 1.43M | if ((RAW == '"') || (RAW == '\'')) { |
5487 | 1.32M | value = xmlParseEntityValue(ctxt, &orig); |
5488 | 1.32M | if (value) { |
5489 | 1.28M | if ((ctxt->sax != NULL) && |
5490 | 1.28M | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5491 | 1.20M | ctxt->sax->entityDecl(ctxt->userData, name, |
5492 | 1.20M | XML_INTERNAL_PARAMETER_ENTITY, |
5493 | 1.20M | NULL, NULL, value); |
5494 | 1.28M | } |
5495 | 1.32M | } else { |
5496 | 106k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5497 | 106k | if ((URI == NULL) && (literal == NULL)) { |
5498 | 2.73k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5499 | 2.73k | } |
5500 | 106k | if (URI) { |
5501 | 103k | xmlURIPtr uri; |
5502 | | |
5503 | 103k | uri = xmlParseURI((const char *) URI); |
5504 | 103k | if (uri == NULL) { |
5505 | 4.93k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5506 | 4.93k | "Invalid URI: %s\n", URI); |
5507 | | /* |
5508 | | * This really ought to be a well formedness error |
5509 | | * but the XML Core WG decided otherwise c.f. issue |
5510 | | * E26 of the XML erratas. |
5511 | | */ |
5512 | 98.4k | } else { |
5513 | 98.4k | if (uri->fragment != NULL) { |
5514 | | /* |
5515 | | * Okay this is foolish to block those but not |
5516 | | * invalid URIs. |
5517 | | */ |
5518 | 172 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5519 | 98.2k | } else { |
5520 | 98.2k | if ((ctxt->sax != NULL) && |
5521 | 98.2k | (!ctxt->disableSAX) && |
5522 | 98.2k | (ctxt->sax->entityDecl != NULL)) |
5523 | 90.5k | ctxt->sax->entityDecl(ctxt->userData, name, |
5524 | 90.5k | XML_EXTERNAL_PARAMETER_ENTITY, |
5525 | 90.5k | literal, URI, NULL); |
5526 | 98.2k | } |
5527 | 98.4k | xmlFreeURI(uri); |
5528 | 98.4k | } |
5529 | 103k | } |
5530 | 106k | } |
5531 | 1.43M | } else { |
5532 | 893k | if ((RAW == '"') || (RAW == '\'')) { |
5533 | 782k | value = xmlParseEntityValue(ctxt, &orig); |
5534 | 782k | if ((ctxt->sax != NULL) && |
5535 | 782k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5536 | 656k | ctxt->sax->entityDecl(ctxt->userData, name, |
5537 | 656k | XML_INTERNAL_GENERAL_ENTITY, |
5538 | 656k | NULL, NULL, value); |
5539 | | /* |
5540 | | * For expat compatibility in SAX mode. |
5541 | | */ |
5542 | 782k | if ((ctxt->myDoc == NULL) || |
5543 | 782k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5544 | 20.4k | if (ctxt->myDoc == NULL) { |
5545 | 1.98k | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5546 | 1.98k | if (ctxt->myDoc == NULL) { |
5547 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5548 | 0 | return; |
5549 | 0 | } |
5550 | 1.98k | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5551 | 1.98k | } |
5552 | 20.4k | if (ctxt->myDoc->intSubset == NULL) |
5553 | 1.98k | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5554 | 1.98k | BAD_CAST "fake", NULL, NULL); |
5555 | | |
5556 | 20.4k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5557 | 20.4k | NULL, NULL, value); |
5558 | 20.4k | } |
5559 | 782k | } else { |
5560 | 111k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5561 | 111k | if ((URI == NULL) && (literal == NULL)) { |
5562 | 8.46k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5563 | 8.46k | } |
5564 | 111k | if (URI) { |
5565 | 100k | xmlURIPtr uri; |
5566 | | |
5567 | 100k | uri = xmlParseURI((const char *)URI); |
5568 | 100k | if (uri == NULL) { |
5569 | 5.19k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5570 | 5.19k | "Invalid URI: %s\n", URI); |
5571 | | /* |
5572 | | * This really ought to be a well formedness error |
5573 | | * but the XML Core WG decided otherwise c.f. issue |
5574 | | * E26 of the XML erratas. |
5575 | | */ |
5576 | 95.4k | } else { |
5577 | 95.4k | if (uri->fragment != NULL) { |
5578 | | /* |
5579 | | * Okay this is foolish to block those but not |
5580 | | * invalid URIs. |
5581 | | */ |
5582 | 1.04k | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5583 | 1.04k | } |
5584 | 95.4k | xmlFreeURI(uri); |
5585 | 95.4k | } |
5586 | 100k | } |
5587 | 111k | if ((RAW != '>') && (SKIP_BLANKS == 0)) { |
5588 | 10.0k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5589 | 10.0k | "Space required before 'NDATA'\n"); |
5590 | 10.0k | } |
5591 | 111k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5592 | 32.6k | SKIP(5); |
5593 | 32.6k | if (SKIP_BLANKS == 0) { |
5594 | 386 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5595 | 386 | "Space required after 'NDATA'\n"); |
5596 | 386 | } |
5597 | 32.6k | ndata = xmlParseName(ctxt); |
5598 | 32.6k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5599 | 32.6k | (ctxt->sax->unparsedEntityDecl != NULL)) |
5600 | 31.4k | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5601 | 31.4k | literal, URI, ndata); |
5602 | 78.4k | } else { |
5603 | 78.4k | if ((ctxt->sax != NULL) && |
5604 | 78.4k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5605 | 67.9k | ctxt->sax->entityDecl(ctxt->userData, name, |
5606 | 67.9k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5607 | 67.9k | literal, URI, NULL); |
5608 | | /* |
5609 | | * For expat compatibility in SAX mode. |
5610 | | * assuming the entity replacement was asked for |
5611 | | */ |
5612 | 78.4k | if ((ctxt->replaceEntities != 0) && |
5613 | 78.4k | ((ctxt->myDoc == NULL) || |
5614 | 39.0k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5615 | 1.45k | if (ctxt->myDoc == NULL) { |
5616 | 822 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5617 | 822 | if (ctxt->myDoc == NULL) { |
5618 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5619 | 0 | return; |
5620 | 0 | } |
5621 | 822 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5622 | 822 | } |
5623 | | |
5624 | 1.45k | if (ctxt->myDoc->intSubset == NULL) |
5625 | 822 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5626 | 822 | BAD_CAST "fake", NULL, NULL); |
5627 | 1.45k | xmlSAX2EntityDecl(ctxt, name, |
5628 | 1.45k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5629 | 1.45k | literal, URI, NULL); |
5630 | 1.45k | } |
5631 | 78.4k | } |
5632 | 111k | } |
5633 | 893k | } |
5634 | 2.32M | if (ctxt->instate == XML_PARSER_EOF) |
5635 | 0 | goto done; |
5636 | 2.32M | SKIP_BLANKS; |
5637 | 2.32M | if (RAW != '>') { |
5638 | 23.9k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5639 | 23.9k | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5640 | 23.9k | xmlHaltParser(ctxt); |
5641 | 2.30M | } else { |
5642 | 2.30M | if (inputid != ctxt->input->id) { |
5643 | 4.63k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5644 | 4.63k | "Entity declaration doesn't start and stop in" |
5645 | 4.63k | " the same entity\n"); |
5646 | 4.63k | } |
5647 | 2.30M | NEXT; |
5648 | 2.30M | } |
5649 | 2.32M | if (orig != NULL) { |
5650 | | /* |
5651 | | * Ugly mechanism to save the raw entity value. |
5652 | | */ |
5653 | 2.08M | xmlEntityPtr cur = NULL; |
5654 | | |
5655 | 2.08M | if (isParameter) { |
5656 | 1.31M | if ((ctxt->sax != NULL) && |
5657 | 1.31M | (ctxt->sax->getParameterEntity != NULL)) |
5658 | 1.31M | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5659 | 1.31M | } else { |
5660 | 770k | if ((ctxt->sax != NULL) && |
5661 | 770k | (ctxt->sax->getEntity != NULL)) |
5662 | 770k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5663 | 770k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5664 | 92.3k | cur = xmlSAX2GetEntity(ctxt, name); |
5665 | 92.3k | } |
5666 | 770k | } |
5667 | 2.08M | if ((cur != NULL) && (cur->orig == NULL)) { |
5668 | 1.47M | cur->orig = orig; |
5669 | 1.47M | orig = NULL; |
5670 | 1.47M | } |
5671 | 2.08M | } |
5672 | | |
5673 | 2.32M | done: |
5674 | 2.32M | if (value != NULL) xmlFree(value); |
5675 | 2.32M | if (URI != NULL) xmlFree(URI); |
5676 | 2.32M | if (literal != NULL) xmlFree(literal); |
5677 | 2.32M | if (orig != NULL) xmlFree(orig); |
5678 | 2.32M | } |
5679 | 2.42M | } |
5680 | | |
5681 | | /** |
5682 | | * xmlParseDefaultDecl: |
5683 | | * @ctxt: an XML parser context |
5684 | | * @value: Receive a possible fixed default value for the attribute |
5685 | | * |
5686 | | * DEPRECATED: Internal function, don't use. |
5687 | | * |
5688 | | * Parse an attribute default declaration |
5689 | | * |
5690 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5691 | | * |
5692 | | * [ VC: Required Attribute ] |
5693 | | * if the default declaration is the keyword #REQUIRED, then the |
5694 | | * attribute must be specified for all elements of the type in the |
5695 | | * attribute-list declaration. |
5696 | | * |
5697 | | * [ VC: Attribute Default Legal ] |
5698 | | * The declared default value must meet the lexical constraints of |
5699 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5700 | | * |
5701 | | * [ VC: Fixed Attribute Default ] |
5702 | | * if an attribute has a default value declared with the #FIXED |
5703 | | * keyword, instances of that attribute must match the default value. |
5704 | | * |
5705 | | * [ WFC: No < in Attribute Values ] |
5706 | | * handled in xmlParseAttValue() |
5707 | | * |
5708 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5709 | | * or XML_ATTRIBUTE_FIXED. |
5710 | | */ |
5711 | | |
5712 | | int |
5713 | 3.74M | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5714 | 3.74M | int val; |
5715 | 3.74M | xmlChar *ret; |
5716 | | |
5717 | 3.74M | *value = NULL; |
5718 | 3.74M | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5719 | 654k | SKIP(9); |
5720 | 654k | return(XML_ATTRIBUTE_REQUIRED); |
5721 | 654k | } |
5722 | 3.09M | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5723 | 2.58M | SKIP(8); |
5724 | 2.58M | return(XML_ATTRIBUTE_IMPLIED); |
5725 | 2.58M | } |
5726 | 508k | val = XML_ATTRIBUTE_NONE; |
5727 | 508k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5728 | 272k | SKIP(6); |
5729 | 272k | val = XML_ATTRIBUTE_FIXED; |
5730 | 272k | if (SKIP_BLANKS == 0) { |
5731 | 351 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5732 | 351 | "Space required after '#FIXED'\n"); |
5733 | 351 | } |
5734 | 272k | } |
5735 | 508k | ret = xmlParseAttValue(ctxt); |
5736 | 508k | ctxt->instate = XML_PARSER_DTD; |
5737 | 508k | if (ret == NULL) { |
5738 | 7.59k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5739 | 7.59k | "Attribute default value declaration error\n"); |
5740 | 7.59k | } else |
5741 | 500k | *value = ret; |
5742 | 508k | return(val); |
5743 | 3.09M | } |
5744 | | |
5745 | | /** |
5746 | | * xmlParseNotationType: |
5747 | | * @ctxt: an XML parser context |
5748 | | * |
5749 | | * DEPRECATED: Internal function, don't use. |
5750 | | * |
5751 | | * parse an Notation attribute type. |
5752 | | * |
5753 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5754 | | * |
5755 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5756 | | * |
5757 | | * [ VC: Notation Attributes ] |
5758 | | * Values of this type must match one of the notation names included |
5759 | | * in the declaration; all notation names in the declaration must be declared. |
5760 | | * |
5761 | | * Returns: the notation attribute tree built while parsing |
5762 | | */ |
5763 | | |
5764 | | xmlEnumerationPtr |
5765 | 5.27k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5766 | 5.27k | const xmlChar *name; |
5767 | 5.27k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5768 | | |
5769 | 5.27k | if (RAW != '(') { |
5770 | 271 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5771 | 271 | return(NULL); |
5772 | 271 | } |
5773 | 4.99k | SHRINK; |
5774 | 6.28k | do { |
5775 | 6.28k | NEXT; |
5776 | 6.28k | SKIP_BLANKS; |
5777 | 6.28k | name = xmlParseName(ctxt); |
5778 | 6.28k | if (name == NULL) { |
5779 | 329 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5780 | 329 | "Name expected in NOTATION declaration\n"); |
5781 | 329 | xmlFreeEnumeration(ret); |
5782 | 329 | return(NULL); |
5783 | 329 | } |
5784 | 5.95k | tmp = ret; |
5785 | 9.00k | while (tmp != NULL) { |
5786 | 3.35k | if (xmlStrEqual(name, tmp->name)) { |
5787 | 308 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5788 | 308 | "standalone: attribute notation value token %s duplicated\n", |
5789 | 308 | name, NULL); |
5790 | 308 | if (!xmlDictOwns(ctxt->dict, name)) |
5791 | 0 | xmlFree((xmlChar *) name); |
5792 | 308 | break; |
5793 | 308 | } |
5794 | 3.04k | tmp = tmp->next; |
5795 | 3.04k | } |
5796 | 5.95k | if (tmp == NULL) { |
5797 | 5.65k | cur = xmlCreateEnumeration(name); |
5798 | 5.65k | if (cur == NULL) { |
5799 | 0 | xmlFreeEnumeration(ret); |
5800 | 0 | return(NULL); |
5801 | 0 | } |
5802 | 5.65k | if (last == NULL) ret = last = cur; |
5803 | 916 | else { |
5804 | 916 | last->next = cur; |
5805 | 916 | last = cur; |
5806 | 916 | } |
5807 | 5.65k | } |
5808 | 5.95k | SKIP_BLANKS; |
5809 | 5.95k | } while (RAW == '|'); |
5810 | 4.67k | if (RAW != ')') { |
5811 | 564 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5812 | 564 | xmlFreeEnumeration(ret); |
5813 | 564 | return(NULL); |
5814 | 564 | } |
5815 | 4.10k | NEXT; |
5816 | 4.10k | return(ret); |
5817 | 4.67k | } |
5818 | | |
5819 | | /** |
5820 | | * xmlParseEnumerationType: |
5821 | | * @ctxt: an XML parser context |
5822 | | * |
5823 | | * DEPRECATED: Internal function, don't use. |
5824 | | * |
5825 | | * parse an Enumeration attribute type. |
5826 | | * |
5827 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5828 | | * |
5829 | | * [ VC: Enumeration ] |
5830 | | * Values of this type must match one of the Nmtoken tokens in |
5831 | | * the declaration |
5832 | | * |
5833 | | * Returns: the enumeration attribute tree built while parsing |
5834 | | */ |
5835 | | |
5836 | | xmlEnumerationPtr |
5837 | 229k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
5838 | 229k | xmlChar *name; |
5839 | 229k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5840 | | |
5841 | 229k | if (RAW != '(') { |
5842 | 13.7k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5843 | 13.7k | return(NULL); |
5844 | 13.7k | } |
5845 | 215k | SHRINK; |
5846 | 749k | do { |
5847 | 749k | NEXT; |
5848 | 749k | SKIP_BLANKS; |
5849 | 749k | name = xmlParseNmtoken(ctxt); |
5850 | 749k | if (name == NULL) { |
5851 | 577 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5852 | 577 | return(ret); |
5853 | 577 | } |
5854 | 748k | tmp = ret; |
5855 | 2.07M | while (tmp != NULL) { |
5856 | 1.32M | if (xmlStrEqual(name, tmp->name)) { |
5857 | 584 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5858 | 584 | "standalone: attribute enumeration value token %s duplicated\n", |
5859 | 584 | name, NULL); |
5860 | 584 | if (!xmlDictOwns(ctxt->dict, name)) |
5861 | 584 | xmlFree(name); |
5862 | 584 | break; |
5863 | 584 | } |
5864 | 1.32M | tmp = tmp->next; |
5865 | 1.32M | } |
5866 | 748k | if (tmp == NULL) { |
5867 | 747k | cur = xmlCreateEnumeration(name); |
5868 | 747k | if (!xmlDictOwns(ctxt->dict, name)) |
5869 | 747k | xmlFree(name); |
5870 | 747k | if (cur == NULL) { |
5871 | 0 | xmlFreeEnumeration(ret); |
5872 | 0 | return(NULL); |
5873 | 0 | } |
5874 | 747k | if (last == NULL) ret = last = cur; |
5875 | 532k | else { |
5876 | 532k | last->next = cur; |
5877 | 532k | last = cur; |
5878 | 532k | } |
5879 | 747k | } |
5880 | 748k | SKIP_BLANKS; |
5881 | 748k | } while (RAW == '|'); |
5882 | 215k | if (RAW != ')') { |
5883 | 2.40k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5884 | 2.40k | return(ret); |
5885 | 2.40k | } |
5886 | 212k | NEXT; |
5887 | 212k | return(ret); |
5888 | 215k | } |
5889 | | |
5890 | | /** |
5891 | | * xmlParseEnumeratedType: |
5892 | | * @ctxt: an XML parser context |
5893 | | * @tree: the enumeration tree built while parsing |
5894 | | * |
5895 | | * DEPRECATED: Internal function, don't use. |
5896 | | * |
5897 | | * parse an Enumerated attribute type. |
5898 | | * |
5899 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5900 | | * |
5901 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5902 | | * |
5903 | | * |
5904 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
5905 | | */ |
5906 | | |
5907 | | int |
5908 | 234k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5909 | 234k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5910 | 5.59k | SKIP(8); |
5911 | 5.59k | if (SKIP_BLANKS == 0) { |
5912 | 325 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5913 | 325 | "Space required after 'NOTATION'\n"); |
5914 | 325 | return(0); |
5915 | 325 | } |
5916 | 5.27k | *tree = xmlParseNotationType(ctxt); |
5917 | 5.27k | if (*tree == NULL) return(0); |
5918 | 4.10k | return(XML_ATTRIBUTE_NOTATION); |
5919 | 5.27k | } |
5920 | 229k | *tree = xmlParseEnumerationType(ctxt); |
5921 | 229k | if (*tree == NULL) return(0); |
5922 | 215k | return(XML_ATTRIBUTE_ENUMERATION); |
5923 | 229k | } |
5924 | | |
5925 | | /** |
5926 | | * xmlParseAttributeType: |
5927 | | * @ctxt: an XML parser context |
5928 | | * @tree: the enumeration tree built while parsing |
5929 | | * |
5930 | | * DEPRECATED: Internal function, don't use. |
5931 | | * |
5932 | | * parse the Attribute list def for an element |
5933 | | * |
5934 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
5935 | | * |
5936 | | * [55] StringType ::= 'CDATA' |
5937 | | * |
5938 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
5939 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
5940 | | * |
5941 | | * Validity constraints for attribute values syntax are checked in |
5942 | | * xmlValidateAttributeValue() |
5943 | | * |
5944 | | * [ VC: ID ] |
5945 | | * Values of type ID must match the Name production. A name must not |
5946 | | * appear more than once in an XML document as a value of this type; |
5947 | | * i.e., ID values must uniquely identify the elements which bear them. |
5948 | | * |
5949 | | * [ VC: One ID per Element Type ] |
5950 | | * No element type may have more than one ID attribute specified. |
5951 | | * |
5952 | | * [ VC: ID Attribute Default ] |
5953 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
5954 | | * |
5955 | | * [ VC: IDREF ] |
5956 | | * Values of type IDREF must match the Name production, and values |
5957 | | * of type IDREFS must match Names; each IDREF Name must match the value |
5958 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
5959 | | * values must match the value of some ID attribute. |
5960 | | * |
5961 | | * [ VC: Entity Name ] |
5962 | | * Values of type ENTITY must match the Name production, values |
5963 | | * of type ENTITIES must match Names; each Entity Name must match the |
5964 | | * name of an unparsed entity declared in the DTD. |
5965 | | * |
5966 | | * [ VC: Name Token ] |
5967 | | * Values of type NMTOKEN must match the Nmtoken production; values |
5968 | | * of type NMTOKENS must match Nmtokens. |
5969 | | * |
5970 | | * Returns the attribute type |
5971 | | */ |
5972 | | int |
5973 | 3.76M | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5974 | 3.76M | SHRINK; |
5975 | 3.76M | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
5976 | 1.34M | SKIP(5); |
5977 | 1.34M | return(XML_ATTRIBUTE_CDATA); |
5978 | 2.42M | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
5979 | 18.4k | SKIP(6); |
5980 | 18.4k | return(XML_ATTRIBUTE_IDREFS); |
5981 | 2.40M | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
5982 | 75.7k | SKIP(5); |
5983 | 75.7k | return(XML_ATTRIBUTE_IDREF); |
5984 | 2.32M | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
5985 | 994k | SKIP(2); |
5986 | 994k | return(XML_ATTRIBUTE_ID); |
5987 | 1.33M | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5988 | 10.0k | SKIP(6); |
5989 | 10.0k | return(XML_ATTRIBUTE_ENTITY); |
5990 | 1.32M | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
5991 | 9.92k | SKIP(8); |
5992 | 9.92k | return(XML_ATTRIBUTE_ENTITIES); |
5993 | 1.31M | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
5994 | 102k | SKIP(8); |
5995 | 102k | return(XML_ATTRIBUTE_NMTOKENS); |
5996 | 1.20M | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
5997 | 974k | SKIP(7); |
5998 | 974k | return(XML_ATTRIBUTE_NMTOKEN); |
5999 | 974k | } |
6000 | 234k | return(xmlParseEnumeratedType(ctxt, tree)); |
6001 | 3.76M | } |
6002 | | |
6003 | | /** |
6004 | | * xmlParseAttributeListDecl: |
6005 | | * @ctxt: an XML parser context |
6006 | | * |
6007 | | * DEPRECATED: Internal function, don't use. |
6008 | | * |
6009 | | * : parse the Attribute list def for an element |
6010 | | * |
6011 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
6012 | | * |
6013 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
6014 | | * |
6015 | | */ |
6016 | | void |
6017 | 1.72M | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
6018 | 1.72M | const xmlChar *elemName; |
6019 | 1.72M | const xmlChar *attrName; |
6020 | 1.72M | xmlEnumerationPtr tree; |
6021 | | |
6022 | 1.72M | if (CMP9(CUR_PTR, '<', '!', 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
6023 | 1.71M | int inputid = ctxt->input->id; |
6024 | | |
6025 | 1.71M | SKIP(9); |
6026 | 1.71M | if (SKIP_BLANKS == 0) { |
6027 | 6.13k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6028 | 6.13k | "Space required after '<!ATTLIST'\n"); |
6029 | 6.13k | } |
6030 | 1.71M | elemName = xmlParseName(ctxt); |
6031 | 1.71M | if (elemName == NULL) { |
6032 | 2.78k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6033 | 2.78k | "ATTLIST: no name for Element\n"); |
6034 | 2.78k | return; |
6035 | 2.78k | } |
6036 | 1.70M | SKIP_BLANKS; |
6037 | 1.70M | GROW; |
6038 | 5.44M | while ((RAW != '>') && (ctxt->instate != XML_PARSER_EOF)) { |
6039 | 3.79M | int type; |
6040 | 3.79M | int def; |
6041 | 3.79M | xmlChar *defaultValue = NULL; |
6042 | | |
6043 | 3.79M | GROW; |
6044 | 3.79M | tree = NULL; |
6045 | 3.79M | attrName = xmlParseName(ctxt); |
6046 | 3.79M | if (attrName == NULL) { |
6047 | 16.7k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6048 | 16.7k | "ATTLIST: no name for Attribute\n"); |
6049 | 16.7k | break; |
6050 | 16.7k | } |
6051 | 3.77M | GROW; |
6052 | 3.77M | if (SKIP_BLANKS == 0) { |
6053 | 8.73k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6054 | 8.73k | "Space required after the attribute name\n"); |
6055 | 8.73k | break; |
6056 | 8.73k | } |
6057 | | |
6058 | 3.76M | type = xmlParseAttributeType(ctxt, &tree); |
6059 | 3.76M | if (type <= 0) { |
6060 | 15.6k | break; |
6061 | 15.6k | } |
6062 | | |
6063 | 3.75M | GROW; |
6064 | 3.75M | if (SKIP_BLANKS == 0) { |
6065 | 5.87k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6066 | 5.87k | "Space required after the attribute type\n"); |
6067 | 5.87k | if (tree != NULL) |
6068 | 2.90k | xmlFreeEnumeration(tree); |
6069 | 5.87k | break; |
6070 | 5.87k | } |
6071 | | |
6072 | 3.74M | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6073 | 3.74M | if (def <= 0) { |
6074 | 0 | if (defaultValue != NULL) |
6075 | 0 | xmlFree(defaultValue); |
6076 | 0 | if (tree != NULL) |
6077 | 0 | xmlFreeEnumeration(tree); |
6078 | 0 | break; |
6079 | 0 | } |
6080 | 3.74M | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6081 | 127k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6082 | | |
6083 | 3.74M | GROW; |
6084 | 3.74M | if (RAW != '>') { |
6085 | 3.19M | if (SKIP_BLANKS == 0) { |
6086 | 15.3k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6087 | 15.3k | "Space required after the attribute default value\n"); |
6088 | 15.3k | if (defaultValue != NULL) |
6089 | 7.57k | xmlFree(defaultValue); |
6090 | 15.3k | if (tree != NULL) |
6091 | 1.61k | xmlFreeEnumeration(tree); |
6092 | 15.3k | break; |
6093 | 15.3k | } |
6094 | 3.19M | } |
6095 | 3.73M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6096 | 3.73M | (ctxt->sax->attributeDecl != NULL)) |
6097 | 3.46M | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6098 | 3.46M | type, def, defaultValue, tree); |
6099 | 271k | else if (tree != NULL) |
6100 | 16.4k | xmlFreeEnumeration(tree); |
6101 | | |
6102 | 3.73M | if ((ctxt->sax2) && (defaultValue != NULL) && |
6103 | 3.73M | (def != XML_ATTRIBUTE_IMPLIED) && |
6104 | 3.73M | (def != XML_ATTRIBUTE_REQUIRED)) { |
6105 | 401k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6106 | 401k | } |
6107 | 3.73M | if (ctxt->sax2) { |
6108 | 2.79M | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6109 | 2.79M | } |
6110 | 3.73M | if (defaultValue != NULL) |
6111 | 492k | xmlFree(defaultValue); |
6112 | 3.73M | GROW; |
6113 | 3.73M | } |
6114 | 1.70M | if (RAW == '>') { |
6115 | 1.65M | if (inputid != ctxt->input->id) { |
6116 | 3.33k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6117 | 3.33k | "Attribute list declaration doesn't start and" |
6118 | 3.33k | " stop in the same entity\n"); |
6119 | 3.33k | } |
6120 | 1.65M | NEXT; |
6121 | 1.65M | } |
6122 | 1.70M | } |
6123 | 1.72M | } |
6124 | | |
6125 | | /** |
6126 | | * xmlParseElementMixedContentDecl: |
6127 | | * @ctxt: an XML parser context |
6128 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6129 | | * |
6130 | | * DEPRECATED: Internal function, don't use. |
6131 | | * |
6132 | | * parse the declaration for a Mixed Element content |
6133 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6134 | | * |
6135 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6136 | | * '(' S? '#PCDATA' S? ')' |
6137 | | * |
6138 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6139 | | * |
6140 | | * [ VC: No Duplicate Types ] |
6141 | | * The same name must not appear more than once in a single |
6142 | | * mixed-content declaration. |
6143 | | * |
6144 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6145 | | */ |
6146 | | xmlElementContentPtr |
6147 | 607k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6148 | 607k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6149 | 607k | const xmlChar *elem = NULL; |
6150 | | |
6151 | 607k | GROW; |
6152 | 607k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6153 | 607k | SKIP(7); |
6154 | 607k | SKIP_BLANKS; |
6155 | 607k | SHRINK; |
6156 | 607k | if (RAW == ')') { |
6157 | 464k | if (ctxt->input->id != inputchk) { |
6158 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6159 | 0 | "Element content declaration doesn't start and" |
6160 | 0 | " stop in the same entity\n"); |
6161 | 0 | } |
6162 | 464k | NEXT; |
6163 | 464k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6164 | 464k | if (ret == NULL) |
6165 | 0 | return(NULL); |
6166 | 464k | if (RAW == '*') { |
6167 | 168 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6168 | 168 | NEXT; |
6169 | 168 | } |
6170 | 464k | return(ret); |
6171 | 464k | } |
6172 | 142k | if ((RAW == '(') || (RAW == '|')) { |
6173 | 140k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6174 | 140k | if (ret == NULL) return(NULL); |
6175 | 140k | } |
6176 | 1.64M | while ((RAW == '|') && (ctxt->instate != XML_PARSER_EOF)) { |
6177 | 1.50M | NEXT; |
6178 | 1.50M | if (elem == NULL) { |
6179 | 140k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6180 | 140k | if (ret == NULL) { |
6181 | 0 | xmlFreeDocElementContent(ctxt->myDoc, cur); |
6182 | 0 | return(NULL); |
6183 | 0 | } |
6184 | 140k | ret->c1 = cur; |
6185 | 140k | if (cur != NULL) |
6186 | 140k | cur->parent = ret; |
6187 | 140k | cur = ret; |
6188 | 1.36M | } else { |
6189 | 1.36M | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6190 | 1.36M | if (n == NULL) { |
6191 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6192 | 0 | return(NULL); |
6193 | 0 | } |
6194 | 1.36M | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6195 | 1.36M | if (n->c1 != NULL) |
6196 | 1.36M | n->c1->parent = n; |
6197 | 1.36M | cur->c2 = n; |
6198 | 1.36M | if (n != NULL) |
6199 | 1.36M | n->parent = cur; |
6200 | 1.36M | cur = n; |
6201 | 1.36M | } |
6202 | 1.50M | SKIP_BLANKS; |
6203 | 1.50M | elem = xmlParseName(ctxt); |
6204 | 1.50M | if (elem == NULL) { |
6205 | 697 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6206 | 697 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6207 | 697 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6208 | 697 | return(NULL); |
6209 | 697 | } |
6210 | 1.50M | SKIP_BLANKS; |
6211 | 1.50M | GROW; |
6212 | 1.50M | } |
6213 | 141k | if ((RAW == ')') && (NXT(1) == '*')) { |
6214 | 138k | if (elem != NULL) { |
6215 | 138k | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6216 | 138k | XML_ELEMENT_CONTENT_ELEMENT); |
6217 | 138k | if (cur->c2 != NULL) |
6218 | 138k | cur->c2->parent = cur; |
6219 | 138k | } |
6220 | 138k | if (ret != NULL) |
6221 | 138k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6222 | 138k | if (ctxt->input->id != inputchk) { |
6223 | 10 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6224 | 10 | "Element content declaration doesn't start and" |
6225 | 10 | " stop in the same entity\n"); |
6226 | 10 | } |
6227 | 138k | SKIP(2); |
6228 | 138k | } else { |
6229 | 3.21k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6230 | 3.21k | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6231 | 3.21k | return(NULL); |
6232 | 3.21k | } |
6233 | | |
6234 | 141k | } else { |
6235 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6236 | 0 | } |
6237 | 138k | return(ret); |
6238 | 607k | } |
6239 | | |
6240 | | /** |
6241 | | * xmlParseElementChildrenContentDeclPriv: |
6242 | | * @ctxt: an XML parser context |
6243 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6244 | | * @depth: the level of recursion |
6245 | | * |
6246 | | * parse the declaration for a Mixed Element content |
6247 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6248 | | * |
6249 | | * |
6250 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6251 | | * |
6252 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6253 | | * |
6254 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6255 | | * |
6256 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6257 | | * |
6258 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6259 | | * TODO Parameter-entity replacement text must be properly nested |
6260 | | * with parenthesized groups. That is to say, if either of the |
6261 | | * opening or closing parentheses in a choice, seq, or Mixed |
6262 | | * construct is contained in the replacement text for a parameter |
6263 | | * entity, both must be contained in the same replacement text. For |
6264 | | * interoperability, if a parameter-entity reference appears in a |
6265 | | * choice, seq, or Mixed construct, its replacement text should not |
6266 | | * be empty, and neither the first nor last non-blank character of |
6267 | | * the replacement text should be a connector (| or ,). |
6268 | | * |
6269 | | * Returns the tree of xmlElementContentPtr describing the element |
6270 | | * hierarchy. |
6271 | | */ |
6272 | | static xmlElementContentPtr |
6273 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6274 | 1.32M | int depth) { |
6275 | 1.32M | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6276 | 1.32M | const xmlChar *elem; |
6277 | 1.32M | xmlChar type = 0; |
6278 | | |
6279 | 1.32M | if (((depth > 128) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
6280 | 1.32M | (depth > 2048)) { |
6281 | 154 | xmlFatalErrMsgInt(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, |
6282 | 154 | "xmlParseElementChildrenContentDecl : depth %d too deep, use XML_PARSE_HUGE\n", |
6283 | 154 | depth); |
6284 | 154 | return(NULL); |
6285 | 154 | } |
6286 | 1.32M | SKIP_BLANKS; |
6287 | 1.32M | GROW; |
6288 | 1.32M | if (RAW == '(') { |
6289 | 161k | int inputid = ctxt->input->id; |
6290 | | |
6291 | | /* Recurse on first child */ |
6292 | 161k | NEXT; |
6293 | 161k | SKIP_BLANKS; |
6294 | 161k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6295 | 161k | depth + 1); |
6296 | 161k | if (cur == NULL) |
6297 | 103k | return(NULL); |
6298 | 57.0k | SKIP_BLANKS; |
6299 | 57.0k | GROW; |
6300 | 1.16M | } else { |
6301 | 1.16M | elem = xmlParseName(ctxt); |
6302 | 1.16M | if (elem == NULL) { |
6303 | 41.6k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6304 | 41.6k | return(NULL); |
6305 | 41.6k | } |
6306 | 1.12M | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6307 | 1.12M | if (cur == NULL) { |
6308 | 0 | xmlErrMemory(ctxt, NULL); |
6309 | 0 | return(NULL); |
6310 | 0 | } |
6311 | 1.12M | GROW; |
6312 | 1.12M | if (RAW == '?') { |
6313 | 67.3k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6314 | 67.3k | NEXT; |
6315 | 1.05M | } else if (RAW == '*') { |
6316 | 149k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6317 | 149k | NEXT; |
6318 | 907k | } else if (RAW == '+') { |
6319 | 151k | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6320 | 151k | NEXT; |
6321 | 755k | } else { |
6322 | 755k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6323 | 755k | } |
6324 | 1.12M | GROW; |
6325 | 1.12M | } |
6326 | 1.18M | SKIP_BLANKS; |
6327 | 1.18M | SHRINK; |
6328 | 4.66M | while ((RAW != ')') && (ctxt->instate != XML_PARSER_EOF)) { |
6329 | | /* |
6330 | | * Each loop we parse one separator and one element. |
6331 | | */ |
6332 | 3.53M | if (RAW == ',') { |
6333 | 908k | if (type == 0) type = CUR; |
6334 | | |
6335 | | /* |
6336 | | * Detect "Name | Name , Name" error |
6337 | | */ |
6338 | 500k | else if (type != CUR) { |
6339 | 258 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6340 | 258 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6341 | 258 | type); |
6342 | 258 | if ((last != NULL) && (last != ret)) |
6343 | 258 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6344 | 258 | if (ret != NULL) |
6345 | 258 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6346 | 258 | return(NULL); |
6347 | 258 | } |
6348 | 908k | NEXT; |
6349 | | |
6350 | 908k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6351 | 908k | if (op == NULL) { |
6352 | 0 | if ((last != NULL) && (last != ret)) |
6353 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6354 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6355 | 0 | return(NULL); |
6356 | 0 | } |
6357 | 908k | if (last == NULL) { |
6358 | 408k | op->c1 = ret; |
6359 | 408k | if (ret != NULL) |
6360 | 408k | ret->parent = op; |
6361 | 408k | ret = cur = op; |
6362 | 500k | } else { |
6363 | 500k | cur->c2 = op; |
6364 | 500k | if (op != NULL) |
6365 | 500k | op->parent = cur; |
6366 | 500k | op->c1 = last; |
6367 | 500k | if (last != NULL) |
6368 | 500k | last->parent = op; |
6369 | 500k | cur =op; |
6370 | 500k | last = NULL; |
6371 | 500k | } |
6372 | 2.62M | } else if (RAW == '|') { |
6373 | 2.61M | if (type == 0) type = CUR; |
6374 | | |
6375 | | /* |
6376 | | * Detect "Name , Name | Name" error |
6377 | | */ |
6378 | 2.27M | else if (type != CUR) { |
6379 | 221 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6380 | 221 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6381 | 221 | type); |
6382 | 221 | if ((last != NULL) && (last != ret)) |
6383 | 221 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6384 | 221 | if (ret != NULL) |
6385 | 221 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6386 | 221 | return(NULL); |
6387 | 221 | } |
6388 | 2.61M | NEXT; |
6389 | | |
6390 | 2.61M | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6391 | 2.61M | if (op == NULL) { |
6392 | 0 | if ((last != NULL) && (last != ret)) |
6393 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6394 | 0 | if (ret != NULL) |
6395 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6396 | 0 | return(NULL); |
6397 | 0 | } |
6398 | 2.61M | if (last == NULL) { |
6399 | 331k | op->c1 = ret; |
6400 | 331k | if (ret != NULL) |
6401 | 331k | ret->parent = op; |
6402 | 331k | ret = cur = op; |
6403 | 2.27M | } else { |
6404 | 2.27M | cur->c2 = op; |
6405 | 2.27M | if (op != NULL) |
6406 | 2.27M | op->parent = cur; |
6407 | 2.27M | op->c1 = last; |
6408 | 2.27M | if (last != NULL) |
6409 | 2.27M | last->parent = op; |
6410 | 2.27M | cur =op; |
6411 | 2.27M | last = NULL; |
6412 | 2.27M | } |
6413 | 2.61M | } else { |
6414 | 15.2k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6415 | 15.2k | if ((last != NULL) && (last != ret)) |
6416 | 6.76k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6417 | 15.2k | if (ret != NULL) |
6418 | 15.2k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6419 | 15.2k | return(NULL); |
6420 | 15.2k | } |
6421 | 3.51M | GROW; |
6422 | 3.51M | SKIP_BLANKS; |
6423 | 3.51M | GROW; |
6424 | 3.51M | if (RAW == '(') { |
6425 | 149k | int inputid = ctxt->input->id; |
6426 | | /* Recurse on second child */ |
6427 | 149k | NEXT; |
6428 | 149k | SKIP_BLANKS; |
6429 | 149k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6430 | 149k | depth + 1); |
6431 | 149k | if (last == NULL) { |
6432 | 1.52k | if (ret != NULL) |
6433 | 1.52k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6434 | 1.52k | return(NULL); |
6435 | 1.52k | } |
6436 | 148k | SKIP_BLANKS; |
6437 | 3.36M | } else { |
6438 | 3.36M | elem = xmlParseName(ctxt); |
6439 | 3.36M | if (elem == NULL) { |
6440 | 32.1k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6441 | 32.1k | if (ret != NULL) |
6442 | 32.1k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6443 | 32.1k | return(NULL); |
6444 | 32.1k | } |
6445 | 3.33M | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6446 | 3.33M | if (last == NULL) { |
6447 | 0 | if (ret != NULL) |
6448 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6449 | 0 | return(NULL); |
6450 | 0 | } |
6451 | 3.33M | if (RAW == '?') { |
6452 | 367k | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6453 | 367k | NEXT; |
6454 | 2.96M | } else if (RAW == '*') { |
6455 | 181k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6456 | 181k | NEXT; |
6457 | 2.78M | } else if (RAW == '+') { |
6458 | 38.6k | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6459 | 38.6k | NEXT; |
6460 | 2.74M | } else { |
6461 | 2.74M | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6462 | 2.74M | } |
6463 | 3.33M | } |
6464 | 3.48M | SKIP_BLANKS; |
6465 | 3.48M | GROW; |
6466 | 3.48M | } |
6467 | 1.13M | if ((cur != NULL) && (last != NULL)) { |
6468 | 698k | cur->c2 = last; |
6469 | 698k | if (last != NULL) |
6470 | 698k | last->parent = cur; |
6471 | 698k | } |
6472 | 1.13M | if (ctxt->input->id != inputchk) { |
6473 | 237 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6474 | 237 | "Element content declaration doesn't start and stop in" |
6475 | 237 | " the same entity\n"); |
6476 | 237 | } |
6477 | 1.13M | NEXT; |
6478 | 1.13M | if (RAW == '?') { |
6479 | 14.1k | if (ret != NULL) { |
6480 | 14.1k | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6481 | 14.1k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6482 | 75 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6483 | 14.1k | else |
6484 | 14.1k | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6485 | 14.1k | } |
6486 | 14.1k | NEXT; |
6487 | 1.11M | } else if (RAW == '*') { |
6488 | 342k | if (ret != NULL) { |
6489 | 342k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6490 | 342k | cur = ret; |
6491 | | /* |
6492 | | * Some normalization: |
6493 | | * (a | b* | c?)* == (a | b | c)* |
6494 | | */ |
6495 | 1.94M | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6496 | 1.60M | if ((cur->c1 != NULL) && |
6497 | 1.60M | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6498 | 1.60M | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6499 | 82.8k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6500 | 1.60M | if ((cur->c2 != NULL) && |
6501 | 1.60M | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6502 | 1.60M | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6503 | 13.7k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6504 | 1.60M | cur = cur->c2; |
6505 | 1.60M | } |
6506 | 342k | } |
6507 | 342k | NEXT; |
6508 | 774k | } else if (RAW == '+') { |
6509 | 118k | if (ret != NULL) { |
6510 | 118k | int found = 0; |
6511 | | |
6512 | 118k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6513 | 118k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6514 | 17 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6515 | 117k | else |
6516 | 117k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6517 | | /* |
6518 | | * Some normalization: |
6519 | | * (a | b*)+ == (a | b)* |
6520 | | * (a | b?)+ == (a | b)* |
6521 | | */ |
6522 | 206k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6523 | 88.3k | if ((cur->c1 != NULL) && |
6524 | 88.3k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6525 | 88.3k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6526 | 523 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6527 | 523 | found = 1; |
6528 | 523 | } |
6529 | 88.3k | if ((cur->c2 != NULL) && |
6530 | 88.3k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6531 | 88.3k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6532 | 313 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6533 | 313 | found = 1; |
6534 | 313 | } |
6535 | 88.3k | cur = cur->c2; |
6536 | 88.3k | } |
6537 | 118k | if (found) |
6538 | 586 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6539 | 118k | } |
6540 | 118k | NEXT; |
6541 | 118k | } |
6542 | 1.13M | return(ret); |
6543 | 1.18M | } |
6544 | | |
6545 | | /** |
6546 | | * xmlParseElementChildrenContentDecl: |
6547 | | * @ctxt: an XML parser context |
6548 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6549 | | * |
6550 | | * DEPRECATED: Internal function, don't use. |
6551 | | * |
6552 | | * parse the declaration for a Mixed Element content |
6553 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6554 | | * |
6555 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6556 | | * |
6557 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6558 | | * |
6559 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6560 | | * |
6561 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6562 | | * |
6563 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6564 | | * TODO Parameter-entity replacement text must be properly nested |
6565 | | * with parenthesized groups. That is to say, if either of the |
6566 | | * opening or closing parentheses in a choice, seq, or Mixed |
6567 | | * construct is contained in the replacement text for a parameter |
6568 | | * entity, both must be contained in the same replacement text. For |
6569 | | * interoperability, if a parameter-entity reference appears in a |
6570 | | * choice, seq, or Mixed construct, its replacement text should not |
6571 | | * be empty, and neither the first nor last non-blank character of |
6572 | | * the replacement text should be a connector (| or ,). |
6573 | | * |
6574 | | * Returns the tree of xmlElementContentPtr describing the element |
6575 | | * hierarchy. |
6576 | | */ |
6577 | | xmlElementContentPtr |
6578 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6579 | | /* stub left for API/ABI compat */ |
6580 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6581 | 0 | } |
6582 | | |
6583 | | /** |
6584 | | * xmlParseElementContentDecl: |
6585 | | * @ctxt: an XML parser context |
6586 | | * @name: the name of the element being defined. |
6587 | | * @result: the Element Content pointer will be stored here if any |
6588 | | * |
6589 | | * DEPRECATED: Internal function, don't use. |
6590 | | * |
6591 | | * parse the declaration for an Element content either Mixed or Children, |
6592 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6593 | | * |
6594 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6595 | | * |
6596 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6597 | | */ |
6598 | | |
6599 | | int |
6600 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6601 | 1.62M | xmlElementContentPtr *result) { |
6602 | | |
6603 | 1.62M | xmlElementContentPtr tree = NULL; |
6604 | 1.62M | int inputid = ctxt->input->id; |
6605 | 1.62M | int res; |
6606 | | |
6607 | 1.62M | *result = NULL; |
6608 | | |
6609 | 1.62M | if (RAW != '(') { |
6610 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6611 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6612 | 0 | return(-1); |
6613 | 0 | } |
6614 | 1.62M | NEXT; |
6615 | 1.62M | GROW; |
6616 | 1.62M | if (ctxt->instate == XML_PARSER_EOF) |
6617 | 0 | return(-1); |
6618 | 1.62M | SKIP_BLANKS; |
6619 | 1.62M | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6620 | 607k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6621 | 607k | res = XML_ELEMENT_TYPE_MIXED; |
6622 | 1.01M | } else { |
6623 | 1.01M | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6624 | 1.01M | res = XML_ELEMENT_TYPE_ELEMENT; |
6625 | 1.01M | } |
6626 | 1.62M | SKIP_BLANKS; |
6627 | 1.62M | *result = tree; |
6628 | 1.62M | return(res); |
6629 | 1.62M | } |
6630 | | |
6631 | | /** |
6632 | | * xmlParseElementDecl: |
6633 | | * @ctxt: an XML parser context |
6634 | | * |
6635 | | * DEPRECATED: Internal function, don't use. |
6636 | | * |
6637 | | * parse an Element declaration. |
6638 | | * |
6639 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6640 | | * |
6641 | | * [ VC: Unique Element Type Declaration ] |
6642 | | * No element type may be declared more than once |
6643 | | * |
6644 | | * Returns the type of the element, or -1 in case of error |
6645 | | */ |
6646 | | int |
6647 | 2.22M | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6648 | 2.22M | const xmlChar *name; |
6649 | 2.22M | int ret = -1; |
6650 | 2.22M | xmlElementContentPtr content = NULL; |
6651 | | |
6652 | | /* GROW; done in the caller */ |
6653 | 2.22M | if (CMP9(CUR_PTR, '<', '!', 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6654 | 2.21M | int inputid = ctxt->input->id; |
6655 | | |
6656 | 2.21M | SKIP(9); |
6657 | 2.21M | if (SKIP_BLANKS == 0) { |
6658 | 3.21k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6659 | 3.21k | "Space required after 'ELEMENT'\n"); |
6660 | 3.21k | return(-1); |
6661 | 3.21k | } |
6662 | 2.21M | name = xmlParseName(ctxt); |
6663 | 2.21M | if (name == NULL) { |
6664 | 13.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6665 | 13.2k | "xmlParseElementDecl: no name for Element\n"); |
6666 | 13.2k | return(-1); |
6667 | 13.2k | } |
6668 | 2.19M | if (SKIP_BLANKS == 0) { |
6669 | 36.9k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6670 | 36.9k | "Space required after the element name\n"); |
6671 | 36.9k | } |
6672 | 2.19M | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6673 | 526k | SKIP(5); |
6674 | | /* |
6675 | | * Element must always be empty. |
6676 | | */ |
6677 | 526k | ret = XML_ELEMENT_TYPE_EMPTY; |
6678 | 1.67M | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6679 | 1.67M | (NXT(2) == 'Y')) { |
6680 | 8.58k | SKIP(3); |
6681 | | /* |
6682 | | * Element is a generic container. |
6683 | | */ |
6684 | 8.58k | ret = XML_ELEMENT_TYPE_ANY; |
6685 | 1.66M | } else if (RAW == '(') { |
6686 | 1.62M | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6687 | 1.62M | } else { |
6688 | | /* |
6689 | | * [ WFC: PEs in Internal Subset ] error handling. |
6690 | | */ |
6691 | 42.0k | if ((RAW == '%') && (ctxt->external == 0) && |
6692 | 42.0k | (ctxt->inputNr == 1)) { |
6693 | 367 | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_IN_INT_SUBSET, |
6694 | 367 | "PEReference: forbidden within markup decl in internal subset\n"); |
6695 | 41.6k | } else { |
6696 | 41.6k | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6697 | 41.6k | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6698 | 41.6k | } |
6699 | 42.0k | return(-1); |
6700 | 42.0k | } |
6701 | | |
6702 | 2.15M | SKIP_BLANKS; |
6703 | | |
6704 | 2.15M | if (RAW != '>') { |
6705 | 64.3k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6706 | 64.3k | if (content != NULL) { |
6707 | 2.58k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6708 | 2.58k | } |
6709 | 2.09M | } else { |
6710 | 2.09M | if (inputid != ctxt->input->id) { |
6711 | 175 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6712 | 175 | "Element declaration doesn't start and stop in" |
6713 | 175 | " the same entity\n"); |
6714 | 175 | } |
6715 | | |
6716 | 2.09M | NEXT; |
6717 | 2.09M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6718 | 2.09M | (ctxt->sax->elementDecl != NULL)) { |
6719 | 1.90M | if (content != NULL) |
6720 | 1.37M | content->parent = NULL; |
6721 | 1.90M | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6722 | 1.90M | content); |
6723 | 1.90M | if ((content != NULL) && (content->parent == NULL)) { |
6724 | | /* |
6725 | | * this is a trick: if xmlAddElementDecl is called, |
6726 | | * instead of copying the full tree it is plugged directly |
6727 | | * if called from the parser. Avoid duplicating the |
6728 | | * interfaces or change the API/ABI |
6729 | | */ |
6730 | 162k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6731 | 162k | } |
6732 | 1.90M | } else if (content != NULL) { |
6733 | 147k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6734 | 147k | } |
6735 | 2.09M | } |
6736 | 2.15M | } |
6737 | 2.16M | return(ret); |
6738 | 2.22M | } |
6739 | | |
6740 | | /** |
6741 | | * xmlParseConditionalSections |
6742 | | * @ctxt: an XML parser context |
6743 | | * |
6744 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6745 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6746 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6747 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6748 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6749 | | */ |
6750 | | |
6751 | | static void |
6752 | 16.5k | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6753 | 16.5k | int *inputIds = NULL; |
6754 | 16.5k | size_t inputIdsSize = 0; |
6755 | 16.5k | size_t depth = 0; |
6756 | | |
6757 | 84.8k | while (ctxt->instate != XML_PARSER_EOF) { |
6758 | 84.6k | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6759 | 45.4k | int id = ctxt->input->id; |
6760 | | |
6761 | 45.4k | SKIP(3); |
6762 | 45.4k | SKIP_BLANKS; |
6763 | | |
6764 | 45.4k | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6765 | 36.7k | SKIP(7); |
6766 | 36.7k | SKIP_BLANKS; |
6767 | 36.7k | if (RAW != '[') { |
6768 | 331 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6769 | 331 | xmlHaltParser(ctxt); |
6770 | 331 | goto error; |
6771 | 331 | } |
6772 | 36.4k | if (ctxt->input->id != id) { |
6773 | 20 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6774 | 20 | "All markup of the conditional section is" |
6775 | 20 | " not in the same entity\n"); |
6776 | 20 | } |
6777 | 36.4k | NEXT; |
6778 | | |
6779 | 36.4k | if (inputIdsSize <= depth) { |
6780 | 11.2k | int *tmp; |
6781 | | |
6782 | 11.2k | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
6783 | 11.2k | tmp = (int *) xmlRealloc(inputIds, |
6784 | 11.2k | inputIdsSize * sizeof(int)); |
6785 | 11.2k | if (tmp == NULL) { |
6786 | 0 | xmlErrMemory(ctxt, NULL); |
6787 | 0 | goto error; |
6788 | 0 | } |
6789 | 11.2k | inputIds = tmp; |
6790 | 11.2k | } |
6791 | 36.4k | inputIds[depth] = id; |
6792 | 36.4k | depth++; |
6793 | 36.4k | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6794 | 6.30k | int state; |
6795 | 6.30k | xmlParserInputState instate; |
6796 | 6.30k | size_t ignoreDepth = 0; |
6797 | | |
6798 | 6.30k | SKIP(6); |
6799 | 6.30k | SKIP_BLANKS; |
6800 | 6.30k | if (RAW != '[') { |
6801 | 153 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6802 | 153 | xmlHaltParser(ctxt); |
6803 | 153 | goto error; |
6804 | 153 | } |
6805 | 6.15k | if (ctxt->input->id != id) { |
6806 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6807 | 0 | "All markup of the conditional section is" |
6808 | 0 | " not in the same entity\n"); |
6809 | 0 | } |
6810 | 6.15k | NEXT; |
6811 | | |
6812 | | /* |
6813 | | * Parse up to the end of the conditional section but disable |
6814 | | * SAX event generating DTD building in the meantime |
6815 | | */ |
6816 | 6.15k | state = ctxt->disableSAX; |
6817 | 6.15k | instate = ctxt->instate; |
6818 | 6.15k | if (ctxt->recovery == 0) ctxt->disableSAX = 1; |
6819 | 6.15k | ctxt->instate = XML_PARSER_IGNORE; |
6820 | | |
6821 | 8.69M | while (RAW != 0) { |
6822 | 8.69M | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6823 | 14.5k | SKIP(3); |
6824 | 14.5k | ignoreDepth++; |
6825 | | /* Check for integer overflow */ |
6826 | 14.5k | if (ignoreDepth == 0) { |
6827 | 0 | xmlErrMemory(ctxt, NULL); |
6828 | 0 | goto error; |
6829 | 0 | } |
6830 | 8.67M | } else if ((RAW == ']') && (NXT(1) == ']') && |
6831 | 8.67M | (NXT(2) == '>')) { |
6832 | 13.3k | if (ignoreDepth == 0) |
6833 | 3.26k | break; |
6834 | 10.0k | SKIP(3); |
6835 | 10.0k | ignoreDepth--; |
6836 | 8.66M | } else { |
6837 | 8.66M | NEXT; |
6838 | 8.66M | } |
6839 | 8.69M | } |
6840 | | |
6841 | 6.15k | ctxt->disableSAX = state; |
6842 | 6.15k | ctxt->instate = instate; |
6843 | | |
6844 | 6.15k | if (RAW == 0) { |
6845 | 2.88k | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6846 | 2.88k | goto error; |
6847 | 2.88k | } |
6848 | 3.26k | if (ctxt->input->id != id) { |
6849 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6850 | 0 | "All markup of the conditional section is" |
6851 | 0 | " not in the same entity\n"); |
6852 | 0 | } |
6853 | 3.26k | SKIP(3); |
6854 | 3.26k | } else { |
6855 | 2.38k | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6856 | 2.38k | xmlHaltParser(ctxt); |
6857 | 2.38k | goto error; |
6858 | 2.38k | } |
6859 | 45.4k | } else if ((depth > 0) && |
6860 | 39.2k | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6861 | 19.4k | depth--; |
6862 | 19.4k | if (ctxt->input->id != inputIds[depth]) { |
6863 | 251 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6864 | 251 | "All markup of the conditional section is not" |
6865 | 251 | " in the same entity\n"); |
6866 | 251 | } |
6867 | 19.4k | SKIP(3); |
6868 | 19.7k | } else { |
6869 | 19.7k | int id = ctxt->input->id; |
6870 | 19.7k | unsigned long cons = CUR_CONSUMED; |
6871 | | |
6872 | 19.7k | xmlParseMarkupDecl(ctxt); |
6873 | | |
6874 | 19.7k | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
6875 | 3.54k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6876 | 3.54k | xmlHaltParser(ctxt); |
6877 | 3.54k | goto error; |
6878 | 3.54k | } |
6879 | 19.7k | } |
6880 | | |
6881 | 75.3k | if (depth == 0) |
6882 | 7.02k | break; |
6883 | | |
6884 | 68.3k | SKIP_BLANKS; |
6885 | 68.3k | GROW; |
6886 | 68.3k | } |
6887 | | |
6888 | 16.5k | error: |
6889 | 16.5k | xmlFree(inputIds); |
6890 | 16.5k | } |
6891 | | |
6892 | | /** |
6893 | | * xmlParseMarkupDecl: |
6894 | | * @ctxt: an XML parser context |
6895 | | * |
6896 | | * DEPRECATED: Internal function, don't use. |
6897 | | * |
6898 | | * parse Markup declarations |
6899 | | * |
6900 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6901 | | * NotationDecl | PI | Comment |
6902 | | * |
6903 | | * [ VC: Proper Declaration/PE Nesting ] |
6904 | | * Parameter-entity replacement text must be properly nested with |
6905 | | * markup declarations. That is to say, if either the first character |
6906 | | * or the last character of a markup declaration (markupdecl above) is |
6907 | | * contained in the replacement text for a parameter-entity reference, |
6908 | | * both must be contained in the same replacement text. |
6909 | | * |
6910 | | * [ WFC: PEs in Internal Subset ] |
6911 | | * In the internal DTD subset, parameter-entity references can occur |
6912 | | * only where markup declarations can occur, not within markup declarations. |
6913 | | * (This does not apply to references that occur in external parameter |
6914 | | * entities or to the external subset.) |
6915 | | */ |
6916 | | void |
6917 | 8.38M | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
6918 | 8.38M | GROW; |
6919 | 8.38M | if (CUR == '<') { |
6920 | 8.05M | if (NXT(1) == '!') { |
6921 | 8.01M | switch (NXT(2)) { |
6922 | 4.65M | case 'E': |
6923 | 4.65M | if (NXT(3) == 'L') |
6924 | 2.22M | xmlParseElementDecl(ctxt); |
6925 | 2.42M | else if (NXT(3) == 'N') |
6926 | 2.42M | xmlParseEntityDecl(ctxt); |
6927 | 4.65M | break; |
6928 | 1.72M | case 'A': |
6929 | 1.72M | xmlParseAttributeListDecl(ctxt); |
6930 | 1.72M | break; |
6931 | 14.8k | case 'N': |
6932 | 14.8k | xmlParseNotationDecl(ctxt); |
6933 | 14.8k | break; |
6934 | 1.61M | case '-': |
6935 | 1.61M | xmlParseComment(ctxt); |
6936 | 1.61M | break; |
6937 | 6.44k | default: |
6938 | | /* there is an error but it will be detected later */ |
6939 | 6.44k | break; |
6940 | 8.01M | } |
6941 | 8.01M | } else if (NXT(1) == '?') { |
6942 | 16.8k | xmlParsePI(ctxt); |
6943 | 16.8k | } |
6944 | 8.05M | } |
6945 | | |
6946 | | /* |
6947 | | * detect requirement to exit there and act accordingly |
6948 | | * and avoid having instate overridden later on |
6949 | | */ |
6950 | 8.38M | if (ctxt->instate == XML_PARSER_EOF) |
6951 | 23.9k | return; |
6952 | | |
6953 | 8.36M | ctxt->instate = XML_PARSER_DTD; |
6954 | 8.36M | } |
6955 | | |
6956 | | /** |
6957 | | * xmlParseTextDecl: |
6958 | | * @ctxt: an XML parser context |
6959 | | * |
6960 | | * DEPRECATED: Internal function, don't use. |
6961 | | * |
6962 | | * parse an XML declaration header for external entities |
6963 | | * |
6964 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
6965 | | */ |
6966 | | |
6967 | | void |
6968 | 67.6k | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
6969 | 67.6k | xmlChar *version; |
6970 | 67.6k | const xmlChar *encoding; |
6971 | 67.6k | int oldstate; |
6972 | | |
6973 | | /* |
6974 | | * We know that '<?xml' is here. |
6975 | | */ |
6976 | 67.6k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
6977 | 67.3k | SKIP(5); |
6978 | 67.3k | } else { |
6979 | 222 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
6980 | 222 | return; |
6981 | 222 | } |
6982 | | |
6983 | | /* Avoid expansion of parameter entities when skipping blanks. */ |
6984 | 67.3k | oldstate = ctxt->instate; |
6985 | 67.3k | ctxt->instate = XML_PARSER_START; |
6986 | | |
6987 | 67.3k | if (SKIP_BLANKS == 0) { |
6988 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6989 | 0 | "Space needed after '<?xml'\n"); |
6990 | 0 | } |
6991 | | |
6992 | | /* |
6993 | | * We may have the VersionInfo here. |
6994 | | */ |
6995 | 67.3k | version = xmlParseVersionInfo(ctxt); |
6996 | 67.3k | if (version == NULL) |
6997 | 12.2k | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
6998 | 55.1k | else { |
6999 | 55.1k | if (SKIP_BLANKS == 0) { |
7000 | 1.49k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7001 | 1.49k | "Space needed here\n"); |
7002 | 1.49k | } |
7003 | 55.1k | } |
7004 | 67.3k | ctxt->input->version = version; |
7005 | | |
7006 | | /* |
7007 | | * We must have the encoding declaration |
7008 | | */ |
7009 | 67.3k | encoding = xmlParseEncodingDecl(ctxt); |
7010 | 67.3k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7011 | | /* |
7012 | | * The XML REC instructs us to stop parsing right here |
7013 | | */ |
7014 | 254 | ctxt->instate = oldstate; |
7015 | 254 | return; |
7016 | 254 | } |
7017 | 67.1k | if ((encoding == NULL) && (ctxt->errNo == XML_ERR_OK)) { |
7018 | 3.12k | xmlFatalErrMsg(ctxt, XML_ERR_MISSING_ENCODING, |
7019 | 3.12k | "Missing encoding in text declaration\n"); |
7020 | 3.12k | } |
7021 | | |
7022 | 67.1k | SKIP_BLANKS; |
7023 | 67.1k | if ((RAW == '?') && (NXT(1) == '>')) { |
7024 | 28.1k | SKIP(2); |
7025 | 39.0k | } else if (RAW == '>') { |
7026 | | /* Deprecated old WD ... */ |
7027 | 430 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7028 | 430 | NEXT; |
7029 | 38.5k | } else { |
7030 | 38.5k | int c; |
7031 | | |
7032 | 38.5k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7033 | 30.4M | while ((c = CUR) != 0) { |
7034 | 30.4M | NEXT; |
7035 | 30.4M | if (c == '>') |
7036 | 38.0k | break; |
7037 | 30.4M | } |
7038 | 38.5k | } |
7039 | | |
7040 | 67.1k | ctxt->instate = oldstate; |
7041 | 67.1k | } |
7042 | | |
7043 | | /** |
7044 | | * xmlParseExternalSubset: |
7045 | | * @ctxt: an XML parser context |
7046 | | * @ExternalID: the external identifier |
7047 | | * @SystemID: the system identifier (or URL) |
7048 | | * |
7049 | | * parse Markup declarations from an external subset |
7050 | | * |
7051 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7052 | | * |
7053 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
7054 | | */ |
7055 | | void |
7056 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
7057 | 68.5k | const xmlChar *SystemID) { |
7058 | 68.5k | xmlDetectSAX2(ctxt); |
7059 | 68.5k | GROW; |
7060 | | |
7061 | 68.5k | if ((ctxt->encoding == NULL) && |
7062 | 68.5k | (ctxt->input->end - ctxt->input->cur >= 4)) { |
7063 | 68.3k | xmlChar start[4]; |
7064 | 68.3k | xmlCharEncoding enc; |
7065 | | |
7066 | 68.3k | start[0] = RAW; |
7067 | 68.3k | start[1] = NXT(1); |
7068 | 68.3k | start[2] = NXT(2); |
7069 | 68.3k | start[3] = NXT(3); |
7070 | 68.3k | enc = xmlDetectCharEncoding(start, 4); |
7071 | 68.3k | if (enc != XML_CHAR_ENCODING_NONE) |
7072 | 10.4k | xmlSwitchEncoding(ctxt, enc); |
7073 | 68.3k | } |
7074 | | |
7075 | 68.5k | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7076 | 9.97k | xmlParseTextDecl(ctxt); |
7077 | 9.97k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7078 | | /* |
7079 | | * The XML REC instructs us to stop parsing right here |
7080 | | */ |
7081 | 182 | xmlHaltParser(ctxt); |
7082 | 182 | return; |
7083 | 182 | } |
7084 | 9.97k | } |
7085 | 68.3k | if (ctxt->myDoc == NULL) { |
7086 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7087 | 0 | if (ctxt->myDoc == NULL) { |
7088 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
7089 | 0 | return; |
7090 | 0 | } |
7091 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7092 | 0 | } |
7093 | 68.3k | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL)) |
7094 | 0 | xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID); |
7095 | | |
7096 | 68.3k | ctxt->instate = XML_PARSER_DTD; |
7097 | 68.3k | ctxt->external = 1; |
7098 | 68.3k | SKIP_BLANKS; |
7099 | 4.07M | while (((RAW == '<') && (NXT(1) == '?')) || |
7100 | 4.07M | ((RAW == '<') && (NXT(1) == '!')) || |
7101 | 4.07M | (RAW == '%')) { |
7102 | 4.01M | int id = ctxt->input->id; |
7103 | 4.01M | unsigned long cons = CUR_CONSUMED; |
7104 | | |
7105 | 4.01M | GROW; |
7106 | 4.01M | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7107 | 16.5k | xmlParseConditionalSections(ctxt); |
7108 | 16.5k | } else |
7109 | 4.00M | xmlParseMarkupDecl(ctxt); |
7110 | 4.01M | SKIP_BLANKS; |
7111 | | |
7112 | 4.01M | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
7113 | 9.71k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7114 | 9.71k | break; |
7115 | 9.71k | } |
7116 | 4.01M | } |
7117 | | |
7118 | 68.3k | if (RAW != 0) { |
7119 | 27.5k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7120 | 27.5k | } |
7121 | | |
7122 | 68.3k | } |
7123 | | |
7124 | | /** |
7125 | | * xmlParseReference: |
7126 | | * @ctxt: an XML parser context |
7127 | | * |
7128 | | * DEPRECATED: Internal function, don't use. |
7129 | | * |
7130 | | * parse and handle entity references in content, depending on the SAX |
7131 | | * interface, this may end-up in a call to character() if this is a |
7132 | | * CharRef, a predefined entity, if there is no reference() callback. |
7133 | | * or if the parser was asked to switch to that mode. |
7134 | | * |
7135 | | * [67] Reference ::= EntityRef | CharRef |
7136 | | */ |
7137 | | void |
7138 | 46.8M | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7139 | 46.8M | xmlEntityPtr ent; |
7140 | 46.8M | xmlChar *val; |
7141 | 46.8M | int was_checked; |
7142 | 46.8M | xmlNodePtr list = NULL; |
7143 | 46.8M | xmlParserErrors ret = XML_ERR_OK; |
7144 | | |
7145 | | |
7146 | 46.8M | if (RAW != '&') |
7147 | 0 | return; |
7148 | | |
7149 | | /* |
7150 | | * Simple case of a CharRef |
7151 | | */ |
7152 | 46.8M | if (NXT(1) == '#') { |
7153 | 9.75M | int i = 0; |
7154 | 9.75M | xmlChar out[16]; |
7155 | 9.75M | int hex = NXT(2); |
7156 | 9.75M | int value = xmlParseCharRef(ctxt); |
7157 | | |
7158 | 9.75M | if (value == 0) |
7159 | 810k | return; |
7160 | 8.94M | if (ctxt->charset != XML_CHAR_ENCODING_UTF8) { |
7161 | | /* |
7162 | | * So we are using non-UTF-8 buffers |
7163 | | * Check that the char fit on 8bits, if not |
7164 | | * generate a CharRef. |
7165 | | */ |
7166 | 7.10M | if (value <= 0xFF) { |
7167 | 6.98M | out[0] = value; |
7168 | 6.98M | out[1] = 0; |
7169 | 6.98M | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7170 | 6.98M | (!ctxt->disableSAX)) |
7171 | 494k | ctxt->sax->characters(ctxt->userData, out, 1); |
7172 | 6.98M | } else { |
7173 | 122k | if ((hex == 'x') || (hex == 'X')) |
7174 | 60.5k | snprintf((char *)out, sizeof(out), "#x%X", value); |
7175 | 61.8k | else |
7176 | 61.8k | snprintf((char *)out, sizeof(out), "#%d", value); |
7177 | 122k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7178 | 122k | (!ctxt->disableSAX)) |
7179 | 16.5k | ctxt->sax->reference(ctxt->userData, out); |
7180 | 122k | } |
7181 | 7.10M | } else { |
7182 | | /* |
7183 | | * Just encode the value in UTF-8 |
7184 | | */ |
7185 | 1.83M | COPY_BUF(0 ,out, i, value); |
7186 | 1.83M | out[i] = 0; |
7187 | 1.83M | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7188 | 1.83M | (!ctxt->disableSAX)) |
7189 | 178k | ctxt->sax->characters(ctxt->userData, out, i); |
7190 | 1.83M | } |
7191 | 8.94M | return; |
7192 | 9.75M | } |
7193 | | |
7194 | | /* |
7195 | | * We are seeing an entity reference |
7196 | | */ |
7197 | 37.0M | ent = xmlParseEntityRef(ctxt); |
7198 | 37.0M | if (ent == NULL) return; |
7199 | 11.1M | if (!ctxt->wellFormed) |
7200 | 5.82M | return; |
7201 | 5.31M | was_checked = ent->checked; |
7202 | | |
7203 | | /* special case of predefined entities */ |
7204 | 5.31M | if ((ent->name == NULL) || |
7205 | 5.31M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7206 | 501k | val = ent->content; |
7207 | 501k | if (val == NULL) return; |
7208 | | /* |
7209 | | * inline the entity. |
7210 | | */ |
7211 | 501k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7212 | 501k | (!ctxt->disableSAX)) |
7213 | 501k | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7214 | 501k | return; |
7215 | 501k | } |
7216 | | |
7217 | | /* |
7218 | | * The first reference to the entity trigger a parsing phase |
7219 | | * where the ent->children is filled with the result from |
7220 | | * the parsing. |
7221 | | * Note: external parsed entities will not be loaded, it is not |
7222 | | * required for a non-validating parser, unless the parsing option |
7223 | | * of validating, or substituting entities were given. Doing so is |
7224 | | * far more secure as the parser will only process data coming from |
7225 | | * the document entity by default. |
7226 | | */ |
7227 | 4.81M | if (((ent->checked == 0) || |
7228 | 4.81M | ((ent->children == NULL) && (ctxt->options & XML_PARSE_NOENT))) && |
7229 | 4.81M | ((ent->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY) || |
7230 | 4.60M | (ctxt->options & (XML_PARSE_NOENT | XML_PARSE_DTDVALID)))) { |
7231 | 4.60M | unsigned long oldnbent = ctxt->nbentities, diff; |
7232 | | |
7233 | | /* |
7234 | | * This is a bit hackish but this seems the best |
7235 | | * way to make sure both SAX and DOM entity support |
7236 | | * behaves okay. |
7237 | | */ |
7238 | 4.60M | void *user_data; |
7239 | 4.60M | if (ctxt->userData == ctxt) |
7240 | 4.60M | user_data = NULL; |
7241 | 0 | else |
7242 | 0 | user_data = ctxt->userData; |
7243 | | |
7244 | | /* |
7245 | | * Check that this entity is well formed |
7246 | | * 4.3.2: An internal general parsed entity is well-formed |
7247 | | * if its replacement text matches the production labeled |
7248 | | * content. |
7249 | | */ |
7250 | 4.60M | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7251 | 262k | ctxt->depth++; |
7252 | 262k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, ent->content, |
7253 | 262k | user_data, &list); |
7254 | 262k | ctxt->depth--; |
7255 | | |
7256 | 4.33M | } else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7257 | 4.33M | ctxt->depth++; |
7258 | 4.33M | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, ctxt->sax, |
7259 | 4.33M | user_data, ctxt->depth, ent->URI, |
7260 | 4.33M | ent->ExternalID, &list); |
7261 | 4.33M | ctxt->depth--; |
7262 | 4.33M | } else { |
7263 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7264 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7265 | 0 | "invalid entity type found\n", NULL); |
7266 | 0 | } |
7267 | | |
7268 | | /* |
7269 | | * Store the number of entities needing parsing for this entity |
7270 | | * content and do checkings |
7271 | | */ |
7272 | 4.60M | diff = ctxt->nbentities - oldnbent + 1; |
7273 | 4.60M | if (diff > INT_MAX / 2) |
7274 | 0 | diff = INT_MAX / 2; |
7275 | 4.60M | ent->checked = diff * 2; |
7276 | 4.60M | if ((ent->content != NULL) && (xmlStrchr(ent->content, '<'))) |
7277 | 91.0k | ent->checked |= 1; |
7278 | 4.60M | if (ret == XML_ERR_ENTITY_LOOP) { |
7279 | 781k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7280 | 781k | xmlHaltParser(ctxt); |
7281 | 781k | xmlFreeNodeList(list); |
7282 | 781k | return; |
7283 | 781k | } |
7284 | 3.82M | if (xmlParserEntityCheck(ctxt, 0, ent, 0)) { |
7285 | 2.09k | xmlFreeNodeList(list); |
7286 | 2.09k | return; |
7287 | 2.09k | } |
7288 | | |
7289 | 3.81M | if ((ret == XML_ERR_OK) && (list != NULL)) { |
7290 | 41.5k | if (((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7291 | 41.5k | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY))&& |
7292 | 41.5k | (ent->children == NULL)) { |
7293 | 41.1k | ent->children = list; |
7294 | | /* |
7295 | | * Prune it directly in the generated document |
7296 | | * except for single text nodes. |
7297 | | */ |
7298 | 41.1k | if ((ctxt->replaceEntities == 0) || |
7299 | 41.1k | (ctxt->parseMode == XML_PARSE_READER) || |
7300 | 41.1k | ((list->type == XML_TEXT_NODE) && |
7301 | 37.7k | (list->next == NULL))) { |
7302 | 37.7k | ent->owner = 1; |
7303 | 91.2k | while (list != NULL) { |
7304 | 53.4k | list->parent = (xmlNodePtr) ent; |
7305 | 53.4k | if (list->doc != ent->doc) |
7306 | 0 | xmlSetTreeDoc(list, ent->doc); |
7307 | 53.4k | if (list->next == NULL) |
7308 | 37.7k | ent->last = list; |
7309 | 53.4k | list = list->next; |
7310 | 53.4k | } |
7311 | 37.7k | list = NULL; |
7312 | 37.7k | } else { |
7313 | 3.40k | ent->owner = 0; |
7314 | 12.3k | while (list != NULL) { |
7315 | 8.96k | list->parent = (xmlNodePtr) ctxt->node; |
7316 | 8.96k | list->doc = ctxt->myDoc; |
7317 | 8.96k | if (list->next == NULL) |
7318 | 3.40k | ent->last = list; |
7319 | 8.96k | list = list->next; |
7320 | 8.96k | } |
7321 | 3.40k | list = ent->children; |
7322 | | #ifdef LIBXML_LEGACY_ENABLED |
7323 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7324 | | xmlAddEntityReference(ent, list, NULL); |
7325 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7326 | 3.40k | } |
7327 | 41.1k | } else { |
7328 | 383 | xmlFreeNodeList(list); |
7329 | 383 | list = NULL; |
7330 | 383 | } |
7331 | 3.77M | } else if ((ret != XML_ERR_OK) && |
7332 | 3.77M | (ret != XML_WAR_UNDECLARED_ENTITY)) { |
7333 | 3.70M | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7334 | 3.70M | "Entity '%s' failed to parse\n", ent->name); |
7335 | 3.70M | if (ent->content != NULL) |
7336 | 83.5k | ent->content[0] = 0; |
7337 | 3.70M | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7338 | 3.70M | } else if (list != NULL) { |
7339 | 0 | xmlFreeNodeList(list); |
7340 | 0 | list = NULL; |
7341 | 0 | } |
7342 | 3.81M | if (ent->checked == 0) |
7343 | 0 | ent->checked = 2; |
7344 | | |
7345 | | /* Prevent entity from being parsed and expanded twice (Bug 760367). */ |
7346 | 3.81M | was_checked = 0; |
7347 | 3.81M | } else if (ent->checked != 1) { |
7348 | 214k | ctxt->nbentities += ent->checked / 2; |
7349 | 214k | } |
7350 | | |
7351 | | /* |
7352 | | * Now that the entity content has been gathered |
7353 | | * provide it to the application, this can take different forms based |
7354 | | * on the parsing modes. |
7355 | | */ |
7356 | 4.03M | if (ent->children == NULL) { |
7357 | | /* |
7358 | | * Probably running in SAX mode and the callbacks don't |
7359 | | * build the entity content. So unless we already went |
7360 | | * though parsing for first checking go though the entity |
7361 | | * content to generate callbacks associated to the entity |
7362 | | */ |
7363 | 3.88M | if (was_checked != 0) { |
7364 | 112k | void *user_data; |
7365 | | /* |
7366 | | * This is a bit hackish but this seems the best |
7367 | | * way to make sure both SAX and DOM entity support |
7368 | | * behaves okay. |
7369 | | */ |
7370 | 112k | if (ctxt->userData == ctxt) |
7371 | 112k | user_data = NULL; |
7372 | 0 | else |
7373 | 0 | user_data = ctxt->userData; |
7374 | | |
7375 | 112k | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7376 | 30.8k | ctxt->depth++; |
7377 | 30.8k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, |
7378 | 30.8k | ent->content, user_data, NULL); |
7379 | 30.8k | ctxt->depth--; |
7380 | 81.9k | } else if (ent->etype == |
7381 | 81.9k | XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7382 | 81.9k | ctxt->depth++; |
7383 | 81.9k | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, |
7384 | 81.9k | ctxt->sax, user_data, ctxt->depth, |
7385 | 81.9k | ent->URI, ent->ExternalID, NULL); |
7386 | 81.9k | ctxt->depth--; |
7387 | 81.9k | } else { |
7388 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7389 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7390 | 0 | "invalid entity type found\n", NULL); |
7391 | 0 | } |
7392 | 112k | if (ret == XML_ERR_ENTITY_LOOP) { |
7393 | 324 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7394 | 324 | return; |
7395 | 324 | } |
7396 | 112k | } |
7397 | 3.88M | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7398 | 3.88M | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7399 | | /* |
7400 | | * Entity reference callback comes second, it's somewhat |
7401 | | * superfluous but a compatibility to historical behaviour |
7402 | | */ |
7403 | 137k | ctxt->sax->reference(ctxt->userData, ent->name); |
7404 | 137k | } |
7405 | 3.88M | return; |
7406 | 3.88M | } |
7407 | | |
7408 | | /* |
7409 | | * If we didn't get any children for the entity being built |
7410 | | */ |
7411 | 147k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7412 | 147k | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7413 | | /* |
7414 | | * Create a node. |
7415 | | */ |
7416 | 96.7k | ctxt->sax->reference(ctxt->userData, ent->name); |
7417 | 96.7k | return; |
7418 | 96.7k | } |
7419 | | |
7420 | 51.1k | if ((ctxt->replaceEntities) || (ent->children == NULL)) { |
7421 | | /* |
7422 | | * There is a problem on the handling of _private for entities |
7423 | | * (bug 155816): Should we copy the content of the field from |
7424 | | * the entity (possibly overwriting some value set by the user |
7425 | | * when a copy is created), should we leave it alone, or should |
7426 | | * we try to take care of different situations? The problem |
7427 | | * is exacerbated by the usage of this field by the xmlReader. |
7428 | | * To fix this bug, we look at _private on the created node |
7429 | | * and, if it's NULL, we copy in whatever was in the entity. |
7430 | | * If it's not NULL we leave it alone. This is somewhat of a |
7431 | | * hack - maybe we should have further tests to determine |
7432 | | * what to do. |
7433 | | */ |
7434 | 41.3k | if ((ctxt->node != NULL) && (ent->children != NULL)) { |
7435 | | /* |
7436 | | * Seems we are generating the DOM content, do |
7437 | | * a simple tree copy for all references except the first |
7438 | | * In the first occurrence list contains the replacement. |
7439 | | */ |
7440 | 41.3k | if (((list == NULL) && (ent->owner == 0)) || |
7441 | 41.3k | (ctxt->parseMode == XML_PARSE_READER)) { |
7442 | 13.1k | xmlNodePtr nw = NULL, cur, firstChild = NULL; |
7443 | | |
7444 | | /* |
7445 | | * We are copying here, make sure there is no abuse |
7446 | | */ |
7447 | 13.1k | ctxt->sizeentcopy += ent->length + 5; |
7448 | 13.1k | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7449 | 0 | return; |
7450 | | |
7451 | | /* |
7452 | | * when operating on a reader, the entities definitions |
7453 | | * are always owning the entities subtree. |
7454 | | if (ctxt->parseMode == XML_PARSE_READER) |
7455 | | ent->owner = 1; |
7456 | | */ |
7457 | | |
7458 | 13.1k | cur = ent->children; |
7459 | 17.6k | while (cur != NULL) { |
7460 | 17.6k | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7461 | 17.6k | if (nw != NULL) { |
7462 | 17.6k | if (nw->_private == NULL) |
7463 | 17.6k | nw->_private = cur->_private; |
7464 | 17.6k | if (firstChild == NULL){ |
7465 | 13.1k | firstChild = nw; |
7466 | 13.1k | } |
7467 | 17.6k | nw = xmlAddChild(ctxt->node, nw); |
7468 | 17.6k | } |
7469 | 17.6k | if (cur == ent->last) { |
7470 | | /* |
7471 | | * needed to detect some strange empty |
7472 | | * node cases in the reader tests |
7473 | | */ |
7474 | 13.1k | if ((ctxt->parseMode == XML_PARSE_READER) && |
7475 | 13.1k | (nw != NULL) && |
7476 | 13.1k | (nw->type == XML_ELEMENT_NODE) && |
7477 | 13.1k | (nw->children == NULL)) |
7478 | 854 | nw->extra = 1; |
7479 | | |
7480 | 13.1k | break; |
7481 | 13.1k | } |
7482 | 4.49k | cur = cur->next; |
7483 | 4.49k | } |
7484 | | #ifdef LIBXML_LEGACY_ENABLED |
7485 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7486 | | xmlAddEntityReference(ent, firstChild, nw); |
7487 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7488 | 28.2k | } else if ((list == NULL) || (ctxt->inputNr > 0)) { |
7489 | 28.2k | xmlNodePtr nw = NULL, cur, next, last, |
7490 | 28.2k | firstChild = NULL; |
7491 | | |
7492 | | /* |
7493 | | * We are copying here, make sure there is no abuse |
7494 | | */ |
7495 | 28.2k | ctxt->sizeentcopy += ent->length + 5; |
7496 | 28.2k | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7497 | 0 | return; |
7498 | | |
7499 | | /* |
7500 | | * Copy the entity child list and make it the new |
7501 | | * entity child list. The goal is to make sure any |
7502 | | * ID or REF referenced will be the one from the |
7503 | | * document content and not the entity copy. |
7504 | | */ |
7505 | 28.2k | cur = ent->children; |
7506 | 28.2k | ent->children = NULL; |
7507 | 28.2k | last = ent->last; |
7508 | 28.2k | ent->last = NULL; |
7509 | 36.4k | while (cur != NULL) { |
7510 | 36.4k | next = cur->next; |
7511 | 36.4k | cur->next = NULL; |
7512 | 36.4k | cur->parent = NULL; |
7513 | 36.4k | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7514 | 36.4k | if (nw != NULL) { |
7515 | 36.4k | if (nw->_private == NULL) |
7516 | 36.4k | nw->_private = cur->_private; |
7517 | 36.4k | if (firstChild == NULL){ |
7518 | 28.2k | firstChild = cur; |
7519 | 28.2k | } |
7520 | 36.4k | xmlAddChild((xmlNodePtr) ent, nw); |
7521 | 36.4k | xmlAddChild(ctxt->node, cur); |
7522 | 36.4k | } |
7523 | 36.4k | if (cur == last) |
7524 | 28.2k | break; |
7525 | 8.24k | cur = next; |
7526 | 8.24k | } |
7527 | 28.2k | if (ent->owner == 0) |
7528 | 3.40k | ent->owner = 1; |
7529 | | #ifdef LIBXML_LEGACY_ENABLED |
7530 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7531 | | xmlAddEntityReference(ent, firstChild, nw); |
7532 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7533 | 28.2k | } else { |
7534 | 0 | const xmlChar *nbktext; |
7535 | | |
7536 | | /* |
7537 | | * the name change is to avoid coalescing of the |
7538 | | * node with a possible previous text one which |
7539 | | * would make ent->children a dangling pointer |
7540 | | */ |
7541 | 0 | nbktext = xmlDictLookup(ctxt->dict, BAD_CAST "nbktext", |
7542 | 0 | -1); |
7543 | 0 | if (ent->children->type == XML_TEXT_NODE) |
7544 | 0 | ent->children->name = nbktext; |
7545 | 0 | if ((ent->last != ent->children) && |
7546 | 0 | (ent->last->type == XML_TEXT_NODE)) |
7547 | 0 | ent->last->name = nbktext; |
7548 | 0 | xmlAddChildList(ctxt->node, ent->children); |
7549 | 0 | } |
7550 | | |
7551 | | /* |
7552 | | * This is to avoid a nasty side effect, see |
7553 | | * characters() in SAX.c |
7554 | | */ |
7555 | 41.3k | ctxt->nodemem = 0; |
7556 | 41.3k | ctxt->nodelen = 0; |
7557 | 41.3k | return; |
7558 | 41.3k | } |
7559 | 41.3k | } |
7560 | 51.1k | } |
7561 | | |
7562 | | /** |
7563 | | * xmlParseEntityRef: |
7564 | | * @ctxt: an XML parser context |
7565 | | * |
7566 | | * DEPRECATED: Internal function, don't use. |
7567 | | * |
7568 | | * parse ENTITY references declarations |
7569 | | * |
7570 | | * [68] EntityRef ::= '&' Name ';' |
7571 | | * |
7572 | | * [ WFC: Entity Declared ] |
7573 | | * In a document without any DTD, a document with only an internal DTD |
7574 | | * subset which contains no parameter entity references, or a document |
7575 | | * with "standalone='yes'", the Name given in the entity reference |
7576 | | * must match that in an entity declaration, except that well-formed |
7577 | | * documents need not declare any of the following entities: amp, lt, |
7578 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7579 | | * any reference to it. Similarly, the declaration of a general entity |
7580 | | * must precede any reference to it which appears in a default value in an |
7581 | | * attribute-list declaration. Note that if entities are declared in the |
7582 | | * external subset or in external parameter entities, a non-validating |
7583 | | * processor is not obligated to read and process their declarations; |
7584 | | * for such documents, the rule that an entity must be declared is a |
7585 | | * well-formedness constraint only if standalone='yes'. |
7586 | | * |
7587 | | * [ WFC: Parsed Entity ] |
7588 | | * An entity reference must not contain the name of an unparsed entity |
7589 | | * |
7590 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7591 | | */ |
7592 | | xmlEntityPtr |
7593 | 43.4M | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7594 | 43.4M | const xmlChar *name; |
7595 | 43.4M | xmlEntityPtr ent = NULL; |
7596 | | |
7597 | 43.4M | GROW; |
7598 | 43.4M | if (ctxt->instate == XML_PARSER_EOF) |
7599 | 0 | return(NULL); |
7600 | | |
7601 | 43.4M | if (RAW != '&') |
7602 | 0 | return(NULL); |
7603 | 43.4M | NEXT; |
7604 | 43.4M | name = xmlParseName(ctxt); |
7605 | 43.4M | if (name == NULL) { |
7606 | 24.5M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7607 | 24.5M | "xmlParseEntityRef: no name\n"); |
7608 | 24.5M | return(NULL); |
7609 | 24.5M | } |
7610 | 18.8M | if (RAW != ';') { |
7611 | 2.22M | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7612 | 2.22M | return(NULL); |
7613 | 2.22M | } |
7614 | 16.6M | NEXT; |
7615 | | |
7616 | | /* |
7617 | | * Predefined entities override any extra definition |
7618 | | */ |
7619 | 16.6M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7620 | 8.52M | ent = xmlGetPredefinedEntity(name); |
7621 | 8.52M | if (ent != NULL) |
7622 | 2.71M | return(ent); |
7623 | 8.52M | } |
7624 | | |
7625 | | /* |
7626 | | * Increase the number of entity references parsed |
7627 | | */ |
7628 | 13.9M | ctxt->nbentities++; |
7629 | | |
7630 | | /* |
7631 | | * Ask first SAX for entity resolution, otherwise try the |
7632 | | * entities which may have stored in the parser context. |
7633 | | */ |
7634 | 13.9M | if (ctxt->sax != NULL) { |
7635 | 13.9M | if (ctxt->sax->getEntity != NULL) |
7636 | 13.9M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7637 | 13.9M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7638 | 13.9M | (ctxt->options & XML_PARSE_OLDSAX)) |
7639 | 7.57k | ent = xmlGetPredefinedEntity(name); |
7640 | 13.9M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7641 | 13.9M | (ctxt->userData==ctxt)) { |
7642 | 22.2k | ent = xmlSAX2GetEntity(ctxt, name); |
7643 | 22.2k | } |
7644 | 13.9M | } |
7645 | 13.9M | if (ctxt->instate == XML_PARSER_EOF) |
7646 | 0 | return(NULL); |
7647 | | /* |
7648 | | * [ WFC: Entity Declared ] |
7649 | | * In a document without any DTD, a document with only an |
7650 | | * internal DTD subset which contains no parameter entity |
7651 | | * references, or a document with "standalone='yes'", the |
7652 | | * Name given in the entity reference must match that in an |
7653 | | * entity declaration, except that well-formed documents |
7654 | | * need not declare any of the following entities: amp, lt, |
7655 | | * gt, apos, quot. |
7656 | | * The declaration of a parameter entity must precede any |
7657 | | * reference to it. |
7658 | | * Similarly, the declaration of a general entity must |
7659 | | * precede any reference to it which appears in a default |
7660 | | * value in an attribute-list declaration. Note that if |
7661 | | * entities are declared in the external subset or in |
7662 | | * external parameter entities, a non-validating processor |
7663 | | * is not obligated to read and process their declarations; |
7664 | | * for such documents, the rule that an entity must be |
7665 | | * declared is a well-formedness constraint only if |
7666 | | * standalone='yes'. |
7667 | | */ |
7668 | 13.9M | if (ent == NULL) { |
7669 | 3.32M | if ((ctxt->standalone == 1) || |
7670 | 3.32M | ((ctxt->hasExternalSubset == 0) && |
7671 | 3.32M | (ctxt->hasPErefs == 0))) { |
7672 | 3.24M | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7673 | 3.24M | "Entity '%s' not defined\n", name); |
7674 | 3.24M | } else { |
7675 | 79.8k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7676 | 79.8k | "Entity '%s' not defined\n", name); |
7677 | 79.8k | if ((ctxt->inSubset == 0) && |
7678 | 79.8k | (ctxt->sax != NULL) && |
7679 | 79.8k | (ctxt->sax->reference != NULL)) { |
7680 | 79.3k | ctxt->sax->reference(ctxt->userData, name); |
7681 | 79.3k | } |
7682 | 79.8k | } |
7683 | 3.32M | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7684 | 3.32M | ctxt->valid = 0; |
7685 | 3.32M | } |
7686 | | |
7687 | | /* |
7688 | | * [ WFC: Parsed Entity ] |
7689 | | * An entity reference must not contain the name of an |
7690 | | * unparsed entity |
7691 | | */ |
7692 | 10.6M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7693 | 5.39k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7694 | 5.39k | "Entity reference to unparsed entity %s\n", name); |
7695 | 5.39k | } |
7696 | | |
7697 | | /* |
7698 | | * [ WFC: No External Entity References ] |
7699 | | * Attribute values cannot contain direct or indirect |
7700 | | * entity references to external entities. |
7701 | | */ |
7702 | 10.6M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7703 | 10.6M | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7704 | 51.6k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7705 | 51.6k | "Attribute references external entity '%s'\n", name); |
7706 | 51.6k | } |
7707 | | /* |
7708 | | * [ WFC: No < in Attribute Values ] |
7709 | | * The replacement text of any entity referred to directly or |
7710 | | * indirectly in an attribute value (other than "<") must |
7711 | | * not contain a <. |
7712 | | */ |
7713 | 10.5M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7714 | 10.5M | (ent != NULL) && |
7715 | 10.5M | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
7716 | 207k | if (((ent->checked & 1) || (ent->checked == 0)) && |
7717 | 207k | (ent->content != NULL) && (xmlStrchr(ent->content, '<'))) { |
7718 | 5.37k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7719 | 5.37k | "'<' in entity '%s' is not allowed in attributes values\n", name); |
7720 | 5.37k | } |
7721 | 207k | } |
7722 | | |
7723 | | /* |
7724 | | * Internal check, no parameter entities here ... |
7725 | | */ |
7726 | 10.3M | else { |
7727 | 10.3M | switch (ent->etype) { |
7728 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7729 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7730 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7731 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7732 | 0 | name); |
7733 | 0 | break; |
7734 | 10.3M | default: |
7735 | 10.3M | break; |
7736 | 10.3M | } |
7737 | 10.3M | } |
7738 | | |
7739 | | /* |
7740 | | * [ WFC: No Recursion ] |
7741 | | * A parsed entity must not contain a recursive reference |
7742 | | * to itself, either directly or indirectly. |
7743 | | * Done somewhere else |
7744 | | */ |
7745 | 13.9M | return(ent); |
7746 | 13.9M | } |
7747 | | |
7748 | | /** |
7749 | | * xmlParseStringEntityRef: |
7750 | | * @ctxt: an XML parser context |
7751 | | * @str: a pointer to an index in the string |
7752 | | * |
7753 | | * parse ENTITY references declarations, but this version parses it from |
7754 | | * a string value. |
7755 | | * |
7756 | | * [68] EntityRef ::= '&' Name ';' |
7757 | | * |
7758 | | * [ WFC: Entity Declared ] |
7759 | | * In a document without any DTD, a document with only an internal DTD |
7760 | | * subset which contains no parameter entity references, or a document |
7761 | | * with "standalone='yes'", the Name given in the entity reference |
7762 | | * must match that in an entity declaration, except that well-formed |
7763 | | * documents need not declare any of the following entities: amp, lt, |
7764 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7765 | | * any reference to it. Similarly, the declaration of a general entity |
7766 | | * must precede any reference to it which appears in a default value in an |
7767 | | * attribute-list declaration. Note that if entities are declared in the |
7768 | | * external subset or in external parameter entities, a non-validating |
7769 | | * processor is not obligated to read and process their declarations; |
7770 | | * for such documents, the rule that an entity must be declared is a |
7771 | | * well-formedness constraint only if standalone='yes'. |
7772 | | * |
7773 | | * [ WFC: Parsed Entity ] |
7774 | | * An entity reference must not contain the name of an unparsed entity |
7775 | | * |
7776 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7777 | | * is updated to the current location in the string. |
7778 | | */ |
7779 | | static xmlEntityPtr |
7780 | 921k | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7781 | 921k | xmlChar *name; |
7782 | 921k | const xmlChar *ptr; |
7783 | 921k | xmlChar cur; |
7784 | 921k | xmlEntityPtr ent = NULL; |
7785 | | |
7786 | 921k | if ((str == NULL) || (*str == NULL)) |
7787 | 0 | return(NULL); |
7788 | 921k | ptr = *str; |
7789 | 921k | cur = *ptr; |
7790 | 921k | if (cur != '&') |
7791 | 0 | return(NULL); |
7792 | | |
7793 | 921k | ptr++; |
7794 | 921k | name = xmlParseStringName(ctxt, &ptr); |
7795 | 921k | if (name == NULL) { |
7796 | 44.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7797 | 44.2k | "xmlParseStringEntityRef: no name\n"); |
7798 | 44.2k | *str = ptr; |
7799 | 44.2k | return(NULL); |
7800 | 44.2k | } |
7801 | 877k | if (*ptr != ';') { |
7802 | 30.7k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7803 | 30.7k | xmlFree(name); |
7804 | 30.7k | *str = ptr; |
7805 | 30.7k | return(NULL); |
7806 | 30.7k | } |
7807 | 846k | ptr++; |
7808 | | |
7809 | | |
7810 | | /* |
7811 | | * Predefined entities override any extra definition |
7812 | | */ |
7813 | 846k | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7814 | 366k | ent = xmlGetPredefinedEntity(name); |
7815 | 366k | if (ent != NULL) { |
7816 | 38.9k | xmlFree(name); |
7817 | 38.9k | *str = ptr; |
7818 | 38.9k | return(ent); |
7819 | 38.9k | } |
7820 | 366k | } |
7821 | | |
7822 | | /* |
7823 | | * Increase the number of entity references parsed |
7824 | | */ |
7825 | 807k | ctxt->nbentities++; |
7826 | | |
7827 | | /* |
7828 | | * Ask first SAX for entity resolution, otherwise try the |
7829 | | * entities which may have stored in the parser context. |
7830 | | */ |
7831 | 807k | if (ctxt->sax != NULL) { |
7832 | 807k | if (ctxt->sax->getEntity != NULL) |
7833 | 807k | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7834 | 807k | if ((ent == NULL) && (ctxt->options & XML_PARSE_OLDSAX)) |
7835 | 37.8k | ent = xmlGetPredefinedEntity(name); |
7836 | 807k | if ((ent == NULL) && (ctxt->userData==ctxt)) { |
7837 | 59.6k | ent = xmlSAX2GetEntity(ctxt, name); |
7838 | 59.6k | } |
7839 | 807k | } |
7840 | 807k | if (ctxt->instate == XML_PARSER_EOF) { |
7841 | 0 | xmlFree(name); |
7842 | 0 | return(NULL); |
7843 | 0 | } |
7844 | | |
7845 | | /* |
7846 | | * [ WFC: Entity Declared ] |
7847 | | * In a document without any DTD, a document with only an |
7848 | | * internal DTD subset which contains no parameter entity |
7849 | | * references, or a document with "standalone='yes'", the |
7850 | | * Name given in the entity reference must match that in an |
7851 | | * entity declaration, except that well-formed documents |
7852 | | * need not declare any of the following entities: amp, lt, |
7853 | | * gt, apos, quot. |
7854 | | * The declaration of a parameter entity must precede any |
7855 | | * reference to it. |
7856 | | * Similarly, the declaration of a general entity must |
7857 | | * precede any reference to it which appears in a default |
7858 | | * value in an attribute-list declaration. Note that if |
7859 | | * entities are declared in the external subset or in |
7860 | | * external parameter entities, a non-validating processor |
7861 | | * is not obligated to read and process their declarations; |
7862 | | * for such documents, the rule that an entity must be |
7863 | | * declared is a well-formedness constraint only if |
7864 | | * standalone='yes'. |
7865 | | */ |
7866 | 807k | if (ent == NULL) { |
7867 | 59.6k | if ((ctxt->standalone == 1) || |
7868 | 59.6k | ((ctxt->hasExternalSubset == 0) && |
7869 | 57.6k | (ctxt->hasPErefs == 0))) { |
7870 | 53.9k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7871 | 53.9k | "Entity '%s' not defined\n", name); |
7872 | 53.9k | } else { |
7873 | 5.72k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7874 | 5.72k | "Entity '%s' not defined\n", |
7875 | 5.72k | name); |
7876 | 5.72k | } |
7877 | 59.6k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7878 | | /* TODO ? check regressions ctxt->valid = 0; */ |
7879 | 59.6k | } |
7880 | | |
7881 | | /* |
7882 | | * [ WFC: Parsed Entity ] |
7883 | | * An entity reference must not contain the name of an |
7884 | | * unparsed entity |
7885 | | */ |
7886 | 748k | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7887 | 4.64k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7888 | 4.64k | "Entity reference to unparsed entity %s\n", name); |
7889 | 4.64k | } |
7890 | | |
7891 | | /* |
7892 | | * [ WFC: No External Entity References ] |
7893 | | * Attribute values cannot contain direct or indirect |
7894 | | * entity references to external entities. |
7895 | | */ |
7896 | 743k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7897 | 743k | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7898 | 14.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7899 | 14.5k | "Attribute references external entity '%s'\n", name); |
7900 | 14.5k | } |
7901 | | /* |
7902 | | * [ WFC: No < in Attribute Values ] |
7903 | | * The replacement text of any entity referred to directly or |
7904 | | * indirectly in an attribute value (other than "<") must |
7905 | | * not contain a <. |
7906 | | */ |
7907 | 728k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7908 | 728k | (ent != NULL) && (ent->content != NULL) && |
7909 | 728k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
7910 | 728k | (xmlStrchr(ent->content, '<'))) { |
7911 | 163k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7912 | 163k | "'<' in entity '%s' is not allowed in attributes values\n", |
7913 | 163k | name); |
7914 | 163k | } |
7915 | | |
7916 | | /* |
7917 | | * Internal check, no parameter entities here ... |
7918 | | */ |
7919 | 565k | else { |
7920 | 565k | switch (ent->etype) { |
7921 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7922 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7923 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7924 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7925 | 0 | name); |
7926 | 0 | break; |
7927 | 565k | default: |
7928 | 565k | break; |
7929 | 565k | } |
7930 | 565k | } |
7931 | | |
7932 | | /* |
7933 | | * [ WFC: No Recursion ] |
7934 | | * A parsed entity must not contain a recursive reference |
7935 | | * to itself, either directly or indirectly. |
7936 | | * Done somewhere else |
7937 | | */ |
7938 | | |
7939 | 807k | xmlFree(name); |
7940 | 807k | *str = ptr; |
7941 | 807k | return(ent); |
7942 | 807k | } |
7943 | | |
7944 | | /** |
7945 | | * xmlParsePEReference: |
7946 | | * @ctxt: an XML parser context |
7947 | | * |
7948 | | * DEPRECATED: Internal function, don't use. |
7949 | | * |
7950 | | * parse PEReference declarations |
7951 | | * The entity content is handled directly by pushing it's content as |
7952 | | * a new input stream. |
7953 | | * |
7954 | | * [69] PEReference ::= '%' Name ';' |
7955 | | * |
7956 | | * [ WFC: No Recursion ] |
7957 | | * A parsed entity must not contain a recursive |
7958 | | * reference to itself, either directly or indirectly. |
7959 | | * |
7960 | | * [ WFC: Entity Declared ] |
7961 | | * In a document without any DTD, a document with only an internal DTD |
7962 | | * subset which contains no parameter entity references, or a document |
7963 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7964 | | * entity must precede any reference to it... |
7965 | | * |
7966 | | * [ VC: Entity Declared ] |
7967 | | * In a document with an external subset or external parameter entities |
7968 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7969 | | * must precede any reference to it... |
7970 | | * |
7971 | | * [ WFC: In DTD ] |
7972 | | * Parameter-entity references may only appear in the DTD. |
7973 | | * NOTE: misleading but this is handled. |
7974 | | */ |
7975 | | void |
7976 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7977 | 8.88M | { |
7978 | 8.88M | const xmlChar *name; |
7979 | 8.88M | xmlEntityPtr entity = NULL; |
7980 | 8.88M | xmlParserInputPtr input; |
7981 | | |
7982 | 8.88M | if (RAW != '%') |
7983 | 4.30M | return; |
7984 | 4.58M | NEXT; |
7985 | 4.58M | name = xmlParseName(ctxt); |
7986 | 4.58M | if (name == NULL) { |
7987 | 2.64M | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7988 | 2.64M | return; |
7989 | 2.64M | } |
7990 | 1.93M | if (xmlParserDebugEntities) |
7991 | 0 | xmlGenericError(xmlGenericErrorContext, |
7992 | 0 | "PEReference: %s\n", name); |
7993 | 1.93M | if (RAW != ';') { |
7994 | 19.5k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7995 | 19.5k | return; |
7996 | 19.5k | } |
7997 | | |
7998 | 1.91M | NEXT; |
7999 | | |
8000 | | /* |
8001 | | * Increase the number of entity references parsed |
8002 | | */ |
8003 | 1.91M | ctxt->nbentities++; |
8004 | | |
8005 | | /* |
8006 | | * Request the entity from SAX |
8007 | | */ |
8008 | 1.91M | if ((ctxt->sax != NULL) && |
8009 | 1.91M | (ctxt->sax->getParameterEntity != NULL)) |
8010 | 1.91M | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8011 | 1.91M | if (ctxt->instate == XML_PARSER_EOF) |
8012 | 0 | return; |
8013 | 1.91M | if (entity == NULL) { |
8014 | | /* |
8015 | | * [ WFC: Entity Declared ] |
8016 | | * In a document without any DTD, a document with only an |
8017 | | * internal DTD subset which contains no parameter entity |
8018 | | * references, or a document with "standalone='yes'", ... |
8019 | | * ... The declaration of a parameter entity must precede |
8020 | | * any reference to it... |
8021 | | */ |
8022 | 160k | if ((ctxt->standalone == 1) || |
8023 | 160k | ((ctxt->hasExternalSubset == 0) && |
8024 | 155k | (ctxt->hasPErefs == 0))) { |
8025 | 6.70k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8026 | 6.70k | "PEReference: %%%s; not found\n", |
8027 | 6.70k | name); |
8028 | 153k | } else { |
8029 | | /* |
8030 | | * [ VC: Entity Declared ] |
8031 | | * In a document with an external subset or external |
8032 | | * parameter entities with "standalone='no'", ... |
8033 | | * ... The declaration of a parameter entity must |
8034 | | * precede any reference to it... |
8035 | | */ |
8036 | 153k | if ((ctxt->validate) && (ctxt->vctxt.error != NULL)) { |
8037 | 19.4k | xmlValidityError(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8038 | 19.4k | "PEReference: %%%s; not found\n", |
8039 | 19.4k | name, NULL); |
8040 | 19.4k | } else |
8041 | 134k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8042 | 134k | "PEReference: %%%s; not found\n", |
8043 | 134k | name, NULL); |
8044 | 153k | ctxt->valid = 0; |
8045 | 153k | } |
8046 | 160k | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
8047 | 1.75M | } else { |
8048 | | /* |
8049 | | * Internal checking in case the entity quest barfed |
8050 | | */ |
8051 | 1.75M | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8052 | 1.75M | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8053 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8054 | 0 | "Internal: %%%s; is not a parameter entity\n", |
8055 | 0 | name, NULL); |
8056 | 1.75M | } else { |
8057 | 1.75M | xmlChar start[4]; |
8058 | 1.75M | xmlCharEncoding enc; |
8059 | | |
8060 | 1.75M | if (xmlParserEntityCheck(ctxt, 0, entity, 0)) |
8061 | 596 | return; |
8062 | | |
8063 | 1.75M | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
8064 | 1.75M | ((ctxt->options & XML_PARSE_NOENT) == 0) && |
8065 | 1.75M | ((ctxt->options & XML_PARSE_DTDVALID) == 0) && |
8066 | 1.75M | ((ctxt->options & XML_PARSE_DTDLOAD) == 0) && |
8067 | 1.75M | ((ctxt->options & XML_PARSE_DTDATTR) == 0) && |
8068 | 1.75M | (ctxt->replaceEntities == 0) && |
8069 | 1.75M | (ctxt->validate == 0)) |
8070 | 28 | return; |
8071 | | |
8072 | 1.75M | input = xmlNewEntityInputStream(ctxt, entity); |
8073 | 1.75M | if (xmlPushInput(ctxt, input) < 0) { |
8074 | 6.45k | xmlFreeInputStream(input); |
8075 | 6.45k | return; |
8076 | 6.45k | } |
8077 | | |
8078 | 1.75M | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
8079 | | /* |
8080 | | * Get the 4 first bytes and decode the charset |
8081 | | * if enc != XML_CHAR_ENCODING_NONE |
8082 | | * plug some encoding conversion routines. |
8083 | | * Note that, since we may have some non-UTF8 |
8084 | | * encoding (like UTF16, bug 135229), the 'length' |
8085 | | * is not known, but we can calculate based upon |
8086 | | * the amount of data in the buffer. |
8087 | | */ |
8088 | 141k | GROW |
8089 | 141k | if (ctxt->instate == XML_PARSER_EOF) |
8090 | 0 | return; |
8091 | 141k | if ((ctxt->input->end - ctxt->input->cur)>=4) { |
8092 | 140k | start[0] = RAW; |
8093 | 140k | start[1] = NXT(1); |
8094 | 140k | start[2] = NXT(2); |
8095 | 140k | start[3] = NXT(3); |
8096 | 140k | enc = xmlDetectCharEncoding(start, 4); |
8097 | 140k | if (enc != XML_CHAR_ENCODING_NONE) { |
8098 | 40.8k | xmlSwitchEncoding(ctxt, enc); |
8099 | 40.8k | } |
8100 | 140k | } |
8101 | | |
8102 | 141k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
8103 | 141k | (IS_BLANK_CH(NXT(5)))) { |
8104 | 39.4k | xmlParseTextDecl(ctxt); |
8105 | 39.4k | } |
8106 | 141k | } |
8107 | 1.75M | } |
8108 | 1.75M | } |
8109 | 1.91M | ctxt->hasPErefs = 1; |
8110 | 1.91M | } |
8111 | | |
8112 | | /** |
8113 | | * xmlLoadEntityContent: |
8114 | | * @ctxt: an XML parser context |
8115 | | * @entity: an unloaded system entity |
8116 | | * |
8117 | | * Load the original content of the given system entity from the |
8118 | | * ExternalID/SystemID given. This is to be used for Included in Literal |
8119 | | * http://www.w3.org/TR/REC-xml/#inliteral processing of entities references |
8120 | | * |
8121 | | * Returns 0 in case of success and -1 in case of failure |
8122 | | */ |
8123 | | static int |
8124 | 10.5k | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
8125 | 10.5k | xmlParserInputPtr input; |
8126 | 10.5k | xmlBufferPtr buf; |
8127 | 10.5k | int l, c; |
8128 | 10.5k | int count = 0; |
8129 | | |
8130 | 10.5k | if ((ctxt == NULL) || (entity == NULL) || |
8131 | 10.5k | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
8132 | 10.5k | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
8133 | 10.5k | (entity->content != NULL)) { |
8134 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8135 | 0 | "xmlLoadEntityContent parameter error"); |
8136 | 0 | return(-1); |
8137 | 0 | } |
8138 | | |
8139 | 10.5k | if (xmlParserDebugEntities) |
8140 | 0 | xmlGenericError(xmlGenericErrorContext, |
8141 | 0 | "Reading %s entity content input\n", entity->name); |
8142 | | |
8143 | 10.5k | buf = xmlBufferCreate(); |
8144 | 10.5k | if (buf == NULL) { |
8145 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8146 | 0 | "xmlLoadEntityContent parameter error"); |
8147 | 0 | return(-1); |
8148 | 0 | } |
8149 | 10.5k | xmlBufferSetAllocationScheme(buf, XML_BUFFER_ALLOC_DOUBLEIT); |
8150 | | |
8151 | 10.5k | input = xmlNewEntityInputStream(ctxt, entity); |
8152 | 10.5k | if (input == NULL) { |
8153 | 411 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8154 | 411 | "xmlLoadEntityContent input error"); |
8155 | 411 | xmlBufferFree(buf); |
8156 | 411 | return(-1); |
8157 | 411 | } |
8158 | | |
8159 | | /* |
8160 | | * Push the entity as the current input, read char by char |
8161 | | * saving to the buffer until the end of the entity or an error |
8162 | | */ |
8163 | 10.1k | if (xmlPushInput(ctxt, input) < 0) { |
8164 | 36 | xmlBufferFree(buf); |
8165 | 36 | xmlFreeInputStream(input); |
8166 | 36 | return(-1); |
8167 | 36 | } |
8168 | | |
8169 | 10.1k | GROW; |
8170 | 10.1k | c = CUR_CHAR(l); |
8171 | 26.4M | while ((ctxt->input == input) && (ctxt->input->cur < ctxt->input->end) && |
8172 | 26.4M | (IS_CHAR(c))) { |
8173 | 26.4M | xmlBufferAdd(buf, ctxt->input->cur, l); |
8174 | 26.4M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
8175 | 254k | count = 0; |
8176 | 254k | GROW; |
8177 | 254k | if (ctxt->instate == XML_PARSER_EOF) { |
8178 | 0 | xmlBufferFree(buf); |
8179 | 0 | return(-1); |
8180 | 0 | } |
8181 | 254k | } |
8182 | 26.4M | NEXTL(l); |
8183 | 26.4M | c = CUR_CHAR(l); |
8184 | 26.4M | if (c == 0) { |
8185 | 8.37k | count = 0; |
8186 | 8.37k | GROW; |
8187 | 8.37k | if (ctxt->instate == XML_PARSER_EOF) { |
8188 | 0 | xmlBufferFree(buf); |
8189 | 0 | return(-1); |
8190 | 0 | } |
8191 | 8.37k | c = CUR_CHAR(l); |
8192 | 8.37k | } |
8193 | 26.4M | } |
8194 | | |
8195 | 10.1k | if ((ctxt->input == input) && (ctxt->input->cur >= ctxt->input->end)) { |
8196 | 4.91k | xmlPopInput(ctxt); |
8197 | 5.19k | } else if (!IS_CHAR(c)) { |
8198 | 5.19k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8199 | 5.19k | "xmlLoadEntityContent: invalid char value %d\n", |
8200 | 5.19k | c); |
8201 | 5.19k | xmlBufferFree(buf); |
8202 | 5.19k | return(-1); |
8203 | 5.19k | } |
8204 | 4.91k | entity->content = buf->content; |
8205 | 4.91k | buf->content = NULL; |
8206 | 4.91k | xmlBufferFree(buf); |
8207 | | |
8208 | 4.91k | return(0); |
8209 | 10.1k | } |
8210 | | |
8211 | | /** |
8212 | | * xmlParseStringPEReference: |
8213 | | * @ctxt: an XML parser context |
8214 | | * @str: a pointer to an index in the string |
8215 | | * |
8216 | | * parse PEReference declarations |
8217 | | * |
8218 | | * [69] PEReference ::= '%' Name ';' |
8219 | | * |
8220 | | * [ WFC: No Recursion ] |
8221 | | * A parsed entity must not contain a recursive |
8222 | | * reference to itself, either directly or indirectly. |
8223 | | * |
8224 | | * [ WFC: Entity Declared ] |
8225 | | * In a document without any DTD, a document with only an internal DTD |
8226 | | * subset which contains no parameter entity references, or a document |
8227 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8228 | | * entity must precede any reference to it... |
8229 | | * |
8230 | | * [ VC: Entity Declared ] |
8231 | | * In a document with an external subset or external parameter entities |
8232 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8233 | | * must precede any reference to it... |
8234 | | * |
8235 | | * [ WFC: In DTD ] |
8236 | | * Parameter-entity references may only appear in the DTD. |
8237 | | * NOTE: misleading but this is handled. |
8238 | | * |
8239 | | * Returns the string of the entity content. |
8240 | | * str is updated to the current value of the index |
8241 | | */ |
8242 | | static xmlEntityPtr |
8243 | 1.91M | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8244 | 1.91M | const xmlChar *ptr; |
8245 | 1.91M | xmlChar cur; |
8246 | 1.91M | xmlChar *name; |
8247 | 1.91M | xmlEntityPtr entity = NULL; |
8248 | | |
8249 | 1.91M | if ((str == NULL) || (*str == NULL)) return(NULL); |
8250 | 1.91M | ptr = *str; |
8251 | 1.91M | cur = *ptr; |
8252 | 1.91M | if (cur != '%') |
8253 | 0 | return(NULL); |
8254 | 1.91M | ptr++; |
8255 | 1.91M | name = xmlParseStringName(ctxt, &ptr); |
8256 | 1.91M | if (name == NULL) { |
8257 | 429k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8258 | 429k | "xmlParseStringPEReference: no name\n"); |
8259 | 429k | *str = ptr; |
8260 | 429k | return(NULL); |
8261 | 429k | } |
8262 | 1.48M | cur = *ptr; |
8263 | 1.48M | if (cur != ';') { |
8264 | 67.8k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8265 | 67.8k | xmlFree(name); |
8266 | 67.8k | *str = ptr; |
8267 | 67.8k | return(NULL); |
8268 | 67.8k | } |
8269 | 1.42M | ptr++; |
8270 | | |
8271 | | /* |
8272 | | * Increase the number of entity references parsed |
8273 | | */ |
8274 | 1.42M | ctxt->nbentities++; |
8275 | | |
8276 | | /* |
8277 | | * Request the entity from SAX |
8278 | | */ |
8279 | 1.42M | if ((ctxt->sax != NULL) && |
8280 | 1.42M | (ctxt->sax->getParameterEntity != NULL)) |
8281 | 1.42M | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8282 | 1.42M | if (ctxt->instate == XML_PARSER_EOF) { |
8283 | 0 | xmlFree(name); |
8284 | 0 | *str = ptr; |
8285 | 0 | return(NULL); |
8286 | 0 | } |
8287 | 1.42M | if (entity == NULL) { |
8288 | | /* |
8289 | | * [ WFC: Entity Declared ] |
8290 | | * In a document without any DTD, a document with only an |
8291 | | * internal DTD subset which contains no parameter entity |
8292 | | * references, or a document with "standalone='yes'", ... |
8293 | | * ... The declaration of a parameter entity must precede |
8294 | | * any reference to it... |
8295 | | */ |
8296 | 135k | if ((ctxt->standalone == 1) || |
8297 | 135k | ((ctxt->hasExternalSubset == 0) && (ctxt->hasPErefs == 0))) { |
8298 | 2.40k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8299 | 2.40k | "PEReference: %%%s; not found\n", name); |
8300 | 132k | } else { |
8301 | | /* |
8302 | | * [ VC: Entity Declared ] |
8303 | | * In a document with an external subset or external |
8304 | | * parameter entities with "standalone='no'", ... |
8305 | | * ... The declaration of a parameter entity must |
8306 | | * precede any reference to it... |
8307 | | */ |
8308 | 132k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8309 | 132k | "PEReference: %%%s; not found\n", |
8310 | 132k | name, NULL); |
8311 | 132k | ctxt->valid = 0; |
8312 | 132k | } |
8313 | 135k | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
8314 | 1.28M | } else { |
8315 | | /* |
8316 | | * Internal checking in case the entity quest barfed |
8317 | | */ |
8318 | 1.28M | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8319 | 1.28M | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8320 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8321 | 0 | "%%%s; is not a parameter entity\n", |
8322 | 0 | name, NULL); |
8323 | 0 | } |
8324 | 1.28M | } |
8325 | 1.42M | ctxt->hasPErefs = 1; |
8326 | 1.42M | xmlFree(name); |
8327 | 1.42M | *str = ptr; |
8328 | 1.42M | return(entity); |
8329 | 1.42M | } |
8330 | | |
8331 | | /** |
8332 | | * xmlParseDocTypeDecl: |
8333 | | * @ctxt: an XML parser context |
8334 | | * |
8335 | | * DEPRECATED: Internal function, don't use. |
8336 | | * |
8337 | | * parse a DOCTYPE declaration |
8338 | | * |
8339 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8340 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8341 | | * |
8342 | | * [ VC: Root Element Type ] |
8343 | | * The Name in the document type declaration must match the element |
8344 | | * type of the root element. |
8345 | | */ |
8346 | | |
8347 | | void |
8348 | 511k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8349 | 511k | const xmlChar *name = NULL; |
8350 | 511k | xmlChar *ExternalID = NULL; |
8351 | 511k | xmlChar *URI = NULL; |
8352 | | |
8353 | | /* |
8354 | | * We know that '<!DOCTYPE' has been detected. |
8355 | | */ |
8356 | 511k | SKIP(9); |
8357 | | |
8358 | 511k | SKIP_BLANKS; |
8359 | | |
8360 | | /* |
8361 | | * Parse the DOCTYPE name. |
8362 | | */ |
8363 | 511k | name = xmlParseName(ctxt); |
8364 | 511k | if (name == NULL) { |
8365 | 2.16k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8366 | 2.16k | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8367 | 2.16k | } |
8368 | 511k | ctxt->intSubName = name; |
8369 | | |
8370 | 511k | SKIP_BLANKS; |
8371 | | |
8372 | | /* |
8373 | | * Check for SystemID and ExternalID |
8374 | | */ |
8375 | 511k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8376 | | |
8377 | 511k | if ((URI != NULL) || (ExternalID != NULL)) { |
8378 | 174k | ctxt->hasExternalSubset = 1; |
8379 | 174k | } |
8380 | 511k | ctxt->extSubURI = URI; |
8381 | 511k | ctxt->extSubSystem = ExternalID; |
8382 | | |
8383 | 511k | SKIP_BLANKS; |
8384 | | |
8385 | | /* |
8386 | | * Create and update the internal subset. |
8387 | | */ |
8388 | 511k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8389 | 511k | (!ctxt->disableSAX)) |
8390 | 495k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8391 | 511k | if (ctxt->instate == XML_PARSER_EOF) |
8392 | 0 | return; |
8393 | | |
8394 | | /* |
8395 | | * Is there any internal subset declarations ? |
8396 | | * they are handled separately in xmlParseInternalSubset() |
8397 | | */ |
8398 | 511k | if (RAW == '[') |
8399 | 392k | return; |
8400 | | |
8401 | | /* |
8402 | | * We should be at the end of the DOCTYPE declaration. |
8403 | | */ |
8404 | 118k | if (RAW != '>') { |
8405 | 19.6k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8406 | 19.6k | } |
8407 | 118k | NEXT; |
8408 | 118k | } |
8409 | | |
8410 | | /** |
8411 | | * xmlParseInternalSubset: |
8412 | | * @ctxt: an XML parser context |
8413 | | * |
8414 | | * parse the internal subset declaration |
8415 | | * |
8416 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8417 | | */ |
8418 | | |
8419 | | static void |
8420 | 344k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8421 | | /* |
8422 | | * Is there any DTD definition ? |
8423 | | */ |
8424 | 344k | if (RAW == '[') { |
8425 | 344k | int baseInputNr = ctxt->inputNr; |
8426 | 344k | ctxt->instate = XML_PARSER_DTD; |
8427 | 344k | NEXT; |
8428 | | /* |
8429 | | * Parse the succession of Markup declarations and |
8430 | | * PEReferences. |
8431 | | * Subsequence (markupdecl | PEReference | S)* |
8432 | | */ |
8433 | 4.62M | while (((RAW != ']') || (ctxt->inputNr > baseInputNr)) && |
8434 | 4.62M | (ctxt->instate != XML_PARSER_EOF)) { |
8435 | 4.36M | int id = ctxt->input->id; |
8436 | 4.36M | unsigned long cons = CUR_CONSUMED; |
8437 | | |
8438 | 4.36M | SKIP_BLANKS; |
8439 | 4.36M | xmlParseMarkupDecl(ctxt); |
8440 | 4.36M | xmlParsePEReference(ctxt); |
8441 | | |
8442 | | /* |
8443 | | * Conditional sections are allowed from external entities included |
8444 | | * by PE References in the internal subset. |
8445 | | */ |
8446 | 4.36M | if ((ctxt->inputNr > 1) && (ctxt->input->filename != NULL) && |
8447 | 4.36M | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8448 | 0 | xmlParseConditionalSections(ctxt); |
8449 | 0 | } |
8450 | | |
8451 | 4.36M | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
8452 | 97.8k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8453 | 97.8k | "xmlParseInternalSubset: error detected in Markup declaration\n"); |
8454 | 97.8k | if (ctxt->inputNr > baseInputNr) |
8455 | 4.70k | xmlPopInput(ctxt); |
8456 | 93.1k | else |
8457 | 93.1k | break; |
8458 | 97.8k | } |
8459 | 4.36M | } |
8460 | 344k | if (RAW == ']') { |
8461 | 234k | NEXT; |
8462 | 234k | SKIP_BLANKS; |
8463 | 234k | } |
8464 | 344k | } |
8465 | | |
8466 | | /* |
8467 | | * We should be at the end of the DOCTYPE declaration. |
8468 | | */ |
8469 | 344k | if (RAW != '>') { |
8470 | 109k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8471 | 109k | return; |
8472 | 109k | } |
8473 | 234k | NEXT; |
8474 | 234k | } |
8475 | | |
8476 | | #ifdef LIBXML_SAX1_ENABLED |
8477 | | /** |
8478 | | * xmlParseAttribute: |
8479 | | * @ctxt: an XML parser context |
8480 | | * @value: a xmlChar ** used to store the value of the attribute |
8481 | | * |
8482 | | * DEPRECATED: Internal function, don't use. |
8483 | | * |
8484 | | * parse an attribute |
8485 | | * |
8486 | | * [41] Attribute ::= Name Eq AttValue |
8487 | | * |
8488 | | * [ WFC: No External Entity References ] |
8489 | | * Attribute values cannot contain direct or indirect entity references |
8490 | | * to external entities. |
8491 | | * |
8492 | | * [ WFC: No < in Attribute Values ] |
8493 | | * The replacement text of any entity referred to directly or indirectly in |
8494 | | * an attribute value (other than "<") must not contain a <. |
8495 | | * |
8496 | | * [ VC: Attribute Value Type ] |
8497 | | * The attribute must have been declared; the value must be of the type |
8498 | | * declared for it. |
8499 | | * |
8500 | | * [25] Eq ::= S? '=' S? |
8501 | | * |
8502 | | * With namespace: |
8503 | | * |
8504 | | * [NS 11] Attribute ::= QName Eq AttValue |
8505 | | * |
8506 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8507 | | * definition. |
8508 | | * |
8509 | | * Returns the attribute name, and the value in *value. |
8510 | | */ |
8511 | | |
8512 | | const xmlChar * |
8513 | 85.3M | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8514 | 85.3M | const xmlChar *name; |
8515 | 85.3M | xmlChar *val; |
8516 | | |
8517 | 85.3M | *value = NULL; |
8518 | 85.3M | GROW; |
8519 | 85.3M | name = xmlParseName(ctxt); |
8520 | 85.3M | if (name == NULL) { |
8521 | 8.76M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8522 | 8.76M | "error parsing attribute name\n"); |
8523 | 8.76M | return(NULL); |
8524 | 8.76M | } |
8525 | | |
8526 | | /* |
8527 | | * read the value |
8528 | | */ |
8529 | 76.5M | SKIP_BLANKS; |
8530 | 76.5M | if (RAW == '=') { |
8531 | 71.4M | NEXT; |
8532 | 71.4M | SKIP_BLANKS; |
8533 | 71.4M | val = xmlParseAttValue(ctxt); |
8534 | 71.4M | ctxt->instate = XML_PARSER_CONTENT; |
8535 | 71.4M | } else { |
8536 | 5.05M | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8537 | 5.05M | "Specification mandates value for attribute %s\n", name); |
8538 | 5.05M | return(NULL); |
8539 | 5.05M | } |
8540 | | |
8541 | | /* |
8542 | | * Check that xml:lang conforms to the specification |
8543 | | * No more registered as an error, just generate a warning now |
8544 | | * since this was deprecated in XML second edition |
8545 | | */ |
8546 | 71.4M | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8547 | 18.7k | if (!xmlCheckLanguageID(val)) { |
8548 | 7.62k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8549 | 7.62k | "Malformed value for xml:lang : %s\n", |
8550 | 7.62k | val, NULL); |
8551 | 7.62k | } |
8552 | 18.7k | } |
8553 | | |
8554 | | /* |
8555 | | * Check that xml:space conforms to the specification |
8556 | | */ |
8557 | 71.4M | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8558 | 1.09k | if (xmlStrEqual(val, BAD_CAST "default")) |
8559 | 0 | *(ctxt->space) = 0; |
8560 | 1.09k | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8561 | 563 | *(ctxt->space) = 1; |
8562 | 527 | else { |
8563 | 527 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8564 | 527 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8565 | 527 | val, NULL); |
8566 | 527 | } |
8567 | 1.09k | } |
8568 | | |
8569 | 71.4M | *value = val; |
8570 | 71.4M | return(name); |
8571 | 76.5M | } |
8572 | | |
8573 | | /** |
8574 | | * xmlParseStartTag: |
8575 | | * @ctxt: an XML parser context |
8576 | | * |
8577 | | * DEPRECATED: Internal function, don't use. |
8578 | | * |
8579 | | * parse a start of tag either for rule element or |
8580 | | * EmptyElement. In both case we don't parse the tag closing chars. |
8581 | | * |
8582 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8583 | | * |
8584 | | * [ WFC: Unique Att Spec ] |
8585 | | * No attribute name may appear more than once in the same start-tag or |
8586 | | * empty-element tag. |
8587 | | * |
8588 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8589 | | * |
8590 | | * [ WFC: Unique Att Spec ] |
8591 | | * No attribute name may appear more than once in the same start-tag or |
8592 | | * empty-element tag. |
8593 | | * |
8594 | | * With namespace: |
8595 | | * |
8596 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8597 | | * |
8598 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8599 | | * |
8600 | | * Returns the element name parsed |
8601 | | */ |
8602 | | |
8603 | | const xmlChar * |
8604 | 149M | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8605 | 149M | const xmlChar *name; |
8606 | 149M | const xmlChar *attname; |
8607 | 149M | xmlChar *attvalue; |
8608 | 149M | const xmlChar **atts = ctxt->atts; |
8609 | 149M | int nbatts = 0; |
8610 | 149M | int maxatts = ctxt->maxatts; |
8611 | 149M | int i; |
8612 | | |
8613 | 149M | if (RAW != '<') return(NULL); |
8614 | 149M | NEXT1; |
8615 | | |
8616 | 149M | name = xmlParseName(ctxt); |
8617 | 149M | if (name == NULL) { |
8618 | 63.6M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8619 | 63.6M | "xmlParseStartTag: invalid element name\n"); |
8620 | 63.6M | return(NULL); |
8621 | 63.6M | } |
8622 | | |
8623 | | /* |
8624 | | * Now parse the attributes, it ends up with the ending |
8625 | | * |
8626 | | * (S Attribute)* S? |
8627 | | */ |
8628 | 85.5M | SKIP_BLANKS; |
8629 | 85.5M | GROW; |
8630 | | |
8631 | 117M | while (((RAW != '>') && |
8632 | 117M | ((RAW != '/') || (NXT(1) != '>')) && |
8633 | 117M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
8634 | 85.3M | int id = ctxt->input->id; |
8635 | 85.3M | unsigned long cons = CUR_CONSUMED; |
8636 | | |
8637 | 85.3M | attname = xmlParseAttribute(ctxt, &attvalue); |
8638 | 85.3M | if ((attname != NULL) && (attvalue != NULL)) { |
8639 | | /* |
8640 | | * [ WFC: Unique Att Spec ] |
8641 | | * No attribute name may appear more than once in the same |
8642 | | * start-tag or empty-element tag. |
8643 | | */ |
8644 | 95.8M | for (i = 0; i < nbatts;i += 2) { |
8645 | 24.8M | if (xmlStrEqual(atts[i], attname)) { |
8646 | 87.1k | xmlErrAttributeDup(ctxt, NULL, attname); |
8647 | 87.1k | xmlFree(attvalue); |
8648 | 87.1k | goto failed; |
8649 | 87.1k | } |
8650 | 24.8M | } |
8651 | | /* |
8652 | | * Add the pair to atts |
8653 | | */ |
8654 | 71.0M | if (atts == NULL) { |
8655 | 4.03M | maxatts = 22; /* allow for 10 attrs by default */ |
8656 | 4.03M | atts = (const xmlChar **) |
8657 | 4.03M | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8658 | 4.03M | if (atts == NULL) { |
8659 | 0 | xmlErrMemory(ctxt, NULL); |
8660 | 0 | if (attvalue != NULL) |
8661 | 0 | xmlFree(attvalue); |
8662 | 0 | goto failed; |
8663 | 0 | } |
8664 | 4.03M | ctxt->atts = atts; |
8665 | 4.03M | ctxt->maxatts = maxatts; |
8666 | 67.0M | } else if (nbatts + 4 > maxatts) { |
8667 | 4.31k | const xmlChar **n; |
8668 | | |
8669 | 4.31k | maxatts *= 2; |
8670 | 4.31k | n = (const xmlChar **) xmlRealloc((void *) atts, |
8671 | 4.31k | maxatts * sizeof(const xmlChar *)); |
8672 | 4.31k | if (n == NULL) { |
8673 | 0 | xmlErrMemory(ctxt, NULL); |
8674 | 0 | if (attvalue != NULL) |
8675 | 0 | xmlFree(attvalue); |
8676 | 0 | goto failed; |
8677 | 0 | } |
8678 | 4.31k | atts = n; |
8679 | 4.31k | ctxt->atts = atts; |
8680 | 4.31k | ctxt->maxatts = maxatts; |
8681 | 4.31k | } |
8682 | 71.0M | atts[nbatts++] = attname; |
8683 | 71.0M | atts[nbatts++] = attvalue; |
8684 | 71.0M | atts[nbatts] = NULL; |
8685 | 71.0M | atts[nbatts + 1] = NULL; |
8686 | 71.0M | } else { |
8687 | 14.1M | if (attvalue != NULL) |
8688 | 0 | xmlFree(attvalue); |
8689 | 14.1M | } |
8690 | | |
8691 | 85.3M | failed: |
8692 | | |
8693 | 85.3M | GROW |
8694 | 85.3M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8695 | 44.4M | break; |
8696 | 40.8M | if (SKIP_BLANKS == 0) { |
8697 | 17.0M | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8698 | 17.0M | "attributes construct error\n"); |
8699 | 17.0M | } |
8700 | 40.8M | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
8701 | 40.8M | (attname == NULL) && (attvalue == NULL)) { |
8702 | 8.76M | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
8703 | 8.76M | "xmlParseStartTag: problem parsing attributes\n"); |
8704 | 8.76M | break; |
8705 | 8.76M | } |
8706 | 32.1M | SHRINK; |
8707 | 32.1M | GROW; |
8708 | 32.1M | } |
8709 | | |
8710 | | /* |
8711 | | * SAX: Start of Element ! |
8712 | | */ |
8713 | 85.5M | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8714 | 85.5M | (!ctxt->disableSAX)) { |
8715 | 18.8M | if (nbatts > 0) |
8716 | 10.8M | ctxt->sax->startElement(ctxt->userData, name, atts); |
8717 | 7.98M | else |
8718 | 7.98M | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8719 | 18.8M | } |
8720 | | |
8721 | 85.5M | if (atts != NULL) { |
8722 | | /* Free only the content strings */ |
8723 | 154M | for (i = 1;i < nbatts;i+=2) |
8724 | 71.0M | if (atts[i] != NULL) |
8725 | 71.0M | xmlFree((xmlChar *) atts[i]); |
8726 | 83.4M | } |
8727 | 85.5M | return(name); |
8728 | 85.5M | } |
8729 | | |
8730 | | /** |
8731 | | * xmlParseEndTag1: |
8732 | | * @ctxt: an XML parser context |
8733 | | * @line: line of the start tag |
8734 | | * @nsNr: number of namespaces on the start tag |
8735 | | * |
8736 | | * parse an end of tag |
8737 | | * |
8738 | | * [42] ETag ::= '</' Name S? '>' |
8739 | | * |
8740 | | * With namespace |
8741 | | * |
8742 | | * [NS 9] ETag ::= '</' QName S? '>' |
8743 | | */ |
8744 | | |
8745 | | static void |
8746 | 34.6M | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8747 | 34.6M | const xmlChar *name; |
8748 | | |
8749 | 34.6M | GROW; |
8750 | 34.6M | if ((RAW != '<') || (NXT(1) != '/')) { |
8751 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8752 | 0 | "xmlParseEndTag: '</' not found\n"); |
8753 | 0 | return; |
8754 | 0 | } |
8755 | 34.6M | SKIP(2); |
8756 | | |
8757 | 34.6M | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8758 | | |
8759 | | /* |
8760 | | * We should definitely be at the ending "S? '>'" part |
8761 | | */ |
8762 | 34.6M | GROW; |
8763 | 34.6M | SKIP_BLANKS; |
8764 | 34.6M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8765 | 3.00M | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8766 | 3.00M | } else |
8767 | 31.6M | NEXT1; |
8768 | | |
8769 | | /* |
8770 | | * [ WFC: Element Type Match ] |
8771 | | * The Name in an element's end-tag must match the element type in the |
8772 | | * start-tag. |
8773 | | * |
8774 | | */ |
8775 | 34.6M | if (name != (xmlChar*)1) { |
8776 | 8.47M | if (name == NULL) name = BAD_CAST "unparsable"; |
8777 | 8.47M | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8778 | 8.47M | "Opening and ending tag mismatch: %s line %d and %s\n", |
8779 | 8.47M | ctxt->name, line, name); |
8780 | 8.47M | } |
8781 | | |
8782 | | /* |
8783 | | * SAX: End of Tag |
8784 | | */ |
8785 | 34.6M | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8786 | 34.6M | (!ctxt->disableSAX)) |
8787 | 6.04M | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8788 | | |
8789 | 34.6M | namePop(ctxt); |
8790 | 34.6M | spacePop(ctxt); |
8791 | 34.6M | return; |
8792 | 34.6M | } |
8793 | | |
8794 | | /** |
8795 | | * xmlParseEndTag: |
8796 | | * @ctxt: an XML parser context |
8797 | | * |
8798 | | * DEPRECATED: Internal function, don't use. |
8799 | | * |
8800 | | * parse an end of tag |
8801 | | * |
8802 | | * [42] ETag ::= '</' Name S? '>' |
8803 | | * |
8804 | | * With namespace |
8805 | | * |
8806 | | * [NS 9] ETag ::= '</' QName S? '>' |
8807 | | */ |
8808 | | |
8809 | | void |
8810 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8811 | 0 | xmlParseEndTag1(ctxt, 0); |
8812 | 0 | } |
8813 | | #endif /* LIBXML_SAX1_ENABLED */ |
8814 | | |
8815 | | /************************************************************************ |
8816 | | * * |
8817 | | * SAX 2 specific operations * |
8818 | | * * |
8819 | | ************************************************************************/ |
8820 | | |
8821 | | /* |
8822 | | * xmlGetNamespace: |
8823 | | * @ctxt: an XML parser context |
8824 | | * @prefix: the prefix to lookup |
8825 | | * |
8826 | | * Lookup the namespace name for the @prefix (which ca be NULL) |
8827 | | * The prefix must come from the @ctxt->dict dictionary |
8828 | | * |
8829 | | * Returns the namespace name or NULL if not bound |
8830 | | */ |
8831 | | static const xmlChar * |
8832 | 20.9M | xmlGetNamespace(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
8833 | 20.9M | int i; |
8834 | | |
8835 | 20.9M | if (prefix == ctxt->str_xml) return(ctxt->str_xml_ns); |
8836 | 26.3M | for (i = ctxt->nsNr - 2;i >= 0;i-=2) |
8837 | 7.99M | if (ctxt->nsTab[i] == prefix) { |
8838 | 2.47M | if ((prefix == NULL) && (*ctxt->nsTab[i + 1] == 0)) |
8839 | 11.0k | return(NULL); |
8840 | 2.46M | return(ctxt->nsTab[i + 1]); |
8841 | 2.47M | } |
8842 | 18.3M | return(NULL); |
8843 | 20.8M | } |
8844 | | |
8845 | | /** |
8846 | | * xmlParseQName: |
8847 | | * @ctxt: an XML parser context |
8848 | | * @prefix: pointer to store the prefix part |
8849 | | * |
8850 | | * parse an XML Namespace QName |
8851 | | * |
8852 | | * [6] QName ::= (Prefix ':')? LocalPart |
8853 | | * [7] Prefix ::= NCName |
8854 | | * [8] LocalPart ::= NCName |
8855 | | * |
8856 | | * Returns the Name parsed or NULL |
8857 | | */ |
8858 | | |
8859 | | static const xmlChar * |
8860 | 54.3M | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8861 | 54.3M | const xmlChar *l, *p; |
8862 | | |
8863 | 54.3M | GROW; |
8864 | | |
8865 | 54.3M | l = xmlParseNCName(ctxt); |
8866 | 54.3M | if (l == NULL) { |
8867 | 5.89M | if (CUR == ':') { |
8868 | 20.5k | l = xmlParseName(ctxt); |
8869 | 20.5k | if (l != NULL) { |
8870 | 20.5k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8871 | 20.5k | "Failed to parse QName '%s'\n", l, NULL, NULL); |
8872 | 20.5k | *prefix = NULL; |
8873 | 20.5k | return(l); |
8874 | 20.5k | } |
8875 | 20.5k | } |
8876 | 5.87M | return(NULL); |
8877 | 5.89M | } |
8878 | 48.4M | if (CUR == ':') { |
8879 | 4.05M | NEXT; |
8880 | 4.05M | p = l; |
8881 | 4.05M | l = xmlParseNCName(ctxt); |
8882 | 4.05M | if (l == NULL) { |
8883 | 51.0k | xmlChar *tmp; |
8884 | | |
8885 | 51.0k | if (ctxt->instate == XML_PARSER_EOF) |
8886 | 0 | return(NULL); |
8887 | 51.0k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8888 | 51.0k | "Failed to parse QName '%s:'\n", p, NULL, NULL); |
8889 | 51.0k | l = xmlParseNmtoken(ctxt); |
8890 | 51.0k | if (l == NULL) { |
8891 | 39.0k | if (ctxt->instate == XML_PARSER_EOF) |
8892 | 0 | return(NULL); |
8893 | 39.0k | tmp = xmlBuildQName(BAD_CAST "", p, NULL, 0); |
8894 | 39.0k | } else { |
8895 | 12.0k | tmp = xmlBuildQName(l, p, NULL, 0); |
8896 | 12.0k | xmlFree((char *)l); |
8897 | 12.0k | } |
8898 | 51.0k | p = xmlDictLookup(ctxt->dict, tmp, -1); |
8899 | 51.0k | if (tmp != NULL) xmlFree(tmp); |
8900 | 51.0k | *prefix = NULL; |
8901 | 51.0k | return(p); |
8902 | 51.0k | } |
8903 | 4.00M | if (CUR == ':') { |
8904 | 23.5k | xmlChar *tmp; |
8905 | | |
8906 | 23.5k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8907 | 23.5k | "Failed to parse QName '%s:%s:'\n", p, l, NULL); |
8908 | 23.5k | NEXT; |
8909 | 23.5k | tmp = (xmlChar *) xmlParseName(ctxt); |
8910 | 23.5k | if (tmp != NULL) { |
8911 | 16.8k | tmp = xmlBuildQName(tmp, l, NULL, 0); |
8912 | 16.8k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8913 | 16.8k | if (tmp != NULL) xmlFree(tmp); |
8914 | 16.8k | *prefix = p; |
8915 | 16.8k | return(l); |
8916 | 16.8k | } |
8917 | 6.68k | if (ctxt->instate == XML_PARSER_EOF) |
8918 | 0 | return(NULL); |
8919 | 6.68k | tmp = xmlBuildQName(BAD_CAST "", l, NULL, 0); |
8920 | 6.68k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8921 | 6.68k | if (tmp != NULL) xmlFree(tmp); |
8922 | 6.68k | *prefix = p; |
8923 | 6.68k | return(l); |
8924 | 6.68k | } |
8925 | 3.97M | *prefix = p; |
8926 | 3.97M | } else |
8927 | 44.4M | *prefix = NULL; |
8928 | 48.3M | return(l); |
8929 | 48.4M | } |
8930 | | |
8931 | | /** |
8932 | | * xmlParseQNameAndCompare: |
8933 | | * @ctxt: an XML parser context |
8934 | | * @name: the localname |
8935 | | * @prefix: the prefix, if any. |
8936 | | * |
8937 | | * parse an XML name and compares for match |
8938 | | * (specialized for endtag parsing) |
8939 | | * |
8940 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8941 | | * and the name for mismatch |
8942 | | */ |
8943 | | |
8944 | | static const xmlChar * |
8945 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8946 | 723k | xmlChar const *prefix) { |
8947 | 723k | const xmlChar *cmp; |
8948 | 723k | const xmlChar *in; |
8949 | 723k | const xmlChar *ret; |
8950 | 723k | const xmlChar *prefix2; |
8951 | | |
8952 | 723k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8953 | | |
8954 | 723k | GROW; |
8955 | 723k | in = ctxt->input->cur; |
8956 | | |
8957 | 723k | cmp = prefix; |
8958 | 2.47M | while (*in != 0 && *in == *cmp) { |
8959 | 1.75M | ++in; |
8960 | 1.75M | ++cmp; |
8961 | 1.75M | } |
8962 | 723k | if ((*cmp == 0) && (*in == ':')) { |
8963 | 656k | in++; |
8964 | 656k | cmp = name; |
8965 | 4.98M | while (*in != 0 && *in == *cmp) { |
8966 | 4.32M | ++in; |
8967 | 4.32M | ++cmp; |
8968 | 4.32M | } |
8969 | 656k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8970 | | /* success */ |
8971 | 506k | ctxt->input->col += in - ctxt->input->cur; |
8972 | 506k | ctxt->input->cur = in; |
8973 | 506k | return((const xmlChar*) 1); |
8974 | 506k | } |
8975 | 656k | } |
8976 | | /* |
8977 | | * all strings coms from the dictionary, equality can be done directly |
8978 | | */ |
8979 | 217k | ret = xmlParseQName (ctxt, &prefix2); |
8980 | 217k | if ((ret == name) && (prefix == prefix2)) |
8981 | 1.50k | return((const xmlChar*) 1); |
8982 | 215k | return ret; |
8983 | 217k | } |
8984 | | |
8985 | | /** |
8986 | | * xmlParseAttValueInternal: |
8987 | | * @ctxt: an XML parser context |
8988 | | * @len: attribute len result |
8989 | | * @alloc: whether the attribute was reallocated as a new string |
8990 | | * @normalize: if 1 then further non-CDATA normalization must be done |
8991 | | * |
8992 | | * parse a value for an attribute. |
8993 | | * NOTE: if no normalization is needed, the routine will return pointers |
8994 | | * directly from the data buffer. |
8995 | | * |
8996 | | * 3.3.3 Attribute-Value Normalization: |
8997 | | * Before the value of an attribute is passed to the application or |
8998 | | * checked for validity, the XML processor must normalize it as follows: |
8999 | | * - a character reference is processed by appending the referenced |
9000 | | * character to the attribute value |
9001 | | * - an entity reference is processed by recursively processing the |
9002 | | * replacement text of the entity |
9003 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
9004 | | * appending #x20 to the normalized value, except that only a single |
9005 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
9006 | | * parsed entity or the literal entity value of an internal parsed entity |
9007 | | * - other characters are processed by appending them to the normalized value |
9008 | | * If the declared value is not CDATA, then the XML processor must further |
9009 | | * process the normalized attribute value by discarding any leading and |
9010 | | * trailing space (#x20) characters, and by replacing sequences of space |
9011 | | * (#x20) characters by a single space (#x20) character. |
9012 | | * All attributes for which no declaration has been read should be treated |
9013 | | * by a non-validating parser as if declared CDATA. |
9014 | | * |
9015 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
9016 | | * caller if it was copied, this can be detected by val[*len] == 0. |
9017 | | */ |
9018 | | |
9019 | | #define GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) \ |
9020 | 42.9k | const xmlChar *oldbase = ctxt->input->base;\ |
9021 | 42.9k | GROW;\ |
9022 | 42.9k | if (ctxt->instate == XML_PARSER_EOF)\ |
9023 | 42.9k | return(NULL);\ |
9024 | 42.9k | if (oldbase != ctxt->input->base) {\ |
9025 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase;\ |
9026 | 0 | start = start + delta;\ |
9027 | 0 | in = in + delta;\ |
9028 | 0 | }\ |
9029 | 42.9k | end = ctxt->input->end; |
9030 | | |
9031 | | static xmlChar * |
9032 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *len, int *alloc, |
9033 | | int normalize) |
9034 | 100M | { |
9035 | 100M | xmlChar limit = 0; |
9036 | 100M | const xmlChar *in = NULL, *start, *end, *last; |
9037 | 100M | xmlChar *ret = NULL; |
9038 | 100M | int line, col; |
9039 | 100M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9040 | 62.1M | XML_MAX_HUGE_LENGTH : |
9041 | 100M | XML_MAX_TEXT_LENGTH; |
9042 | | |
9043 | 100M | GROW; |
9044 | 100M | in = (xmlChar *) CUR_PTR; |
9045 | 100M | line = ctxt->input->line; |
9046 | 100M | col = ctxt->input->col; |
9047 | 100M | if (*in != '"' && *in != '\'') { |
9048 | 404k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
9049 | 404k | return (NULL); |
9050 | 404k | } |
9051 | 99.7M | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
9052 | | |
9053 | | /* |
9054 | | * try to handle in this routine the most common case where no |
9055 | | * allocation of a new string is required and where content is |
9056 | | * pure ASCII. |
9057 | | */ |
9058 | 99.7M | limit = *in++; |
9059 | 99.7M | col++; |
9060 | 99.7M | end = ctxt->input->end; |
9061 | 99.7M | start = in; |
9062 | 99.7M | if (in >= end) { |
9063 | 9.87k | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9064 | 9.87k | } |
9065 | 99.7M | if (normalize) { |
9066 | | /* |
9067 | | * Skip any leading spaces |
9068 | | */ |
9069 | 693k | while ((in < end) && (*in != limit) && |
9070 | 693k | ((*in == 0x20) || (*in == 0x9) || |
9071 | 686k | (*in == 0xA) || (*in == 0xD))) { |
9072 | 360k | if (*in == 0xA) { |
9073 | 198k | line++; col = 1; |
9074 | 198k | } else { |
9075 | 162k | col++; |
9076 | 162k | } |
9077 | 360k | in++; |
9078 | 360k | start = in; |
9079 | 360k | if (in >= end) { |
9080 | 168 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9081 | 168 | if ((in - start) > maxLength) { |
9082 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9083 | 0 | "AttValue length too long\n"); |
9084 | 0 | return(NULL); |
9085 | 0 | } |
9086 | 168 | } |
9087 | 360k | } |
9088 | 2.06M | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9089 | 2.06M | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9090 | 1.74M | col++; |
9091 | 1.74M | if ((*in++ == 0x20) && (*in == 0x20)) break; |
9092 | 1.73M | if (in >= end) { |
9093 | 227 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9094 | 227 | if ((in - start) > maxLength) { |
9095 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9096 | 0 | "AttValue length too long\n"); |
9097 | 0 | return(NULL); |
9098 | 0 | } |
9099 | 227 | } |
9100 | 1.73M | } |
9101 | 332k | last = in; |
9102 | | /* |
9103 | | * skip the trailing blanks |
9104 | | */ |
9105 | 344k | while ((last[-1] == 0x20) && (last > start)) last--; |
9106 | 738k | while ((in < end) && (*in != limit) && |
9107 | 738k | ((*in == 0x20) || (*in == 0x9) || |
9108 | 476k | (*in == 0xA) || (*in == 0xD))) { |
9109 | 406k | if (*in == 0xA) { |
9110 | 135k | line++, col = 1; |
9111 | 271k | } else { |
9112 | 271k | col++; |
9113 | 271k | } |
9114 | 406k | in++; |
9115 | 406k | if (in >= end) { |
9116 | 256 | const xmlChar *oldbase = ctxt->input->base; |
9117 | 256 | GROW; |
9118 | 256 | if (ctxt->instate == XML_PARSER_EOF) |
9119 | 0 | return(NULL); |
9120 | 256 | if (oldbase != ctxt->input->base) { |
9121 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase; |
9122 | 0 | start = start + delta; |
9123 | 0 | in = in + delta; |
9124 | 0 | last = last + delta; |
9125 | 0 | } |
9126 | 256 | end = ctxt->input->end; |
9127 | 256 | if ((in - start) > maxLength) { |
9128 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9129 | 0 | "AttValue length too long\n"); |
9130 | 0 | return(NULL); |
9131 | 0 | } |
9132 | 256 | } |
9133 | 406k | } |
9134 | 332k | if ((in - start) > maxLength) { |
9135 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9136 | 0 | "AttValue length too long\n"); |
9137 | 0 | return(NULL); |
9138 | 0 | } |
9139 | 332k | if (*in != limit) goto need_complex; |
9140 | 99.4M | } else { |
9141 | 1.03G | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9142 | 1.03G | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9143 | 939M | in++; |
9144 | 939M | col++; |
9145 | 939M | if (in >= end) { |
9146 | 32.7k | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9147 | 32.7k | if ((in - start) > maxLength) { |
9148 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9149 | 0 | "AttValue length too long\n"); |
9150 | 0 | return(NULL); |
9151 | 0 | } |
9152 | 32.7k | } |
9153 | 939M | } |
9154 | 99.4M | last = in; |
9155 | 99.4M | if ((in - start) > maxLength) { |
9156 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9157 | 0 | "AttValue length too long\n"); |
9158 | 0 | return(NULL); |
9159 | 0 | } |
9160 | 99.4M | if (*in != limit) goto need_complex; |
9161 | 99.4M | } |
9162 | 92.8M | in++; |
9163 | 92.8M | col++; |
9164 | 92.8M | if (len != NULL) { |
9165 | 27.2M | if (alloc) *alloc = 0; |
9166 | 27.2M | *len = last - start; |
9167 | 27.2M | ret = (xmlChar *) start; |
9168 | 65.5M | } else { |
9169 | 65.5M | if (alloc) *alloc = 1; |
9170 | 65.5M | ret = xmlStrndup(start, last - start); |
9171 | 65.5M | } |
9172 | 92.8M | CUR_PTR = in; |
9173 | 92.8M | ctxt->input->line = line; |
9174 | 92.8M | ctxt->input->col = col; |
9175 | 92.8M | return ret; |
9176 | 6.92M | need_complex: |
9177 | 6.92M | if (alloc) *alloc = 1; |
9178 | 6.92M | return xmlParseAttValueComplex(ctxt, len, normalize); |
9179 | 99.7M | } |
9180 | | |
9181 | | /** |
9182 | | * xmlParseAttribute2: |
9183 | | * @ctxt: an XML parser context |
9184 | | * @pref: the element prefix |
9185 | | * @elem: the element name |
9186 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
9187 | | * @value: a xmlChar ** used to store the value of the attribute |
9188 | | * @len: an int * to save the length of the attribute |
9189 | | * @alloc: an int * to indicate if the attribute was allocated |
9190 | | * |
9191 | | * parse an attribute in the new SAX2 framework. |
9192 | | * |
9193 | | * Returns the attribute name, and the value in *value, . |
9194 | | */ |
9195 | | |
9196 | | static const xmlChar * |
9197 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
9198 | | const xmlChar * pref, const xmlChar * elem, |
9199 | | const xmlChar ** prefix, xmlChar ** value, |
9200 | | int *len, int *alloc) |
9201 | 28.8M | { |
9202 | 28.8M | const xmlChar *name; |
9203 | 28.8M | xmlChar *val, *internal_val = NULL; |
9204 | 28.8M | int normalize = 0; |
9205 | | |
9206 | 28.8M | *value = NULL; |
9207 | 28.8M | GROW; |
9208 | 28.8M | name = xmlParseQName(ctxt, prefix); |
9209 | 28.8M | if (name == NULL) { |
9210 | 451k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9211 | 451k | "error parsing attribute name\n"); |
9212 | 451k | return (NULL); |
9213 | 451k | } |
9214 | | |
9215 | | /* |
9216 | | * get the type if needed |
9217 | | */ |
9218 | 28.4M | if (ctxt->attsSpecial != NULL) { |
9219 | 2.04M | int type; |
9220 | | |
9221 | 2.04M | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
9222 | 2.04M | pref, elem, *prefix, name); |
9223 | 2.04M | if (type != 0) |
9224 | 334k | normalize = 1; |
9225 | 2.04M | } |
9226 | | |
9227 | | /* |
9228 | | * read the value |
9229 | | */ |
9230 | 28.4M | SKIP_BLANKS; |
9231 | 28.4M | if (RAW == '=') { |
9232 | 28.1M | NEXT; |
9233 | 28.1M | SKIP_BLANKS; |
9234 | 28.1M | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize); |
9235 | 28.1M | if (normalize) { |
9236 | | /* |
9237 | | * Sometimes a second normalisation pass for spaces is needed |
9238 | | * but that only happens if charrefs or entities references |
9239 | | * have been used in the attribute value, i.e. the attribute |
9240 | | * value have been extracted in an allocated string already. |
9241 | | */ |
9242 | 333k | if (*alloc) { |
9243 | 70.5k | const xmlChar *val2; |
9244 | | |
9245 | 70.5k | val2 = xmlAttrNormalizeSpace2(ctxt, val, len); |
9246 | 70.5k | if ((val2 != NULL) && (val2 != val)) { |
9247 | 10.4k | xmlFree(val); |
9248 | 10.4k | val = (xmlChar *) val2; |
9249 | 10.4k | } |
9250 | 70.5k | } |
9251 | 333k | } |
9252 | 28.1M | ctxt->instate = XML_PARSER_CONTENT; |
9253 | 28.1M | } else { |
9254 | 251k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
9255 | 251k | "Specification mandates value for attribute %s\n", |
9256 | 251k | name); |
9257 | 251k | return (NULL); |
9258 | 251k | } |
9259 | | |
9260 | 28.1M | if (*prefix == ctxt->str_xml) { |
9261 | | /* |
9262 | | * Check that xml:lang conforms to the specification |
9263 | | * No more registered as an error, just generate a warning now |
9264 | | * since this was deprecated in XML second edition |
9265 | | */ |
9266 | 76.5k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
9267 | 19.7k | internal_val = xmlStrndup(val, *len); |
9268 | 19.7k | if (!xmlCheckLanguageID(internal_val)) { |
9269 | 9.02k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
9270 | 9.02k | "Malformed value for xml:lang : %s\n", |
9271 | 9.02k | internal_val, NULL); |
9272 | 9.02k | } |
9273 | 19.7k | } |
9274 | | |
9275 | | /* |
9276 | | * Check that xml:space conforms to the specification |
9277 | | */ |
9278 | 76.5k | if (xmlStrEqual(name, BAD_CAST "space")) { |
9279 | 814 | internal_val = xmlStrndup(val, *len); |
9280 | 814 | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
9281 | 0 | *(ctxt->space) = 0; |
9282 | 814 | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
9283 | 273 | *(ctxt->space) = 1; |
9284 | 541 | else { |
9285 | 541 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
9286 | 541 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
9287 | 541 | internal_val, NULL); |
9288 | 541 | } |
9289 | 814 | } |
9290 | 76.5k | if (internal_val) { |
9291 | 19.9k | xmlFree(internal_val); |
9292 | 19.9k | } |
9293 | 76.5k | } |
9294 | | |
9295 | 28.1M | *value = val; |
9296 | 28.1M | return (name); |
9297 | 28.4M | } |
9298 | | /** |
9299 | | * xmlParseStartTag2: |
9300 | | * @ctxt: an XML parser context |
9301 | | * |
9302 | | * parse a start of tag either for rule element or |
9303 | | * EmptyElement. In both case we don't parse the tag closing chars. |
9304 | | * This routine is called when running SAX2 parsing |
9305 | | * |
9306 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
9307 | | * |
9308 | | * [ WFC: Unique Att Spec ] |
9309 | | * No attribute name may appear more than once in the same start-tag or |
9310 | | * empty-element tag. |
9311 | | * |
9312 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
9313 | | * |
9314 | | * [ WFC: Unique Att Spec ] |
9315 | | * No attribute name may appear more than once in the same start-tag or |
9316 | | * empty-element tag. |
9317 | | * |
9318 | | * With namespace: |
9319 | | * |
9320 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
9321 | | * |
9322 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
9323 | | * |
9324 | | * Returns the element name parsed |
9325 | | */ |
9326 | | |
9327 | | static const xmlChar * |
9328 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
9329 | 25.2M | const xmlChar **URI, int *tlen) { |
9330 | 25.2M | const xmlChar *localname; |
9331 | 25.2M | const xmlChar *prefix; |
9332 | 25.2M | const xmlChar *attname; |
9333 | 25.2M | const xmlChar *aprefix; |
9334 | 25.2M | const xmlChar *nsname; |
9335 | 25.2M | xmlChar *attvalue; |
9336 | 25.2M | const xmlChar **atts = ctxt->atts; |
9337 | 25.2M | int maxatts = ctxt->maxatts; |
9338 | 25.2M | int nratts, nbatts, nbdef, inputid; |
9339 | 25.2M | int i, j, nbNs, attval; |
9340 | 25.2M | unsigned long cur; |
9341 | 25.2M | int nsNr = ctxt->nsNr; |
9342 | | |
9343 | 25.2M | if (RAW != '<') return(NULL); |
9344 | 25.2M | NEXT1; |
9345 | | |
9346 | | /* |
9347 | | * NOTE: it is crucial with the SAX2 API to never call SHRINK beyond that |
9348 | | * point since the attribute values may be stored as pointers to |
9349 | | * the buffer and calling SHRINK would destroy them ! |
9350 | | * The Shrinking is only possible once the full set of attribute |
9351 | | * callbacks have been done. |
9352 | | */ |
9353 | 25.2M | SHRINK; |
9354 | 25.2M | cur = ctxt->input->cur - ctxt->input->base; |
9355 | 25.2M | inputid = ctxt->input->id; |
9356 | 25.2M | nbatts = 0; |
9357 | 25.2M | nratts = 0; |
9358 | 25.2M | nbdef = 0; |
9359 | 25.2M | nbNs = 0; |
9360 | 25.2M | attval = 0; |
9361 | | /* Forget any namespaces added during an earlier parse of this element. */ |
9362 | 25.2M | ctxt->nsNr = nsNr; |
9363 | | |
9364 | 25.2M | localname = xmlParseQName(ctxt, &prefix); |
9365 | 25.2M | if (localname == NULL) { |
9366 | 5.41M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9367 | 5.41M | "StartTag: invalid element name\n"); |
9368 | 5.41M | return(NULL); |
9369 | 5.41M | } |
9370 | 19.8M | *tlen = ctxt->input->cur - ctxt->input->base - cur; |
9371 | | |
9372 | | /* |
9373 | | * Now parse the attributes, it ends up with the ending |
9374 | | * |
9375 | | * (S Attribute)* S? |
9376 | | */ |
9377 | 19.8M | SKIP_BLANKS; |
9378 | 19.8M | GROW; |
9379 | | |
9380 | 32.5M | while (((RAW != '>') && |
9381 | 32.5M | ((RAW != '/') || (NXT(1) != '>')) && |
9382 | 32.5M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
9383 | 28.8M | int id = ctxt->input->id; |
9384 | 28.8M | unsigned long cons = CUR_CONSUMED; |
9385 | 28.8M | int len = -1, alloc = 0; |
9386 | | |
9387 | 28.8M | attname = xmlParseAttribute2(ctxt, prefix, localname, |
9388 | 28.8M | &aprefix, &attvalue, &len, &alloc); |
9389 | 28.8M | if ((attname == NULL) || (attvalue == NULL)) |
9390 | 737k | goto next_attr; |
9391 | 28.1M | if (len < 0) len = xmlStrlen(attvalue); |
9392 | | |
9393 | 28.1M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9394 | 115k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9395 | 115k | xmlURIPtr uri; |
9396 | | |
9397 | 115k | if (URL == NULL) { |
9398 | 0 | xmlErrMemory(ctxt, "dictionary allocation failure"); |
9399 | 0 | if ((attvalue != NULL) && (alloc != 0)) |
9400 | 0 | xmlFree(attvalue); |
9401 | 0 | localname = NULL; |
9402 | 0 | goto done; |
9403 | 0 | } |
9404 | 115k | if (*URL != 0) { |
9405 | 111k | uri = xmlParseURI((const char *) URL); |
9406 | 111k | if (uri == NULL) { |
9407 | 21.4k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9408 | 21.4k | "xmlns: '%s' is not a valid URI\n", |
9409 | 21.4k | URL, NULL, NULL); |
9410 | 90.4k | } else { |
9411 | 90.4k | if (uri->scheme == NULL) { |
9412 | 44.0k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9413 | 44.0k | "xmlns: URI %s is not absolute\n", |
9414 | 44.0k | URL, NULL, NULL); |
9415 | 44.0k | } |
9416 | 90.4k | xmlFreeURI(uri); |
9417 | 90.4k | } |
9418 | 111k | if (URL == ctxt->str_xml_ns) { |
9419 | 0 | if (attname != ctxt->str_xml) { |
9420 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9421 | 0 | "xml namespace URI cannot be the default namespace\n", |
9422 | 0 | NULL, NULL, NULL); |
9423 | 0 | } |
9424 | 0 | goto next_attr; |
9425 | 0 | } |
9426 | 111k | if ((len == 29) && |
9427 | 111k | (xmlStrEqual(URL, |
9428 | 1.74k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9429 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9430 | 0 | "reuse of the xmlns namespace name is forbidden\n", |
9431 | 0 | NULL, NULL, NULL); |
9432 | 0 | goto next_attr; |
9433 | 0 | } |
9434 | 111k | } |
9435 | | /* |
9436 | | * check that it's not a defined namespace |
9437 | | */ |
9438 | 134k | for (j = 1;j <= nbNs;j++) |
9439 | 26.7k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9440 | 7.51k | break; |
9441 | 115k | if (j <= nbNs) |
9442 | 7.51k | xmlErrAttributeDup(ctxt, NULL, attname); |
9443 | 108k | else |
9444 | 108k | if (nsPush(ctxt, NULL, URL) > 0) nbNs++; |
9445 | | |
9446 | 28.0M | } else if (aprefix == ctxt->str_xmlns) { |
9447 | 408k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9448 | 408k | xmlURIPtr uri; |
9449 | | |
9450 | 408k | if (attname == ctxt->str_xml) { |
9451 | 937 | if (URL != ctxt->str_xml_ns) { |
9452 | 937 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9453 | 937 | "xml namespace prefix mapped to wrong URI\n", |
9454 | 937 | NULL, NULL, NULL); |
9455 | 937 | } |
9456 | | /* |
9457 | | * Do not keep a namespace definition node |
9458 | | */ |
9459 | 937 | goto next_attr; |
9460 | 937 | } |
9461 | 407k | if (URL == ctxt->str_xml_ns) { |
9462 | 0 | if (attname != ctxt->str_xml) { |
9463 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9464 | 0 | "xml namespace URI mapped to wrong prefix\n", |
9465 | 0 | NULL, NULL, NULL); |
9466 | 0 | } |
9467 | 0 | goto next_attr; |
9468 | 0 | } |
9469 | 407k | if (attname == ctxt->str_xmlns) { |
9470 | 1.22k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9471 | 1.22k | "redefinition of the xmlns prefix is forbidden\n", |
9472 | 1.22k | NULL, NULL, NULL); |
9473 | 1.22k | goto next_attr; |
9474 | 1.22k | } |
9475 | 405k | if ((len == 29) && |
9476 | 405k | (xmlStrEqual(URL, |
9477 | 5.05k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9478 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9479 | 0 | "reuse of the xmlns namespace name is forbidden\n", |
9480 | 0 | NULL, NULL, NULL); |
9481 | 0 | goto next_attr; |
9482 | 0 | } |
9483 | 405k | if ((URL == NULL) || (URL[0] == 0)) { |
9484 | 5.05k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9485 | 5.05k | "xmlns:%s: Empty XML namespace is not allowed\n", |
9486 | 5.05k | attname, NULL, NULL); |
9487 | 5.05k | goto next_attr; |
9488 | 400k | } else { |
9489 | 400k | uri = xmlParseURI((const char *) URL); |
9490 | 400k | if (uri == NULL) { |
9491 | 73.4k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9492 | 73.4k | "xmlns:%s: '%s' is not a valid URI\n", |
9493 | 73.4k | attname, URL, NULL); |
9494 | 327k | } else { |
9495 | 327k | if ((ctxt->pedantic) && (uri->scheme == NULL)) { |
9496 | 7.92k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9497 | 7.92k | "xmlns:%s: URI %s is not absolute\n", |
9498 | 7.92k | attname, URL, NULL); |
9499 | 7.92k | } |
9500 | 327k | xmlFreeURI(uri); |
9501 | 327k | } |
9502 | 400k | } |
9503 | | |
9504 | | /* |
9505 | | * check that it's not a defined namespace |
9506 | | */ |
9507 | 466k | for (j = 1;j <= nbNs;j++) |
9508 | 72.3k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9509 | 6.80k | break; |
9510 | 400k | if (j <= nbNs) |
9511 | 6.80k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9512 | 393k | else |
9513 | 393k | if (nsPush(ctxt, attname, URL) > 0) nbNs++; |
9514 | | |
9515 | 27.6M | } else { |
9516 | | /* |
9517 | | * Add the pair to atts |
9518 | | */ |
9519 | 27.6M | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9520 | 339k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9521 | 0 | goto next_attr; |
9522 | 0 | } |
9523 | 339k | maxatts = ctxt->maxatts; |
9524 | 339k | atts = ctxt->atts; |
9525 | 339k | } |
9526 | 27.6M | ctxt->attallocs[nratts++] = alloc; |
9527 | 27.6M | atts[nbatts++] = attname; |
9528 | 27.6M | atts[nbatts++] = aprefix; |
9529 | | /* |
9530 | | * The namespace URI field is used temporarily to point at the |
9531 | | * base of the current input buffer for non-alloced attributes. |
9532 | | * When the input buffer is reallocated, all the pointers become |
9533 | | * invalid, but they can be reconstructed later. |
9534 | | */ |
9535 | 27.6M | if (alloc) |
9536 | 772k | atts[nbatts++] = NULL; |
9537 | 26.8M | else |
9538 | 26.8M | atts[nbatts++] = ctxt->input->base; |
9539 | 27.6M | atts[nbatts++] = attvalue; |
9540 | 27.6M | attvalue += len; |
9541 | 27.6M | atts[nbatts++] = attvalue; |
9542 | | /* |
9543 | | * tag if some deallocation is needed |
9544 | | */ |
9545 | 27.6M | if (alloc != 0) attval = 1; |
9546 | 27.6M | attvalue = NULL; /* moved into atts */ |
9547 | 27.6M | } |
9548 | | |
9549 | 28.8M | next_attr: |
9550 | 28.8M | if ((attvalue != NULL) && (alloc != 0)) { |
9551 | 95.5k | xmlFree(attvalue); |
9552 | 95.5k | attvalue = NULL; |
9553 | 95.5k | } |
9554 | | |
9555 | 28.8M | GROW |
9556 | 28.8M | if (ctxt->instate == XML_PARSER_EOF) |
9557 | 0 | break; |
9558 | 28.8M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9559 | 14.9M | break; |
9560 | 13.8M | if (SKIP_BLANKS == 0) { |
9561 | 1.14M | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9562 | 1.14M | "attributes construct error\n"); |
9563 | 1.14M | break; |
9564 | 1.14M | } |
9565 | 12.7M | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
9566 | 12.7M | (attname == NULL) && (attvalue == NULL)) { |
9567 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9568 | 0 | "xmlParseStartTag: problem parsing attributes\n"); |
9569 | 0 | break; |
9570 | 0 | } |
9571 | 12.7M | GROW; |
9572 | 12.7M | } |
9573 | | |
9574 | 19.8M | if (ctxt->input->id != inputid) { |
9575 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9576 | 0 | "Unexpected change of input\n"); |
9577 | 0 | localname = NULL; |
9578 | 0 | goto done; |
9579 | 0 | } |
9580 | | |
9581 | | /* Reconstruct attribute value pointers. */ |
9582 | 47.4M | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9583 | 27.6M | if (atts[i+2] != NULL) { |
9584 | | /* |
9585 | | * Arithmetic on dangling pointers is technically undefined |
9586 | | * behavior, but well... |
9587 | | */ |
9588 | 26.8M | ptrdiff_t offset = ctxt->input->base - atts[i+2]; |
9589 | 26.8M | atts[i+2] = NULL; /* Reset repurposed namespace URI */ |
9590 | 26.8M | atts[i+3] += offset; /* value */ |
9591 | 26.8M | atts[i+4] += offset; /* valuend */ |
9592 | 26.8M | } |
9593 | 27.6M | } |
9594 | | |
9595 | | /* |
9596 | | * The attributes defaulting |
9597 | | */ |
9598 | 19.8M | if (ctxt->attsDefault != NULL) { |
9599 | 1.52M | xmlDefAttrsPtr defaults; |
9600 | | |
9601 | 1.52M | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9602 | 1.52M | if (defaults != NULL) { |
9603 | 191k | for (i = 0;i < defaults->nbAttrs;i++) { |
9604 | 128k | attname = defaults->values[5 * i]; |
9605 | 128k | aprefix = defaults->values[5 * i + 1]; |
9606 | | |
9607 | | /* |
9608 | | * special work for namespaces defaulted defs |
9609 | | */ |
9610 | 128k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9611 | | /* |
9612 | | * check that it's not a defined namespace |
9613 | | */ |
9614 | 1.64k | for (j = 1;j <= nbNs;j++) |
9615 | 985 | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9616 | 516 | break; |
9617 | 1.17k | if (j <= nbNs) continue; |
9618 | | |
9619 | 662 | nsname = xmlGetNamespace(ctxt, NULL); |
9620 | 662 | if (nsname != defaults->values[5 * i + 2]) { |
9621 | 594 | if (nsPush(ctxt, NULL, |
9622 | 594 | defaults->values[5 * i + 2]) > 0) |
9623 | 592 | nbNs++; |
9624 | 594 | } |
9625 | 126k | } else if (aprefix == ctxt->str_xmlns) { |
9626 | | /* |
9627 | | * check that it's not a defined namespace |
9628 | | */ |
9629 | 17.2k | for (j = 1;j <= nbNs;j++) |
9630 | 13.1k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9631 | 12.2k | break; |
9632 | 16.3k | if (j <= nbNs) continue; |
9633 | | |
9634 | 4.14k | nsname = xmlGetNamespace(ctxt, attname); |
9635 | 4.14k | if (nsname != defaults->values[2]) { |
9636 | 3.42k | if (nsPush(ctxt, attname, |
9637 | 3.42k | defaults->values[5 * i + 2]) > 0) |
9638 | 3.04k | nbNs++; |
9639 | 3.42k | } |
9640 | 110k | } else { |
9641 | | /* |
9642 | | * check that it's not a defined attribute |
9643 | | */ |
9644 | 298k | for (j = 0;j < nbatts;j+=5) { |
9645 | 189k | if ((attname == atts[j]) && (aprefix == atts[j+1])) |
9646 | 1.69k | break; |
9647 | 189k | } |
9648 | 110k | if (j < nbatts) continue; |
9649 | | |
9650 | 108k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9651 | 4.09k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9652 | 0 | localname = NULL; |
9653 | 0 | goto done; |
9654 | 0 | } |
9655 | 4.09k | maxatts = ctxt->maxatts; |
9656 | 4.09k | atts = ctxt->atts; |
9657 | 4.09k | } |
9658 | 108k | atts[nbatts++] = attname; |
9659 | 108k | atts[nbatts++] = aprefix; |
9660 | 108k | if (aprefix == NULL) |
9661 | 81.4k | atts[nbatts++] = NULL; |
9662 | 27.3k | else |
9663 | 27.3k | atts[nbatts++] = xmlGetNamespace(ctxt, aprefix); |
9664 | 108k | atts[nbatts++] = defaults->values[5 * i + 2]; |
9665 | 108k | atts[nbatts++] = defaults->values[5 * i + 3]; |
9666 | 108k | if ((ctxt->standalone == 1) && |
9667 | 108k | (defaults->values[5 * i + 4] != NULL)) { |
9668 | 407 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9669 | 407 | "standalone: attribute %s on %s defaulted from external subset\n", |
9670 | 407 | attname, localname); |
9671 | 407 | } |
9672 | 108k | nbdef++; |
9673 | 108k | } |
9674 | 128k | } |
9675 | 63.0k | } |
9676 | 1.52M | } |
9677 | | |
9678 | | /* |
9679 | | * The attributes checkings |
9680 | | */ |
9681 | 47.5M | for (i = 0; i < nbatts;i += 5) { |
9682 | | /* |
9683 | | * The default namespace does not apply to attribute names. |
9684 | | */ |
9685 | 27.7M | if (atts[i + 1] != NULL) { |
9686 | 1.06M | nsname = xmlGetNamespace(ctxt, atts[i + 1]); |
9687 | 1.06M | if (nsname == NULL) { |
9688 | 305k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9689 | 305k | "Namespace prefix %s for %s on %s is not defined\n", |
9690 | 305k | atts[i + 1], atts[i], localname); |
9691 | 305k | } |
9692 | 1.06M | atts[i + 2] = nsname; |
9693 | 1.06M | } else |
9694 | 26.6M | nsname = NULL; |
9695 | | /* |
9696 | | * [ WFC: Unique Att Spec ] |
9697 | | * No attribute name may appear more than once in the same |
9698 | | * start-tag or empty-element tag. |
9699 | | * As extended by the Namespace in XML REC. |
9700 | | */ |
9701 | 40.7M | for (j = 0; j < i;j += 5) { |
9702 | 13.0M | if (atts[i] == atts[j]) { |
9703 | 38.2k | if (atts[i+1] == atts[j+1]) { |
9704 | 18.1k | xmlErrAttributeDup(ctxt, atts[i+1], atts[i]); |
9705 | 18.1k | break; |
9706 | 18.1k | } |
9707 | 20.1k | if ((nsname != NULL) && (atts[j + 2] == nsname)) { |
9708 | 193 | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9709 | 193 | "Namespaced Attribute %s in '%s' redefined\n", |
9710 | 193 | atts[i], nsname, NULL); |
9711 | 193 | break; |
9712 | 193 | } |
9713 | 20.1k | } |
9714 | 13.0M | } |
9715 | 27.7M | } |
9716 | | |
9717 | 19.8M | nsname = xmlGetNamespace(ctxt, prefix); |
9718 | 19.8M | if ((prefix != NULL) && (nsname == NULL)) { |
9719 | 868k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9720 | 868k | "Namespace prefix %s on %s is not defined\n", |
9721 | 868k | prefix, localname, NULL); |
9722 | 868k | } |
9723 | 19.8M | *pref = prefix; |
9724 | 19.8M | *URI = nsname; |
9725 | | |
9726 | | /* |
9727 | | * SAX: Start of Element ! |
9728 | | */ |
9729 | 19.8M | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9730 | 19.8M | (!ctxt->disableSAX)) { |
9731 | 8.06M | if (nbNs > 0) |
9732 | 336k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9733 | 336k | nsname, nbNs, &ctxt->nsTab[ctxt->nsNr - 2 * nbNs], |
9734 | 336k | nbatts / 5, nbdef, atts); |
9735 | 7.72M | else |
9736 | 7.72M | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9737 | 7.72M | nsname, 0, NULL, nbatts / 5, nbdef, atts); |
9738 | 8.06M | } |
9739 | | |
9740 | 19.8M | done: |
9741 | | /* |
9742 | | * Free up attribute allocated strings if needed |
9743 | | */ |
9744 | 19.8M | if (attval != 0) { |
9745 | 1.61M | for (i = 3,j = 0; j < nratts;i += 5,j++) |
9746 | 883k | if ((ctxt->attallocs[j] != 0) && (atts[i] != NULL)) |
9747 | 772k | xmlFree((xmlChar *) atts[i]); |
9748 | 734k | } |
9749 | | |
9750 | 19.8M | return(localname); |
9751 | 19.8M | } |
9752 | | |
9753 | | /** |
9754 | | * xmlParseEndTag2: |
9755 | | * @ctxt: an XML parser context |
9756 | | * @line: line of the start tag |
9757 | | * @nsNr: number of namespaces on the start tag |
9758 | | * |
9759 | | * parse an end of tag |
9760 | | * |
9761 | | * [42] ETag ::= '</' Name S? '>' |
9762 | | * |
9763 | | * With namespace |
9764 | | * |
9765 | | * [NS 9] ETag ::= '</' QName S? '>' |
9766 | | */ |
9767 | | |
9768 | | static void |
9769 | 3.71M | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9770 | 3.71M | const xmlChar *name; |
9771 | | |
9772 | 3.71M | GROW; |
9773 | 3.71M | if ((RAW != '<') || (NXT(1) != '/')) { |
9774 | 0 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9775 | 0 | return; |
9776 | 0 | } |
9777 | 3.71M | SKIP(2); |
9778 | | |
9779 | 3.71M | if (tag->prefix == NULL) |
9780 | 2.99M | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9781 | 723k | else |
9782 | 723k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9783 | | |
9784 | | /* |
9785 | | * We should definitely be at the ending "S? '>'" part |
9786 | | */ |
9787 | 3.71M | GROW; |
9788 | 3.71M | if (ctxt->instate == XML_PARSER_EOF) |
9789 | 0 | return; |
9790 | 3.71M | SKIP_BLANKS; |
9791 | 3.71M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9792 | 201k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9793 | 201k | } else |
9794 | 3.51M | NEXT1; |
9795 | | |
9796 | | /* |
9797 | | * [ WFC: Element Type Match ] |
9798 | | * The Name in an element's end-tag must match the element type in the |
9799 | | * start-tag. |
9800 | | * |
9801 | | */ |
9802 | 3.71M | if (name != (xmlChar*)1) { |
9803 | 558k | if (name == NULL) name = BAD_CAST "unparsable"; |
9804 | 558k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9805 | 558k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9806 | 558k | ctxt->name, tag->line, name); |
9807 | 558k | } |
9808 | | |
9809 | | /* |
9810 | | * SAX: End of Tag |
9811 | | */ |
9812 | 3.71M | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9813 | 3.71M | (!ctxt->disableSAX)) |
9814 | 2.29M | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9815 | 2.29M | tag->URI); |
9816 | | |
9817 | 3.71M | spacePop(ctxt); |
9818 | 3.71M | if (tag->nsNr != 0) |
9819 | 114k | nsPop(ctxt, tag->nsNr); |
9820 | 3.71M | } |
9821 | | |
9822 | | /** |
9823 | | * xmlParseCDSect: |
9824 | | * @ctxt: an XML parser context |
9825 | | * |
9826 | | * DEPRECATED: Internal function, don't use. |
9827 | | * |
9828 | | * Parse escaped pure raw content. |
9829 | | * |
9830 | | * [18] CDSect ::= CDStart CData CDEnd |
9831 | | * |
9832 | | * [19] CDStart ::= '<![CDATA[' |
9833 | | * |
9834 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9835 | | * |
9836 | | * [21] CDEnd ::= ']]>' |
9837 | | */ |
9838 | | void |
9839 | 736k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9840 | 736k | xmlChar *buf = NULL; |
9841 | 736k | int len = 0; |
9842 | 736k | int size = XML_PARSER_BUFFER_SIZE; |
9843 | 736k | int r, rl; |
9844 | 736k | int s, sl; |
9845 | 736k | int cur, l; |
9846 | 736k | int count = 0; |
9847 | 736k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9848 | 583k | XML_MAX_HUGE_LENGTH : |
9849 | 736k | XML_MAX_TEXT_LENGTH; |
9850 | | |
9851 | | /* Check 2.6.0 was NXT(0) not RAW */ |
9852 | 736k | if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9853 | 736k | SKIP(9); |
9854 | 736k | } else |
9855 | 0 | return; |
9856 | | |
9857 | 736k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
9858 | 736k | r = CUR_CHAR(rl); |
9859 | 736k | if (!IS_CHAR(r)) { |
9860 | 66.9k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9861 | 66.9k | ctxt->instate = XML_PARSER_CONTENT; |
9862 | 66.9k | return; |
9863 | 66.9k | } |
9864 | 669k | NEXTL(rl); |
9865 | 669k | s = CUR_CHAR(sl); |
9866 | 669k | if (!IS_CHAR(s)) { |
9867 | 16.1k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9868 | 16.1k | ctxt->instate = XML_PARSER_CONTENT; |
9869 | 16.1k | return; |
9870 | 16.1k | } |
9871 | 653k | NEXTL(sl); |
9872 | 653k | cur = CUR_CHAR(l); |
9873 | 653k | buf = (xmlChar *) xmlMallocAtomic(size); |
9874 | 653k | if (buf == NULL) { |
9875 | 0 | xmlErrMemory(ctxt, NULL); |
9876 | 0 | return; |
9877 | 0 | } |
9878 | 215M | while (IS_CHAR(cur) && |
9879 | 215M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9880 | 215M | if (len + 5 >= size) { |
9881 | 749k | xmlChar *tmp; |
9882 | | |
9883 | 749k | tmp = (xmlChar *) xmlRealloc(buf, size * 2); |
9884 | 749k | if (tmp == NULL) { |
9885 | 0 | xmlFree(buf); |
9886 | 0 | xmlErrMemory(ctxt, NULL); |
9887 | 0 | return; |
9888 | 0 | } |
9889 | 749k | buf = tmp; |
9890 | 749k | size *= 2; |
9891 | 749k | } |
9892 | 215M | COPY_BUF(rl,buf,len,r); |
9893 | 215M | r = s; |
9894 | 215M | rl = sl; |
9895 | 215M | s = cur; |
9896 | 215M | sl = l; |
9897 | 215M | count++; |
9898 | 215M | if (count > 50) { |
9899 | 4.01M | SHRINK; |
9900 | 4.01M | GROW; |
9901 | 4.01M | if (ctxt->instate == XML_PARSER_EOF) { |
9902 | 0 | xmlFree(buf); |
9903 | 0 | return; |
9904 | 0 | } |
9905 | 4.01M | count = 0; |
9906 | 4.01M | } |
9907 | 215M | NEXTL(l); |
9908 | 215M | cur = CUR_CHAR(l); |
9909 | 215M | if (len > maxLength) { |
9910 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9911 | 0 | "CData section too big found\n"); |
9912 | 0 | xmlFree(buf); |
9913 | 0 | return; |
9914 | 0 | } |
9915 | 215M | } |
9916 | 653k | buf[len] = 0; |
9917 | 653k | ctxt->instate = XML_PARSER_CONTENT; |
9918 | 653k | if (cur != '>') { |
9919 | 180k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9920 | 180k | "CData section not finished\n%.50s\n", buf); |
9921 | 180k | xmlFree(buf); |
9922 | 180k | return; |
9923 | 180k | } |
9924 | 472k | NEXTL(l); |
9925 | | |
9926 | | /* |
9927 | | * OK the buffer is to be consumed as cdata. |
9928 | | */ |
9929 | 472k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9930 | 63.5k | if (ctxt->sax->cdataBlock != NULL) |
9931 | 33.1k | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9932 | 30.3k | else if (ctxt->sax->characters != NULL) |
9933 | 30.3k | ctxt->sax->characters(ctxt->userData, buf, len); |
9934 | 63.5k | } |
9935 | 472k | xmlFree(buf); |
9936 | 472k | } |
9937 | | |
9938 | | /** |
9939 | | * xmlParseContentInternal: |
9940 | | * @ctxt: an XML parser context |
9941 | | * |
9942 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9943 | | * unexpected EOF to the caller. |
9944 | | */ |
9945 | | |
9946 | | static void |
9947 | 4.82M | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9948 | 4.82M | int nameNr = ctxt->nameNr; |
9949 | | |
9950 | 4.82M | GROW; |
9951 | 551M | while ((RAW != 0) && |
9952 | 551M | (ctxt->instate != XML_PARSER_EOF)) { |
9953 | 547M | int id = ctxt->input->id; |
9954 | 547M | unsigned long cons = CUR_CONSUMED; |
9955 | 547M | const xmlChar *cur = ctxt->input->cur; |
9956 | | |
9957 | | /* |
9958 | | * First case : a Processing Instruction. |
9959 | | */ |
9960 | 547M | if ((*cur == '<') && (cur[1] == '?')) { |
9961 | 3.23M | xmlParsePI(ctxt); |
9962 | 3.23M | } |
9963 | | |
9964 | | /* |
9965 | | * Second case : a CDSection |
9966 | | */ |
9967 | | /* 2.6.0 test was *cur not RAW */ |
9968 | 544M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9969 | 736k | xmlParseCDSect(ctxt); |
9970 | 736k | } |
9971 | | |
9972 | | /* |
9973 | | * Third case : a comment |
9974 | | */ |
9975 | 543M | else if ((*cur == '<') && (NXT(1) == '!') && |
9976 | 543M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9977 | 11.0M | xmlParseComment(ctxt); |
9978 | 11.0M | ctxt->instate = XML_PARSER_CONTENT; |
9979 | 11.0M | } |
9980 | | |
9981 | | /* |
9982 | | * Fourth case : a sub-element. |
9983 | | */ |
9984 | 532M | else if (*cur == '<') { |
9985 | 203M | if (NXT(1) == '/') { |
9986 | 36.5M | if (ctxt->nameNr <= nameNr) |
9987 | 686k | break; |
9988 | 35.8M | xmlParseElementEnd(ctxt); |
9989 | 166M | } else { |
9990 | 166M | xmlParseElementStart(ctxt); |
9991 | 166M | } |
9992 | 203M | } |
9993 | | |
9994 | | /* |
9995 | | * Fifth case : a reference. If if has not been resolved, |
9996 | | * parsing returns it's Name, create the node |
9997 | | */ |
9998 | | |
9999 | 329M | else if (*cur == '&') { |
10000 | 38.3M | xmlParseReference(ctxt); |
10001 | 38.3M | } |
10002 | | |
10003 | | /* |
10004 | | * Last case, text. Note that References are handled directly. |
10005 | | */ |
10006 | 291M | else { |
10007 | 291M | xmlParseCharData(ctxt, 0); |
10008 | 291M | } |
10009 | | |
10010 | 547M | GROW; |
10011 | 547M | SHRINK; |
10012 | | |
10013 | 547M | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
10014 | 82.9k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
10015 | 82.9k | "detected an error in element content\n"); |
10016 | 82.9k | xmlHaltParser(ctxt); |
10017 | 82.9k | break; |
10018 | 82.9k | } |
10019 | 547M | } |
10020 | 4.82M | } |
10021 | | |
10022 | | /** |
10023 | | * xmlParseContent: |
10024 | | * @ctxt: an XML parser context |
10025 | | * |
10026 | | * Parse a content sequence. Stops at EOF or '</'. |
10027 | | * |
10028 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
10029 | | */ |
10030 | | |
10031 | | void |
10032 | 4.54M | xmlParseContent(xmlParserCtxtPtr ctxt) { |
10033 | 4.54M | int nameNr = ctxt->nameNr; |
10034 | | |
10035 | 4.54M | xmlParseContentInternal(ctxt); |
10036 | | |
10037 | 4.54M | if ((ctxt->instate != XML_PARSER_EOF) && (ctxt->nameNr > nameNr)) { |
10038 | 2.05M | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
10039 | 2.05M | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
10040 | 2.05M | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
10041 | 2.05M | "Premature end of data in tag %s line %d\n", |
10042 | 2.05M | name, line, NULL); |
10043 | 2.05M | } |
10044 | 4.54M | } |
10045 | | |
10046 | | /** |
10047 | | * xmlParseElement: |
10048 | | * @ctxt: an XML parser context |
10049 | | * |
10050 | | * DEPRECATED: Internal function, don't use. |
10051 | | * |
10052 | | * parse an XML element |
10053 | | * |
10054 | | * [39] element ::= EmptyElemTag | STag content ETag |
10055 | | * |
10056 | | * [ WFC: Element Type Match ] |
10057 | | * The Name in an element's end-tag must match the element type in the |
10058 | | * start-tag. |
10059 | | * |
10060 | | */ |
10061 | | |
10062 | | void |
10063 | 379k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
10064 | 379k | if (xmlParseElementStart(ctxt) != 0) |
10065 | 97.3k | return; |
10066 | | |
10067 | 282k | xmlParseContentInternal(ctxt); |
10068 | 282k | if (ctxt->instate == XML_PARSER_EOF) |
10069 | 2.76k | return; |
10070 | | |
10071 | 279k | if (CUR == 0) { |
10072 | 169k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
10073 | 169k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
10074 | 169k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
10075 | 169k | "Premature end of data in tag %s line %d\n", |
10076 | 169k | name, line, NULL); |
10077 | 169k | return; |
10078 | 169k | } |
10079 | | |
10080 | 109k | xmlParseElementEnd(ctxt); |
10081 | 109k | } |
10082 | | |
10083 | | /** |
10084 | | * xmlParseElementStart: |
10085 | | * @ctxt: an XML parser context |
10086 | | * |
10087 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
10088 | | * opening tag was parsed, 1 if an empty element was parsed. |
10089 | | */ |
10090 | | static int |
10091 | 166M | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
10092 | 166M | const xmlChar *name; |
10093 | 166M | const xmlChar *prefix = NULL; |
10094 | 166M | const xmlChar *URI = NULL; |
10095 | 166M | xmlParserNodeInfo node_info; |
10096 | 166M | int line, tlen = 0; |
10097 | 166M | xmlNodePtr ret; |
10098 | 166M | int nsNr = ctxt->nsNr; |
10099 | | |
10100 | 166M | if (((unsigned int) ctxt->nameNr > xmlParserMaxDepth) && |
10101 | 166M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
10102 | 21 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
10103 | 21 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
10104 | 21 | xmlParserMaxDepth); |
10105 | 21 | xmlHaltParser(ctxt); |
10106 | 21 | return(-1); |
10107 | 21 | } |
10108 | | |
10109 | | /* Capture start position */ |
10110 | 166M | if (ctxt->record_info) { |
10111 | 0 | node_info.begin_pos = ctxt->input->consumed + |
10112 | 0 | (CUR_PTR - ctxt->input->base); |
10113 | 0 | node_info.begin_line = ctxt->input->line; |
10114 | 0 | } |
10115 | | |
10116 | 166M | if (ctxt->spaceNr == 0) |
10117 | 0 | spacePush(ctxt, -1); |
10118 | 166M | else if (*ctxt->space == -2) |
10119 | 35.9M | spacePush(ctxt, -1); |
10120 | 130M | else |
10121 | 130M | spacePush(ctxt, *ctxt->space); |
10122 | | |
10123 | 166M | line = ctxt->input->line; |
10124 | 166M | #ifdef LIBXML_SAX1_ENABLED |
10125 | 166M | if (ctxt->sax2) |
10126 | 21.6M | #endif /* LIBXML_SAX1_ENABLED */ |
10127 | 21.6M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
10128 | 145M | #ifdef LIBXML_SAX1_ENABLED |
10129 | 145M | else |
10130 | 145M | name = xmlParseStartTag(ctxt); |
10131 | 166M | #endif /* LIBXML_SAX1_ENABLED */ |
10132 | 166M | if (ctxt->instate == XML_PARSER_EOF) |
10133 | 7.16k | return(-1); |
10134 | 166M | if (name == NULL) { |
10135 | 69.0M | spacePop(ctxt); |
10136 | 69.0M | return(-1); |
10137 | 69.0M | } |
10138 | 97.8M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
10139 | 97.8M | ret = ctxt->node; |
10140 | | |
10141 | 97.8M | #ifdef LIBXML_VALID_ENABLED |
10142 | | /* |
10143 | | * [ VC: Root Element Type ] |
10144 | | * The Name in the document type declaration must match the element |
10145 | | * type of the root element. |
10146 | | */ |
10147 | 97.8M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10148 | 97.8M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10149 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10150 | 97.8M | #endif /* LIBXML_VALID_ENABLED */ |
10151 | | |
10152 | | /* |
10153 | | * Check for an Empty Element. |
10154 | | */ |
10155 | 97.8M | if ((RAW == '/') && (NXT(1) == '>')) { |
10156 | 42.0M | SKIP(2); |
10157 | 42.0M | if (ctxt->sax2) { |
10158 | 11.8M | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
10159 | 11.8M | (!ctxt->disableSAX)) |
10160 | 1.94M | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
10161 | 11.8M | #ifdef LIBXML_SAX1_ENABLED |
10162 | 30.2M | } else { |
10163 | 30.2M | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
10164 | 30.2M | (!ctxt->disableSAX)) |
10165 | 5.35M | ctxt->sax->endElement(ctxt->userData, name); |
10166 | 30.2M | #endif /* LIBXML_SAX1_ENABLED */ |
10167 | 30.2M | } |
10168 | 42.0M | namePop(ctxt); |
10169 | 42.0M | spacePop(ctxt); |
10170 | 42.0M | if (nsNr != ctxt->nsNr) |
10171 | 37.3k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10172 | 42.0M | if ( ret != NULL && ctxt->record_info ) { |
10173 | 0 | node_info.end_pos = ctxt->input->consumed + |
10174 | 0 | (CUR_PTR - ctxt->input->base); |
10175 | 0 | node_info.end_line = ctxt->input->line; |
10176 | 0 | node_info.node = ret; |
10177 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10178 | 0 | } |
10179 | 42.0M | return(1); |
10180 | 42.0M | } |
10181 | 55.7M | if (RAW == '>') { |
10182 | 45.4M | NEXT1; |
10183 | 45.4M | } else { |
10184 | 10.3M | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
10185 | 10.3M | "Couldn't find end of Start Tag %s line %d\n", |
10186 | 10.3M | name, line, NULL); |
10187 | | |
10188 | | /* |
10189 | | * end of parsing of this node. |
10190 | | */ |
10191 | 10.3M | nodePop(ctxt); |
10192 | 10.3M | namePop(ctxt); |
10193 | 10.3M | spacePop(ctxt); |
10194 | 10.3M | if (nsNr != ctxt->nsNr) |
10195 | 65.5k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10196 | | |
10197 | | /* |
10198 | | * Capture end position and add node |
10199 | | */ |
10200 | 10.3M | if ( ret != NULL && ctxt->record_info ) { |
10201 | 0 | node_info.end_pos = ctxt->input->consumed + |
10202 | 0 | (CUR_PTR - ctxt->input->base); |
10203 | 0 | node_info.end_line = ctxt->input->line; |
10204 | 0 | node_info.node = ret; |
10205 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10206 | 0 | } |
10207 | 10.3M | return(-1); |
10208 | 10.3M | } |
10209 | | |
10210 | 45.4M | return(0); |
10211 | 55.7M | } |
10212 | | |
10213 | | /** |
10214 | | * xmlParseElementEnd: |
10215 | | * @ctxt: an XML parser context |
10216 | | * |
10217 | | * Parse the end of an XML element. |
10218 | | */ |
10219 | | static void |
10220 | 35.9M | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
10221 | 35.9M | xmlParserNodeInfo node_info; |
10222 | 35.9M | xmlNodePtr ret = ctxt->node; |
10223 | | |
10224 | 35.9M | if (ctxt->nameNr <= 0) |
10225 | 0 | return; |
10226 | | |
10227 | | /* |
10228 | | * parse the end of tag: '</' should be here. |
10229 | | */ |
10230 | 35.9M | if (ctxt->sax2) { |
10231 | 2.63M | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
10232 | 2.63M | namePop(ctxt); |
10233 | 2.63M | } |
10234 | 33.2M | #ifdef LIBXML_SAX1_ENABLED |
10235 | 33.2M | else |
10236 | 33.2M | xmlParseEndTag1(ctxt, 0); |
10237 | 35.9M | #endif /* LIBXML_SAX1_ENABLED */ |
10238 | | |
10239 | | /* |
10240 | | * Capture end position and add node |
10241 | | */ |
10242 | 35.9M | if ( ret != NULL && ctxt->record_info ) { |
10243 | 0 | node_info.end_pos = ctxt->input->consumed + |
10244 | 0 | (CUR_PTR - ctxt->input->base); |
10245 | 0 | node_info.end_line = ctxt->input->line; |
10246 | 0 | node_info.node = ret; |
10247 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10248 | 0 | } |
10249 | 35.9M | } |
10250 | | |
10251 | | /** |
10252 | | * xmlParseVersionNum: |
10253 | | * @ctxt: an XML parser context |
10254 | | * |
10255 | | * DEPRECATED: Internal function, don't use. |
10256 | | * |
10257 | | * parse the XML version value. |
10258 | | * |
10259 | | * [26] VersionNum ::= '1.' [0-9]+ |
10260 | | * |
10261 | | * In practice allow [0-9].[0-9]+ at that level |
10262 | | * |
10263 | | * Returns the string giving the XML version number, or NULL |
10264 | | */ |
10265 | | xmlChar * |
10266 | 394k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10267 | 394k | xmlChar *buf = NULL; |
10268 | 394k | int len = 0; |
10269 | 394k | int size = 10; |
10270 | 394k | xmlChar cur; |
10271 | | |
10272 | 394k | buf = (xmlChar *) xmlMallocAtomic(size); |
10273 | 394k | if (buf == NULL) { |
10274 | 0 | xmlErrMemory(ctxt, NULL); |
10275 | 0 | return(NULL); |
10276 | 0 | } |
10277 | 394k | cur = CUR; |
10278 | 394k | if (!((cur >= '0') && (cur <= '9'))) { |
10279 | 4.96k | xmlFree(buf); |
10280 | 4.96k | return(NULL); |
10281 | 4.96k | } |
10282 | 389k | buf[len++] = cur; |
10283 | 389k | NEXT; |
10284 | 389k | cur=CUR; |
10285 | 389k | if (cur != '.') { |
10286 | 8.51k | xmlFree(buf); |
10287 | 8.51k | return(NULL); |
10288 | 8.51k | } |
10289 | 381k | buf[len++] = cur; |
10290 | 381k | NEXT; |
10291 | 381k | cur=CUR; |
10292 | 1.52M | while ((cur >= '0') && (cur <= '9')) { |
10293 | 1.14M | if (len + 1 >= size) { |
10294 | 2.22k | xmlChar *tmp; |
10295 | | |
10296 | 2.22k | size *= 2; |
10297 | 2.22k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10298 | 2.22k | if (tmp == NULL) { |
10299 | 0 | xmlFree(buf); |
10300 | 0 | xmlErrMemory(ctxt, NULL); |
10301 | 0 | return(NULL); |
10302 | 0 | } |
10303 | 2.22k | buf = tmp; |
10304 | 2.22k | } |
10305 | 1.14M | buf[len++] = cur; |
10306 | 1.14M | NEXT; |
10307 | 1.14M | cur=CUR; |
10308 | 1.14M | } |
10309 | 381k | buf[len] = 0; |
10310 | 381k | return(buf); |
10311 | 381k | } |
10312 | | |
10313 | | /** |
10314 | | * xmlParseVersionInfo: |
10315 | | * @ctxt: an XML parser context |
10316 | | * |
10317 | | * DEPRECATED: Internal function, don't use. |
10318 | | * |
10319 | | * parse the XML version. |
10320 | | * |
10321 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10322 | | * |
10323 | | * [25] Eq ::= S? '=' S? |
10324 | | * |
10325 | | * Returns the version string, e.g. "1.0" |
10326 | | */ |
10327 | | |
10328 | | xmlChar * |
10329 | 466k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10330 | 466k | xmlChar *version = NULL; |
10331 | | |
10332 | 466k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10333 | 407k | SKIP(7); |
10334 | 407k | SKIP_BLANKS; |
10335 | 407k | if (RAW != '=') { |
10336 | 6.36k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10337 | 6.36k | return(NULL); |
10338 | 6.36k | } |
10339 | 401k | NEXT; |
10340 | 401k | SKIP_BLANKS; |
10341 | 401k | if (RAW == '"') { |
10342 | 320k | NEXT; |
10343 | 320k | version = xmlParseVersionNum(ctxt); |
10344 | 320k | if (RAW != '"') { |
10345 | 16.0k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10346 | 16.0k | } else |
10347 | 304k | NEXT; |
10348 | 320k | } else if (RAW == '\''){ |
10349 | 73.8k | NEXT; |
10350 | 73.8k | version = xmlParseVersionNum(ctxt); |
10351 | 73.8k | if (RAW != '\'') { |
10352 | 6.85k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10353 | 6.85k | } else |
10354 | 66.9k | NEXT; |
10355 | 73.8k | } else { |
10356 | 6.34k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10357 | 6.34k | } |
10358 | 401k | } |
10359 | 459k | return(version); |
10360 | 466k | } |
10361 | | |
10362 | | /** |
10363 | | * xmlParseEncName: |
10364 | | * @ctxt: an XML parser context |
10365 | | * |
10366 | | * DEPRECATED: Internal function, don't use. |
10367 | | * |
10368 | | * parse the XML encoding name |
10369 | | * |
10370 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10371 | | * |
10372 | | * Returns the encoding name value or NULL |
10373 | | */ |
10374 | | xmlChar * |
10375 | 194k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10376 | 194k | xmlChar *buf = NULL; |
10377 | 194k | int len = 0; |
10378 | 194k | int size = 10; |
10379 | 194k | xmlChar cur; |
10380 | | |
10381 | 194k | cur = CUR; |
10382 | 194k | if (((cur >= 'a') && (cur <= 'z')) || |
10383 | 194k | ((cur >= 'A') && (cur <= 'Z'))) { |
10384 | 193k | buf = (xmlChar *) xmlMallocAtomic(size); |
10385 | 193k | if (buf == NULL) { |
10386 | 0 | xmlErrMemory(ctxt, NULL); |
10387 | 0 | return(NULL); |
10388 | 0 | } |
10389 | | |
10390 | 193k | buf[len++] = cur; |
10391 | 193k | NEXT; |
10392 | 193k | cur = CUR; |
10393 | 2.95M | while (((cur >= 'a') && (cur <= 'z')) || |
10394 | 2.95M | ((cur >= 'A') && (cur <= 'Z')) || |
10395 | 2.95M | ((cur >= '0') && (cur <= '9')) || |
10396 | 2.95M | (cur == '.') || (cur == '_') || |
10397 | 2.95M | (cur == '-')) { |
10398 | 2.75M | if (len + 1 >= size) { |
10399 | 69.1k | xmlChar *tmp; |
10400 | | |
10401 | 69.1k | size *= 2; |
10402 | 69.1k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10403 | 69.1k | if (tmp == NULL) { |
10404 | 0 | xmlErrMemory(ctxt, NULL); |
10405 | 0 | xmlFree(buf); |
10406 | 0 | return(NULL); |
10407 | 0 | } |
10408 | 69.1k | buf = tmp; |
10409 | 69.1k | } |
10410 | 2.75M | buf[len++] = cur; |
10411 | 2.75M | NEXT; |
10412 | 2.75M | cur = CUR; |
10413 | 2.75M | if (cur == 0) { |
10414 | 957 | SHRINK; |
10415 | 957 | GROW; |
10416 | 957 | cur = CUR; |
10417 | 957 | } |
10418 | 2.75M | } |
10419 | 193k | buf[len] = 0; |
10420 | 193k | } else { |
10421 | 1.07k | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10422 | 1.07k | } |
10423 | 194k | return(buf); |
10424 | 194k | } |
10425 | | |
10426 | | /** |
10427 | | * xmlParseEncodingDecl: |
10428 | | * @ctxt: an XML parser context |
10429 | | * |
10430 | | * DEPRECATED: Internal function, don't use. |
10431 | | * |
10432 | | * parse the XML encoding declaration |
10433 | | * |
10434 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10435 | | * |
10436 | | * this setups the conversion filters. |
10437 | | * |
10438 | | * Returns the encoding value or NULL |
10439 | | */ |
10440 | | |
10441 | | const xmlChar * |
10442 | 364k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10443 | 364k | xmlChar *encoding = NULL; |
10444 | | |
10445 | 364k | SKIP_BLANKS; |
10446 | 364k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g')) { |
10447 | 198k | SKIP(8); |
10448 | 198k | SKIP_BLANKS; |
10449 | 198k | if (RAW != '=') { |
10450 | 2.24k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10451 | 2.24k | return(NULL); |
10452 | 2.24k | } |
10453 | 196k | NEXT; |
10454 | 196k | SKIP_BLANKS; |
10455 | 196k | if (RAW == '"') { |
10456 | 163k | NEXT; |
10457 | 163k | encoding = xmlParseEncName(ctxt); |
10458 | 163k | if (RAW != '"') { |
10459 | 7.79k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10460 | 7.79k | xmlFree((xmlChar *) encoding); |
10461 | 7.79k | return(NULL); |
10462 | 7.79k | } else |
10463 | 155k | NEXT; |
10464 | 163k | } else if (RAW == '\''){ |
10465 | 30.9k | NEXT; |
10466 | 30.9k | encoding = xmlParseEncName(ctxt); |
10467 | 30.9k | if (RAW != '\'') { |
10468 | 2.11k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10469 | 2.11k | xmlFree((xmlChar *) encoding); |
10470 | 2.11k | return(NULL); |
10471 | 2.11k | } else |
10472 | 28.8k | NEXT; |
10473 | 30.9k | } else { |
10474 | 1.45k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10475 | 1.45k | } |
10476 | | |
10477 | | /* |
10478 | | * Non standard parsing, allowing the user to ignore encoding |
10479 | | */ |
10480 | 186k | if (ctxt->options & XML_PARSE_IGNORE_ENC) { |
10481 | 29.4k | xmlFree((xmlChar *) encoding); |
10482 | 29.4k | return(NULL); |
10483 | 29.4k | } |
10484 | | |
10485 | | /* |
10486 | | * UTF-16 encoding switch has already taken place at this stage, |
10487 | | * more over the little-endian/big-endian selection is already done |
10488 | | */ |
10489 | 156k | if ((encoding != NULL) && |
10490 | 156k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-16")) || |
10491 | 155k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF16")))) { |
10492 | | /* |
10493 | | * If no encoding was passed to the parser, that we are |
10494 | | * using UTF-16 and no decoder is present i.e. the |
10495 | | * document is apparently UTF-8 compatible, then raise an |
10496 | | * encoding mismatch fatal error |
10497 | | */ |
10498 | 3.85k | if ((ctxt->encoding == NULL) && |
10499 | 3.85k | (ctxt->input->buf != NULL) && |
10500 | 3.85k | (ctxt->input->buf->encoder == NULL)) { |
10501 | 3.85k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_ENCODING, |
10502 | 3.85k | "Document labelled UTF-16 but has UTF-8 content\n"); |
10503 | 3.85k | } |
10504 | 3.85k | if (ctxt->encoding != NULL) |
10505 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10506 | 3.85k | ctxt->encoding = encoding; |
10507 | 3.85k | } |
10508 | | /* |
10509 | | * UTF-8 encoding is handled natively |
10510 | | */ |
10511 | 152k | else if ((encoding != NULL) && |
10512 | 152k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-8")) || |
10513 | 151k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF8")))) { |
10514 | 88.0k | if (ctxt->encoding != NULL) |
10515 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10516 | 88.0k | ctxt->encoding = encoding; |
10517 | 88.0k | } |
10518 | 64.8k | else if (encoding != NULL) { |
10519 | 63.6k | xmlCharEncodingHandlerPtr handler; |
10520 | | |
10521 | 63.6k | if (ctxt->input->encoding != NULL) |
10522 | 0 | xmlFree((xmlChar *) ctxt->input->encoding); |
10523 | 63.6k | ctxt->input->encoding = encoding; |
10524 | | |
10525 | 63.6k | handler = xmlFindCharEncodingHandler((const char *) encoding); |
10526 | 63.6k | if (handler != NULL) { |
10527 | 62.0k | if (xmlSwitchToEncoding(ctxt, handler) < 0) { |
10528 | | /* failed to convert */ |
10529 | 141 | ctxt->errNo = XML_ERR_UNSUPPORTED_ENCODING; |
10530 | 141 | return(NULL); |
10531 | 141 | } |
10532 | 62.0k | } else { |
10533 | 1.68k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
10534 | 1.68k | "Unsupported encoding %s\n", encoding); |
10535 | 1.68k | return(NULL); |
10536 | 1.68k | } |
10537 | 63.6k | } |
10538 | 156k | } |
10539 | 321k | return(encoding); |
10540 | 364k | } |
10541 | | |
10542 | | /** |
10543 | | * xmlParseSDDecl: |
10544 | | * @ctxt: an XML parser context |
10545 | | * |
10546 | | * DEPRECATED: Internal function, don't use. |
10547 | | * |
10548 | | * parse the XML standalone declaration |
10549 | | * |
10550 | | * [32] SDDecl ::= S 'standalone' Eq |
10551 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10552 | | * |
10553 | | * [ VC: Standalone Document Declaration ] |
10554 | | * TODO The standalone document declaration must have the value "no" |
10555 | | * if any external markup declarations contain declarations of: |
10556 | | * - attributes with default values, if elements to which these |
10557 | | * attributes apply appear in the document without specifications |
10558 | | * of values for these attributes, or |
10559 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10560 | | * to those entities appear in the document, or |
10561 | | * - attributes with values subject to normalization, where the |
10562 | | * attribute appears in the document with a value which will change |
10563 | | * as a result of normalization, or |
10564 | | * - element types with element content, if white space occurs directly |
10565 | | * within any instance of those types. |
10566 | | * |
10567 | | * Returns: |
10568 | | * 1 if standalone="yes" |
10569 | | * 0 if standalone="no" |
10570 | | * -2 if standalone attribute is missing or invalid |
10571 | | * (A standalone value of -2 means that the XML declaration was found, |
10572 | | * but no value was specified for the standalone attribute). |
10573 | | */ |
10574 | | |
10575 | | int |
10576 | 257k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10577 | 257k | int standalone = -2; |
10578 | | |
10579 | 257k | SKIP_BLANKS; |
10580 | 257k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10581 | 32.4k | SKIP(10); |
10582 | 32.4k | SKIP_BLANKS; |
10583 | 32.4k | if (RAW != '=') { |
10584 | 372 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10585 | 372 | return(standalone); |
10586 | 372 | } |
10587 | 32.0k | NEXT; |
10588 | 32.0k | SKIP_BLANKS; |
10589 | 32.0k | if (RAW == '\''){ |
10590 | 14.4k | NEXT; |
10591 | 14.4k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10592 | 9.33k | standalone = 0; |
10593 | 9.33k | SKIP(2); |
10594 | 9.33k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10595 | 5.13k | (NXT(2) == 's')) { |
10596 | 4.70k | standalone = 1; |
10597 | 4.70k | SKIP(3); |
10598 | 4.70k | } else { |
10599 | 438 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10600 | 438 | } |
10601 | 14.4k | if (RAW != '\'') { |
10602 | 666 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10603 | 666 | } else |
10604 | 13.8k | NEXT; |
10605 | 17.5k | } else if (RAW == '"'){ |
10606 | 17.2k | NEXT; |
10607 | 17.2k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10608 | 7.31k | standalone = 0; |
10609 | 7.31k | SKIP(2); |
10610 | 9.97k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10611 | 9.97k | (NXT(2) == 's')) { |
10612 | 9.41k | standalone = 1; |
10613 | 9.41k | SKIP(3); |
10614 | 9.41k | } else { |
10615 | 559 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10616 | 559 | } |
10617 | 17.2k | if (RAW != '"') { |
10618 | 907 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10619 | 907 | } else |
10620 | 16.3k | NEXT; |
10621 | 17.2k | } else { |
10622 | 284 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10623 | 284 | } |
10624 | 32.0k | } |
10625 | 257k | return(standalone); |
10626 | 257k | } |
10627 | | |
10628 | | /** |
10629 | | * xmlParseXMLDecl: |
10630 | | * @ctxt: an XML parser context |
10631 | | * |
10632 | | * DEPRECATED: Internal function, don't use. |
10633 | | * |
10634 | | * parse an XML declaration header |
10635 | | * |
10636 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10637 | | */ |
10638 | | |
10639 | | void |
10640 | 398k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10641 | 398k | xmlChar *version; |
10642 | | |
10643 | | /* |
10644 | | * This value for standalone indicates that the document has an |
10645 | | * XML declaration but it does not have a standalone attribute. |
10646 | | * It will be overwritten later if a standalone attribute is found. |
10647 | | */ |
10648 | 398k | ctxt->input->standalone = -2; |
10649 | | |
10650 | | /* |
10651 | | * We know that '<?xml' is here. |
10652 | | */ |
10653 | 398k | SKIP(5); |
10654 | | |
10655 | 398k | if (!IS_BLANK_CH(RAW)) { |
10656 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10657 | 0 | "Blank needed after '<?xml'\n"); |
10658 | 0 | } |
10659 | 398k | SKIP_BLANKS; |
10660 | | |
10661 | | /* |
10662 | | * We must have the VersionInfo here. |
10663 | | */ |
10664 | 398k | version = xmlParseVersionInfo(ctxt); |
10665 | 398k | if (version == NULL) { |
10666 | 72.4k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10667 | 326k | } else { |
10668 | 326k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10669 | | /* |
10670 | | * Changed here for XML-1.0 5th edition |
10671 | | */ |
10672 | 8.17k | if (ctxt->options & XML_PARSE_OLD10) { |
10673 | 1.53k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10674 | 1.53k | "Unsupported version '%s'\n", |
10675 | 1.53k | version); |
10676 | 6.64k | } else { |
10677 | 6.64k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10678 | 5.56k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10679 | 5.56k | "Unsupported version '%s'\n", |
10680 | 5.56k | version, NULL); |
10681 | 5.56k | } else { |
10682 | 1.07k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10683 | 1.07k | "Unsupported version '%s'\n", |
10684 | 1.07k | version); |
10685 | 1.07k | } |
10686 | 6.64k | } |
10687 | 8.17k | } |
10688 | 326k | if (ctxt->version != NULL) |
10689 | 0 | xmlFree((void *) ctxt->version); |
10690 | 326k | ctxt->version = version; |
10691 | 326k | } |
10692 | | |
10693 | | /* |
10694 | | * We may have the encoding declaration |
10695 | | */ |
10696 | 398k | if (!IS_BLANK_CH(RAW)) { |
10697 | 185k | if ((RAW == '?') && (NXT(1) == '>')) { |
10698 | 101k | SKIP(2); |
10699 | 101k | return; |
10700 | 101k | } |
10701 | 83.8k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10702 | 83.8k | } |
10703 | 297k | xmlParseEncodingDecl(ctxt); |
10704 | 297k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10705 | 297k | (ctxt->instate == XML_PARSER_EOF)) { |
10706 | | /* |
10707 | | * The XML REC instructs us to stop parsing right here |
10708 | | */ |
10709 | 1.57k | return; |
10710 | 1.57k | } |
10711 | | |
10712 | | /* |
10713 | | * We may have the standalone status. |
10714 | | */ |
10715 | 295k | if ((ctxt->input->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10716 | 40.8k | if ((RAW == '?') && (NXT(1) == '>')) { |
10717 | 38.1k | SKIP(2); |
10718 | 38.1k | return; |
10719 | 38.1k | } |
10720 | 2.78k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10721 | 2.78k | } |
10722 | | |
10723 | | /* |
10724 | | * We can grow the input buffer freely at that point |
10725 | | */ |
10726 | 257k | GROW; |
10727 | | |
10728 | 257k | SKIP_BLANKS; |
10729 | 257k | ctxt->input->standalone = xmlParseSDDecl(ctxt); |
10730 | | |
10731 | 257k | SKIP_BLANKS; |
10732 | 257k | if ((RAW == '?') && (NXT(1) == '>')) { |
10733 | 132k | SKIP(2); |
10734 | 132k | } else if (RAW == '>') { |
10735 | | /* Deprecated old WD ... */ |
10736 | 1.36k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10737 | 1.36k | NEXT; |
10738 | 123k | } else { |
10739 | 123k | int c; |
10740 | | |
10741 | 123k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10742 | 5.38M | while ((c = CUR) != 0) { |
10743 | 5.37M | NEXT; |
10744 | 5.37M | if (c == '>') |
10745 | 111k | break; |
10746 | 5.37M | } |
10747 | 123k | } |
10748 | 257k | } |
10749 | | |
10750 | | /** |
10751 | | * xmlParseMisc: |
10752 | | * @ctxt: an XML parser context |
10753 | | * |
10754 | | * DEPRECATED: Internal function, don't use. |
10755 | | * |
10756 | | * parse an XML Misc* optional field. |
10757 | | * |
10758 | | * [27] Misc ::= Comment | PI | S |
10759 | | */ |
10760 | | |
10761 | | void |
10762 | 1.04M | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10763 | 1.14M | while (ctxt->instate != XML_PARSER_EOF) { |
10764 | 1.14M | SKIP_BLANKS; |
10765 | 1.14M | GROW; |
10766 | 1.14M | if ((RAW == '<') && (NXT(1) == '?')) { |
10767 | 57.1k | xmlParsePI(ctxt); |
10768 | 1.08M | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10769 | 37.2k | xmlParseComment(ctxt); |
10770 | 1.04M | } else { |
10771 | 1.04M | break; |
10772 | 1.04M | } |
10773 | 1.14M | } |
10774 | 1.04M | } |
10775 | | |
10776 | | /** |
10777 | | * xmlParseDocument: |
10778 | | * @ctxt: an XML parser context |
10779 | | * |
10780 | | * parse an XML document (and build a tree if using the standard SAX |
10781 | | * interface). |
10782 | | * |
10783 | | * [1] document ::= prolog element Misc* |
10784 | | * |
10785 | | * [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? |
10786 | | * |
10787 | | * Returns 0, -1 in case of error. the parser context is augmented |
10788 | | * as a result of the parsing. |
10789 | | */ |
10790 | | |
10791 | | int |
10792 | 489k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10793 | 489k | xmlChar start[4]; |
10794 | 489k | xmlCharEncoding enc; |
10795 | | |
10796 | 489k | xmlInitParser(); |
10797 | | |
10798 | 489k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10799 | 0 | return(-1); |
10800 | | |
10801 | 489k | GROW; |
10802 | | |
10803 | | /* |
10804 | | * SAX: detecting the level. |
10805 | | */ |
10806 | 489k | xmlDetectSAX2(ctxt); |
10807 | | |
10808 | | /* |
10809 | | * SAX: beginning of the document processing. |
10810 | | */ |
10811 | 489k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10812 | 489k | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10813 | 489k | if (ctxt->instate == XML_PARSER_EOF) |
10814 | 0 | return(-1); |
10815 | | |
10816 | 489k | if ((ctxt->encoding == NULL) && |
10817 | 489k | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
10818 | | /* |
10819 | | * Get the 4 first bytes and decode the charset |
10820 | | * if enc != XML_CHAR_ENCODING_NONE |
10821 | | * plug some encoding conversion routines. |
10822 | | */ |
10823 | 483k | start[0] = RAW; |
10824 | 483k | start[1] = NXT(1); |
10825 | 483k | start[2] = NXT(2); |
10826 | 483k | start[3] = NXT(3); |
10827 | 483k | enc = xmlDetectCharEncoding(&start[0], 4); |
10828 | 483k | if (enc != XML_CHAR_ENCODING_NONE) { |
10829 | 151k | xmlSwitchEncoding(ctxt, enc); |
10830 | 151k | } |
10831 | 483k | } |
10832 | | |
10833 | | |
10834 | 489k | if (CUR == 0) { |
10835 | 3.77k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10836 | 3.77k | return(-1); |
10837 | 3.77k | } |
10838 | | |
10839 | | /* |
10840 | | * Check for the XMLDecl in the Prolog. |
10841 | | * do not GROW here to avoid the detected encoder to decode more |
10842 | | * than just the first line, unless the amount of data is really |
10843 | | * too small to hold "<?xml version="1.0" encoding="foo" |
10844 | | */ |
10845 | 485k | if ((ctxt->input->end - ctxt->input->cur) < 35) { |
10846 | 21.2k | GROW; |
10847 | 21.2k | } |
10848 | 485k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10849 | | |
10850 | | /* |
10851 | | * Note that we will switch encoding on the fly. |
10852 | | */ |
10853 | 137k | xmlParseXMLDecl(ctxt); |
10854 | 137k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10855 | 137k | (ctxt->instate == XML_PARSER_EOF)) { |
10856 | | /* |
10857 | | * The XML REC instructs us to stop parsing right here |
10858 | | */ |
10859 | 599 | return(-1); |
10860 | 599 | } |
10861 | 137k | ctxt->standalone = ctxt->input->standalone; |
10862 | 137k | SKIP_BLANKS; |
10863 | 347k | } else { |
10864 | 347k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10865 | 347k | } |
10866 | 485k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10867 | 463k | ctxt->sax->startDocument(ctxt->userData); |
10868 | 485k | if (ctxt->instate == XML_PARSER_EOF) |
10869 | 0 | return(-1); |
10870 | 485k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10871 | 485k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10872 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10873 | 0 | } |
10874 | | |
10875 | | /* |
10876 | | * The Misc part of the Prolog |
10877 | | */ |
10878 | 485k | xmlParseMisc(ctxt); |
10879 | | |
10880 | | /* |
10881 | | * Then possibly doc type declaration(s) and more Misc |
10882 | | * (doctypedecl Misc*)? |
10883 | | */ |
10884 | 485k | GROW; |
10885 | 485k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10886 | | |
10887 | 201k | ctxt->inSubset = 1; |
10888 | 201k | xmlParseDocTypeDecl(ctxt); |
10889 | 201k | if (RAW == '[') { |
10890 | 159k | ctxt->instate = XML_PARSER_DTD; |
10891 | 159k | xmlParseInternalSubset(ctxt); |
10892 | 159k | if (ctxt->instate == XML_PARSER_EOF) |
10893 | 11.1k | return(-1); |
10894 | 159k | } |
10895 | | |
10896 | | /* |
10897 | | * Create and update the external subset. |
10898 | | */ |
10899 | 189k | ctxt->inSubset = 2; |
10900 | 189k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10901 | 189k | (!ctxt->disableSAX)) |
10902 | 151k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10903 | 151k | ctxt->extSubSystem, ctxt->extSubURI); |
10904 | 189k | if (ctxt->instate == XML_PARSER_EOF) |
10905 | 4.40k | return(-1); |
10906 | 185k | ctxt->inSubset = 0; |
10907 | | |
10908 | 185k | xmlCleanSpecialAttr(ctxt); |
10909 | | |
10910 | 185k | ctxt->instate = XML_PARSER_PROLOG; |
10911 | 185k | xmlParseMisc(ctxt); |
10912 | 185k | } |
10913 | | |
10914 | | /* |
10915 | | * Time to start parsing the tree itself |
10916 | | */ |
10917 | 469k | GROW; |
10918 | 469k | if (RAW != '<') { |
10919 | 90.1k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10920 | 90.1k | "Start tag expected, '<' not found\n"); |
10921 | 379k | } else { |
10922 | 379k | ctxt->instate = XML_PARSER_CONTENT; |
10923 | 379k | xmlParseElement(ctxt); |
10924 | 379k | ctxt->instate = XML_PARSER_EPILOG; |
10925 | | |
10926 | | |
10927 | | /* |
10928 | | * The Misc part at the end |
10929 | | */ |
10930 | 379k | xmlParseMisc(ctxt); |
10931 | | |
10932 | 379k | if (RAW != 0) { |
10933 | 122k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10934 | 122k | } |
10935 | 379k | ctxt->instate = XML_PARSER_EOF; |
10936 | 379k | } |
10937 | | |
10938 | | /* |
10939 | | * SAX: end of the document processing. |
10940 | | */ |
10941 | 469k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10942 | 469k | ctxt->sax->endDocument(ctxt->userData); |
10943 | | |
10944 | | /* |
10945 | | * Remove locally kept entity definitions if the tree was not built |
10946 | | */ |
10947 | 469k | if ((ctxt->myDoc != NULL) && |
10948 | 469k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
10949 | 1.79k | xmlFreeDoc(ctxt->myDoc); |
10950 | 1.79k | ctxt->myDoc = NULL; |
10951 | 1.79k | } |
10952 | | |
10953 | 469k | if ((ctxt->wellFormed) && (ctxt->myDoc != NULL)) { |
10954 | 33.3k | ctxt->myDoc->properties |= XML_DOC_WELLFORMED; |
10955 | 33.3k | if (ctxt->valid) |
10956 | 21.0k | ctxt->myDoc->properties |= XML_DOC_DTDVALID; |
10957 | 33.3k | if (ctxt->nsWellFormed) |
10958 | 31.7k | ctxt->myDoc->properties |= XML_DOC_NSVALID; |
10959 | 33.3k | if (ctxt->options & XML_PARSE_OLD10) |
10960 | 9.82k | ctxt->myDoc->properties |= XML_DOC_OLD10; |
10961 | 33.3k | } |
10962 | 469k | if (! ctxt->wellFormed) { |
10963 | 436k | ctxt->valid = 0; |
10964 | 436k | return(-1); |
10965 | 436k | } |
10966 | 33.3k | return(0); |
10967 | 469k | } |
10968 | | |
10969 | | /** |
10970 | | * xmlParseExtParsedEnt: |
10971 | | * @ctxt: an XML parser context |
10972 | | * |
10973 | | * parse a general parsed entity |
10974 | | * An external general parsed entity is well-formed if it matches the |
10975 | | * production labeled extParsedEnt. |
10976 | | * |
10977 | | * [78] extParsedEnt ::= TextDecl? content |
10978 | | * |
10979 | | * Returns 0, -1 in case of error. the parser context is augmented |
10980 | | * as a result of the parsing. |
10981 | | */ |
10982 | | |
10983 | | int |
10984 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10985 | 0 | xmlChar start[4]; |
10986 | 0 | xmlCharEncoding enc; |
10987 | |
|
10988 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10989 | 0 | return(-1); |
10990 | | |
10991 | 0 | xmlDetectSAX2(ctxt); |
10992 | |
|
10993 | 0 | GROW; |
10994 | | |
10995 | | /* |
10996 | | * SAX: beginning of the document processing. |
10997 | | */ |
10998 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10999 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
11000 | | |
11001 | | /* |
11002 | | * Get the 4 first bytes and decode the charset |
11003 | | * if enc != XML_CHAR_ENCODING_NONE |
11004 | | * plug some encoding conversion routines. |
11005 | | */ |
11006 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
11007 | 0 | start[0] = RAW; |
11008 | 0 | start[1] = NXT(1); |
11009 | 0 | start[2] = NXT(2); |
11010 | 0 | start[3] = NXT(3); |
11011 | 0 | enc = xmlDetectCharEncoding(start, 4); |
11012 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11013 | 0 | xmlSwitchEncoding(ctxt, enc); |
11014 | 0 | } |
11015 | 0 | } |
11016 | | |
11017 | |
|
11018 | 0 | if (CUR == 0) { |
11019 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11020 | 0 | } |
11021 | | |
11022 | | /* |
11023 | | * Check for the XMLDecl in the Prolog. |
11024 | | */ |
11025 | 0 | GROW; |
11026 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
11027 | | |
11028 | | /* |
11029 | | * Note that we will switch encoding on the fly. |
11030 | | */ |
11031 | 0 | xmlParseXMLDecl(ctxt); |
11032 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11033 | | /* |
11034 | | * The XML REC instructs us to stop parsing right here |
11035 | | */ |
11036 | 0 | return(-1); |
11037 | 0 | } |
11038 | 0 | SKIP_BLANKS; |
11039 | 0 | } else { |
11040 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11041 | 0 | } |
11042 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
11043 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11044 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11045 | 0 | return(-1); |
11046 | | |
11047 | | /* |
11048 | | * Doing validity checking on chunk doesn't make sense |
11049 | | */ |
11050 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11051 | 0 | ctxt->validate = 0; |
11052 | 0 | ctxt->loadsubset = 0; |
11053 | 0 | ctxt->depth = 0; |
11054 | |
|
11055 | 0 | xmlParseContent(ctxt); |
11056 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11057 | 0 | return(-1); |
11058 | | |
11059 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
11060 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
11061 | 0 | } else if (RAW != 0) { |
11062 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
11063 | 0 | } |
11064 | | |
11065 | | /* |
11066 | | * SAX: end of the document processing. |
11067 | | */ |
11068 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11069 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11070 | |
|
11071 | 0 | if (! ctxt->wellFormed) return(-1); |
11072 | 0 | return(0); |
11073 | 0 | } |
11074 | | |
11075 | | #ifdef LIBXML_PUSH_ENABLED |
11076 | | /************************************************************************ |
11077 | | * * |
11078 | | * Progressive parsing interfaces * |
11079 | | * * |
11080 | | ************************************************************************/ |
11081 | | |
11082 | | /** |
11083 | | * xmlParseLookupSequence: |
11084 | | * @ctxt: an XML parser context |
11085 | | * @first: the first char to lookup |
11086 | | * @next: the next char to lookup or zero |
11087 | | * @third: the next char to lookup or zero |
11088 | | * |
11089 | | * Try to find if a sequence (first, next, third) or just (first next) or |
11090 | | * (first) is available in the input stream. |
11091 | | * This function has a side effect of (possibly) incrementing ctxt->checkIndex |
11092 | | * to avoid rescanning sequences of bytes, it DOES change the state of the |
11093 | | * parser, do not use liberally. |
11094 | | * |
11095 | | * Returns the index to the current parsing point if the full sequence |
11096 | | * is available, -1 otherwise. |
11097 | | */ |
11098 | | static int |
11099 | | xmlParseLookupSequence(xmlParserCtxtPtr ctxt, xmlChar first, |
11100 | 11.3M | xmlChar next, xmlChar third) { |
11101 | 11.3M | int base, len; |
11102 | 11.3M | xmlParserInputPtr in; |
11103 | 11.3M | const xmlChar *buf; |
11104 | | |
11105 | 11.3M | in = ctxt->input; |
11106 | 11.3M | if (in == NULL) return(-1); |
11107 | 11.3M | base = in->cur - in->base; |
11108 | 11.3M | if (base < 0) return(-1); |
11109 | 11.3M | if (ctxt->checkIndex > base) |
11110 | 1.17M | base = ctxt->checkIndex; |
11111 | 11.3M | if (in->buf == NULL) { |
11112 | 0 | buf = in->base; |
11113 | 0 | len = in->length; |
11114 | 11.3M | } else { |
11115 | 11.3M | buf = xmlBufContent(in->buf->buffer); |
11116 | 11.3M | len = xmlBufUse(in->buf->buffer); |
11117 | 11.3M | } |
11118 | | /* take into account the sequence length */ |
11119 | 11.3M | if (third) len -= 2; |
11120 | 10.1M | else if (next) len --; |
11121 | 150G | for (;base < len;base++) { |
11122 | 150G | if (buf[base] == first) { |
11123 | 13.9M | if (third != 0) { |
11124 | 4.11M | if ((buf[base + 1] != next) || |
11125 | 4.11M | (buf[base + 2] != third)) continue; |
11126 | 9.80M | } else if (next != 0) { |
11127 | 1.10M | if (buf[base + 1] != next) continue; |
11128 | 1.10M | } |
11129 | 9.71M | ctxt->checkIndex = 0; |
11130 | | #ifdef DEBUG_PUSH |
11131 | | if (next == 0) |
11132 | | xmlGenericError(xmlGenericErrorContext, |
11133 | | "PP: lookup '%c' found at %d\n", |
11134 | | first, base); |
11135 | | else if (third == 0) |
11136 | | xmlGenericError(xmlGenericErrorContext, |
11137 | | "PP: lookup '%c%c' found at %d\n", |
11138 | | first, next, base); |
11139 | | else |
11140 | | xmlGenericError(xmlGenericErrorContext, |
11141 | | "PP: lookup '%c%c%c' found at %d\n", |
11142 | | first, next, third, base); |
11143 | | #endif |
11144 | 9.71M | return(base - (in->cur - in->base)); |
11145 | 13.9M | } |
11146 | 150G | } |
11147 | 1.62M | ctxt->checkIndex = base; |
11148 | | #ifdef DEBUG_PUSH |
11149 | | if (next == 0) |
11150 | | xmlGenericError(xmlGenericErrorContext, |
11151 | | "PP: lookup '%c' failed\n", first); |
11152 | | else if (third == 0) |
11153 | | xmlGenericError(xmlGenericErrorContext, |
11154 | | "PP: lookup '%c%c' failed\n", first, next); |
11155 | | else |
11156 | | xmlGenericError(xmlGenericErrorContext, |
11157 | | "PP: lookup '%c%c%c' failed\n", first, next, third); |
11158 | | #endif |
11159 | 1.62M | return(-1); |
11160 | 11.3M | } |
11161 | | |
11162 | | /** |
11163 | | * xmlParseGetLasts: |
11164 | | * @ctxt: an XML parser context |
11165 | | * @lastlt: pointer to store the last '<' from the input |
11166 | | * @lastgt: pointer to store the last '>' from the input |
11167 | | * |
11168 | | * Lookup the last < and > in the current chunk |
11169 | | */ |
11170 | | static void |
11171 | | xmlParseGetLasts(xmlParserCtxtPtr ctxt, const xmlChar **lastlt, |
11172 | 9.78M | const xmlChar **lastgt) { |
11173 | 9.78M | const xmlChar *tmp; |
11174 | | |
11175 | 9.78M | if ((ctxt == NULL) || (lastlt == NULL) || (lastgt == NULL)) { |
11176 | 0 | xmlGenericError(xmlGenericErrorContext, |
11177 | 0 | "Internal error: xmlParseGetLasts\n"); |
11178 | 0 | return; |
11179 | 0 | } |
11180 | 9.78M | if ((ctxt->progressive != 0) && (ctxt->inputNr == 1)) { |
11181 | 4.15M | tmp = ctxt->input->end; |
11182 | 4.15M | tmp--; |
11183 | 1.29G | while ((tmp >= ctxt->input->base) && (*tmp != '<')) tmp--; |
11184 | 4.15M | if (tmp < ctxt->input->base) { |
11185 | 113k | *lastlt = NULL; |
11186 | 113k | *lastgt = NULL; |
11187 | 4.04M | } else { |
11188 | 4.04M | *lastlt = tmp; |
11189 | 4.04M | tmp++; |
11190 | 215M | while ((tmp < ctxt->input->end) && (*tmp != '>')) { |
11191 | 211M | if (*tmp == '\'') { |
11192 | 1.41M | tmp++; |
11193 | 57.2M | while ((tmp < ctxt->input->end) && (*tmp != '\'')) tmp++; |
11194 | 1.41M | if (tmp < ctxt->input->end) tmp++; |
11195 | 209M | } else if (*tmp == '"') { |
11196 | 2.11M | tmp++; |
11197 | 141M | while ((tmp < ctxt->input->end) && (*tmp != '"')) tmp++; |
11198 | 2.11M | if (tmp < ctxt->input->end) tmp++; |
11199 | 2.11M | } else |
11200 | 207M | tmp++; |
11201 | 211M | } |
11202 | 4.04M | if (tmp < ctxt->input->end) |
11203 | 1.77M | *lastgt = tmp; |
11204 | 2.27M | else { |
11205 | 2.27M | tmp = *lastlt; |
11206 | 2.27M | tmp--; |
11207 | 59.0M | while ((tmp >= ctxt->input->base) && (*tmp != '>')) tmp--; |
11208 | 2.27M | if (tmp >= ctxt->input->base) |
11209 | 2.22M | *lastgt = tmp; |
11210 | 51.0k | else |
11211 | 51.0k | *lastgt = NULL; |
11212 | 2.27M | } |
11213 | 4.04M | } |
11214 | 5.62M | } else { |
11215 | 5.62M | *lastlt = NULL; |
11216 | 5.62M | *lastgt = NULL; |
11217 | 5.62M | } |
11218 | 9.78M | } |
11219 | | /** |
11220 | | * xmlCheckCdataPush: |
11221 | | * @cur: pointer to the block of characters |
11222 | | * @len: length of the block in bytes |
11223 | | * @complete: 1 if complete CDATA block is passed in, 0 if partial block |
11224 | | * |
11225 | | * Check that the block of characters is okay as SCdata content [20] |
11226 | | * |
11227 | | * Returns the number of bytes to pass if okay, a negative index where an |
11228 | | * UTF-8 error occurred otherwise |
11229 | | */ |
11230 | | static int |
11231 | 655k | xmlCheckCdataPush(const xmlChar *utf, int len, int complete) { |
11232 | 655k | int ix; |
11233 | 655k | unsigned char c; |
11234 | 655k | int codepoint; |
11235 | | |
11236 | 655k | if ((utf == NULL) || (len <= 0)) |
11237 | 1.18k | return(0); |
11238 | | |
11239 | 14.0M | for (ix = 0; ix < len;) { /* string is 0-terminated */ |
11240 | 14.0M | c = utf[ix]; |
11241 | 14.0M | if ((c & 0x80) == 0x00) { /* 1-byte code, starts with 10 */ |
11242 | 9.84M | if (c >= 0x20) |
11243 | 9.26M | ix++; |
11244 | 572k | else if ((c == 0xA) || (c == 0xD) || (c == 0x9)) |
11245 | 433k | ix++; |
11246 | 139k | else |
11247 | 139k | return(-ix); |
11248 | 9.84M | } else if ((c & 0xe0) == 0xc0) {/* 2-byte code, starts with 110 */ |
11249 | 1.65M | if (ix + 2 > len) return(complete ? -ix : ix); |
11250 | 1.65M | if ((utf[ix+1] & 0xc0 ) != 0x80) |
11251 | 54.6k | return(-ix); |
11252 | 1.59M | codepoint = (utf[ix] & 0x1f) << 6; |
11253 | 1.59M | codepoint |= utf[ix+1] & 0x3f; |
11254 | 1.59M | if (!xmlIsCharQ(codepoint)) |
11255 | 6.64k | return(-ix); |
11256 | 1.58M | ix += 2; |
11257 | 2.50M | } else if ((c & 0xf0) == 0xe0) {/* 3-byte code, starts with 1110 */ |
11258 | 907k | if (ix + 3 > len) return(complete ? -ix : ix); |
11259 | 900k | if (((utf[ix+1] & 0xc0) != 0x80) || |
11260 | 900k | ((utf[ix+2] & 0xc0) != 0x80)) |
11261 | 30.6k | return(-ix); |
11262 | 869k | codepoint = (utf[ix] & 0xf) << 12; |
11263 | 869k | codepoint |= (utf[ix+1] & 0x3f) << 6; |
11264 | 869k | codepoint |= utf[ix+2] & 0x3f; |
11265 | 869k | if (!xmlIsCharQ(codepoint)) |
11266 | 5.20k | return(-ix); |
11267 | 864k | ix += 3; |
11268 | 1.59M | } else if ((c & 0xf8) == 0xf0) {/* 4-byte code, starts with 11110 */ |
11269 | 1.34M | if (ix + 4 > len) return(complete ? -ix : ix); |
11270 | 1.34M | if (((utf[ix+1] & 0xc0) != 0x80) || |
11271 | 1.34M | ((utf[ix+2] & 0xc0) != 0x80) || |
11272 | 1.34M | ((utf[ix+3] & 0xc0) != 0x80)) |
11273 | 76.4k | return(-ix); |
11274 | 1.26M | codepoint = (utf[ix] & 0x7) << 18; |
11275 | 1.26M | codepoint |= (utf[ix+1] & 0x3f) << 12; |
11276 | 1.26M | codepoint |= (utf[ix+2] & 0x3f) << 6; |
11277 | 1.26M | codepoint |= utf[ix+3] & 0x3f; |
11278 | 1.26M | if (!xmlIsCharQ(codepoint)) |
11279 | 12.3k | return(-ix); |
11280 | 1.25M | ix += 4; |
11281 | 1.25M | } else /* unknown encoding */ |
11282 | 250k | return(-ix); |
11283 | 14.0M | } |
11284 | 59.5k | return(ix); |
11285 | 654k | } |
11286 | | |
11287 | | /** |
11288 | | * xmlParseTryOrFinish: |
11289 | | * @ctxt: an XML parser context |
11290 | | * @terminate: last chunk indicator |
11291 | | * |
11292 | | * Try to progress on parsing |
11293 | | * |
11294 | | * Returns zero if no parsing was possible |
11295 | | */ |
11296 | | static int |
11297 | 9.19M | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11298 | 9.19M | int ret = 0; |
11299 | 9.19M | int avail, tlen; |
11300 | 9.19M | xmlChar cur, next; |
11301 | 9.19M | const xmlChar *lastlt, *lastgt; |
11302 | | |
11303 | 9.19M | if (ctxt->input == NULL) |
11304 | 0 | return(0); |
11305 | | |
11306 | | #ifdef DEBUG_PUSH |
11307 | | switch (ctxt->instate) { |
11308 | | case XML_PARSER_EOF: |
11309 | | xmlGenericError(xmlGenericErrorContext, |
11310 | | "PP: try EOF\n"); break; |
11311 | | case XML_PARSER_START: |
11312 | | xmlGenericError(xmlGenericErrorContext, |
11313 | | "PP: try START\n"); break; |
11314 | | case XML_PARSER_MISC: |
11315 | | xmlGenericError(xmlGenericErrorContext, |
11316 | | "PP: try MISC\n");break; |
11317 | | case XML_PARSER_COMMENT: |
11318 | | xmlGenericError(xmlGenericErrorContext, |
11319 | | "PP: try COMMENT\n");break; |
11320 | | case XML_PARSER_PROLOG: |
11321 | | xmlGenericError(xmlGenericErrorContext, |
11322 | | "PP: try PROLOG\n");break; |
11323 | | case XML_PARSER_START_TAG: |
11324 | | xmlGenericError(xmlGenericErrorContext, |
11325 | | "PP: try START_TAG\n");break; |
11326 | | case XML_PARSER_CONTENT: |
11327 | | xmlGenericError(xmlGenericErrorContext, |
11328 | | "PP: try CONTENT\n");break; |
11329 | | case XML_PARSER_CDATA_SECTION: |
11330 | | xmlGenericError(xmlGenericErrorContext, |
11331 | | "PP: try CDATA_SECTION\n");break; |
11332 | | case XML_PARSER_END_TAG: |
11333 | | xmlGenericError(xmlGenericErrorContext, |
11334 | | "PP: try END_TAG\n");break; |
11335 | | case XML_PARSER_ENTITY_DECL: |
11336 | | xmlGenericError(xmlGenericErrorContext, |
11337 | | "PP: try ENTITY_DECL\n");break; |
11338 | | case XML_PARSER_ENTITY_VALUE: |
11339 | | xmlGenericError(xmlGenericErrorContext, |
11340 | | "PP: try ENTITY_VALUE\n");break; |
11341 | | case XML_PARSER_ATTRIBUTE_VALUE: |
11342 | | xmlGenericError(xmlGenericErrorContext, |
11343 | | "PP: try ATTRIBUTE_VALUE\n");break; |
11344 | | case XML_PARSER_DTD: |
11345 | | xmlGenericError(xmlGenericErrorContext, |
11346 | | "PP: try DTD\n");break; |
11347 | | case XML_PARSER_EPILOG: |
11348 | | xmlGenericError(xmlGenericErrorContext, |
11349 | | "PP: try EPILOG\n");break; |
11350 | | case XML_PARSER_PI: |
11351 | | xmlGenericError(xmlGenericErrorContext, |
11352 | | "PP: try PI\n");break; |
11353 | | case XML_PARSER_IGNORE: |
11354 | | xmlGenericError(xmlGenericErrorContext, |
11355 | | "PP: try IGNORE\n");break; |
11356 | | } |
11357 | | #endif |
11358 | | |
11359 | 9.19M | if ((ctxt->input != NULL) && |
11360 | 9.19M | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11361 | 51.9k | xmlSHRINK(ctxt); |
11362 | 51.9k | ctxt->checkIndex = 0; |
11363 | 51.9k | } |
11364 | 9.19M | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11365 | | |
11366 | 51.3M | while (ctxt->instate != XML_PARSER_EOF) { |
11367 | 51.1M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
11368 | 223k | return(0); |
11369 | | |
11370 | 50.9M | if (ctxt->input == NULL) break; |
11371 | 50.9M | if (ctxt->input->buf == NULL) |
11372 | 0 | avail = ctxt->input->length - |
11373 | 0 | (ctxt->input->cur - ctxt->input->base); |
11374 | 50.9M | else { |
11375 | | /* |
11376 | | * If we are operating on converted input, try to flush |
11377 | | * remaining chars to avoid them stalling in the non-converted |
11378 | | * buffer. But do not do this in document start where |
11379 | | * encoding="..." may not have been read and we work on a |
11380 | | * guessed encoding. |
11381 | | */ |
11382 | 50.9M | if ((ctxt->instate != XML_PARSER_START) && |
11383 | 50.9M | (ctxt->input->buf->raw != NULL) && |
11384 | 50.9M | (xmlBufIsEmpty(ctxt->input->buf->raw) == 0)) { |
11385 | 180k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
11386 | 180k | ctxt->input); |
11387 | 180k | size_t current = ctxt->input->cur - ctxt->input->base; |
11388 | | |
11389 | 180k | xmlParserInputBufferPush(ctxt->input->buf, 0, ""); |
11390 | 180k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
11391 | 180k | base, current); |
11392 | 180k | } |
11393 | 50.9M | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11394 | 50.9M | (ctxt->input->cur - ctxt->input->base); |
11395 | 50.9M | } |
11396 | 50.9M | if (avail < 1) |
11397 | 276k | goto done; |
11398 | 50.6M | switch (ctxt->instate) { |
11399 | 0 | case XML_PARSER_EOF: |
11400 | | /* |
11401 | | * Document parsing is done ! |
11402 | | */ |
11403 | 0 | goto done; |
11404 | 1.72M | case XML_PARSER_START: |
11405 | 1.72M | if (ctxt->charset == XML_CHAR_ENCODING_NONE) { |
11406 | 420k | xmlChar start[4]; |
11407 | 420k | xmlCharEncoding enc; |
11408 | | |
11409 | | /* |
11410 | | * Very first chars read from the document flow. |
11411 | | */ |
11412 | 420k | if (avail < 4) |
11413 | 25.3k | goto done; |
11414 | | |
11415 | | /* |
11416 | | * Get the 4 first bytes and decode the charset |
11417 | | * if enc != XML_CHAR_ENCODING_NONE |
11418 | | * plug some encoding conversion routines, |
11419 | | * else xmlSwitchEncoding will set to (default) |
11420 | | * UTF8. |
11421 | | */ |
11422 | 394k | start[0] = RAW; |
11423 | 394k | start[1] = NXT(1); |
11424 | 394k | start[2] = NXT(2); |
11425 | 394k | start[3] = NXT(3); |
11426 | 394k | enc = xmlDetectCharEncoding(start, 4); |
11427 | 394k | xmlSwitchEncoding(ctxt, enc); |
11428 | 394k | break; |
11429 | 420k | } |
11430 | | |
11431 | 1.30M | if (avail < 2) |
11432 | 380 | goto done; |
11433 | 1.30M | cur = ctxt->input->cur[0]; |
11434 | 1.30M | next = ctxt->input->cur[1]; |
11435 | 1.30M | if (cur == 0) { |
11436 | 4.70k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11437 | 4.70k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11438 | 4.70k | &xmlDefaultSAXLocator); |
11439 | 4.70k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11440 | 4.70k | xmlHaltParser(ctxt); |
11441 | | #ifdef DEBUG_PUSH |
11442 | | xmlGenericError(xmlGenericErrorContext, |
11443 | | "PP: entering EOF\n"); |
11444 | | #endif |
11445 | 4.70k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11446 | 4.70k | ctxt->sax->endDocument(ctxt->userData); |
11447 | 4.70k | goto done; |
11448 | 4.70k | } |
11449 | 1.30M | if ((cur == '<') && (next == '?')) { |
11450 | | /* PI or XML decl */ |
11451 | 829k | if (avail < 5) return(ret); |
11452 | 828k | if ((!terminate) && |
11453 | 828k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
11454 | 525k | return(ret); |
11455 | 302k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11456 | 302k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11457 | 302k | &xmlDefaultSAXLocator); |
11458 | 302k | if ((ctxt->input->cur[2] == 'x') && |
11459 | 302k | (ctxt->input->cur[3] == 'm') && |
11460 | 302k | (ctxt->input->cur[4] == 'l') && |
11461 | 302k | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11462 | 260k | ret += 5; |
11463 | | #ifdef DEBUG_PUSH |
11464 | | xmlGenericError(xmlGenericErrorContext, |
11465 | | "PP: Parsing XML Decl\n"); |
11466 | | #endif |
11467 | 260k | xmlParseXMLDecl(ctxt); |
11468 | 260k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11469 | | /* |
11470 | | * The XML REC instructs us to stop parsing right |
11471 | | * here |
11472 | | */ |
11473 | 974 | xmlHaltParser(ctxt); |
11474 | 974 | return(0); |
11475 | 974 | } |
11476 | 259k | ctxt->standalone = ctxt->input->standalone; |
11477 | 259k | if ((ctxt->encoding == NULL) && |
11478 | 259k | (ctxt->input->encoding != NULL)) |
11479 | 34.3k | ctxt->encoding = xmlStrdup(ctxt->input->encoding); |
11480 | 259k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11481 | 259k | (!ctxt->disableSAX)) |
11482 | 220k | ctxt->sax->startDocument(ctxt->userData); |
11483 | 259k | ctxt->instate = XML_PARSER_MISC; |
11484 | | #ifdef DEBUG_PUSH |
11485 | | xmlGenericError(xmlGenericErrorContext, |
11486 | | "PP: entering MISC\n"); |
11487 | | #endif |
11488 | 259k | } else { |
11489 | 41.9k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11490 | 41.9k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11491 | 41.9k | (!ctxt->disableSAX)) |
11492 | 41.9k | ctxt->sax->startDocument(ctxt->userData); |
11493 | 41.9k | ctxt->instate = XML_PARSER_MISC; |
11494 | | #ifdef DEBUG_PUSH |
11495 | | xmlGenericError(xmlGenericErrorContext, |
11496 | | "PP: entering MISC\n"); |
11497 | | #endif |
11498 | 41.9k | } |
11499 | 473k | } else { |
11500 | 473k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11501 | 473k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11502 | 473k | &xmlDefaultSAXLocator); |
11503 | 473k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11504 | 473k | if (ctxt->version == NULL) { |
11505 | 0 | xmlErrMemory(ctxt, NULL); |
11506 | 0 | break; |
11507 | 0 | } |
11508 | 473k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11509 | 473k | (!ctxt->disableSAX)) |
11510 | 473k | ctxt->sax->startDocument(ctxt->userData); |
11511 | 473k | ctxt->instate = XML_PARSER_MISC; |
11512 | | #ifdef DEBUG_PUSH |
11513 | | xmlGenericError(xmlGenericErrorContext, |
11514 | | "PP: entering MISC\n"); |
11515 | | #endif |
11516 | 473k | } |
11517 | 775k | break; |
11518 | 8.62M | case XML_PARSER_START_TAG: { |
11519 | 8.62M | const xmlChar *name; |
11520 | 8.62M | const xmlChar *prefix = NULL; |
11521 | 8.62M | const xmlChar *URI = NULL; |
11522 | 8.62M | int line = ctxt->input->line; |
11523 | 8.62M | int nsNr = ctxt->nsNr; |
11524 | | |
11525 | 8.62M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11526 | 0 | goto done; |
11527 | 8.62M | cur = ctxt->input->cur[0]; |
11528 | 8.62M | if (cur != '<') { |
11529 | 42.0k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11530 | 42.0k | xmlHaltParser(ctxt); |
11531 | 42.0k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11532 | 42.0k | ctxt->sax->endDocument(ctxt->userData); |
11533 | 42.0k | goto done; |
11534 | 42.0k | } |
11535 | 8.58M | if (!terminate) { |
11536 | 8.01M | if (ctxt->progressive) { |
11537 | | /* > can be found unescaped in attribute values */ |
11538 | 8.01M | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11539 | 1.01M | goto done; |
11540 | 8.01M | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11541 | 0 | goto done; |
11542 | 0 | } |
11543 | 8.01M | } |
11544 | 7.56M | if (ctxt->spaceNr == 0) |
11545 | 90.0k | spacePush(ctxt, -1); |
11546 | 7.47M | else if (*ctxt->space == -2) |
11547 | 1.12M | spacePush(ctxt, -1); |
11548 | 6.35M | else |
11549 | 6.35M | spacePush(ctxt, *ctxt->space); |
11550 | 7.56M | #ifdef LIBXML_SAX1_ENABLED |
11551 | 7.56M | if (ctxt->sax2) |
11552 | 3.61M | #endif /* LIBXML_SAX1_ENABLED */ |
11553 | 3.61M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
11554 | 3.95M | #ifdef LIBXML_SAX1_ENABLED |
11555 | 3.95M | else |
11556 | 3.95M | name = xmlParseStartTag(ctxt); |
11557 | 7.56M | #endif /* LIBXML_SAX1_ENABLED */ |
11558 | 7.56M | if (ctxt->instate == XML_PARSER_EOF) |
11559 | 3 | goto done; |
11560 | 7.56M | if (name == NULL) { |
11561 | 52.1k | spacePop(ctxt); |
11562 | 52.1k | xmlHaltParser(ctxt); |
11563 | 52.1k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11564 | 52.1k | ctxt->sax->endDocument(ctxt->userData); |
11565 | 52.1k | goto done; |
11566 | 52.1k | } |
11567 | 7.51M | #ifdef LIBXML_VALID_ENABLED |
11568 | | /* |
11569 | | * [ VC: Root Element Type ] |
11570 | | * The Name in the document type declaration must match |
11571 | | * the element type of the root element. |
11572 | | */ |
11573 | 7.51M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11574 | 7.51M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11575 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11576 | 7.51M | #endif /* LIBXML_VALID_ENABLED */ |
11577 | | |
11578 | | /* |
11579 | | * Check for an Empty Element. |
11580 | | */ |
11581 | 7.51M | if ((RAW == '/') && (NXT(1) == '>')) { |
11582 | 3.02M | SKIP(2); |
11583 | | |
11584 | 3.02M | if (ctxt->sax2) { |
11585 | 1.66M | if ((ctxt->sax != NULL) && |
11586 | 1.66M | (ctxt->sax->endElementNs != NULL) && |
11587 | 1.66M | (!ctxt->disableSAX)) |
11588 | 1.66M | ctxt->sax->endElementNs(ctxt->userData, name, |
11589 | 1.66M | prefix, URI); |
11590 | 1.66M | if (ctxt->nsNr - nsNr > 0) |
11591 | 11.6k | nsPop(ctxt, ctxt->nsNr - nsNr); |
11592 | 1.66M | #ifdef LIBXML_SAX1_ENABLED |
11593 | 1.66M | } else { |
11594 | 1.35M | if ((ctxt->sax != NULL) && |
11595 | 1.35M | (ctxt->sax->endElement != NULL) && |
11596 | 1.35M | (!ctxt->disableSAX)) |
11597 | 1.35M | ctxt->sax->endElement(ctxt->userData, name); |
11598 | 1.35M | #endif /* LIBXML_SAX1_ENABLED */ |
11599 | 1.35M | } |
11600 | 3.02M | if (ctxt->instate == XML_PARSER_EOF) |
11601 | 0 | goto done; |
11602 | 3.02M | spacePop(ctxt); |
11603 | 3.02M | if (ctxt->nameNr == 0) { |
11604 | 12.1k | ctxt->instate = XML_PARSER_EPILOG; |
11605 | 3.01M | } else { |
11606 | 3.01M | ctxt->instate = XML_PARSER_CONTENT; |
11607 | 3.01M | } |
11608 | 3.02M | ctxt->progressive = 1; |
11609 | 3.02M | break; |
11610 | 3.02M | } |
11611 | 4.49M | if (RAW == '>') { |
11612 | 3.50M | NEXT; |
11613 | 3.50M | } else { |
11614 | 988k | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11615 | 988k | "Couldn't find end of Start Tag %s\n", |
11616 | 988k | name); |
11617 | 988k | nodePop(ctxt); |
11618 | 988k | spacePop(ctxt); |
11619 | 988k | } |
11620 | 4.49M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
11621 | | |
11622 | 4.49M | ctxt->instate = XML_PARSER_CONTENT; |
11623 | 4.49M | ctxt->progressive = 1; |
11624 | 4.49M | break; |
11625 | 7.51M | } |
11626 | 31.3M | case XML_PARSER_CONTENT: { |
11627 | 31.3M | int id; |
11628 | 31.3M | unsigned long cons; |
11629 | 31.3M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11630 | 118k | goto done; |
11631 | 31.2M | cur = ctxt->input->cur[0]; |
11632 | 31.2M | next = ctxt->input->cur[1]; |
11633 | | |
11634 | 31.2M | id = ctxt->input->id; |
11635 | 31.2M | cons = CUR_CONSUMED; |
11636 | 31.2M | if ((cur == '<') && (next == '/')) { |
11637 | 2.44M | ctxt->instate = XML_PARSER_END_TAG; |
11638 | 2.44M | break; |
11639 | 28.7M | } else if ((cur == '<') && (next == '?')) { |
11640 | 223k | if ((!terminate) && |
11641 | 223k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11642 | 113k | ctxt->progressive = XML_PARSER_PI; |
11643 | 113k | goto done; |
11644 | 113k | } |
11645 | 109k | xmlParsePI(ctxt); |
11646 | 109k | ctxt->instate = XML_PARSER_CONTENT; |
11647 | 109k | ctxt->progressive = 1; |
11648 | 28.5M | } else if ((cur == '<') && (next != '!')) { |
11649 | 7.05M | ctxt->instate = XML_PARSER_START_TAG; |
11650 | 7.05M | break; |
11651 | 21.4M | } else if ((cur == '<') && (next == '!') && |
11652 | 21.4M | (ctxt->input->cur[2] == '-') && |
11653 | 21.4M | (ctxt->input->cur[3] == '-')) { |
11654 | 413k | int term; |
11655 | | |
11656 | 413k | if (avail < 4) |
11657 | 0 | goto done; |
11658 | 413k | ctxt->input->cur += 4; |
11659 | 413k | term = xmlParseLookupSequence(ctxt, '-', '-', '>'); |
11660 | 413k | ctxt->input->cur -= 4; |
11661 | 413k | if ((!terminate) && (term < 0)) { |
11662 | 173k | ctxt->progressive = XML_PARSER_COMMENT; |
11663 | 173k | goto done; |
11664 | 173k | } |
11665 | 240k | xmlParseComment(ctxt); |
11666 | 240k | ctxt->instate = XML_PARSER_CONTENT; |
11667 | 240k | ctxt->progressive = 1; |
11668 | 21.0M | } else if ((cur == '<') && (ctxt->input->cur[1] == '!') && |
11669 | 21.0M | (ctxt->input->cur[2] == '[') && |
11670 | 21.0M | (ctxt->input->cur[3] == 'C') && |
11671 | 21.0M | (ctxt->input->cur[4] == 'D') && |
11672 | 21.0M | (ctxt->input->cur[5] == 'A') && |
11673 | 21.0M | (ctxt->input->cur[6] == 'T') && |
11674 | 21.0M | (ctxt->input->cur[7] == 'A') && |
11675 | 21.0M | (ctxt->input->cur[8] == '[')) { |
11676 | 50.7k | SKIP(9); |
11677 | 50.7k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11678 | 50.7k | break; |
11679 | 21.0M | } else if ((cur == '<') && (next == '!') && |
11680 | 21.0M | (avail < 9)) { |
11681 | 21.9k | goto done; |
11682 | 20.9M | } else if (cur == '&') { |
11683 | 8.76M | if ((!terminate) && |
11684 | 8.76M | (xmlParseLookupSequence(ctxt, ';', 0, 0) < 0)) |
11685 | 253k | goto done; |
11686 | 8.51M | xmlParseReference(ctxt); |
11687 | 12.2M | } else { |
11688 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11689 | | /* |
11690 | | * Goal of the following test is: |
11691 | | * - minimize calls to the SAX 'character' callback |
11692 | | * when they are mergeable |
11693 | | * - handle an problem for isBlank when we only parse |
11694 | | * a sequence of blank chars and the next one is |
11695 | | * not available to check against '<' presence. |
11696 | | * - tries to homogenize the differences in SAX |
11697 | | * callbacks between the push and pull versions |
11698 | | * of the parser. |
11699 | | */ |
11700 | 12.2M | if ((ctxt->inputNr == 1) && |
11701 | 12.2M | (avail < XML_PARSER_BIG_BUFFER_SIZE)) { |
11702 | 9.08M | if (!terminate) { |
11703 | 8.98M | if (ctxt->progressive) { |
11704 | 8.98M | if ((lastlt == NULL) || |
11705 | 8.98M | (ctxt->input->cur > lastlt)) |
11706 | 849k | goto done; |
11707 | 8.98M | } else if (xmlParseLookupSequence(ctxt, |
11708 | 0 | '<', 0, 0) < 0) { |
11709 | 0 | goto done; |
11710 | 0 | } |
11711 | 8.98M | } |
11712 | 9.08M | } |
11713 | 11.3M | ctxt->checkIndex = 0; |
11714 | 11.3M | xmlParseCharData(ctxt, 0); |
11715 | 11.3M | } |
11716 | 20.2M | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
11717 | 173k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
11718 | 173k | "detected an error in element content\n"); |
11719 | 173k | xmlHaltParser(ctxt); |
11720 | 173k | break; |
11721 | 173k | } |
11722 | 20.0M | break; |
11723 | 20.2M | } |
11724 | 20.0M | case XML_PARSER_END_TAG: |
11725 | 2.54M | if (avail < 2) |
11726 | 0 | goto done; |
11727 | 2.54M | if (!terminate) { |
11728 | 2.47M | if (ctxt->progressive) { |
11729 | | /* > can be found unescaped in attribute values */ |
11730 | 2.47M | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11731 | 102k | goto done; |
11732 | 2.47M | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11733 | 0 | goto done; |
11734 | 0 | } |
11735 | 2.47M | } |
11736 | 2.44M | if (ctxt->sax2) { |
11737 | 1.07M | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11738 | 1.07M | nameNsPop(ctxt); |
11739 | 1.07M | } |
11740 | 1.36M | #ifdef LIBXML_SAX1_ENABLED |
11741 | 1.36M | else |
11742 | 1.36M | xmlParseEndTag1(ctxt, 0); |
11743 | 2.44M | #endif /* LIBXML_SAX1_ENABLED */ |
11744 | 2.44M | if (ctxt->instate == XML_PARSER_EOF) { |
11745 | | /* Nothing */ |
11746 | 2.44M | } else if (ctxt->nameNr == 0) { |
11747 | 75.9k | ctxt->instate = XML_PARSER_EPILOG; |
11748 | 2.36M | } else { |
11749 | 2.36M | ctxt->instate = XML_PARSER_CONTENT; |
11750 | 2.36M | } |
11751 | 2.44M | break; |
11752 | 697k | case XML_PARSER_CDATA_SECTION: { |
11753 | | /* |
11754 | | * The Push mode need to have the SAX callback for |
11755 | | * cdataBlock merge back contiguous callbacks. |
11756 | | */ |
11757 | 697k | int base; |
11758 | | |
11759 | 697k | base = xmlParseLookupSequence(ctxt, ']', ']', '>'); |
11760 | 697k | if (base < 0) { |
11761 | 426k | if (avail >= XML_PARSER_BIG_BUFFER_SIZE + 2) { |
11762 | 383k | int tmp; |
11763 | | |
11764 | 383k | tmp = xmlCheckCdataPush(ctxt->input->cur, |
11765 | 383k | XML_PARSER_BIG_BUFFER_SIZE, 0); |
11766 | 383k | if (tmp < 0) { |
11767 | 6.19k | tmp = -tmp; |
11768 | 6.19k | ctxt->input->cur += tmp; |
11769 | 6.19k | goto encoding_error; |
11770 | 6.19k | } |
11771 | 377k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
11772 | 377k | if (ctxt->sax->cdataBlock != NULL) |
11773 | 114k | ctxt->sax->cdataBlock(ctxt->userData, |
11774 | 114k | ctxt->input->cur, tmp); |
11775 | 263k | else if (ctxt->sax->characters != NULL) |
11776 | 263k | ctxt->sax->characters(ctxt->userData, |
11777 | 263k | ctxt->input->cur, tmp); |
11778 | 377k | } |
11779 | 377k | if (ctxt->instate == XML_PARSER_EOF) |
11780 | 0 | goto done; |
11781 | 377k | SKIPL(tmp); |
11782 | 377k | ctxt->checkIndex = 0; |
11783 | 377k | } |
11784 | 419k | goto done; |
11785 | 426k | } else { |
11786 | 271k | int tmp; |
11787 | | |
11788 | 271k | tmp = xmlCheckCdataPush(ctxt->input->cur, base, 1); |
11789 | 271k | if ((tmp < 0) || (tmp != base)) { |
11790 | 240k | tmp = -tmp; |
11791 | 240k | ctxt->input->cur += tmp; |
11792 | 240k | goto encoding_error; |
11793 | 240k | } |
11794 | 31.0k | if ((ctxt->sax != NULL) && (base == 0) && |
11795 | 31.0k | (ctxt->sax->cdataBlock != NULL) && |
11796 | 31.0k | (!ctxt->disableSAX)) { |
11797 | | /* |
11798 | | * Special case to provide identical behaviour |
11799 | | * between pull and push parsers on enpty CDATA |
11800 | | * sections |
11801 | | */ |
11802 | 683 | if ((ctxt->input->cur - ctxt->input->base >= 9) && |
11803 | 683 | (!strncmp((const char *)&ctxt->input->cur[-9], |
11804 | 683 | "<![CDATA[", 9))) |
11805 | 683 | ctxt->sax->cdataBlock(ctxt->userData, |
11806 | 683 | BAD_CAST "", 0); |
11807 | 30.4k | } else if ((ctxt->sax != NULL) && (base > 0) && |
11808 | 30.4k | (!ctxt->disableSAX)) { |
11809 | 29.9k | if (ctxt->sax->cdataBlock != NULL) |
11810 | 17.0k | ctxt->sax->cdataBlock(ctxt->userData, |
11811 | 17.0k | ctxt->input->cur, base); |
11812 | 12.8k | else if (ctxt->sax->characters != NULL) |
11813 | 12.8k | ctxt->sax->characters(ctxt->userData, |
11814 | 12.8k | ctxt->input->cur, base); |
11815 | 29.9k | } |
11816 | 31.0k | if (ctxt->instate == XML_PARSER_EOF) |
11817 | 0 | goto done; |
11818 | 31.0k | SKIPL(base + 3); |
11819 | 31.0k | ctxt->checkIndex = 0; |
11820 | 31.0k | ctxt->instate = XML_PARSER_CONTENT; |
11821 | | #ifdef DEBUG_PUSH |
11822 | | xmlGenericError(xmlGenericErrorContext, |
11823 | | "PP: entering CONTENT\n"); |
11824 | | #endif |
11825 | 31.0k | } |
11826 | 31.0k | break; |
11827 | 697k | } |
11828 | 889k | case XML_PARSER_MISC: |
11829 | 889k | SKIP_BLANKS; |
11830 | 889k | if (ctxt->input->buf == NULL) |
11831 | 0 | avail = ctxt->input->length - |
11832 | 0 | (ctxt->input->cur - ctxt->input->base); |
11833 | 889k | else |
11834 | 889k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11835 | 889k | (ctxt->input->cur - ctxt->input->base); |
11836 | 889k | if (avail < 2) |
11837 | 9.64k | goto done; |
11838 | 879k | cur = ctxt->input->cur[0]; |
11839 | 879k | next = ctxt->input->cur[1]; |
11840 | 879k | if ((cur == '<') && (next == '?')) { |
11841 | 56.1k | if ((!terminate) && |
11842 | 56.1k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11843 | 8.78k | ctxt->progressive = XML_PARSER_PI; |
11844 | 8.78k | goto done; |
11845 | 8.78k | } |
11846 | | #ifdef DEBUG_PUSH |
11847 | | xmlGenericError(xmlGenericErrorContext, |
11848 | | "PP: Parsing PI\n"); |
11849 | | #endif |
11850 | 47.4k | xmlParsePI(ctxt); |
11851 | 47.4k | if (ctxt->instate == XML_PARSER_EOF) |
11852 | 0 | goto done; |
11853 | 47.4k | ctxt->instate = XML_PARSER_MISC; |
11854 | 47.4k | ctxt->progressive = 1; |
11855 | 47.4k | ctxt->checkIndex = 0; |
11856 | 823k | } else if ((cur == '<') && (next == '!') && |
11857 | 823k | (ctxt->input->cur[2] == '-') && |
11858 | 823k | (ctxt->input->cur[3] == '-')) { |
11859 | 52.8k | if ((!terminate) && |
11860 | 52.8k | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11861 | 31.2k | ctxt->progressive = XML_PARSER_COMMENT; |
11862 | 31.2k | goto done; |
11863 | 31.2k | } |
11864 | | #ifdef DEBUG_PUSH |
11865 | | xmlGenericError(xmlGenericErrorContext, |
11866 | | "PP: Parsing Comment\n"); |
11867 | | #endif |
11868 | 21.6k | xmlParseComment(ctxt); |
11869 | 21.6k | if (ctxt->instate == XML_PARSER_EOF) |
11870 | 0 | goto done; |
11871 | 21.6k | ctxt->instate = XML_PARSER_MISC; |
11872 | 21.6k | ctxt->progressive = 1; |
11873 | 21.6k | ctxt->checkIndex = 0; |
11874 | 770k | } else if ((cur == '<') && (next == '!') && |
11875 | 770k | (ctxt->input->cur[2] == 'D') && |
11876 | 770k | (ctxt->input->cur[3] == 'O') && |
11877 | 770k | (ctxt->input->cur[4] == 'C') && |
11878 | 770k | (ctxt->input->cur[5] == 'T') && |
11879 | 770k | (ctxt->input->cur[6] == 'Y') && |
11880 | 770k | (ctxt->input->cur[7] == 'P') && |
11881 | 770k | (ctxt->input->cur[8] == 'E')) { |
11882 | 355k | if ((!terminate) && |
11883 | 355k | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) { |
11884 | 45.4k | ctxt->progressive = XML_PARSER_DTD; |
11885 | 45.4k | goto done; |
11886 | 45.4k | } |
11887 | | #ifdef DEBUG_PUSH |
11888 | | xmlGenericError(xmlGenericErrorContext, |
11889 | | "PP: Parsing internal subset\n"); |
11890 | | #endif |
11891 | 310k | ctxt->inSubset = 1; |
11892 | 310k | ctxt->progressive = 0; |
11893 | 310k | ctxt->checkIndex = 0; |
11894 | 310k | xmlParseDocTypeDecl(ctxt); |
11895 | 310k | if (ctxt->instate == XML_PARSER_EOF) |
11896 | 0 | goto done; |
11897 | 310k | if (RAW == '[') { |
11898 | 236k | ctxt->instate = XML_PARSER_DTD; |
11899 | | #ifdef DEBUG_PUSH |
11900 | | xmlGenericError(xmlGenericErrorContext, |
11901 | | "PP: entering DTD\n"); |
11902 | | #endif |
11903 | 236k | } else { |
11904 | | /* |
11905 | | * Create and update the external subset. |
11906 | | */ |
11907 | 74.1k | ctxt->inSubset = 2; |
11908 | 74.1k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11909 | 74.1k | (ctxt->sax->externalSubset != NULL)) |
11910 | 67.8k | ctxt->sax->externalSubset(ctxt->userData, |
11911 | 67.8k | ctxt->intSubName, ctxt->extSubSystem, |
11912 | 67.8k | ctxt->extSubURI); |
11913 | 74.1k | ctxt->inSubset = 0; |
11914 | 74.1k | xmlCleanSpecialAttr(ctxt); |
11915 | 74.1k | ctxt->instate = XML_PARSER_PROLOG; |
11916 | | #ifdef DEBUG_PUSH |
11917 | | xmlGenericError(xmlGenericErrorContext, |
11918 | | "PP: entering PROLOG\n"); |
11919 | | #endif |
11920 | 74.1k | } |
11921 | 414k | } else if ((cur == '<') && (next == '!') && |
11922 | 414k | (avail < 9)) { |
11923 | 9.58k | goto done; |
11924 | 405k | } else { |
11925 | 405k | ctxt->instate = XML_PARSER_START_TAG; |
11926 | 405k | ctxt->progressive = XML_PARSER_START_TAG; |
11927 | 405k | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11928 | | #ifdef DEBUG_PUSH |
11929 | | xmlGenericError(xmlGenericErrorContext, |
11930 | | "PP: entering START_TAG\n"); |
11931 | | #endif |
11932 | 405k | } |
11933 | 784k | break; |
11934 | 784k | case XML_PARSER_PROLOG: |
11935 | 287k | SKIP_BLANKS; |
11936 | 287k | if (ctxt->input->buf == NULL) |
11937 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11938 | 287k | else |
11939 | 287k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11940 | 287k | (ctxt->input->cur - ctxt->input->base); |
11941 | 287k | if (avail < 2) |
11942 | 10.7k | goto done; |
11943 | 277k | cur = ctxt->input->cur[0]; |
11944 | 277k | next = ctxt->input->cur[1]; |
11945 | 277k | if ((cur == '<') && (next == '?')) { |
11946 | 49.5k | if ((!terminate) && |
11947 | 49.5k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11948 | 16.9k | ctxt->progressive = XML_PARSER_PI; |
11949 | 16.9k | goto done; |
11950 | 16.9k | } |
11951 | | #ifdef DEBUG_PUSH |
11952 | | xmlGenericError(xmlGenericErrorContext, |
11953 | | "PP: Parsing PI\n"); |
11954 | | #endif |
11955 | 32.6k | xmlParsePI(ctxt); |
11956 | 32.6k | if (ctxt->instate == XML_PARSER_EOF) |
11957 | 0 | goto done; |
11958 | 32.6k | ctxt->instate = XML_PARSER_PROLOG; |
11959 | 32.6k | ctxt->progressive = 1; |
11960 | 227k | } else if ((cur == '<') && (next == '!') && |
11961 | 227k | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
11962 | 42.2k | if ((!terminate) && |
11963 | 42.2k | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11964 | 9.39k | ctxt->progressive = XML_PARSER_COMMENT; |
11965 | 9.39k | goto done; |
11966 | 9.39k | } |
11967 | | #ifdef DEBUG_PUSH |
11968 | | xmlGenericError(xmlGenericErrorContext, |
11969 | | "PP: Parsing Comment\n"); |
11970 | | #endif |
11971 | 32.8k | xmlParseComment(ctxt); |
11972 | 32.8k | if (ctxt->instate == XML_PARSER_EOF) |
11973 | 0 | goto done; |
11974 | 32.8k | ctxt->instate = XML_PARSER_PROLOG; |
11975 | 32.8k | ctxt->progressive = 1; |
11976 | 185k | } else if ((cur == '<') && (next == '!') && |
11977 | 185k | (avail < 4)) { |
11978 | 677 | goto done; |
11979 | 184k | } else { |
11980 | 184k | ctxt->instate = XML_PARSER_START_TAG; |
11981 | 184k | if (ctxt->progressive == 0) |
11982 | 153k | ctxt->progressive = XML_PARSER_START_TAG; |
11983 | 184k | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11984 | | #ifdef DEBUG_PUSH |
11985 | | xmlGenericError(xmlGenericErrorContext, |
11986 | | "PP: entering START_TAG\n"); |
11987 | | #endif |
11988 | 184k | } |
11989 | 250k | break; |
11990 | 250k | case XML_PARSER_EPILOG: |
11991 | 101k | SKIP_BLANKS; |
11992 | 101k | if (ctxt->input->buf == NULL) |
11993 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11994 | 101k | else |
11995 | 101k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11996 | 101k | (ctxt->input->cur - ctxt->input->base); |
11997 | 101k | if (avail < 2) |
11998 | 68.8k | goto done; |
11999 | 32.4k | cur = ctxt->input->cur[0]; |
12000 | 32.4k | next = ctxt->input->cur[1]; |
12001 | 32.4k | if ((cur == '<') && (next == '?')) { |
12002 | 10.4k | if ((!terminate) && |
12003 | 10.4k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
12004 | 8.35k | ctxt->progressive = XML_PARSER_PI; |
12005 | 8.35k | goto done; |
12006 | 8.35k | } |
12007 | | #ifdef DEBUG_PUSH |
12008 | | xmlGenericError(xmlGenericErrorContext, |
12009 | | "PP: Parsing PI\n"); |
12010 | | #endif |
12011 | 2.14k | xmlParsePI(ctxt); |
12012 | 2.14k | if (ctxt->instate == XML_PARSER_EOF) |
12013 | 0 | goto done; |
12014 | 2.14k | ctxt->instate = XML_PARSER_EPILOG; |
12015 | 2.14k | ctxt->progressive = 1; |
12016 | 21.9k | } else if ((cur == '<') && (next == '!') && |
12017 | 21.9k | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
12018 | 7.73k | if ((!terminate) && |
12019 | 7.73k | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
12020 | 6.49k | ctxt->progressive = XML_PARSER_COMMENT; |
12021 | 6.49k | goto done; |
12022 | 6.49k | } |
12023 | | #ifdef DEBUG_PUSH |
12024 | | xmlGenericError(xmlGenericErrorContext, |
12025 | | "PP: Parsing Comment\n"); |
12026 | | #endif |
12027 | 1.24k | xmlParseComment(ctxt); |
12028 | 1.24k | if (ctxt->instate == XML_PARSER_EOF) |
12029 | 0 | goto done; |
12030 | 1.24k | ctxt->instate = XML_PARSER_EPILOG; |
12031 | 1.24k | ctxt->progressive = 1; |
12032 | 14.2k | } else if ((cur == '<') && (next == '!') && |
12033 | 14.2k | (avail < 4)) { |
12034 | 305 | goto done; |
12035 | 13.9k | } else { |
12036 | 13.9k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12037 | 13.9k | xmlHaltParser(ctxt); |
12038 | | #ifdef DEBUG_PUSH |
12039 | | xmlGenericError(xmlGenericErrorContext, |
12040 | | "PP: entering EOF\n"); |
12041 | | #endif |
12042 | 13.9k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
12043 | 13.9k | ctxt->sax->endDocument(ctxt->userData); |
12044 | 13.9k | goto done; |
12045 | 13.9k | } |
12046 | 3.38k | break; |
12047 | 4.48M | case XML_PARSER_DTD: { |
12048 | | /* |
12049 | | * Sorry but progressive parsing of the internal subset |
12050 | | * is not expected to be supported. We first check that |
12051 | | * the full content of the internal subset is available and |
12052 | | * the parsing is launched only at that point. |
12053 | | * Internal subset ends up with "']' S? '>'" in an unescaped |
12054 | | * section and not in a ']]>' sequence which are conditional |
12055 | | * sections (whoever argued to keep that crap in XML deserve |
12056 | | * a place in hell !). |
12057 | | */ |
12058 | 4.48M | int base, i; |
12059 | 4.48M | xmlChar *buf; |
12060 | 4.48M | xmlChar quote = 0; |
12061 | 4.48M | size_t use; |
12062 | | |
12063 | 4.48M | base = ctxt->input->cur - ctxt->input->base; |
12064 | 4.48M | if (base < 0) return(0); |
12065 | 4.48M | if (ctxt->checkIndex > base) |
12066 | 2.04M | base = ctxt->checkIndex; |
12067 | 4.48M | buf = xmlBufContent(ctxt->input->buf->buffer); |
12068 | 4.48M | use = xmlBufUse(ctxt->input->buf->buffer); |
12069 | 293G | for (;(unsigned int) base < use; base++) { |
12070 | 293G | if (quote != 0) { |
12071 | 181G | if (buf[base] == quote) |
12072 | 13.0G | quote = 0; |
12073 | 181G | continue; |
12074 | 181G | } |
12075 | 111G | if ((quote == 0) && (buf[base] == '<')) { |
12076 | 3.35G | int found = 0; |
12077 | | /* special handling of comments */ |
12078 | 3.35G | if (((unsigned int) base + 4 < use) && |
12079 | 3.35G | (buf[base + 1] == '!') && |
12080 | 3.35G | (buf[base + 2] == '-') && |
12081 | 3.35G | (buf[base + 3] == '-')) { |
12082 | 8.46G | for (;(unsigned int) base + 3 < use; base++) { |
12083 | 8.46G | if ((buf[base] == '-') && |
12084 | 8.46G | (buf[base + 1] == '-') && |
12085 | 8.46G | (buf[base + 2] == '>')) { |
12086 | 10.3M | found = 1; |
12087 | 10.3M | base += 2; |
12088 | 10.3M | break; |
12089 | 10.3M | } |
12090 | 8.46G | } |
12091 | 10.5M | if (!found) { |
12092 | | #if 0 |
12093 | | fprintf(stderr, "unfinished comment\n"); |
12094 | | #endif |
12095 | 209k | break; /* for */ |
12096 | 209k | } |
12097 | 10.3M | continue; |
12098 | 10.5M | } |
12099 | 3.35G | } |
12100 | 111G | if (buf[base] == '"') { |
12101 | 13.0G | quote = '"'; |
12102 | 13.0G | continue; |
12103 | 13.0G | } |
12104 | 98.4G | if (buf[base] == '\'') { |
12105 | 30.6M | quote = '\''; |
12106 | 30.6M | continue; |
12107 | 30.6M | } |
12108 | 98.4G | if (buf[base] == ']') { |
12109 | | #if 0 |
12110 | | fprintf(stderr, "%c%c%c%c: ", buf[base], |
12111 | | buf[base + 1], buf[base + 2], buf[base + 3]); |
12112 | | #endif |
12113 | 4.89M | if ((unsigned int) base +1 >= use) |
12114 | 1.02k | break; |
12115 | 4.89M | if (buf[base + 1] == ']') { |
12116 | | /* conditional crap, skip both ']' ! */ |
12117 | 3.57M | base++; |
12118 | 3.57M | continue; |
12119 | 3.57M | } |
12120 | 2.29M | for (i = 1; (unsigned int) base + i < use; i++) { |
12121 | 2.29M | if (buf[base + i] == '>') { |
12122 | | #if 0 |
12123 | | fprintf(stderr, "found\n"); |
12124 | | #endif |
12125 | 184k | goto found_end_int_subset; |
12126 | 184k | } |
12127 | 2.10M | if (!IS_BLANK_CH(buf[base + i])) { |
12128 | | #if 0 |
12129 | | fprintf(stderr, "not found\n"); |
12130 | | #endif |
12131 | 1.13M | goto not_end_of_int_subset; |
12132 | 1.13M | } |
12133 | 2.10M | } |
12134 | | #if 0 |
12135 | | fprintf(stderr, "end of stream\n"); |
12136 | | #endif |
12137 | 384 | break; |
12138 | | |
12139 | 1.32M | } |
12140 | 98.4G | not_end_of_int_subset: |
12141 | 98.4G | continue; /* for */ |
12142 | 98.4G | } |
12143 | | /* |
12144 | | * We didn't found the end of the Internal subset |
12145 | | */ |
12146 | 4.29M | if (quote == 0) |
12147 | 2.07M | ctxt->checkIndex = base; |
12148 | 2.22M | else |
12149 | 2.22M | ctxt->checkIndex = 0; |
12150 | | #ifdef DEBUG_PUSH |
12151 | | if (next == 0) |
12152 | | xmlGenericError(xmlGenericErrorContext, |
12153 | | "PP: lookup of int subset end filed\n"); |
12154 | | #endif |
12155 | 4.29M | goto done; |
12156 | | |
12157 | 184k | found_end_int_subset: |
12158 | 184k | ctxt->checkIndex = 0; |
12159 | 184k | xmlParseInternalSubset(ctxt); |
12160 | 184k | if (ctxt->instate == XML_PARSER_EOF) |
12161 | 5.46k | goto done; |
12162 | 179k | ctxt->inSubset = 2; |
12163 | 179k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
12164 | 179k | (ctxt->sax->externalSubset != NULL)) |
12165 | 150k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
12166 | 150k | ctxt->extSubSystem, ctxt->extSubURI); |
12167 | 179k | ctxt->inSubset = 0; |
12168 | 179k | xmlCleanSpecialAttr(ctxt); |
12169 | 179k | if (ctxt->instate == XML_PARSER_EOF) |
12170 | 4.70k | goto done; |
12171 | 174k | ctxt->instate = XML_PARSER_PROLOG; |
12172 | 174k | ctxt->checkIndex = 0; |
12173 | | #ifdef DEBUG_PUSH |
12174 | | xmlGenericError(xmlGenericErrorContext, |
12175 | | "PP: entering PROLOG\n"); |
12176 | | #endif |
12177 | 174k | break; |
12178 | 179k | } |
12179 | 0 | case XML_PARSER_COMMENT: |
12180 | 0 | xmlGenericError(xmlGenericErrorContext, |
12181 | 0 | "PP: internal error, state == COMMENT\n"); |
12182 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12183 | | #ifdef DEBUG_PUSH |
12184 | | xmlGenericError(xmlGenericErrorContext, |
12185 | | "PP: entering CONTENT\n"); |
12186 | | #endif |
12187 | 0 | break; |
12188 | 0 | case XML_PARSER_IGNORE: |
12189 | 0 | xmlGenericError(xmlGenericErrorContext, |
12190 | 0 | "PP: internal error, state == IGNORE"); |
12191 | 0 | ctxt->instate = XML_PARSER_DTD; |
12192 | | #ifdef DEBUG_PUSH |
12193 | | xmlGenericError(xmlGenericErrorContext, |
12194 | | "PP: entering DTD\n"); |
12195 | | #endif |
12196 | 0 | break; |
12197 | 0 | case XML_PARSER_PI: |
12198 | 0 | xmlGenericError(xmlGenericErrorContext, |
12199 | 0 | "PP: internal error, state == PI\n"); |
12200 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12201 | | #ifdef DEBUG_PUSH |
12202 | | xmlGenericError(xmlGenericErrorContext, |
12203 | | "PP: entering CONTENT\n"); |
12204 | | #endif |
12205 | 0 | break; |
12206 | 0 | case XML_PARSER_ENTITY_DECL: |
12207 | 0 | xmlGenericError(xmlGenericErrorContext, |
12208 | 0 | "PP: internal error, state == ENTITY_DECL\n"); |
12209 | 0 | ctxt->instate = XML_PARSER_DTD; |
12210 | | #ifdef DEBUG_PUSH |
12211 | | xmlGenericError(xmlGenericErrorContext, |
12212 | | "PP: entering DTD\n"); |
12213 | | #endif |
12214 | 0 | break; |
12215 | 0 | case XML_PARSER_ENTITY_VALUE: |
12216 | 0 | xmlGenericError(xmlGenericErrorContext, |
12217 | 0 | "PP: internal error, state == ENTITY_VALUE\n"); |
12218 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12219 | | #ifdef DEBUG_PUSH |
12220 | | xmlGenericError(xmlGenericErrorContext, |
12221 | | "PP: entering DTD\n"); |
12222 | | #endif |
12223 | 0 | break; |
12224 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
12225 | 0 | xmlGenericError(xmlGenericErrorContext, |
12226 | 0 | "PP: internal error, state == ATTRIBUTE_VALUE\n"); |
12227 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12228 | | #ifdef DEBUG_PUSH |
12229 | | xmlGenericError(xmlGenericErrorContext, |
12230 | | "PP: entering START_TAG\n"); |
12231 | | #endif |
12232 | 0 | break; |
12233 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
12234 | 0 | xmlGenericError(xmlGenericErrorContext, |
12235 | 0 | "PP: internal error, state == SYSTEM_LITERAL\n"); |
12236 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12237 | | #ifdef DEBUG_PUSH |
12238 | | xmlGenericError(xmlGenericErrorContext, |
12239 | | "PP: entering START_TAG\n"); |
12240 | | #endif |
12241 | 0 | break; |
12242 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
12243 | 0 | xmlGenericError(xmlGenericErrorContext, |
12244 | 0 | "PP: internal error, state == PUBLIC_LITERAL\n"); |
12245 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12246 | | #ifdef DEBUG_PUSH |
12247 | | xmlGenericError(xmlGenericErrorContext, |
12248 | | "PP: entering START_TAG\n"); |
12249 | | #endif |
12250 | 0 | break; |
12251 | 50.6M | } |
12252 | 50.6M | } |
12253 | 8.19M | done: |
12254 | | #ifdef DEBUG_PUSH |
12255 | | xmlGenericError(xmlGenericErrorContext, "PP: done %d\n", ret); |
12256 | | #endif |
12257 | 8.19M | return(ret); |
12258 | 246k | encoding_error: |
12259 | 246k | { |
12260 | 246k | char buffer[150]; |
12261 | | |
12262 | 246k | snprintf(buffer, 149, "Bytes: 0x%02X 0x%02X 0x%02X 0x%02X\n", |
12263 | 246k | ctxt->input->cur[0], ctxt->input->cur[1], |
12264 | 246k | ctxt->input->cur[2], ctxt->input->cur[3]); |
12265 | 246k | __xmlErrEncoding(ctxt, XML_ERR_INVALID_CHAR, |
12266 | 246k | "Input is not proper UTF-8, indicate encoding !\n%s", |
12267 | 246k | BAD_CAST buffer, NULL); |
12268 | 246k | } |
12269 | 246k | return(0); |
12270 | 9.19M | } |
12271 | | |
12272 | | /** |
12273 | | * xmlParseCheckTransition: |
12274 | | * @ctxt: an XML parser context |
12275 | | * @chunk: a char array |
12276 | | * @size: the size in byte of the chunk |
12277 | | * |
12278 | | * Check depending on the current parser state if the chunk given must be |
12279 | | * processed immediately or one need more data to advance on parsing. |
12280 | | * |
12281 | | * Returns -1 in case of error, 0 if the push is not needed and 1 if needed |
12282 | | */ |
12283 | | static int |
12284 | 10.2M | xmlParseCheckTransition(xmlParserCtxtPtr ctxt, const char *chunk, int size) { |
12285 | 10.2M | if ((ctxt == NULL) || (chunk == NULL) || (size < 0)) |
12286 | 0 | return(-1); |
12287 | 10.2M | if (ctxt->instate == XML_PARSER_START_TAG) { |
12288 | 1.56M | if (memchr(chunk, '>', size) != NULL) |
12289 | 948k | return(1); |
12290 | 617k | return(0); |
12291 | 1.56M | } |
12292 | 8.72M | if (ctxt->progressive == XML_PARSER_COMMENT) { |
12293 | 303k | if (memchr(chunk, '>', size) != NULL) |
12294 | 208k | return(1); |
12295 | 94.8k | return(0); |
12296 | 303k | } |
12297 | 8.42M | if (ctxt->instate == XML_PARSER_CDATA_SECTION) { |
12298 | 877k | if (memchr(chunk, '>', size) != NULL) |
12299 | 639k | return(1); |
12300 | 238k | return(0); |
12301 | 877k | } |
12302 | 7.54M | if (ctxt->progressive == XML_PARSER_PI) { |
12303 | 177k | if (memchr(chunk, '>', size) != NULL) |
12304 | 138k | return(1); |
12305 | 39.2k | return(0); |
12306 | 177k | } |
12307 | 7.36M | if (ctxt->instate == XML_PARSER_END_TAG) { |
12308 | 106k | if (memchr(chunk, '>', size) != NULL) |
12309 | 90.8k | return(1); |
12310 | 15.5k | return(0); |
12311 | 106k | } |
12312 | 7.26M | if ((ctxt->progressive == XML_PARSER_DTD) || |
12313 | 7.26M | (ctxt->instate == XML_PARSER_DTD)) { |
12314 | 5.10M | if (memchr(chunk, '>', size) != NULL) |
12315 | 4.22M | return(1); |
12316 | 876k | return(0); |
12317 | 5.10M | } |
12318 | 2.16M | return(1); |
12319 | 7.26M | } |
12320 | | |
12321 | | /** |
12322 | | * xmlParseChunk: |
12323 | | * @ctxt: an XML parser context |
12324 | | * @chunk: an char array |
12325 | | * @size: the size in byte of the chunk |
12326 | | * @terminate: last chunk indicator |
12327 | | * |
12328 | | * Parse a Chunk of memory |
12329 | | * |
12330 | | * Returns zero if no error, the xmlParserErrors otherwise. |
12331 | | */ |
12332 | | int |
12333 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
12334 | 18.3M | int terminate) { |
12335 | 18.3M | int end_in_lf = 0; |
12336 | 18.3M | int remain = 0; |
12337 | 18.3M | size_t old_avail = 0; |
12338 | 18.3M | size_t avail = 0; |
12339 | | |
12340 | 18.3M | if (ctxt == NULL) |
12341 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12342 | 18.3M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12343 | 7.33M | return(ctxt->errNo); |
12344 | 11.0M | if (ctxt->instate == XML_PARSER_EOF) |
12345 | 1.63k | return(-1); |
12346 | 11.0M | if (ctxt->instate == XML_PARSER_START) |
12347 | 1.28M | xmlDetectSAX2(ctxt); |
12348 | 11.0M | if ((size > 0) && (chunk != NULL) && (!terminate) && |
12349 | 11.0M | (chunk[size - 1] == '\r')) { |
12350 | 151k | end_in_lf = 1; |
12351 | 151k | size--; |
12352 | 151k | } |
12353 | | |
12354 | 11.0M | xmldecl_done: |
12355 | | |
12356 | 11.0M | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
12357 | 11.0M | (ctxt->input->buf != NULL) && (ctxt->instate != XML_PARSER_EOF)) { |
12358 | 10.7M | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12359 | 10.7M | size_t cur = ctxt->input->cur - ctxt->input->base; |
12360 | 10.7M | int res; |
12361 | | |
12362 | 10.7M | old_avail = xmlBufUse(ctxt->input->buf->buffer); |
12363 | | /* |
12364 | | * Specific handling if we autodetected an encoding, we should not |
12365 | | * push more than the first line ... which depend on the encoding |
12366 | | * And only push the rest once the final encoding was detected |
12367 | | */ |
12368 | 10.7M | if ((ctxt->instate == XML_PARSER_START) && (ctxt->input != NULL) && |
12369 | 10.7M | (ctxt->input->buf != NULL) && (ctxt->input->buf->encoder != NULL)) { |
12370 | 86.3k | unsigned int len = 45; |
12371 | | |
12372 | 86.3k | if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12373 | 86.3k | BAD_CAST "UTF-16")) || |
12374 | 86.3k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12375 | 30.1k | BAD_CAST "UTF16"))) |
12376 | 56.2k | len = 90; |
12377 | 30.1k | else if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12378 | 30.1k | BAD_CAST "UCS-4")) || |
12379 | 30.1k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12380 | 29.7k | BAD_CAST "UCS4"))) |
12381 | 405 | len = 180; |
12382 | | |
12383 | 86.3k | if (ctxt->input->buf->rawconsumed < len) |
12384 | 11.5k | len -= ctxt->input->buf->rawconsumed; |
12385 | | |
12386 | | /* |
12387 | | * Change size for reading the initial declaration only |
12388 | | * if size is greater than len. Otherwise, memmove in xmlBufferAdd |
12389 | | * will blindly copy extra bytes from memory. |
12390 | | */ |
12391 | 86.3k | if ((unsigned int) size > len) { |
12392 | 58.7k | remain = size - len; |
12393 | 58.7k | size = len; |
12394 | 58.7k | } else { |
12395 | 27.6k | remain = 0; |
12396 | 27.6k | } |
12397 | 86.3k | } |
12398 | 10.7M | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12399 | 10.7M | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12400 | 10.7M | if (res < 0) { |
12401 | 4.35k | ctxt->errNo = XML_PARSER_EOF; |
12402 | 4.35k | xmlHaltParser(ctxt); |
12403 | 4.35k | return (XML_PARSER_EOF); |
12404 | 4.35k | } |
12405 | | #ifdef DEBUG_PUSH |
12406 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12407 | | #endif |
12408 | | |
12409 | 10.7M | } else if (ctxt->instate != XML_PARSER_EOF) { |
12410 | 315k | if ((ctxt->input != NULL) && ctxt->input->buf != NULL) { |
12411 | 315k | xmlParserInputBufferPtr in = ctxt->input->buf; |
12412 | 315k | if ((in->encoder != NULL) && (in->buffer != NULL) && |
12413 | 315k | (in->raw != NULL)) { |
12414 | 25.2k | int nbchars; |
12415 | 25.2k | size_t base = xmlBufGetInputBase(in->buffer, ctxt->input); |
12416 | 25.2k | size_t current = ctxt->input->cur - ctxt->input->base; |
12417 | | |
12418 | 25.2k | nbchars = xmlCharEncInput(in, terminate); |
12419 | 25.2k | xmlBufSetInputBaseCur(in->buffer, ctxt->input, base, current); |
12420 | 25.2k | if (nbchars < 0) { |
12421 | | /* TODO 2.6.0 */ |
12422 | 4.02k | xmlGenericError(xmlGenericErrorContext, |
12423 | 4.02k | "xmlParseChunk: encoder error\n"); |
12424 | 4.02k | xmlHaltParser(ctxt); |
12425 | 4.02k | return(XML_ERR_INVALID_ENCODING); |
12426 | 4.02k | } |
12427 | 25.2k | } |
12428 | 315k | } |
12429 | 315k | } |
12430 | 11.0M | if (remain != 0) { |
12431 | 57.0k | xmlParseTryOrFinish(ctxt, 0); |
12432 | 11.0M | } else { |
12433 | 11.0M | if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) |
12434 | 11.0M | avail = xmlBufUse(ctxt->input->buf->buffer); |
12435 | | /* |
12436 | | * Depending on the current state it may not be such |
12437 | | * a good idea to try parsing if there is nothing in the chunk |
12438 | | * which would be worth doing a parser state transition and we |
12439 | | * need to wait for more data |
12440 | | */ |
12441 | 11.0M | if ((terminate) || (avail > XML_MAX_TEXT_LENGTH) || |
12442 | 11.0M | (old_avail == 0) || (avail == 0) || |
12443 | 11.0M | (xmlParseCheckTransition(ctxt, |
12444 | 10.2M | (const char *)&ctxt->input->base[old_avail], |
12445 | 10.2M | avail - old_avail))) |
12446 | 9.13M | xmlParseTryOrFinish(ctxt, terminate); |
12447 | 11.0M | } |
12448 | 11.0M | if (ctxt->instate == XML_PARSER_EOF) |
12449 | 301k | return(ctxt->errNo); |
12450 | | |
12451 | 10.7M | if ((ctxt->input != NULL) && |
12452 | 10.7M | (((ctxt->input->end - ctxt->input->cur) > XML_MAX_LOOKUP_LIMIT) || |
12453 | 10.7M | ((ctxt->input->cur - ctxt->input->base) > XML_MAX_LOOKUP_LIMIT)) && |
12454 | 10.7M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
12455 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
12456 | 0 | xmlHaltParser(ctxt); |
12457 | 0 | } |
12458 | 10.7M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12459 | 230k | return(ctxt->errNo); |
12460 | | |
12461 | 10.5M | if (remain != 0) { |
12462 | 55.0k | chunk += size; |
12463 | 55.0k | size = remain; |
12464 | 55.0k | remain = 0; |
12465 | 55.0k | goto xmldecl_done; |
12466 | 55.0k | } |
12467 | 10.4M | if ((end_in_lf == 1) && (ctxt->input != NULL) && |
12468 | 10.4M | (ctxt->input->buf != NULL)) { |
12469 | 150k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
12470 | 150k | ctxt->input); |
12471 | 150k | size_t current = ctxt->input->cur - ctxt->input->base; |
12472 | | |
12473 | 150k | xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
12474 | | |
12475 | 150k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
12476 | 150k | base, current); |
12477 | 150k | } |
12478 | 10.4M | if (terminate) { |
12479 | | /* |
12480 | | * Check for termination |
12481 | | */ |
12482 | 166k | int cur_avail = 0; |
12483 | | |
12484 | 166k | if (ctxt->input != NULL) { |
12485 | 166k | if (ctxt->input->buf == NULL) |
12486 | 0 | cur_avail = ctxt->input->length - |
12487 | 0 | (ctxt->input->cur - ctxt->input->base); |
12488 | 166k | else |
12489 | 166k | cur_avail = xmlBufUse(ctxt->input->buf->buffer) - |
12490 | 166k | (ctxt->input->cur - ctxt->input->base); |
12491 | 166k | } |
12492 | | |
12493 | 166k | if ((ctxt->instate != XML_PARSER_EOF) && |
12494 | 166k | (ctxt->instate != XML_PARSER_EPILOG)) { |
12495 | 105k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12496 | 105k | } |
12497 | 166k | if ((ctxt->instate == XML_PARSER_EPILOG) && (cur_avail > 0)) { |
12498 | 828 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12499 | 828 | } |
12500 | 166k | if (ctxt->instate != XML_PARSER_EOF) { |
12501 | 166k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
12502 | 166k | ctxt->sax->endDocument(ctxt->userData); |
12503 | 166k | } |
12504 | 166k | ctxt->instate = XML_PARSER_EOF; |
12505 | 166k | } |
12506 | 10.4M | if (ctxt->wellFormed == 0) |
12507 | 3.78M | return((xmlParserErrors) ctxt->errNo); |
12508 | 6.70M | else |
12509 | 6.70M | return(0); |
12510 | 10.4M | } |
12511 | | |
12512 | | /************************************************************************ |
12513 | | * * |
12514 | | * I/O front end functions to the parser * |
12515 | | * * |
12516 | | ************************************************************************/ |
12517 | | |
12518 | | /** |
12519 | | * xmlCreatePushParserCtxt: |
12520 | | * @sax: a SAX handler |
12521 | | * @user_data: The user data returned on SAX callbacks |
12522 | | * @chunk: a pointer to an array of chars |
12523 | | * @size: number of chars in the array |
12524 | | * @filename: an optional file name or URI |
12525 | | * |
12526 | | * Create a parser context for using the XML parser in push mode. |
12527 | | * If @buffer and @size are non-NULL, the data is used to detect |
12528 | | * the encoding. The remaining characters will be parsed so they |
12529 | | * don't need to be fed in again through xmlParseChunk. |
12530 | | * To allow content encoding detection, @size should be >= 4 |
12531 | | * The value of @filename is used for fetching external entities |
12532 | | * and error/warning reports. |
12533 | | * |
12534 | | * Returns the new parser context or NULL |
12535 | | */ |
12536 | | |
12537 | | xmlParserCtxtPtr |
12538 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12539 | 800k | const char *chunk, int size, const char *filename) { |
12540 | 800k | xmlParserCtxtPtr ctxt; |
12541 | 800k | xmlParserInputPtr inputStream; |
12542 | 800k | xmlParserInputBufferPtr buf; |
12543 | 800k | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
12544 | | |
12545 | | /* |
12546 | | * plug some encoding conversion routines |
12547 | | */ |
12548 | 800k | if ((chunk != NULL) && (size >= 4)) |
12549 | 393k | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
12550 | | |
12551 | 800k | buf = xmlAllocParserInputBuffer(enc); |
12552 | 800k | if (buf == NULL) return(NULL); |
12553 | | |
12554 | 800k | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12555 | 800k | if (ctxt == NULL) { |
12556 | 0 | xmlErrMemory(NULL, "creating parser: out of memory\n"); |
12557 | 0 | xmlFreeParserInputBuffer(buf); |
12558 | 0 | return(NULL); |
12559 | 0 | } |
12560 | 800k | ctxt->dictNames = 1; |
12561 | 800k | if (filename == NULL) { |
12562 | 399k | ctxt->directory = NULL; |
12563 | 400k | } else { |
12564 | 400k | ctxt->directory = xmlParserGetDirectory(filename); |
12565 | 400k | } |
12566 | | |
12567 | 800k | inputStream = xmlNewInputStream(ctxt); |
12568 | 800k | if (inputStream == NULL) { |
12569 | 0 | xmlFreeParserCtxt(ctxt); |
12570 | 0 | xmlFreeParserInputBuffer(buf); |
12571 | 0 | return(NULL); |
12572 | 0 | } |
12573 | | |
12574 | 800k | if (filename == NULL) |
12575 | 399k | inputStream->filename = NULL; |
12576 | 400k | else { |
12577 | 400k | inputStream->filename = (char *) |
12578 | 400k | xmlCanonicPath((const xmlChar *) filename); |
12579 | 400k | if (inputStream->filename == NULL) { |
12580 | 0 | xmlFreeParserCtxt(ctxt); |
12581 | 0 | xmlFreeParserInputBuffer(buf); |
12582 | 0 | return(NULL); |
12583 | 0 | } |
12584 | 400k | } |
12585 | 800k | inputStream->buf = buf; |
12586 | 800k | xmlBufResetInput(inputStream->buf->buffer, inputStream); |
12587 | 800k | inputPush(ctxt, inputStream); |
12588 | | |
12589 | | /* |
12590 | | * If the caller didn't provide an initial 'chunk' for determining |
12591 | | * the encoding, we set the context to XML_CHAR_ENCODING_NONE so |
12592 | | * that it can be automatically determined later |
12593 | | */ |
12594 | 800k | if ((size == 0) || (chunk == NULL)) { |
12595 | 406k | ctxt->charset = XML_CHAR_ENCODING_NONE; |
12596 | 406k | } else if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) { |
12597 | 393k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12598 | 393k | size_t cur = ctxt->input->cur - ctxt->input->base; |
12599 | | |
12600 | 393k | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12601 | | |
12602 | 393k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12603 | | #ifdef DEBUG_PUSH |
12604 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12605 | | #endif |
12606 | 393k | } |
12607 | | |
12608 | 800k | if (enc != XML_CHAR_ENCODING_NONE) { |
12609 | 145k | xmlSwitchEncoding(ctxt, enc); |
12610 | 145k | } |
12611 | | |
12612 | 800k | return(ctxt); |
12613 | 800k | } |
12614 | | #endif /* LIBXML_PUSH_ENABLED */ |
12615 | | |
12616 | | /** |
12617 | | * xmlHaltParser: |
12618 | | * @ctxt: an XML parser context |
12619 | | * |
12620 | | * Blocks further parser processing don't override error |
12621 | | * for internal use |
12622 | | */ |
12623 | | static void |
12624 | 1.59M | xmlHaltParser(xmlParserCtxtPtr ctxt) { |
12625 | 1.59M | if (ctxt == NULL) |
12626 | 0 | return; |
12627 | 1.59M | ctxt->instate = XML_PARSER_EOF; |
12628 | 1.59M | ctxt->disableSAX = 1; |
12629 | 1.59M | while (ctxt->inputNr > 1) |
12630 | 3.74k | xmlFreeInputStream(inputPop(ctxt)); |
12631 | 1.59M | if (ctxt->input != NULL) { |
12632 | | /* |
12633 | | * in case there was a specific allocation deallocate before |
12634 | | * overriding base |
12635 | | */ |
12636 | 1.59M | if (ctxt->input->free != NULL) { |
12637 | 0 | ctxt->input->free((xmlChar *) ctxt->input->base); |
12638 | 0 | ctxt->input->free = NULL; |
12639 | 0 | } |
12640 | 1.59M | if (ctxt->input->buf != NULL) { |
12641 | 1.45M | xmlFreeParserInputBuffer(ctxt->input->buf); |
12642 | 1.45M | ctxt->input->buf = NULL; |
12643 | 1.45M | } |
12644 | 1.59M | ctxt->input->cur = BAD_CAST""; |
12645 | 1.59M | ctxt->input->length = 0; |
12646 | 1.59M | ctxt->input->base = ctxt->input->cur; |
12647 | 1.59M | ctxt->input->end = ctxt->input->cur; |
12648 | 1.59M | } |
12649 | 1.59M | } |
12650 | | |
12651 | | /** |
12652 | | * xmlStopParser: |
12653 | | * @ctxt: an XML parser context |
12654 | | * |
12655 | | * Blocks further parser processing |
12656 | | */ |
12657 | | void |
12658 | 399k | xmlStopParser(xmlParserCtxtPtr ctxt) { |
12659 | 399k | if (ctxt == NULL) |
12660 | 0 | return; |
12661 | 399k | xmlHaltParser(ctxt); |
12662 | 399k | ctxt->errNo = XML_ERR_USER_STOP; |
12663 | 399k | } |
12664 | | |
12665 | | /** |
12666 | | * xmlCreateIOParserCtxt: |
12667 | | * @sax: a SAX handler |
12668 | | * @user_data: The user data returned on SAX callbacks |
12669 | | * @ioread: an I/O read function |
12670 | | * @ioclose: an I/O close function |
12671 | | * @ioctx: an I/O handler |
12672 | | * @enc: the charset encoding if known |
12673 | | * |
12674 | | * Create a parser context for using the XML parser with an existing |
12675 | | * I/O stream |
12676 | | * |
12677 | | * Returns the new parser context or NULL |
12678 | | */ |
12679 | | xmlParserCtxtPtr |
12680 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12681 | | xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
12682 | 0 | void *ioctx, xmlCharEncoding enc) { |
12683 | 0 | xmlParserCtxtPtr ctxt; |
12684 | 0 | xmlParserInputPtr inputStream; |
12685 | 0 | xmlParserInputBufferPtr buf; |
12686 | |
|
12687 | 0 | if (ioread == NULL) return(NULL); |
12688 | | |
12689 | 0 | buf = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, enc); |
12690 | 0 | if (buf == NULL) { |
12691 | 0 | if (ioclose != NULL) |
12692 | 0 | ioclose(ioctx); |
12693 | 0 | return (NULL); |
12694 | 0 | } |
12695 | | |
12696 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12697 | 0 | if (ctxt == NULL) { |
12698 | 0 | xmlFreeParserInputBuffer(buf); |
12699 | 0 | return(NULL); |
12700 | 0 | } |
12701 | | |
12702 | 0 | inputStream = xmlNewIOInputStream(ctxt, buf, enc); |
12703 | 0 | if (inputStream == NULL) { |
12704 | 0 | xmlFreeParserCtxt(ctxt); |
12705 | 0 | return(NULL); |
12706 | 0 | } |
12707 | 0 | inputPush(ctxt, inputStream); |
12708 | |
|
12709 | 0 | return(ctxt); |
12710 | 0 | } |
12711 | | |
12712 | | #ifdef LIBXML_VALID_ENABLED |
12713 | | /************************************************************************ |
12714 | | * * |
12715 | | * Front ends when parsing a DTD * |
12716 | | * * |
12717 | | ************************************************************************/ |
12718 | | |
12719 | | /** |
12720 | | * xmlIOParseDTD: |
12721 | | * @sax: the SAX handler block or NULL |
12722 | | * @input: an Input Buffer |
12723 | | * @enc: the charset encoding if known |
12724 | | * |
12725 | | * Load and parse a DTD |
12726 | | * |
12727 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12728 | | * @input will be freed by the function in any case. |
12729 | | */ |
12730 | | |
12731 | | xmlDtdPtr |
12732 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
12733 | 0 | xmlCharEncoding enc) { |
12734 | 0 | xmlDtdPtr ret = NULL; |
12735 | 0 | xmlParserCtxtPtr ctxt; |
12736 | 0 | xmlParserInputPtr pinput = NULL; |
12737 | 0 | xmlChar start[4]; |
12738 | |
|
12739 | 0 | if (input == NULL) |
12740 | 0 | return(NULL); |
12741 | | |
12742 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12743 | 0 | if (ctxt == NULL) { |
12744 | 0 | xmlFreeParserInputBuffer(input); |
12745 | 0 | return(NULL); |
12746 | 0 | } |
12747 | | |
12748 | | /* We are loading a DTD */ |
12749 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12750 | |
|
12751 | 0 | xmlDetectSAX2(ctxt); |
12752 | | |
12753 | | /* |
12754 | | * generate a parser input from the I/O handler |
12755 | | */ |
12756 | |
|
12757 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
12758 | 0 | if (pinput == NULL) { |
12759 | 0 | xmlFreeParserInputBuffer(input); |
12760 | 0 | xmlFreeParserCtxt(ctxt); |
12761 | 0 | return(NULL); |
12762 | 0 | } |
12763 | | |
12764 | | /* |
12765 | | * plug some encoding conversion routines here. |
12766 | | */ |
12767 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
12768 | 0 | xmlFreeParserCtxt(ctxt); |
12769 | 0 | return(NULL); |
12770 | 0 | } |
12771 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12772 | 0 | xmlSwitchEncoding(ctxt, enc); |
12773 | 0 | } |
12774 | |
|
12775 | 0 | pinput->filename = NULL; |
12776 | 0 | pinput->line = 1; |
12777 | 0 | pinput->col = 1; |
12778 | 0 | pinput->base = ctxt->input->cur; |
12779 | 0 | pinput->cur = ctxt->input->cur; |
12780 | 0 | pinput->free = NULL; |
12781 | | |
12782 | | /* |
12783 | | * let's parse that entity knowing it's an external subset. |
12784 | | */ |
12785 | 0 | ctxt->inSubset = 2; |
12786 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12787 | 0 | if (ctxt->myDoc == NULL) { |
12788 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12789 | 0 | return(NULL); |
12790 | 0 | } |
12791 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12792 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12793 | 0 | BAD_CAST "none", BAD_CAST "none"); |
12794 | |
|
12795 | 0 | if ((enc == XML_CHAR_ENCODING_NONE) && |
12796 | 0 | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
12797 | | /* |
12798 | | * Get the 4 first bytes and decode the charset |
12799 | | * if enc != XML_CHAR_ENCODING_NONE |
12800 | | * plug some encoding conversion routines. |
12801 | | */ |
12802 | 0 | start[0] = RAW; |
12803 | 0 | start[1] = NXT(1); |
12804 | 0 | start[2] = NXT(2); |
12805 | 0 | start[3] = NXT(3); |
12806 | 0 | enc = xmlDetectCharEncoding(start, 4); |
12807 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12808 | 0 | xmlSwitchEncoding(ctxt, enc); |
12809 | 0 | } |
12810 | 0 | } |
12811 | |
|
12812 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
12813 | |
|
12814 | 0 | if (ctxt->myDoc != NULL) { |
12815 | 0 | if (ctxt->wellFormed) { |
12816 | 0 | ret = ctxt->myDoc->extSubset; |
12817 | 0 | ctxt->myDoc->extSubset = NULL; |
12818 | 0 | if (ret != NULL) { |
12819 | 0 | xmlNodePtr tmp; |
12820 | |
|
12821 | 0 | ret->doc = NULL; |
12822 | 0 | tmp = ret->children; |
12823 | 0 | while (tmp != NULL) { |
12824 | 0 | tmp->doc = NULL; |
12825 | 0 | tmp = tmp->next; |
12826 | 0 | } |
12827 | 0 | } |
12828 | 0 | } else { |
12829 | 0 | ret = NULL; |
12830 | 0 | } |
12831 | 0 | xmlFreeDoc(ctxt->myDoc); |
12832 | 0 | ctxt->myDoc = NULL; |
12833 | 0 | } |
12834 | 0 | xmlFreeParserCtxt(ctxt); |
12835 | |
|
12836 | 0 | return(ret); |
12837 | 0 | } |
12838 | | |
12839 | | /** |
12840 | | * xmlSAXParseDTD: |
12841 | | * @sax: the SAX handler block |
12842 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12843 | | * @SystemID: a NAME* containing the URL to the DTD |
12844 | | * |
12845 | | * DEPRECATED: Don't use. |
12846 | | * |
12847 | | * Load and parse an external subset. |
12848 | | * |
12849 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12850 | | */ |
12851 | | |
12852 | | xmlDtdPtr |
12853 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
12854 | 0 | const xmlChar *SystemID) { |
12855 | 0 | xmlDtdPtr ret = NULL; |
12856 | 0 | xmlParserCtxtPtr ctxt; |
12857 | 0 | xmlParserInputPtr input = NULL; |
12858 | 0 | xmlCharEncoding enc; |
12859 | 0 | xmlChar* systemIdCanonic; |
12860 | |
|
12861 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
12862 | | |
12863 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12864 | 0 | if (ctxt == NULL) { |
12865 | 0 | return(NULL); |
12866 | 0 | } |
12867 | | |
12868 | | /* We are loading a DTD */ |
12869 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12870 | | |
12871 | | /* |
12872 | | * Canonicalise the system ID |
12873 | | */ |
12874 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
12875 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
12876 | 0 | xmlFreeParserCtxt(ctxt); |
12877 | 0 | return(NULL); |
12878 | 0 | } |
12879 | | |
12880 | | /* |
12881 | | * Ask the Entity resolver to load the damn thing |
12882 | | */ |
12883 | | |
12884 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
12885 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
12886 | 0 | systemIdCanonic); |
12887 | 0 | if (input == NULL) { |
12888 | 0 | xmlFreeParserCtxt(ctxt); |
12889 | 0 | if (systemIdCanonic != NULL) |
12890 | 0 | xmlFree(systemIdCanonic); |
12891 | 0 | return(NULL); |
12892 | 0 | } |
12893 | | |
12894 | | /* |
12895 | | * plug some encoding conversion routines here. |
12896 | | */ |
12897 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
12898 | 0 | xmlFreeParserCtxt(ctxt); |
12899 | 0 | if (systemIdCanonic != NULL) |
12900 | 0 | xmlFree(systemIdCanonic); |
12901 | 0 | return(NULL); |
12902 | 0 | } |
12903 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
12904 | 0 | enc = xmlDetectCharEncoding(ctxt->input->cur, 4); |
12905 | 0 | xmlSwitchEncoding(ctxt, enc); |
12906 | 0 | } |
12907 | |
|
12908 | 0 | if (input->filename == NULL) |
12909 | 0 | input->filename = (char *) systemIdCanonic; |
12910 | 0 | else |
12911 | 0 | xmlFree(systemIdCanonic); |
12912 | 0 | input->line = 1; |
12913 | 0 | input->col = 1; |
12914 | 0 | input->base = ctxt->input->cur; |
12915 | 0 | input->cur = ctxt->input->cur; |
12916 | 0 | input->free = NULL; |
12917 | | |
12918 | | /* |
12919 | | * let's parse that entity knowing it's an external subset. |
12920 | | */ |
12921 | 0 | ctxt->inSubset = 2; |
12922 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12923 | 0 | if (ctxt->myDoc == NULL) { |
12924 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12925 | 0 | xmlFreeParserCtxt(ctxt); |
12926 | 0 | return(NULL); |
12927 | 0 | } |
12928 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12929 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12930 | 0 | ExternalID, SystemID); |
12931 | 0 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
12932 | |
|
12933 | 0 | if (ctxt->myDoc != NULL) { |
12934 | 0 | if (ctxt->wellFormed) { |
12935 | 0 | ret = ctxt->myDoc->extSubset; |
12936 | 0 | ctxt->myDoc->extSubset = NULL; |
12937 | 0 | if (ret != NULL) { |
12938 | 0 | xmlNodePtr tmp; |
12939 | |
|
12940 | 0 | ret->doc = NULL; |
12941 | 0 | tmp = ret->children; |
12942 | 0 | while (tmp != NULL) { |
12943 | 0 | tmp->doc = NULL; |
12944 | 0 | tmp = tmp->next; |
12945 | 0 | } |
12946 | 0 | } |
12947 | 0 | } else { |
12948 | 0 | ret = NULL; |
12949 | 0 | } |
12950 | 0 | xmlFreeDoc(ctxt->myDoc); |
12951 | 0 | ctxt->myDoc = NULL; |
12952 | 0 | } |
12953 | 0 | xmlFreeParserCtxt(ctxt); |
12954 | |
|
12955 | 0 | return(ret); |
12956 | 0 | } |
12957 | | |
12958 | | |
12959 | | /** |
12960 | | * xmlParseDTD: |
12961 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12962 | | * @SystemID: a NAME* containing the URL to the DTD |
12963 | | * |
12964 | | * Load and parse an external subset. |
12965 | | * |
12966 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12967 | | */ |
12968 | | |
12969 | | xmlDtdPtr |
12970 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
12971 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
12972 | 0 | } |
12973 | | #endif /* LIBXML_VALID_ENABLED */ |
12974 | | |
12975 | | /************************************************************************ |
12976 | | * * |
12977 | | * Front ends when parsing an Entity * |
12978 | | * * |
12979 | | ************************************************************************/ |
12980 | | |
12981 | | /** |
12982 | | * xmlParseCtxtExternalEntity: |
12983 | | * @ctx: the existing parsing context |
12984 | | * @URL: the URL for the entity to load |
12985 | | * @ID: the System ID for the entity to load |
12986 | | * @lst: the return value for the set of parsed nodes |
12987 | | * |
12988 | | * Parse an external general entity within an existing parsing context |
12989 | | * An external general parsed entity is well-formed if it matches the |
12990 | | * production labeled extParsedEnt. |
12991 | | * |
12992 | | * [78] extParsedEnt ::= TextDecl? content |
12993 | | * |
12994 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12995 | | * the parser error code otherwise |
12996 | | */ |
12997 | | |
12998 | | int |
12999 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx, const xmlChar *URL, |
13000 | 0 | const xmlChar *ID, xmlNodePtr *lst) { |
13001 | 0 | void *userData; |
13002 | |
|
13003 | 0 | if (ctx == NULL) return(-1); |
13004 | | /* |
13005 | | * If the user provided their own SAX callbacks, then reuse the |
13006 | | * userData callback field, otherwise the expected setup in a |
13007 | | * DOM builder is to have userData == ctxt |
13008 | | */ |
13009 | 0 | if (ctx->userData == ctx) |
13010 | 0 | userData = NULL; |
13011 | 0 | else |
13012 | 0 | userData = ctx->userData; |
13013 | 0 | return xmlParseExternalEntityPrivate(ctx->myDoc, ctx, ctx->sax, |
13014 | 0 | userData, ctx->depth + 1, |
13015 | 0 | URL, ID, lst); |
13016 | 0 | } |
13017 | | |
13018 | | /** |
13019 | | * xmlParseExternalEntityPrivate: |
13020 | | * @doc: the document the chunk pertains to |
13021 | | * @oldctxt: the previous parser context if available |
13022 | | * @sax: the SAX handler block (possibly NULL) |
13023 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13024 | | * @depth: Used for loop detection, use 0 |
13025 | | * @URL: the URL for the entity to load |
13026 | | * @ID: the System ID for the entity to load |
13027 | | * @list: the return value for the set of parsed nodes |
13028 | | * |
13029 | | * Private version of xmlParseExternalEntity() |
13030 | | * |
13031 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13032 | | * the parser error code otherwise |
13033 | | */ |
13034 | | |
13035 | | static xmlParserErrors |
13036 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
13037 | | xmlSAXHandlerPtr sax, |
13038 | | void *user_data, int depth, const xmlChar *URL, |
13039 | 4.42M | const xmlChar *ID, xmlNodePtr *list) { |
13040 | 4.42M | xmlParserCtxtPtr ctxt; |
13041 | 4.42M | xmlDocPtr newDoc; |
13042 | 4.42M | xmlNodePtr newRoot; |
13043 | 4.42M | xmlParserErrors ret = XML_ERR_OK; |
13044 | 4.42M | xmlChar start[4]; |
13045 | 4.42M | xmlCharEncoding enc; |
13046 | | |
13047 | 4.42M | if (((depth > 40) && |
13048 | 4.42M | ((oldctxt == NULL) || (oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13049 | 4.42M | (depth > 1024)) { |
13050 | 4.58k | return(XML_ERR_ENTITY_LOOP); |
13051 | 4.58k | } |
13052 | | |
13053 | 4.41M | if (list != NULL) |
13054 | 4.33M | *list = NULL; |
13055 | 4.41M | if ((URL == NULL) && (ID == NULL)) |
13056 | 320 | return(XML_ERR_INTERNAL_ERROR); |
13057 | 4.41M | if (doc == NULL) |
13058 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13059 | | |
13060 | 4.41M | ctxt = xmlCreateEntityParserCtxtInternal(sax, user_data, URL, ID, NULL, |
13061 | 4.41M | oldctxt); |
13062 | 4.41M | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13063 | 4.30M | xmlDetectSAX2(ctxt); |
13064 | | |
13065 | 4.30M | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13066 | 4.30M | if (newDoc == NULL) { |
13067 | 0 | xmlFreeParserCtxt(ctxt); |
13068 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13069 | 0 | } |
13070 | 4.30M | newDoc->properties = XML_DOC_INTERNAL; |
13071 | 4.30M | if (doc) { |
13072 | 4.30M | newDoc->intSubset = doc->intSubset; |
13073 | 4.30M | newDoc->extSubset = doc->extSubset; |
13074 | 4.30M | if (doc->dict) { |
13075 | 3.01M | newDoc->dict = doc->dict; |
13076 | 3.01M | xmlDictReference(newDoc->dict); |
13077 | 3.01M | } |
13078 | 4.30M | if (doc->URL != NULL) { |
13079 | 2.20M | newDoc->URL = xmlStrdup(doc->URL); |
13080 | 2.20M | } |
13081 | 4.30M | } |
13082 | 4.30M | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13083 | 4.30M | if (newRoot == NULL) { |
13084 | 0 | if (sax != NULL) |
13085 | 0 | xmlFreeParserCtxt(ctxt); |
13086 | 0 | newDoc->intSubset = NULL; |
13087 | 0 | newDoc->extSubset = NULL; |
13088 | 0 | xmlFreeDoc(newDoc); |
13089 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13090 | 0 | } |
13091 | 4.30M | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13092 | 4.30M | nodePush(ctxt, newDoc->children); |
13093 | 4.30M | if (doc == NULL) { |
13094 | 0 | ctxt->myDoc = newDoc; |
13095 | 4.30M | } else { |
13096 | 4.30M | ctxt->myDoc = doc; |
13097 | 4.30M | newRoot->doc = doc; |
13098 | 4.30M | } |
13099 | | |
13100 | | /* |
13101 | | * Get the 4 first bytes and decode the charset |
13102 | | * if enc != XML_CHAR_ENCODING_NONE |
13103 | | * plug some encoding conversion routines. |
13104 | | */ |
13105 | 4.30M | GROW; |
13106 | 4.30M | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
13107 | 4.29M | start[0] = RAW; |
13108 | 4.29M | start[1] = NXT(1); |
13109 | 4.29M | start[2] = NXT(2); |
13110 | 4.29M | start[3] = NXT(3); |
13111 | 4.29M | enc = xmlDetectCharEncoding(start, 4); |
13112 | 4.29M | if (enc != XML_CHAR_ENCODING_NONE) { |
13113 | 20.2k | xmlSwitchEncoding(ctxt, enc); |
13114 | 20.2k | } |
13115 | 4.29M | } |
13116 | | |
13117 | | /* |
13118 | | * Parse a possible text declaration first |
13119 | | */ |
13120 | 4.30M | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
13121 | 18.2k | xmlParseTextDecl(ctxt); |
13122 | | /* |
13123 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
13124 | | */ |
13125 | 18.2k | if ((xmlStrEqual(oldctxt->version, BAD_CAST "1.0")) && |
13126 | 18.2k | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
13127 | 289 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
13128 | 289 | "Version mismatch between document and entity\n"); |
13129 | 289 | } |
13130 | 18.2k | } |
13131 | | |
13132 | 4.30M | ctxt->instate = XML_PARSER_CONTENT; |
13133 | 4.30M | ctxt->depth = depth; |
13134 | 4.30M | if (oldctxt != NULL) { |
13135 | 4.30M | ctxt->_private = oldctxt->_private; |
13136 | 4.30M | ctxt->loadsubset = oldctxt->loadsubset; |
13137 | 4.30M | ctxt->validate = oldctxt->validate; |
13138 | 4.30M | ctxt->valid = oldctxt->valid; |
13139 | 4.30M | ctxt->replaceEntities = oldctxt->replaceEntities; |
13140 | 4.30M | if (oldctxt->validate) { |
13141 | 3.90M | ctxt->vctxt.error = oldctxt->vctxt.error; |
13142 | 3.90M | ctxt->vctxt.warning = oldctxt->vctxt.warning; |
13143 | 3.90M | ctxt->vctxt.userData = oldctxt->vctxt.userData; |
13144 | 3.90M | } |
13145 | 4.30M | ctxt->external = oldctxt->external; |
13146 | 4.30M | if (ctxt->dict) xmlDictFree(ctxt->dict); |
13147 | 4.30M | ctxt->dict = oldctxt->dict; |
13148 | 4.30M | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13149 | 4.30M | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13150 | 4.30M | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13151 | 4.30M | ctxt->dictNames = oldctxt->dictNames; |
13152 | 4.30M | ctxt->attsDefault = oldctxt->attsDefault; |
13153 | 4.30M | ctxt->attsSpecial = oldctxt->attsSpecial; |
13154 | 4.30M | ctxt->linenumbers = oldctxt->linenumbers; |
13155 | 4.30M | ctxt->record_info = oldctxt->record_info; |
13156 | 4.30M | ctxt->node_seq.maximum = oldctxt->node_seq.maximum; |
13157 | 4.30M | ctxt->node_seq.length = oldctxt->node_seq.length; |
13158 | 4.30M | ctxt->node_seq.buffer = oldctxt->node_seq.buffer; |
13159 | 4.30M | } else { |
13160 | | /* |
13161 | | * Doing validity checking on chunk without context |
13162 | | * doesn't make sense |
13163 | | */ |
13164 | 0 | ctxt->_private = NULL; |
13165 | 0 | ctxt->validate = 0; |
13166 | 0 | ctxt->external = 2; |
13167 | 0 | ctxt->loadsubset = 0; |
13168 | 0 | } |
13169 | | |
13170 | 4.30M | xmlParseContent(ctxt); |
13171 | | |
13172 | 4.30M | if ((RAW == '<') && (NXT(1) == '/')) { |
13173 | 571k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13174 | 3.72M | } else if (RAW != 0) { |
13175 | 6.92k | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13176 | 6.92k | } |
13177 | 4.30M | if (ctxt->node != newDoc->children) { |
13178 | 3.27M | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13179 | 3.27M | } |
13180 | | |
13181 | 4.30M | if (!ctxt->wellFormed) { |
13182 | 4.28M | if (ctxt->errNo == 0) |
13183 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13184 | 4.28M | else |
13185 | 4.28M | ret = (xmlParserErrors)ctxt->errNo; |
13186 | 4.28M | } else { |
13187 | 17.9k | if (list != NULL) { |
13188 | 6.44k | xmlNodePtr cur; |
13189 | | |
13190 | | /* |
13191 | | * Return the newly created nodeset after unlinking it from |
13192 | | * they pseudo parent. |
13193 | | */ |
13194 | 6.44k | cur = newDoc->children->children; |
13195 | 6.44k | *list = cur; |
13196 | 12.7k | while (cur != NULL) { |
13197 | 6.32k | cur->parent = NULL; |
13198 | 6.32k | cur = cur->next; |
13199 | 6.32k | } |
13200 | 6.44k | newDoc->children->children = NULL; |
13201 | 6.44k | } |
13202 | 17.9k | ret = XML_ERR_OK; |
13203 | 17.9k | } |
13204 | | |
13205 | | /* |
13206 | | * Record in the parent context the number of entities replacement |
13207 | | * done when parsing that reference. |
13208 | | */ |
13209 | 4.30M | if (oldctxt != NULL) |
13210 | 4.29M | oldctxt->nbentities += ctxt->nbentities; |
13211 | | |
13212 | | /* |
13213 | | * Also record the size of the entity parsed |
13214 | | */ |
13215 | 4.30M | if (ctxt->input != NULL && oldctxt != NULL) { |
13216 | 4.29M | oldctxt->sizeentities += ctxt->input->consumed; |
13217 | 4.29M | oldctxt->sizeentities += (ctxt->input->cur - ctxt->input->base); |
13218 | 4.29M | } |
13219 | | /* |
13220 | | * And record the last error if any |
13221 | | */ |
13222 | 4.30M | if ((oldctxt != NULL) && (ctxt->lastError.code != XML_ERR_OK)) |
13223 | 4.28M | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13224 | | |
13225 | 4.30M | if (oldctxt != NULL) { |
13226 | 4.29M | ctxt->dict = NULL; |
13227 | 4.29M | ctxt->attsDefault = NULL; |
13228 | 4.29M | ctxt->attsSpecial = NULL; |
13229 | 4.29M | oldctxt->validate = ctxt->validate; |
13230 | 4.29M | oldctxt->valid = ctxt->valid; |
13231 | 4.29M | oldctxt->node_seq.maximum = ctxt->node_seq.maximum; |
13232 | 4.29M | oldctxt->node_seq.length = ctxt->node_seq.length; |
13233 | 4.29M | oldctxt->node_seq.buffer = ctxt->node_seq.buffer; |
13234 | 4.29M | } |
13235 | 4.30M | ctxt->node_seq.maximum = 0; |
13236 | 4.30M | ctxt->node_seq.length = 0; |
13237 | 4.30M | ctxt->node_seq.buffer = NULL; |
13238 | 4.30M | xmlFreeParserCtxt(ctxt); |
13239 | 4.30M | newDoc->intSubset = NULL; |
13240 | 4.30M | newDoc->extSubset = NULL; |
13241 | 4.30M | xmlFreeDoc(newDoc); |
13242 | | |
13243 | 4.30M | return(ret); |
13244 | 4.30M | } |
13245 | | |
13246 | | #ifdef LIBXML_SAX1_ENABLED |
13247 | | /** |
13248 | | * xmlParseExternalEntity: |
13249 | | * @doc: the document the chunk pertains to |
13250 | | * @sax: the SAX handler block (possibly NULL) |
13251 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13252 | | * @depth: Used for loop detection, use 0 |
13253 | | * @URL: the URL for the entity to load |
13254 | | * @ID: the System ID for the entity to load |
13255 | | * @lst: the return value for the set of parsed nodes |
13256 | | * |
13257 | | * Parse an external general entity |
13258 | | * An external general parsed entity is well-formed if it matches the |
13259 | | * production labeled extParsedEnt. |
13260 | | * |
13261 | | * [78] extParsedEnt ::= TextDecl? content |
13262 | | * |
13263 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13264 | | * the parser error code otherwise |
13265 | | */ |
13266 | | |
13267 | | int |
13268 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
13269 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *lst) { |
13270 | 0 | return(xmlParseExternalEntityPrivate(doc, NULL, sax, user_data, depth, URL, |
13271 | 0 | ID, lst)); |
13272 | 0 | } |
13273 | | |
13274 | | /** |
13275 | | * xmlParseBalancedChunkMemory: |
13276 | | * @doc: the document the chunk pertains to (must not be NULL) |
13277 | | * @sax: the SAX handler block (possibly NULL) |
13278 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13279 | | * @depth: Used for loop detection, use 0 |
13280 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13281 | | * @lst: the return value for the set of parsed nodes |
13282 | | * |
13283 | | * Parse a well-balanced chunk of an XML document |
13284 | | * called by the parser |
13285 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13286 | | * the content production in the XML grammar: |
13287 | | * |
13288 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13289 | | * |
13290 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13291 | | * the parser error code otherwise |
13292 | | */ |
13293 | | |
13294 | | int |
13295 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13296 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
13297 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
13298 | 0 | depth, string, lst, 0 ); |
13299 | 0 | } |
13300 | | #endif /* LIBXML_SAX1_ENABLED */ |
13301 | | |
13302 | | /** |
13303 | | * xmlParseBalancedChunkMemoryInternal: |
13304 | | * @oldctxt: the existing parsing context |
13305 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13306 | | * @user_data: the user data field for the parser context |
13307 | | * @lst: the return value for the set of parsed nodes |
13308 | | * |
13309 | | * |
13310 | | * Parse a well-balanced chunk of an XML document |
13311 | | * called by the parser |
13312 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13313 | | * the content production in the XML grammar: |
13314 | | * |
13315 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13316 | | * |
13317 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13318 | | * error code otherwise |
13319 | | * |
13320 | | * In case recover is set to 1, the nodelist will not be empty even if |
13321 | | * the parsed chunk is not well balanced. |
13322 | | */ |
13323 | | static xmlParserErrors |
13324 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
13325 | 293k | const xmlChar *string, void *user_data, xmlNodePtr *lst) { |
13326 | 293k | xmlParserCtxtPtr ctxt; |
13327 | 293k | xmlDocPtr newDoc = NULL; |
13328 | 293k | xmlNodePtr newRoot; |
13329 | 293k | xmlSAXHandlerPtr oldsax = NULL; |
13330 | 293k | xmlNodePtr content = NULL; |
13331 | 293k | xmlNodePtr last = NULL; |
13332 | 293k | int size; |
13333 | 293k | xmlParserErrors ret = XML_ERR_OK; |
13334 | 293k | #ifdef SAX2 |
13335 | 293k | int i; |
13336 | 293k | #endif |
13337 | | |
13338 | 293k | if (((oldctxt->depth > 40) && ((oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13339 | 293k | (oldctxt->depth > 1024)) { |
13340 | 920 | return(XML_ERR_ENTITY_LOOP); |
13341 | 920 | } |
13342 | | |
13343 | | |
13344 | 292k | if (lst != NULL) |
13345 | 261k | *lst = NULL; |
13346 | 292k | if (string == NULL) |
13347 | 115 | return(XML_ERR_INTERNAL_ERROR); |
13348 | | |
13349 | 292k | size = xmlStrlen(string); |
13350 | | |
13351 | 292k | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13352 | 292k | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13353 | 245k | if (user_data != NULL) |
13354 | 0 | ctxt->userData = user_data; |
13355 | 245k | else |
13356 | 245k | ctxt->userData = ctxt; |
13357 | 245k | if (ctxt->dict != NULL) xmlDictFree(ctxt->dict); |
13358 | 245k | ctxt->dict = oldctxt->dict; |
13359 | 245k | ctxt->input_id = oldctxt->input_id + 1; |
13360 | 245k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13361 | 245k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13362 | 245k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13363 | | |
13364 | 245k | #ifdef SAX2 |
13365 | | /* propagate namespaces down the entity */ |
13366 | 1.03M | for (i = 0;i < oldctxt->nsNr;i += 2) { |
13367 | 790k | nsPush(ctxt, oldctxt->nsTab[i], oldctxt->nsTab[i+1]); |
13368 | 790k | } |
13369 | 245k | #endif |
13370 | | |
13371 | 245k | oldsax = ctxt->sax; |
13372 | 245k | ctxt->sax = oldctxt->sax; |
13373 | 245k | xmlDetectSAX2(ctxt); |
13374 | 245k | ctxt->replaceEntities = oldctxt->replaceEntities; |
13375 | 245k | ctxt->options = oldctxt->options; |
13376 | | |
13377 | 245k | ctxt->_private = oldctxt->_private; |
13378 | 245k | if (oldctxt->myDoc == NULL) { |
13379 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13380 | 0 | if (newDoc == NULL) { |
13381 | 0 | ctxt->sax = oldsax; |
13382 | 0 | ctxt->dict = NULL; |
13383 | 0 | xmlFreeParserCtxt(ctxt); |
13384 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13385 | 0 | } |
13386 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13387 | 0 | newDoc->dict = ctxt->dict; |
13388 | 0 | xmlDictReference(newDoc->dict); |
13389 | 0 | ctxt->myDoc = newDoc; |
13390 | 245k | } else { |
13391 | 245k | ctxt->myDoc = oldctxt->myDoc; |
13392 | 245k | content = ctxt->myDoc->children; |
13393 | 245k | last = ctxt->myDoc->last; |
13394 | 245k | } |
13395 | 245k | newRoot = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13396 | 245k | if (newRoot == NULL) { |
13397 | 0 | ctxt->sax = oldsax; |
13398 | 0 | ctxt->dict = NULL; |
13399 | 0 | xmlFreeParserCtxt(ctxt); |
13400 | 0 | if (newDoc != NULL) { |
13401 | 0 | xmlFreeDoc(newDoc); |
13402 | 0 | } |
13403 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13404 | 0 | } |
13405 | 245k | ctxt->myDoc->children = NULL; |
13406 | 245k | ctxt->myDoc->last = NULL; |
13407 | 245k | xmlAddChild((xmlNodePtr) ctxt->myDoc, newRoot); |
13408 | 245k | nodePush(ctxt, ctxt->myDoc->children); |
13409 | 245k | ctxt->instate = XML_PARSER_CONTENT; |
13410 | 245k | ctxt->depth = oldctxt->depth + 1; |
13411 | | |
13412 | 245k | ctxt->validate = 0; |
13413 | 245k | ctxt->loadsubset = oldctxt->loadsubset; |
13414 | 245k | if ((oldctxt->validate) || (oldctxt->replaceEntities != 0)) { |
13415 | | /* |
13416 | | * ID/IDREF registration will be done in xmlValidateElement below |
13417 | | */ |
13418 | 104k | ctxt->loadsubset |= XML_SKIP_IDS; |
13419 | 104k | } |
13420 | 245k | ctxt->dictNames = oldctxt->dictNames; |
13421 | 245k | ctxt->attsDefault = oldctxt->attsDefault; |
13422 | 245k | ctxt->attsSpecial = oldctxt->attsSpecial; |
13423 | | |
13424 | 245k | xmlParseContent(ctxt); |
13425 | 245k | if ((RAW == '<') && (NXT(1) == '/')) { |
13426 | 5.74k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13427 | 239k | } else if (RAW != 0) { |
13428 | 237 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13429 | 237 | } |
13430 | 245k | if (ctxt->node != ctxt->myDoc->children) { |
13431 | 85.9k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13432 | 85.9k | } |
13433 | | |
13434 | 245k | if (!ctxt->wellFormed) { |
13435 | 199k | if (ctxt->errNo == 0) |
13436 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13437 | 199k | else |
13438 | 199k | ret = (xmlParserErrors)ctxt->errNo; |
13439 | 199k | } else { |
13440 | 46.1k | ret = XML_ERR_OK; |
13441 | 46.1k | } |
13442 | | |
13443 | 245k | if ((lst != NULL) && (ret == XML_ERR_OK)) { |
13444 | 39.1k | xmlNodePtr cur; |
13445 | | |
13446 | | /* |
13447 | | * Return the newly created nodeset after unlinking it from |
13448 | | * they pseudo parent. |
13449 | | */ |
13450 | 39.1k | cur = ctxt->myDoc->children->children; |
13451 | 39.1k | *lst = cur; |
13452 | 95.6k | while (cur != NULL) { |
13453 | 56.4k | #ifdef LIBXML_VALID_ENABLED |
13454 | 56.4k | if ((oldctxt->validate) && (oldctxt->wellFormed) && |
13455 | 56.4k | (oldctxt->myDoc) && (oldctxt->myDoc->intSubset) && |
13456 | 56.4k | (cur->type == XML_ELEMENT_NODE)) { |
13457 | 5.84k | oldctxt->valid &= xmlValidateElement(&oldctxt->vctxt, |
13458 | 5.84k | oldctxt->myDoc, cur); |
13459 | 5.84k | } |
13460 | 56.4k | #endif /* LIBXML_VALID_ENABLED */ |
13461 | 56.4k | cur->parent = NULL; |
13462 | 56.4k | cur = cur->next; |
13463 | 56.4k | } |
13464 | 39.1k | ctxt->myDoc->children->children = NULL; |
13465 | 39.1k | } |
13466 | 245k | if (ctxt->myDoc != NULL) { |
13467 | 245k | xmlFreeNode(ctxt->myDoc->children); |
13468 | 245k | ctxt->myDoc->children = content; |
13469 | 245k | ctxt->myDoc->last = last; |
13470 | 245k | } |
13471 | | |
13472 | | /* |
13473 | | * Record in the parent context the number of entities replacement |
13474 | | * done when parsing that reference. |
13475 | | */ |
13476 | 245k | if (oldctxt != NULL) |
13477 | 245k | oldctxt->nbentities += ctxt->nbentities; |
13478 | | |
13479 | | /* |
13480 | | * Also record the last error if any |
13481 | | */ |
13482 | 245k | if (ctxt->lastError.code != XML_ERR_OK) |
13483 | 200k | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13484 | | |
13485 | 245k | ctxt->sax = oldsax; |
13486 | 245k | ctxt->dict = NULL; |
13487 | 245k | ctxt->attsDefault = NULL; |
13488 | 245k | ctxt->attsSpecial = NULL; |
13489 | 245k | xmlFreeParserCtxt(ctxt); |
13490 | 245k | if (newDoc != NULL) { |
13491 | 0 | xmlFreeDoc(newDoc); |
13492 | 0 | } |
13493 | | |
13494 | 245k | return(ret); |
13495 | 245k | } |
13496 | | |
13497 | | /** |
13498 | | * xmlParseInNodeContext: |
13499 | | * @node: the context node |
13500 | | * @data: the input string |
13501 | | * @datalen: the input string length in bytes |
13502 | | * @options: a combination of xmlParserOption |
13503 | | * @lst: the return value for the set of parsed nodes |
13504 | | * |
13505 | | * Parse a well-balanced chunk of an XML document |
13506 | | * within the context (DTD, namespaces, etc ...) of the given node. |
13507 | | * |
13508 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
13509 | | * the content production in the XML grammar: |
13510 | | * |
13511 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13512 | | * |
13513 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13514 | | * error code otherwise |
13515 | | */ |
13516 | | xmlParserErrors |
13517 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
13518 | 0 | int options, xmlNodePtr *lst) { |
13519 | 0 | #ifdef SAX2 |
13520 | 0 | xmlParserCtxtPtr ctxt; |
13521 | 0 | xmlDocPtr doc = NULL; |
13522 | 0 | xmlNodePtr fake, cur; |
13523 | 0 | int nsnr = 0; |
13524 | |
|
13525 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13526 | | |
13527 | | /* |
13528 | | * check all input parameters, grab the document |
13529 | | */ |
13530 | 0 | if ((lst == NULL) || (node == NULL) || (data == NULL) || (datalen < 0)) |
13531 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13532 | 0 | switch (node->type) { |
13533 | 0 | case XML_ELEMENT_NODE: |
13534 | 0 | case XML_ATTRIBUTE_NODE: |
13535 | 0 | case XML_TEXT_NODE: |
13536 | 0 | case XML_CDATA_SECTION_NODE: |
13537 | 0 | case XML_ENTITY_REF_NODE: |
13538 | 0 | case XML_PI_NODE: |
13539 | 0 | case XML_COMMENT_NODE: |
13540 | 0 | case XML_DOCUMENT_NODE: |
13541 | 0 | case XML_HTML_DOCUMENT_NODE: |
13542 | 0 | break; |
13543 | 0 | default: |
13544 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13545 | |
|
13546 | 0 | } |
13547 | 0 | while ((node != NULL) && (node->type != XML_ELEMENT_NODE) && |
13548 | 0 | (node->type != XML_DOCUMENT_NODE) && |
13549 | 0 | (node->type != XML_HTML_DOCUMENT_NODE)) |
13550 | 0 | node = node->parent; |
13551 | 0 | if (node == NULL) |
13552 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13553 | 0 | if (node->type == XML_ELEMENT_NODE) |
13554 | 0 | doc = node->doc; |
13555 | 0 | else |
13556 | 0 | doc = (xmlDocPtr) node; |
13557 | 0 | if (doc == NULL) |
13558 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13559 | | |
13560 | | /* |
13561 | | * allocate a context and set-up everything not related to the |
13562 | | * node position in the tree |
13563 | | */ |
13564 | 0 | if (doc->type == XML_DOCUMENT_NODE) |
13565 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) data, datalen); |
13566 | 0 | #ifdef LIBXML_HTML_ENABLED |
13567 | 0 | else if (doc->type == XML_HTML_DOCUMENT_NODE) { |
13568 | 0 | ctxt = htmlCreateMemoryParserCtxt((char *) data, datalen); |
13569 | | /* |
13570 | | * When parsing in context, it makes no sense to add implied |
13571 | | * elements like html/body/etc... |
13572 | | */ |
13573 | 0 | options |= HTML_PARSE_NOIMPLIED; |
13574 | 0 | } |
13575 | 0 | #endif |
13576 | 0 | else |
13577 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13578 | | |
13579 | 0 | if (ctxt == NULL) |
13580 | 0 | return(XML_ERR_NO_MEMORY); |
13581 | | |
13582 | | /* |
13583 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
13584 | | * We need a dictionary for xmlDetectSAX2, so if there's no doc dict |
13585 | | * we must wait until the last moment to free the original one. |
13586 | | */ |
13587 | 0 | if (doc->dict != NULL) { |
13588 | 0 | if (ctxt->dict != NULL) |
13589 | 0 | xmlDictFree(ctxt->dict); |
13590 | 0 | ctxt->dict = doc->dict; |
13591 | 0 | } else |
13592 | 0 | options |= XML_PARSE_NODICT; |
13593 | |
|
13594 | 0 | if (doc->encoding != NULL) { |
13595 | 0 | xmlCharEncodingHandlerPtr hdlr; |
13596 | |
|
13597 | 0 | if (ctxt->encoding != NULL) |
13598 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
13599 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) doc->encoding); |
13600 | |
|
13601 | 0 | hdlr = xmlFindCharEncodingHandler((const char *) doc->encoding); |
13602 | 0 | if (hdlr != NULL) { |
13603 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
13604 | 0 | } else { |
13605 | 0 | return(XML_ERR_UNSUPPORTED_ENCODING); |
13606 | 0 | } |
13607 | 0 | } |
13608 | | |
13609 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
13610 | 0 | xmlDetectSAX2(ctxt); |
13611 | 0 | ctxt->myDoc = doc; |
13612 | | /* parsing in context, i.e. as within existing content */ |
13613 | 0 | ctxt->input_id = 2; |
13614 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13615 | |
|
13616 | 0 | fake = xmlNewDocComment(node->doc, NULL); |
13617 | 0 | if (fake == NULL) { |
13618 | 0 | xmlFreeParserCtxt(ctxt); |
13619 | 0 | return(XML_ERR_NO_MEMORY); |
13620 | 0 | } |
13621 | 0 | xmlAddChild(node, fake); |
13622 | |
|
13623 | 0 | if (node->type == XML_ELEMENT_NODE) { |
13624 | 0 | nodePush(ctxt, node); |
13625 | | /* |
13626 | | * initialize the SAX2 namespaces stack |
13627 | | */ |
13628 | 0 | cur = node; |
13629 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
13630 | 0 | xmlNsPtr ns = cur->nsDef; |
13631 | 0 | const xmlChar *iprefix, *ihref; |
13632 | |
|
13633 | 0 | while (ns != NULL) { |
13634 | 0 | if (ctxt->dict) { |
13635 | 0 | iprefix = xmlDictLookup(ctxt->dict, ns->prefix, -1); |
13636 | 0 | ihref = xmlDictLookup(ctxt->dict, ns->href, -1); |
13637 | 0 | } else { |
13638 | 0 | iprefix = ns->prefix; |
13639 | 0 | ihref = ns->href; |
13640 | 0 | } |
13641 | |
|
13642 | 0 | if (xmlGetNamespace(ctxt, iprefix) == NULL) { |
13643 | 0 | nsPush(ctxt, iprefix, ihref); |
13644 | 0 | nsnr++; |
13645 | 0 | } |
13646 | 0 | ns = ns->next; |
13647 | 0 | } |
13648 | 0 | cur = cur->parent; |
13649 | 0 | } |
13650 | 0 | } |
13651 | |
|
13652 | 0 | if ((ctxt->validate) || (ctxt->replaceEntities != 0)) { |
13653 | | /* |
13654 | | * ID/IDREF registration will be done in xmlValidateElement below |
13655 | | */ |
13656 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13657 | 0 | } |
13658 | |
|
13659 | 0 | #ifdef LIBXML_HTML_ENABLED |
13660 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) |
13661 | 0 | __htmlParseContent(ctxt); |
13662 | 0 | else |
13663 | 0 | #endif |
13664 | 0 | xmlParseContent(ctxt); |
13665 | |
|
13666 | 0 | nsPop(ctxt, nsnr); |
13667 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13668 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13669 | 0 | } else if (RAW != 0) { |
13670 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13671 | 0 | } |
13672 | 0 | if ((ctxt->node != NULL) && (ctxt->node != node)) { |
13673 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13674 | 0 | ctxt->wellFormed = 0; |
13675 | 0 | } |
13676 | |
|
13677 | 0 | if (!ctxt->wellFormed) { |
13678 | 0 | if (ctxt->errNo == 0) |
13679 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13680 | 0 | else |
13681 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13682 | 0 | } else { |
13683 | 0 | ret = XML_ERR_OK; |
13684 | 0 | } |
13685 | | |
13686 | | /* |
13687 | | * Return the newly created nodeset after unlinking it from |
13688 | | * the pseudo sibling. |
13689 | | */ |
13690 | |
|
13691 | 0 | cur = fake->next; |
13692 | 0 | fake->next = NULL; |
13693 | 0 | node->last = fake; |
13694 | |
|
13695 | 0 | if (cur != NULL) { |
13696 | 0 | cur->prev = NULL; |
13697 | 0 | } |
13698 | |
|
13699 | 0 | *lst = cur; |
13700 | |
|
13701 | 0 | while (cur != NULL) { |
13702 | 0 | cur->parent = NULL; |
13703 | 0 | cur = cur->next; |
13704 | 0 | } |
13705 | |
|
13706 | 0 | xmlUnlinkNode(fake); |
13707 | 0 | xmlFreeNode(fake); |
13708 | | |
13709 | |
|
13710 | 0 | if (ret != XML_ERR_OK) { |
13711 | 0 | xmlFreeNodeList(*lst); |
13712 | 0 | *lst = NULL; |
13713 | 0 | } |
13714 | |
|
13715 | 0 | if (doc->dict != NULL) |
13716 | 0 | ctxt->dict = NULL; |
13717 | 0 | xmlFreeParserCtxt(ctxt); |
13718 | |
|
13719 | 0 | return(ret); |
13720 | | #else /* !SAX2 */ |
13721 | | return(XML_ERR_INTERNAL_ERROR); |
13722 | | #endif |
13723 | 0 | } |
13724 | | |
13725 | | #ifdef LIBXML_SAX1_ENABLED |
13726 | | /** |
13727 | | * xmlParseBalancedChunkMemoryRecover: |
13728 | | * @doc: the document the chunk pertains to (must not be NULL) |
13729 | | * @sax: the SAX handler block (possibly NULL) |
13730 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13731 | | * @depth: Used for loop detection, use 0 |
13732 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13733 | | * @lst: the return value for the set of parsed nodes |
13734 | | * @recover: return nodes even if the data is broken (use 0) |
13735 | | * |
13736 | | * |
13737 | | * Parse a well-balanced chunk of an XML document |
13738 | | * called by the parser |
13739 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13740 | | * the content production in the XML grammar: |
13741 | | * |
13742 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13743 | | * |
13744 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13745 | | * the parser error code otherwise |
13746 | | * |
13747 | | * In case recover is set to 1, the nodelist will not be empty even if |
13748 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
13749 | | * some extent. |
13750 | | */ |
13751 | | int |
13752 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13753 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst, |
13754 | 0 | int recover) { |
13755 | 0 | xmlParserCtxtPtr ctxt; |
13756 | 0 | xmlDocPtr newDoc; |
13757 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13758 | 0 | xmlNodePtr content, newRoot; |
13759 | 0 | int size; |
13760 | 0 | int ret = 0; |
13761 | |
|
13762 | 0 | if (depth > 40) { |
13763 | 0 | return(XML_ERR_ENTITY_LOOP); |
13764 | 0 | } |
13765 | | |
13766 | | |
13767 | 0 | if (lst != NULL) |
13768 | 0 | *lst = NULL; |
13769 | 0 | if (string == NULL) |
13770 | 0 | return(-1); |
13771 | | |
13772 | 0 | size = xmlStrlen(string); |
13773 | |
|
13774 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13775 | 0 | if (ctxt == NULL) return(-1); |
13776 | 0 | ctxt->userData = ctxt; |
13777 | 0 | if (sax != NULL) { |
13778 | 0 | oldsax = ctxt->sax; |
13779 | 0 | ctxt->sax = sax; |
13780 | 0 | if (user_data != NULL) |
13781 | 0 | ctxt->userData = user_data; |
13782 | 0 | } |
13783 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13784 | 0 | if (newDoc == NULL) { |
13785 | 0 | xmlFreeParserCtxt(ctxt); |
13786 | 0 | return(-1); |
13787 | 0 | } |
13788 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13789 | 0 | if ((doc != NULL) && (doc->dict != NULL)) { |
13790 | 0 | xmlDictFree(ctxt->dict); |
13791 | 0 | ctxt->dict = doc->dict; |
13792 | 0 | xmlDictReference(ctxt->dict); |
13793 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13794 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13795 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13796 | 0 | ctxt->dictNames = 1; |
13797 | 0 | } else { |
13798 | 0 | xmlCtxtUseOptionsInternal(ctxt, XML_PARSE_NODICT, NULL); |
13799 | 0 | } |
13800 | | /* doc == NULL is only supported for historic reasons */ |
13801 | 0 | if (doc != NULL) { |
13802 | 0 | newDoc->intSubset = doc->intSubset; |
13803 | 0 | newDoc->extSubset = doc->extSubset; |
13804 | 0 | } |
13805 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13806 | 0 | if (newRoot == NULL) { |
13807 | 0 | if (sax != NULL) |
13808 | 0 | ctxt->sax = oldsax; |
13809 | 0 | xmlFreeParserCtxt(ctxt); |
13810 | 0 | newDoc->intSubset = NULL; |
13811 | 0 | newDoc->extSubset = NULL; |
13812 | 0 | xmlFreeDoc(newDoc); |
13813 | 0 | return(-1); |
13814 | 0 | } |
13815 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13816 | 0 | nodePush(ctxt, newRoot); |
13817 | | /* doc == NULL is only supported for historic reasons */ |
13818 | 0 | if (doc == NULL) { |
13819 | 0 | ctxt->myDoc = newDoc; |
13820 | 0 | } else { |
13821 | 0 | ctxt->myDoc = newDoc; |
13822 | 0 | newDoc->children->doc = doc; |
13823 | | /* Ensure that doc has XML spec namespace */ |
13824 | 0 | xmlSearchNsByHref(doc, (xmlNodePtr)doc, XML_XML_NAMESPACE); |
13825 | 0 | newDoc->oldNs = doc->oldNs; |
13826 | 0 | } |
13827 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13828 | 0 | ctxt->input_id = 2; |
13829 | 0 | ctxt->depth = depth; |
13830 | | |
13831 | | /* |
13832 | | * Doing validity checking on chunk doesn't make sense |
13833 | | */ |
13834 | 0 | ctxt->validate = 0; |
13835 | 0 | ctxt->loadsubset = 0; |
13836 | 0 | xmlDetectSAX2(ctxt); |
13837 | |
|
13838 | 0 | if ( doc != NULL ){ |
13839 | 0 | content = doc->children; |
13840 | 0 | doc->children = NULL; |
13841 | 0 | xmlParseContent(ctxt); |
13842 | 0 | doc->children = content; |
13843 | 0 | } |
13844 | 0 | else { |
13845 | 0 | xmlParseContent(ctxt); |
13846 | 0 | } |
13847 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13848 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13849 | 0 | } else if (RAW != 0) { |
13850 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13851 | 0 | } |
13852 | 0 | if (ctxt->node != newDoc->children) { |
13853 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13854 | 0 | } |
13855 | |
|
13856 | 0 | if (!ctxt->wellFormed) { |
13857 | 0 | if (ctxt->errNo == 0) |
13858 | 0 | ret = 1; |
13859 | 0 | else |
13860 | 0 | ret = ctxt->errNo; |
13861 | 0 | } else { |
13862 | 0 | ret = 0; |
13863 | 0 | } |
13864 | |
|
13865 | 0 | if ((lst != NULL) && ((ret == 0) || (recover == 1))) { |
13866 | 0 | xmlNodePtr cur; |
13867 | | |
13868 | | /* |
13869 | | * Return the newly created nodeset after unlinking it from |
13870 | | * they pseudo parent. |
13871 | | */ |
13872 | 0 | cur = newDoc->children->children; |
13873 | 0 | *lst = cur; |
13874 | 0 | while (cur != NULL) { |
13875 | 0 | xmlSetTreeDoc(cur, doc); |
13876 | 0 | cur->parent = NULL; |
13877 | 0 | cur = cur->next; |
13878 | 0 | } |
13879 | 0 | newDoc->children->children = NULL; |
13880 | 0 | } |
13881 | |
|
13882 | 0 | if (sax != NULL) |
13883 | 0 | ctxt->sax = oldsax; |
13884 | 0 | xmlFreeParserCtxt(ctxt); |
13885 | 0 | newDoc->intSubset = NULL; |
13886 | 0 | newDoc->extSubset = NULL; |
13887 | | /* This leaks the namespace list if doc == NULL */ |
13888 | 0 | newDoc->oldNs = NULL; |
13889 | 0 | xmlFreeDoc(newDoc); |
13890 | |
|
13891 | 0 | return(ret); |
13892 | 0 | } |
13893 | | |
13894 | | /** |
13895 | | * xmlSAXParseEntity: |
13896 | | * @sax: the SAX handler block |
13897 | | * @filename: the filename |
13898 | | * |
13899 | | * DEPRECATED: Don't use. |
13900 | | * |
13901 | | * parse an XML external entity out of context and build a tree. |
13902 | | * It use the given SAX function block to handle the parsing callback. |
13903 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13904 | | * |
13905 | | * [78] extParsedEnt ::= TextDecl? content |
13906 | | * |
13907 | | * This correspond to a "Well Balanced" chunk |
13908 | | * |
13909 | | * Returns the resulting document tree |
13910 | | */ |
13911 | | |
13912 | | xmlDocPtr |
13913 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
13914 | 0 | xmlDocPtr ret; |
13915 | 0 | xmlParserCtxtPtr ctxt; |
13916 | |
|
13917 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13918 | 0 | if (ctxt == NULL) { |
13919 | 0 | return(NULL); |
13920 | 0 | } |
13921 | 0 | if (sax != NULL) { |
13922 | 0 | if (ctxt->sax != NULL) |
13923 | 0 | xmlFree(ctxt->sax); |
13924 | 0 | ctxt->sax = sax; |
13925 | 0 | ctxt->userData = NULL; |
13926 | 0 | } |
13927 | |
|
13928 | 0 | xmlParseExtParsedEnt(ctxt); |
13929 | |
|
13930 | 0 | if (ctxt->wellFormed) |
13931 | 0 | ret = ctxt->myDoc; |
13932 | 0 | else { |
13933 | 0 | ret = NULL; |
13934 | 0 | xmlFreeDoc(ctxt->myDoc); |
13935 | 0 | ctxt->myDoc = NULL; |
13936 | 0 | } |
13937 | 0 | if (sax != NULL) |
13938 | 0 | ctxt->sax = NULL; |
13939 | 0 | xmlFreeParserCtxt(ctxt); |
13940 | |
|
13941 | 0 | return(ret); |
13942 | 0 | } |
13943 | | |
13944 | | /** |
13945 | | * xmlParseEntity: |
13946 | | * @filename: the filename |
13947 | | * |
13948 | | * parse an XML external entity out of context and build a tree. |
13949 | | * |
13950 | | * [78] extParsedEnt ::= TextDecl? content |
13951 | | * |
13952 | | * This correspond to a "Well Balanced" chunk |
13953 | | * |
13954 | | * Returns the resulting document tree |
13955 | | */ |
13956 | | |
13957 | | xmlDocPtr |
13958 | 0 | xmlParseEntity(const char *filename) { |
13959 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
13960 | 0 | } |
13961 | | #endif /* LIBXML_SAX1_ENABLED */ |
13962 | | |
13963 | | /** |
13964 | | * xmlCreateEntityParserCtxtInternal: |
13965 | | * @URL: the entity URL |
13966 | | * @ID: the entity PUBLIC ID |
13967 | | * @base: a possible base for the target URI |
13968 | | * @pctx: parser context used to set options on new context |
13969 | | * |
13970 | | * Create a parser context for an external entity |
13971 | | * Automatic support for ZLIB/Compress compressed document is provided |
13972 | | * by default if found at compile-time. |
13973 | | * |
13974 | | * Returns the new parser context or NULL |
13975 | | */ |
13976 | | static xmlParserCtxtPtr |
13977 | | xmlCreateEntityParserCtxtInternal(xmlSAXHandlerPtr sax, void *userData, |
13978 | | const xmlChar *URL, const xmlChar *ID, const xmlChar *base, |
13979 | 4.41M | xmlParserCtxtPtr pctx) { |
13980 | 4.41M | xmlParserCtxtPtr ctxt; |
13981 | 4.41M | xmlParserInputPtr inputStream; |
13982 | 4.41M | char *directory = NULL; |
13983 | 4.41M | xmlChar *uri; |
13984 | | |
13985 | 4.41M | ctxt = xmlNewSAXParserCtxt(sax, userData); |
13986 | 4.41M | if (ctxt == NULL) { |
13987 | 0 | return(NULL); |
13988 | 0 | } |
13989 | | |
13990 | 4.41M | if (pctx != NULL) { |
13991 | 4.41M | ctxt->options = pctx->options; |
13992 | 4.41M | ctxt->_private = pctx->_private; |
13993 | | /* |
13994 | | * this is a subparser of pctx, so the input_id should be |
13995 | | * incremented to distinguish from main entity |
13996 | | */ |
13997 | 4.41M | ctxt->input_id = pctx->input_id + 1; |
13998 | 4.41M | } |
13999 | | |
14000 | | /* Don't read from stdin. */ |
14001 | 4.41M | if (xmlStrcmp(URL, BAD_CAST "-") == 0) |
14002 | 0 | URL = BAD_CAST "./-"; |
14003 | | |
14004 | 4.41M | uri = xmlBuildURI(URL, base); |
14005 | | |
14006 | 4.41M | if (uri == NULL) { |
14007 | 77.2k | inputStream = xmlLoadExternalEntity((char *)URL, (char *)ID, ctxt); |
14008 | 77.2k | if (inputStream == NULL) { |
14009 | 77.0k | xmlFreeParserCtxt(ctxt); |
14010 | 77.0k | return(NULL); |
14011 | 77.0k | } |
14012 | | |
14013 | 211 | inputPush(ctxt, inputStream); |
14014 | | |
14015 | 211 | if ((ctxt->directory == NULL) && (directory == NULL)) |
14016 | 211 | directory = xmlParserGetDirectory((char *)URL); |
14017 | 211 | if ((ctxt->directory == NULL) && (directory != NULL)) |
14018 | 211 | ctxt->directory = directory; |
14019 | 4.33M | } else { |
14020 | 4.33M | inputStream = xmlLoadExternalEntity((char *)uri, (char *)ID, ctxt); |
14021 | 4.33M | if (inputStream == NULL) { |
14022 | 39.1k | xmlFree(uri); |
14023 | 39.1k | xmlFreeParserCtxt(ctxt); |
14024 | 39.1k | return(NULL); |
14025 | 39.1k | } |
14026 | | |
14027 | 4.29M | inputPush(ctxt, inputStream); |
14028 | | |
14029 | 4.29M | if ((ctxt->directory == NULL) && (directory == NULL)) |
14030 | 4.29M | directory = xmlParserGetDirectory((char *)uri); |
14031 | 4.29M | if ((ctxt->directory == NULL) && (directory != NULL)) |
14032 | 4.29M | ctxt->directory = directory; |
14033 | 4.29M | xmlFree(uri); |
14034 | 4.29M | } |
14035 | 4.30M | return(ctxt); |
14036 | 4.41M | } |
14037 | | |
14038 | | /** |
14039 | | * xmlCreateEntityParserCtxt: |
14040 | | * @URL: the entity URL |
14041 | | * @ID: the entity PUBLIC ID |
14042 | | * @base: a possible base for the target URI |
14043 | | * |
14044 | | * Create a parser context for an external entity |
14045 | | * Automatic support for ZLIB/Compress compressed document is provided |
14046 | | * by default if found at compile-time. |
14047 | | * |
14048 | | * Returns the new parser context or NULL |
14049 | | */ |
14050 | | xmlParserCtxtPtr |
14051 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
14052 | 0 | const xmlChar *base) { |
14053 | 0 | return xmlCreateEntityParserCtxtInternal(NULL, NULL, URL, ID, base, NULL); |
14054 | |
|
14055 | 0 | } |
14056 | | |
14057 | | /************************************************************************ |
14058 | | * * |
14059 | | * Front ends when parsing from a file * |
14060 | | * * |
14061 | | ************************************************************************/ |
14062 | | |
14063 | | /** |
14064 | | * xmlCreateURLParserCtxt: |
14065 | | * @filename: the filename or URL |
14066 | | * @options: a combination of xmlParserOption |
14067 | | * |
14068 | | * Create a parser context for a file or URL content. |
14069 | | * Automatic support for ZLIB/Compress compressed document is provided |
14070 | | * by default if found at compile-time and for file accesses |
14071 | | * |
14072 | | * Returns the new parser context or NULL |
14073 | | */ |
14074 | | xmlParserCtxtPtr |
14075 | | xmlCreateURLParserCtxt(const char *filename, int options) |
14076 | 0 | { |
14077 | 0 | xmlParserCtxtPtr ctxt; |
14078 | 0 | xmlParserInputPtr inputStream; |
14079 | 0 | char *directory = NULL; |
14080 | |
|
14081 | 0 | ctxt = xmlNewParserCtxt(); |
14082 | 0 | if (ctxt == NULL) { |
14083 | 0 | xmlErrMemory(NULL, "cannot allocate parser context"); |
14084 | 0 | return(NULL); |
14085 | 0 | } |
14086 | | |
14087 | 0 | if (options) |
14088 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
14089 | 0 | ctxt->linenumbers = 1; |
14090 | |
|
14091 | 0 | inputStream = xmlLoadExternalEntity(filename, NULL, ctxt); |
14092 | 0 | if (inputStream == NULL) { |
14093 | 0 | xmlFreeParserCtxt(ctxt); |
14094 | 0 | return(NULL); |
14095 | 0 | } |
14096 | | |
14097 | 0 | inputPush(ctxt, inputStream); |
14098 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
14099 | 0 | directory = xmlParserGetDirectory(filename); |
14100 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
14101 | 0 | ctxt->directory = directory; |
14102 | |
|
14103 | 0 | return(ctxt); |
14104 | 0 | } |
14105 | | |
14106 | | /** |
14107 | | * xmlCreateFileParserCtxt: |
14108 | | * @filename: the filename |
14109 | | * |
14110 | | * Create a parser context for a file content. |
14111 | | * Automatic support for ZLIB/Compress compressed document is provided |
14112 | | * by default if found at compile-time. |
14113 | | * |
14114 | | * Returns the new parser context or NULL |
14115 | | */ |
14116 | | xmlParserCtxtPtr |
14117 | | xmlCreateFileParserCtxt(const char *filename) |
14118 | 0 | { |
14119 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
14120 | 0 | } |
14121 | | |
14122 | | #ifdef LIBXML_SAX1_ENABLED |
14123 | | /** |
14124 | | * xmlSAXParseFileWithData: |
14125 | | * @sax: the SAX handler block |
14126 | | * @filename: the filename |
14127 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14128 | | * documents |
14129 | | * @data: the userdata |
14130 | | * |
14131 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
14132 | | * |
14133 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14134 | | * compressed document is provided by default if found at compile-time. |
14135 | | * It use the given SAX function block to handle the parsing callback. |
14136 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14137 | | * |
14138 | | * User data (void *) is stored within the parser context in the |
14139 | | * context's _private member, so it is available nearly everywhere in libxml |
14140 | | * |
14141 | | * Returns the resulting document tree |
14142 | | */ |
14143 | | |
14144 | | xmlDocPtr |
14145 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
14146 | 0 | int recovery, void *data) { |
14147 | 0 | xmlDocPtr ret; |
14148 | 0 | xmlParserCtxtPtr ctxt; |
14149 | |
|
14150 | 0 | xmlInitParser(); |
14151 | |
|
14152 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14153 | 0 | if (ctxt == NULL) { |
14154 | 0 | return(NULL); |
14155 | 0 | } |
14156 | 0 | if (sax != NULL) { |
14157 | 0 | if (ctxt->sax != NULL) |
14158 | 0 | xmlFree(ctxt->sax); |
14159 | 0 | ctxt->sax = sax; |
14160 | 0 | } |
14161 | 0 | xmlDetectSAX2(ctxt); |
14162 | 0 | if (data!=NULL) { |
14163 | 0 | ctxt->_private = data; |
14164 | 0 | } |
14165 | |
|
14166 | 0 | if (ctxt->directory == NULL) |
14167 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14168 | |
|
14169 | 0 | ctxt->recovery = recovery; |
14170 | |
|
14171 | 0 | xmlParseDocument(ctxt); |
14172 | |
|
14173 | 0 | if ((ctxt->wellFormed) || recovery) { |
14174 | 0 | ret = ctxt->myDoc; |
14175 | 0 | if ((ret != NULL) && (ctxt->input->buf != NULL)) { |
14176 | 0 | if (ctxt->input->buf->compressed > 0) |
14177 | 0 | ret->compression = 9; |
14178 | 0 | else |
14179 | 0 | ret->compression = ctxt->input->buf->compressed; |
14180 | 0 | } |
14181 | 0 | } |
14182 | 0 | else { |
14183 | 0 | ret = NULL; |
14184 | 0 | xmlFreeDoc(ctxt->myDoc); |
14185 | 0 | ctxt->myDoc = NULL; |
14186 | 0 | } |
14187 | 0 | if (sax != NULL) |
14188 | 0 | ctxt->sax = NULL; |
14189 | 0 | xmlFreeParserCtxt(ctxt); |
14190 | |
|
14191 | 0 | return(ret); |
14192 | 0 | } |
14193 | | |
14194 | | /** |
14195 | | * xmlSAXParseFile: |
14196 | | * @sax: the SAX handler block |
14197 | | * @filename: the filename |
14198 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14199 | | * documents |
14200 | | * |
14201 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
14202 | | * |
14203 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14204 | | * compressed document is provided by default if found at compile-time. |
14205 | | * It use the given SAX function block to handle the parsing callback. |
14206 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14207 | | * |
14208 | | * Returns the resulting document tree |
14209 | | */ |
14210 | | |
14211 | | xmlDocPtr |
14212 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
14213 | 0 | int recovery) { |
14214 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
14215 | 0 | } |
14216 | | |
14217 | | /** |
14218 | | * xmlRecoverDoc: |
14219 | | * @cur: a pointer to an array of xmlChar |
14220 | | * |
14221 | | * DEPRECATED: Use xmlReadDoc with XML_PARSE_RECOVER. |
14222 | | * |
14223 | | * parse an XML in-memory document and build a tree. |
14224 | | * In the case the document is not Well Formed, a attempt to build a |
14225 | | * tree is tried anyway |
14226 | | * |
14227 | | * Returns the resulting document tree or NULL in case of failure |
14228 | | */ |
14229 | | |
14230 | | xmlDocPtr |
14231 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
14232 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
14233 | 0 | } |
14234 | | |
14235 | | /** |
14236 | | * xmlParseFile: |
14237 | | * @filename: the filename |
14238 | | * |
14239 | | * DEPRECATED: Use xmlReadFile. |
14240 | | * |
14241 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14242 | | * compressed document is provided by default if found at compile-time. |
14243 | | * |
14244 | | * Returns the resulting document tree if the file was wellformed, |
14245 | | * NULL otherwise. |
14246 | | */ |
14247 | | |
14248 | | xmlDocPtr |
14249 | 0 | xmlParseFile(const char *filename) { |
14250 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
14251 | 0 | } |
14252 | | |
14253 | | /** |
14254 | | * xmlRecoverFile: |
14255 | | * @filename: the filename |
14256 | | * |
14257 | | * DEPRECATED: Use xmlReadFile with XML_PARSE_RECOVER. |
14258 | | * |
14259 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14260 | | * compressed document is provided by default if found at compile-time. |
14261 | | * In the case the document is not Well Formed, it attempts to build |
14262 | | * a tree anyway |
14263 | | * |
14264 | | * Returns the resulting document tree or NULL in case of failure |
14265 | | */ |
14266 | | |
14267 | | xmlDocPtr |
14268 | 0 | xmlRecoverFile(const char *filename) { |
14269 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
14270 | 0 | } |
14271 | | |
14272 | | |
14273 | | /** |
14274 | | * xmlSetupParserForBuffer: |
14275 | | * @ctxt: an XML parser context |
14276 | | * @buffer: a xmlChar * buffer |
14277 | | * @filename: a file name |
14278 | | * |
14279 | | * DEPRECATED: Don't use. |
14280 | | * |
14281 | | * Setup the parser context to parse a new buffer; Clears any prior |
14282 | | * contents from the parser context. The buffer parameter must not be |
14283 | | * NULL, but the filename parameter can be |
14284 | | */ |
14285 | | void |
14286 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
14287 | | const char* filename) |
14288 | 0 | { |
14289 | 0 | xmlParserInputPtr input; |
14290 | |
|
14291 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
14292 | 0 | return; |
14293 | | |
14294 | 0 | input = xmlNewInputStream(ctxt); |
14295 | 0 | if (input == NULL) { |
14296 | 0 | xmlErrMemory(NULL, "parsing new buffer: out of memory\n"); |
14297 | 0 | xmlClearParserCtxt(ctxt); |
14298 | 0 | return; |
14299 | 0 | } |
14300 | | |
14301 | 0 | xmlClearParserCtxt(ctxt); |
14302 | 0 | if (filename != NULL) |
14303 | 0 | input->filename = (char *) xmlCanonicPath((const xmlChar *)filename); |
14304 | 0 | input->base = buffer; |
14305 | 0 | input->cur = buffer; |
14306 | 0 | input->end = &buffer[xmlStrlen(buffer)]; |
14307 | 0 | inputPush(ctxt, input); |
14308 | 0 | } |
14309 | | |
14310 | | /** |
14311 | | * xmlSAXUserParseFile: |
14312 | | * @sax: a SAX handler |
14313 | | * @user_data: The user data returned on SAX callbacks |
14314 | | * @filename: a file name |
14315 | | * |
14316 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
14317 | | * |
14318 | | * parse an XML file and call the given SAX handler routines. |
14319 | | * Automatic support for ZLIB/Compress compressed document is provided |
14320 | | * |
14321 | | * Returns 0 in case of success or a error number otherwise |
14322 | | */ |
14323 | | int |
14324 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
14325 | 0 | const char *filename) { |
14326 | 0 | int ret = 0; |
14327 | 0 | xmlParserCtxtPtr ctxt; |
14328 | |
|
14329 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14330 | 0 | if (ctxt == NULL) return -1; |
14331 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14332 | 0 | xmlFree(ctxt->sax); |
14333 | 0 | ctxt->sax = sax; |
14334 | 0 | xmlDetectSAX2(ctxt); |
14335 | |
|
14336 | 0 | if (user_data != NULL) |
14337 | 0 | ctxt->userData = user_data; |
14338 | |
|
14339 | 0 | xmlParseDocument(ctxt); |
14340 | |
|
14341 | 0 | if (ctxt->wellFormed) |
14342 | 0 | ret = 0; |
14343 | 0 | else { |
14344 | 0 | if (ctxt->errNo != 0) |
14345 | 0 | ret = ctxt->errNo; |
14346 | 0 | else |
14347 | 0 | ret = -1; |
14348 | 0 | } |
14349 | 0 | if (sax != NULL) |
14350 | 0 | ctxt->sax = NULL; |
14351 | 0 | if (ctxt->myDoc != NULL) { |
14352 | 0 | xmlFreeDoc(ctxt->myDoc); |
14353 | 0 | ctxt->myDoc = NULL; |
14354 | 0 | } |
14355 | 0 | xmlFreeParserCtxt(ctxt); |
14356 | |
|
14357 | 0 | return ret; |
14358 | 0 | } |
14359 | | #endif /* LIBXML_SAX1_ENABLED */ |
14360 | | |
14361 | | /************************************************************************ |
14362 | | * * |
14363 | | * Front ends when parsing from memory * |
14364 | | * * |
14365 | | ************************************************************************/ |
14366 | | |
14367 | | /** |
14368 | | * xmlCreateMemoryParserCtxt: |
14369 | | * @buffer: a pointer to a char array |
14370 | | * @size: the size of the array |
14371 | | * |
14372 | | * Create a parser context for an XML in-memory document. |
14373 | | * |
14374 | | * Returns the new parser context or NULL |
14375 | | */ |
14376 | | xmlParserCtxtPtr |
14377 | 699k | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
14378 | 699k | xmlParserCtxtPtr ctxt; |
14379 | 699k | xmlParserInputPtr input; |
14380 | 699k | xmlParserInputBufferPtr buf; |
14381 | | |
14382 | 699k | if (buffer == NULL) |
14383 | 0 | return(NULL); |
14384 | 699k | if (size <= 0) |
14385 | 47.4k | return(NULL); |
14386 | | |
14387 | 651k | ctxt = xmlNewParserCtxt(); |
14388 | 651k | if (ctxt == NULL) |
14389 | 0 | return(NULL); |
14390 | | |
14391 | | /* TODO: xmlParserInputBufferCreateStatic, requires some serious changes */ |
14392 | 651k | buf = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
14393 | 651k | if (buf == NULL) { |
14394 | 0 | xmlFreeParserCtxt(ctxt); |
14395 | 0 | return(NULL); |
14396 | 0 | } |
14397 | | |
14398 | 651k | input = xmlNewInputStream(ctxt); |
14399 | 651k | if (input == NULL) { |
14400 | 0 | xmlFreeParserInputBuffer(buf); |
14401 | 0 | xmlFreeParserCtxt(ctxt); |
14402 | 0 | return(NULL); |
14403 | 0 | } |
14404 | | |
14405 | 651k | input->filename = NULL; |
14406 | 651k | input->buf = buf; |
14407 | 651k | xmlBufResetInput(input->buf->buffer, input); |
14408 | | |
14409 | 651k | inputPush(ctxt, input); |
14410 | 651k | return(ctxt); |
14411 | 651k | } |
14412 | | |
14413 | | #ifdef LIBXML_SAX1_ENABLED |
14414 | | /** |
14415 | | * xmlSAXParseMemoryWithData: |
14416 | | * @sax: the SAX handler block |
14417 | | * @buffer: an pointer to a char array |
14418 | | * @size: the size of the array |
14419 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14420 | | * documents |
14421 | | * @data: the userdata |
14422 | | * |
14423 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14424 | | * |
14425 | | * parse an XML in-memory block and use the given SAX function block |
14426 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14427 | | * DOM tree building routines. |
14428 | | * |
14429 | | * User data (void *) is stored within the parser context in the |
14430 | | * context's _private member, so it is available nearly everywhere in libxml |
14431 | | * |
14432 | | * Returns the resulting document tree |
14433 | | */ |
14434 | | |
14435 | | xmlDocPtr |
14436 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
14437 | 0 | int size, int recovery, void *data) { |
14438 | 0 | xmlDocPtr ret; |
14439 | 0 | xmlParserCtxtPtr ctxt; |
14440 | |
|
14441 | 0 | xmlInitParser(); |
14442 | |
|
14443 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14444 | 0 | if (ctxt == NULL) return(NULL); |
14445 | 0 | if (sax != NULL) { |
14446 | 0 | if (ctxt->sax != NULL) |
14447 | 0 | xmlFree(ctxt->sax); |
14448 | 0 | ctxt->sax = sax; |
14449 | 0 | } |
14450 | 0 | xmlDetectSAX2(ctxt); |
14451 | 0 | if (data!=NULL) { |
14452 | 0 | ctxt->_private=data; |
14453 | 0 | } |
14454 | |
|
14455 | 0 | ctxt->recovery = recovery; |
14456 | |
|
14457 | 0 | xmlParseDocument(ctxt); |
14458 | |
|
14459 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14460 | 0 | else { |
14461 | 0 | ret = NULL; |
14462 | 0 | xmlFreeDoc(ctxt->myDoc); |
14463 | 0 | ctxt->myDoc = NULL; |
14464 | 0 | } |
14465 | 0 | if (sax != NULL) |
14466 | 0 | ctxt->sax = NULL; |
14467 | 0 | xmlFreeParserCtxt(ctxt); |
14468 | |
|
14469 | 0 | return(ret); |
14470 | 0 | } |
14471 | | |
14472 | | /** |
14473 | | * xmlSAXParseMemory: |
14474 | | * @sax: the SAX handler block |
14475 | | * @buffer: an pointer to a char array |
14476 | | * @size: the size of the array |
14477 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
14478 | | * documents |
14479 | | * |
14480 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14481 | | * |
14482 | | * parse an XML in-memory block and use the given SAX function block |
14483 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14484 | | * DOM tree building routines. |
14485 | | * |
14486 | | * Returns the resulting document tree |
14487 | | */ |
14488 | | xmlDocPtr |
14489 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
14490 | 0 | int size, int recovery) { |
14491 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
14492 | 0 | } |
14493 | | |
14494 | | /** |
14495 | | * xmlParseMemory: |
14496 | | * @buffer: an pointer to a char array |
14497 | | * @size: the size of the array |
14498 | | * |
14499 | | * DEPRECATED: Use xmlReadMemory. |
14500 | | * |
14501 | | * parse an XML in-memory block and build a tree. |
14502 | | * |
14503 | | * Returns the resulting document tree |
14504 | | */ |
14505 | | |
14506 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
14507 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
14508 | 0 | } |
14509 | | |
14510 | | /** |
14511 | | * xmlRecoverMemory: |
14512 | | * @buffer: an pointer to a char array |
14513 | | * @size: the size of the array |
14514 | | * |
14515 | | * DEPRECATED: Use xmlReadMemory with XML_PARSE_RECOVER. |
14516 | | * |
14517 | | * parse an XML in-memory block and build a tree. |
14518 | | * In the case the document is not Well Formed, an attempt to |
14519 | | * build a tree is tried anyway |
14520 | | * |
14521 | | * Returns the resulting document tree or NULL in case of error |
14522 | | */ |
14523 | | |
14524 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
14525 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
14526 | 0 | } |
14527 | | |
14528 | | /** |
14529 | | * xmlSAXUserParseMemory: |
14530 | | * @sax: a SAX handler |
14531 | | * @user_data: The user data returned on SAX callbacks |
14532 | | * @buffer: an in-memory XML document input |
14533 | | * @size: the length of the XML document in bytes |
14534 | | * |
14535 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14536 | | * |
14537 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
14538 | | * |
14539 | | * Returns 0 in case of success or a error number otherwise |
14540 | | */ |
14541 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
14542 | 0 | const char *buffer, int size) { |
14543 | 0 | int ret = 0; |
14544 | 0 | xmlParserCtxtPtr ctxt; |
14545 | |
|
14546 | 0 | xmlInitParser(); |
14547 | |
|
14548 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14549 | 0 | if (ctxt == NULL) return -1; |
14550 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14551 | 0 | xmlFree(ctxt->sax); |
14552 | 0 | ctxt->sax = sax; |
14553 | 0 | xmlDetectSAX2(ctxt); |
14554 | |
|
14555 | 0 | if (user_data != NULL) |
14556 | 0 | ctxt->userData = user_data; |
14557 | |
|
14558 | 0 | xmlParseDocument(ctxt); |
14559 | |
|
14560 | 0 | if (ctxt->wellFormed) |
14561 | 0 | ret = 0; |
14562 | 0 | else { |
14563 | 0 | if (ctxt->errNo != 0) |
14564 | 0 | ret = ctxt->errNo; |
14565 | 0 | else |
14566 | 0 | ret = -1; |
14567 | 0 | } |
14568 | 0 | if (sax != NULL) |
14569 | 0 | ctxt->sax = NULL; |
14570 | 0 | if (ctxt->myDoc != NULL) { |
14571 | 0 | xmlFreeDoc(ctxt->myDoc); |
14572 | 0 | ctxt->myDoc = NULL; |
14573 | 0 | } |
14574 | 0 | xmlFreeParserCtxt(ctxt); |
14575 | |
|
14576 | 0 | return ret; |
14577 | 0 | } |
14578 | | #endif /* LIBXML_SAX1_ENABLED */ |
14579 | | |
14580 | | /** |
14581 | | * xmlCreateDocParserCtxt: |
14582 | | * @cur: a pointer to an array of xmlChar |
14583 | | * |
14584 | | * Creates a parser context for an XML in-memory document. |
14585 | | * |
14586 | | * Returns the new parser context or NULL |
14587 | | */ |
14588 | | xmlParserCtxtPtr |
14589 | 0 | xmlCreateDocParserCtxt(const xmlChar *cur) { |
14590 | 0 | int len; |
14591 | |
|
14592 | 0 | if (cur == NULL) |
14593 | 0 | return(NULL); |
14594 | 0 | len = xmlStrlen(cur); |
14595 | 0 | return(xmlCreateMemoryParserCtxt((const char *)cur, len)); |
14596 | 0 | } |
14597 | | |
14598 | | #ifdef LIBXML_SAX1_ENABLED |
14599 | | /** |
14600 | | * xmlSAXParseDoc: |
14601 | | * @sax: the SAX handler block |
14602 | | * @cur: a pointer to an array of xmlChar |
14603 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14604 | | * documents |
14605 | | * |
14606 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadDoc. |
14607 | | * |
14608 | | * parse an XML in-memory document and build a tree. |
14609 | | * It use the given SAX function block to handle the parsing callback. |
14610 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14611 | | * |
14612 | | * Returns the resulting document tree |
14613 | | */ |
14614 | | |
14615 | | xmlDocPtr |
14616 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
14617 | 0 | xmlDocPtr ret; |
14618 | 0 | xmlParserCtxtPtr ctxt; |
14619 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
14620 | |
|
14621 | 0 | if (cur == NULL) return(NULL); |
14622 | | |
14623 | | |
14624 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
14625 | 0 | if (ctxt == NULL) return(NULL); |
14626 | 0 | if (sax != NULL) { |
14627 | 0 | oldsax = ctxt->sax; |
14628 | 0 | ctxt->sax = sax; |
14629 | 0 | ctxt->userData = NULL; |
14630 | 0 | } |
14631 | 0 | xmlDetectSAX2(ctxt); |
14632 | |
|
14633 | 0 | xmlParseDocument(ctxt); |
14634 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14635 | 0 | else { |
14636 | 0 | ret = NULL; |
14637 | 0 | xmlFreeDoc(ctxt->myDoc); |
14638 | 0 | ctxt->myDoc = NULL; |
14639 | 0 | } |
14640 | 0 | if (sax != NULL) |
14641 | 0 | ctxt->sax = oldsax; |
14642 | 0 | xmlFreeParserCtxt(ctxt); |
14643 | |
|
14644 | 0 | return(ret); |
14645 | 0 | } |
14646 | | |
14647 | | /** |
14648 | | * xmlParseDoc: |
14649 | | * @cur: a pointer to an array of xmlChar |
14650 | | * |
14651 | | * DEPRECATED: Use xmlReadDoc. |
14652 | | * |
14653 | | * parse an XML in-memory document and build a tree. |
14654 | | * |
14655 | | * Returns the resulting document tree |
14656 | | */ |
14657 | | |
14658 | | xmlDocPtr |
14659 | 0 | xmlParseDoc(const xmlChar *cur) { |
14660 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
14661 | 0 | } |
14662 | | #endif /* LIBXML_SAX1_ENABLED */ |
14663 | | |
14664 | | #ifdef LIBXML_LEGACY_ENABLED |
14665 | | /************************************************************************ |
14666 | | * * |
14667 | | * Specific function to keep track of entities references * |
14668 | | * and used by the XSLT debugger * |
14669 | | * * |
14670 | | ************************************************************************/ |
14671 | | |
14672 | | static xmlEntityReferenceFunc xmlEntityRefFunc = NULL; |
14673 | | |
14674 | | /** |
14675 | | * xmlAddEntityReference: |
14676 | | * @ent : A valid entity |
14677 | | * @firstNode : A valid first node for children of entity |
14678 | | * @lastNode : A valid last node of children entity |
14679 | | * |
14680 | | * Notify of a reference to an entity of type XML_EXTERNAL_GENERAL_PARSED_ENTITY |
14681 | | */ |
14682 | | static void |
14683 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
14684 | | xmlNodePtr lastNode) |
14685 | | { |
14686 | | if (xmlEntityRefFunc != NULL) { |
14687 | | (*xmlEntityRefFunc) (ent, firstNode, lastNode); |
14688 | | } |
14689 | | } |
14690 | | |
14691 | | |
14692 | | /** |
14693 | | * xmlSetEntityReferenceFunc: |
14694 | | * @func: A valid function |
14695 | | * |
14696 | | * Set the function to call call back when a xml reference has been made |
14697 | | */ |
14698 | | void |
14699 | | xmlSetEntityReferenceFunc(xmlEntityReferenceFunc func) |
14700 | | { |
14701 | | xmlEntityRefFunc = func; |
14702 | | } |
14703 | | #endif /* LIBXML_LEGACY_ENABLED */ |
14704 | | |
14705 | | /************************************************************************ |
14706 | | * * |
14707 | | * Miscellaneous * |
14708 | | * * |
14709 | | ************************************************************************/ |
14710 | | |
14711 | | static int xmlParserInitialized = 0; |
14712 | | |
14713 | | /** |
14714 | | * xmlInitParser: |
14715 | | * |
14716 | | * Initialization function for the XML parser. |
14717 | | * This is not reentrant. Call once before processing in case of |
14718 | | * use in multithreaded programs. |
14719 | | */ |
14720 | | |
14721 | | void |
14722 | 7.82M | xmlInitParser(void) { |
14723 | 7.82M | if (xmlParserInitialized != 0) |
14724 | 7.81M | return; |
14725 | | |
14726 | | #if defined(_WIN32) && (!defined(LIBXML_STATIC) || defined(LIBXML_STATIC_FOR_DLL)) |
14727 | | if (xmlFree == free) |
14728 | | atexit(xmlCleanupParser); |
14729 | | #endif |
14730 | | |
14731 | 10.8k | #ifdef LIBXML_THREAD_ENABLED |
14732 | 10.8k | __xmlGlobalInitMutexLock(); |
14733 | 10.8k | if (xmlParserInitialized == 0) { |
14734 | 10.8k | #endif |
14735 | 10.8k | xmlInitThreads(); |
14736 | 10.8k | xmlInitGlobals(); |
14737 | 10.8k | xmlInitMemory(); |
14738 | 10.8k | xmlInitializeDict(); |
14739 | 10.8k | xmlInitCharEncodingHandlers(); |
14740 | 10.8k | xmlDefaultSAXHandlerInit(); |
14741 | 10.8k | xmlRegisterDefaultInputCallbacks(); |
14742 | 10.8k | #ifdef LIBXML_OUTPUT_ENABLED |
14743 | 10.8k | xmlRegisterDefaultOutputCallbacks(); |
14744 | 10.8k | #endif /* LIBXML_OUTPUT_ENABLED */ |
14745 | 10.8k | #ifdef LIBXML_HTML_ENABLED |
14746 | 10.8k | htmlInitAutoClose(); |
14747 | 10.8k | htmlDefaultSAXHandlerInit(); |
14748 | 10.8k | #endif |
14749 | 10.8k | #if defined(LIBXML_XPATH_ENABLED) || defined(LIBXML_SCHEMAS_ENABLED) |
14750 | 10.8k | xmlXPathInit(); |
14751 | 10.8k | #endif |
14752 | 10.8k | xmlParserInitialized = 1; |
14753 | 10.8k | #ifdef LIBXML_THREAD_ENABLED |
14754 | 10.8k | } |
14755 | 10.8k | __xmlGlobalInitMutexUnlock(); |
14756 | 10.8k | #endif |
14757 | 10.8k | } |
14758 | | |
14759 | | /** |
14760 | | * xmlCleanupParser: |
14761 | | * |
14762 | | * This function name is somewhat misleading. It does not clean up |
14763 | | * parser state, it cleans up memory allocated by the library itself. |
14764 | | * It is a cleanup function for the XML library. It tries to reclaim all |
14765 | | * related global memory allocated for the library processing. |
14766 | | * It doesn't deallocate any document related memory. One should |
14767 | | * call xmlCleanupParser() only when the process has finished using |
14768 | | * the library and all XML/HTML documents built with it. |
14769 | | * See also xmlInitParser() which has the opposite function of preparing |
14770 | | * the library for operations. |
14771 | | * |
14772 | | * WARNING: if your application is multithreaded or has plugin support |
14773 | | * calling this may crash the application if another thread or |
14774 | | * a plugin is still using libxml2. It's sometimes very hard to |
14775 | | * guess if libxml2 is in use in the application, some libraries |
14776 | | * or plugins may use it without notice. In case of doubt abstain |
14777 | | * from calling this function or do it just before calling exit() |
14778 | | * to avoid leak reports from valgrind ! |
14779 | | */ |
14780 | | |
14781 | | void |
14782 | 0 | xmlCleanupParser(void) { |
14783 | 0 | if (!xmlParserInitialized) |
14784 | 0 | return; |
14785 | | |
14786 | 0 | xmlCleanupCharEncodingHandlers(); |
14787 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14788 | 0 | xmlCatalogCleanup(); |
14789 | 0 | #endif |
14790 | 0 | xmlDictCleanup(); |
14791 | 0 | xmlCleanupInputCallbacks(); |
14792 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
14793 | 0 | xmlCleanupOutputCallbacks(); |
14794 | 0 | #endif |
14795 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
14796 | 0 | xmlSchemaCleanupTypes(); |
14797 | 0 | xmlRelaxNGCleanupTypes(); |
14798 | 0 | #endif |
14799 | 0 | xmlCleanupGlobals(); |
14800 | 0 | xmlCleanupThreads(); /* must be last if called not from the main thread */ |
14801 | 0 | xmlCleanupMemory(); |
14802 | 0 | xmlParserInitialized = 0; |
14803 | 0 | } |
14804 | | |
14805 | | #if defined(HAVE_ATTRIBUTE_DESTRUCTOR) && !defined(LIBXML_STATIC) && \ |
14806 | | !defined(_WIN32) |
14807 | | static void |
14808 | | ATTRIBUTE_DESTRUCTOR |
14809 | | xmlDestructor(void) { |
14810 | | /* |
14811 | | * Calling custom deallocation functions in a destructor can cause |
14812 | | * problems, for example with Nokogiri. |
14813 | | */ |
14814 | | if (xmlFree == free) |
14815 | | xmlCleanupParser(); |
14816 | | } |
14817 | | #endif |
14818 | | |
14819 | | /************************************************************************ |
14820 | | * * |
14821 | | * New set (2.6.0) of simpler and more flexible APIs * |
14822 | | * * |
14823 | | ************************************************************************/ |
14824 | | |
14825 | | /** |
14826 | | * DICT_FREE: |
14827 | | * @str: a string |
14828 | | * |
14829 | | * Free a string if it is not owned by the "dict" dictionary in the |
14830 | | * current scope |
14831 | | */ |
14832 | | #define DICT_FREE(str) \ |
14833 | 0 | if ((str) && ((!dict) || \ |
14834 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
14835 | 0 | xmlFree((char *)(str)); |
14836 | | |
14837 | | /** |
14838 | | * xmlCtxtReset: |
14839 | | * @ctxt: an XML parser context |
14840 | | * |
14841 | | * Reset a parser context |
14842 | | */ |
14843 | | void |
14844 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
14845 | 0 | { |
14846 | 0 | xmlParserInputPtr input; |
14847 | 0 | xmlDictPtr dict; |
14848 | |
|
14849 | 0 | if (ctxt == NULL) |
14850 | 0 | return; |
14851 | | |
14852 | 0 | dict = ctxt->dict; |
14853 | |
|
14854 | 0 | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
14855 | 0 | xmlFreeInputStream(input); |
14856 | 0 | } |
14857 | 0 | ctxt->inputNr = 0; |
14858 | 0 | ctxt->input = NULL; |
14859 | |
|
14860 | 0 | ctxt->spaceNr = 0; |
14861 | 0 | if (ctxt->spaceTab != NULL) { |
14862 | 0 | ctxt->spaceTab[0] = -1; |
14863 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
14864 | 0 | } else { |
14865 | 0 | ctxt->space = NULL; |
14866 | 0 | } |
14867 | | |
14868 | |
|
14869 | 0 | ctxt->nodeNr = 0; |
14870 | 0 | ctxt->node = NULL; |
14871 | |
|
14872 | 0 | ctxt->nameNr = 0; |
14873 | 0 | ctxt->name = NULL; |
14874 | |
|
14875 | 0 | ctxt->nsNr = 0; |
14876 | |
|
14877 | 0 | DICT_FREE(ctxt->version); |
14878 | 0 | ctxt->version = NULL; |
14879 | 0 | DICT_FREE(ctxt->encoding); |
14880 | 0 | ctxt->encoding = NULL; |
14881 | 0 | DICT_FREE(ctxt->directory); |
14882 | 0 | ctxt->directory = NULL; |
14883 | 0 | DICT_FREE(ctxt->extSubURI); |
14884 | 0 | ctxt->extSubURI = NULL; |
14885 | 0 | DICT_FREE(ctxt->extSubSystem); |
14886 | 0 | ctxt->extSubSystem = NULL; |
14887 | 0 | if (ctxt->myDoc != NULL) |
14888 | 0 | xmlFreeDoc(ctxt->myDoc); |
14889 | 0 | ctxt->myDoc = NULL; |
14890 | |
|
14891 | 0 | ctxt->standalone = -1; |
14892 | 0 | ctxt->hasExternalSubset = 0; |
14893 | 0 | ctxt->hasPErefs = 0; |
14894 | 0 | ctxt->html = 0; |
14895 | 0 | ctxt->external = 0; |
14896 | 0 | ctxt->instate = XML_PARSER_START; |
14897 | 0 | ctxt->token = 0; |
14898 | |
|
14899 | 0 | ctxt->wellFormed = 1; |
14900 | 0 | ctxt->nsWellFormed = 1; |
14901 | 0 | ctxt->disableSAX = 0; |
14902 | 0 | ctxt->valid = 1; |
14903 | | #if 0 |
14904 | | ctxt->vctxt.userData = ctxt; |
14905 | | ctxt->vctxt.error = xmlParserValidityError; |
14906 | | ctxt->vctxt.warning = xmlParserValidityWarning; |
14907 | | #endif |
14908 | 0 | ctxt->record_info = 0; |
14909 | 0 | ctxt->checkIndex = 0; |
14910 | 0 | ctxt->inSubset = 0; |
14911 | 0 | ctxt->errNo = XML_ERR_OK; |
14912 | 0 | ctxt->depth = 0; |
14913 | 0 | ctxt->charset = XML_CHAR_ENCODING_UTF8; |
14914 | 0 | ctxt->catalogs = NULL; |
14915 | 0 | ctxt->nbentities = 0; |
14916 | 0 | ctxt->sizeentities = 0; |
14917 | 0 | ctxt->sizeentcopy = 0; |
14918 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
14919 | |
|
14920 | 0 | if (ctxt->attsDefault != NULL) { |
14921 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
14922 | 0 | ctxt->attsDefault = NULL; |
14923 | 0 | } |
14924 | 0 | if (ctxt->attsSpecial != NULL) { |
14925 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
14926 | 0 | ctxt->attsSpecial = NULL; |
14927 | 0 | } |
14928 | |
|
14929 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14930 | 0 | if (ctxt->catalogs != NULL) |
14931 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
14932 | 0 | #endif |
14933 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
14934 | 0 | xmlResetError(&ctxt->lastError); |
14935 | 0 | } |
14936 | | |
14937 | | /** |
14938 | | * xmlCtxtResetPush: |
14939 | | * @ctxt: an XML parser context |
14940 | | * @chunk: a pointer to an array of chars |
14941 | | * @size: number of chars in the array |
14942 | | * @filename: an optional file name or URI |
14943 | | * @encoding: the document encoding, or NULL |
14944 | | * |
14945 | | * Reset a push parser context |
14946 | | * |
14947 | | * Returns 0 in case of success and 1 in case of error |
14948 | | */ |
14949 | | int |
14950 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
14951 | | int size, const char *filename, const char *encoding) |
14952 | 0 | { |
14953 | 0 | xmlParserInputPtr inputStream; |
14954 | 0 | xmlParserInputBufferPtr buf; |
14955 | 0 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
14956 | |
|
14957 | 0 | if (ctxt == NULL) |
14958 | 0 | return(1); |
14959 | | |
14960 | 0 | if ((encoding == NULL) && (chunk != NULL) && (size >= 4)) |
14961 | 0 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
14962 | |
|
14963 | 0 | buf = xmlAllocParserInputBuffer(enc); |
14964 | 0 | if (buf == NULL) |
14965 | 0 | return(1); |
14966 | | |
14967 | 0 | if (ctxt == NULL) { |
14968 | 0 | xmlFreeParserInputBuffer(buf); |
14969 | 0 | return(1); |
14970 | 0 | } |
14971 | | |
14972 | 0 | xmlCtxtReset(ctxt); |
14973 | |
|
14974 | 0 | if (filename == NULL) { |
14975 | 0 | ctxt->directory = NULL; |
14976 | 0 | } else { |
14977 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14978 | 0 | } |
14979 | |
|
14980 | 0 | inputStream = xmlNewInputStream(ctxt); |
14981 | 0 | if (inputStream == NULL) { |
14982 | 0 | xmlFreeParserInputBuffer(buf); |
14983 | 0 | return(1); |
14984 | 0 | } |
14985 | | |
14986 | 0 | if (filename == NULL) |
14987 | 0 | inputStream->filename = NULL; |
14988 | 0 | else |
14989 | 0 | inputStream->filename = (char *) |
14990 | 0 | xmlCanonicPath((const xmlChar *) filename); |
14991 | 0 | inputStream->buf = buf; |
14992 | 0 | xmlBufResetInput(buf->buffer, inputStream); |
14993 | |
|
14994 | 0 | inputPush(ctxt, inputStream); |
14995 | |
|
14996 | 0 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
14997 | 0 | (ctxt->input->buf != NULL)) { |
14998 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
14999 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
15000 | |
|
15001 | 0 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
15002 | |
|
15003 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
15004 | | #ifdef DEBUG_PUSH |
15005 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
15006 | | #endif |
15007 | 0 | } |
15008 | |
|
15009 | 0 | if (encoding != NULL) { |
15010 | 0 | xmlCharEncodingHandlerPtr hdlr; |
15011 | |
|
15012 | 0 | if (ctxt->encoding != NULL) |
15013 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
15014 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
15015 | |
|
15016 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
15017 | 0 | if (hdlr != NULL) { |
15018 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
15019 | 0 | } else { |
15020 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
15021 | 0 | "Unsupported encoding %s\n", BAD_CAST encoding); |
15022 | 0 | } |
15023 | 0 | } else if (enc != XML_CHAR_ENCODING_NONE) { |
15024 | 0 | xmlSwitchEncoding(ctxt, enc); |
15025 | 0 | } |
15026 | |
|
15027 | 0 | return(0); |
15028 | 0 | } |
15029 | | |
15030 | | |
15031 | | /** |
15032 | | * xmlCtxtUseOptionsInternal: |
15033 | | * @ctxt: an XML parser context |
15034 | | * @options: a combination of xmlParserOption |
15035 | | * @encoding: the user provided encoding to use |
15036 | | * |
15037 | | * Applies the options to the parser context |
15038 | | * |
15039 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
15040 | | * in case of error. |
15041 | | */ |
15042 | | static int |
15043 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, const char *encoding) |
15044 | 1.47M | { |
15045 | 1.47M | if (ctxt == NULL) |
15046 | 0 | return(-1); |
15047 | 1.47M | if (encoding != NULL) { |
15048 | 0 | if (ctxt->encoding != NULL) |
15049 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
15050 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
15051 | 0 | } |
15052 | 1.47M | if (options & XML_PARSE_RECOVER) { |
15053 | 887k | ctxt->recovery = 1; |
15054 | 887k | options -= XML_PARSE_RECOVER; |
15055 | 887k | ctxt->options |= XML_PARSE_RECOVER; |
15056 | 887k | } else |
15057 | 592k | ctxt->recovery = 0; |
15058 | 1.47M | if (options & XML_PARSE_DTDLOAD) { |
15059 | 1.26M | ctxt->loadsubset = XML_DETECT_IDS; |
15060 | 1.26M | options -= XML_PARSE_DTDLOAD; |
15061 | 1.26M | ctxt->options |= XML_PARSE_DTDLOAD; |
15062 | 1.26M | } else |
15063 | 212k | ctxt->loadsubset = 0; |
15064 | 1.47M | if (options & XML_PARSE_DTDATTR) { |
15065 | 537k | ctxt->loadsubset |= XML_COMPLETE_ATTRS; |
15066 | 537k | options -= XML_PARSE_DTDATTR; |
15067 | 537k | ctxt->options |= XML_PARSE_DTDATTR; |
15068 | 537k | } |
15069 | 1.47M | if (options & XML_PARSE_NOENT) { |
15070 | 953k | ctxt->replaceEntities = 1; |
15071 | | /* ctxt->loadsubset |= XML_DETECT_IDS; */ |
15072 | 953k | options -= XML_PARSE_NOENT; |
15073 | 953k | ctxt->options |= XML_PARSE_NOENT; |
15074 | 953k | } else |
15075 | 525k | ctxt->replaceEntities = 0; |
15076 | 1.47M | if (options & XML_PARSE_PEDANTIC) { |
15077 | 373k | ctxt->pedantic = 1; |
15078 | 373k | options -= XML_PARSE_PEDANTIC; |
15079 | 373k | ctxt->options |= XML_PARSE_PEDANTIC; |
15080 | 373k | } else |
15081 | 1.10M | ctxt->pedantic = 0; |
15082 | 1.47M | if (options & XML_PARSE_NOBLANKS) { |
15083 | 621k | ctxt->keepBlanks = 0; |
15084 | 621k | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
15085 | 621k | options -= XML_PARSE_NOBLANKS; |
15086 | 621k | ctxt->options |= XML_PARSE_NOBLANKS; |
15087 | 621k | } else |
15088 | 858k | ctxt->keepBlanks = 1; |
15089 | 1.47M | if (options & XML_PARSE_DTDVALID) { |
15090 | 558k | ctxt->validate = 1; |
15091 | 558k | if (options & XML_PARSE_NOWARNING) |
15092 | 372k | ctxt->vctxt.warning = NULL; |
15093 | 558k | if (options & XML_PARSE_NOERROR) |
15094 | 403k | ctxt->vctxt.error = NULL; |
15095 | 558k | options -= XML_PARSE_DTDVALID; |
15096 | 558k | ctxt->options |= XML_PARSE_DTDVALID; |
15097 | 558k | } else |
15098 | 921k | ctxt->validate = 0; |
15099 | 1.47M | if (options & XML_PARSE_NOWARNING) { |
15100 | 493k | ctxt->sax->warning = NULL; |
15101 | 493k | options -= XML_PARSE_NOWARNING; |
15102 | 493k | } |
15103 | 1.47M | if (options & XML_PARSE_NOERROR) { |
15104 | 598k | ctxt->sax->error = NULL; |
15105 | 598k | ctxt->sax->fatalError = NULL; |
15106 | 598k | options -= XML_PARSE_NOERROR; |
15107 | 598k | } |
15108 | 1.47M | #ifdef LIBXML_SAX1_ENABLED |
15109 | 1.47M | if (options & XML_PARSE_SAX1) { |
15110 | 673k | ctxt->sax->startElement = xmlSAX2StartElement; |
15111 | 673k | ctxt->sax->endElement = xmlSAX2EndElement; |
15112 | 673k | ctxt->sax->startElementNs = NULL; |
15113 | 673k | ctxt->sax->endElementNs = NULL; |
15114 | 673k | ctxt->sax->initialized = 1; |
15115 | 673k | options -= XML_PARSE_SAX1; |
15116 | 673k | ctxt->options |= XML_PARSE_SAX1; |
15117 | 673k | } |
15118 | 1.47M | #endif /* LIBXML_SAX1_ENABLED */ |
15119 | 1.47M | if (options & XML_PARSE_NODICT) { |
15120 | 655k | ctxt->dictNames = 0; |
15121 | 655k | options -= XML_PARSE_NODICT; |
15122 | 655k | ctxt->options |= XML_PARSE_NODICT; |
15123 | 824k | } else { |
15124 | 824k | ctxt->dictNames = 1; |
15125 | 824k | } |
15126 | 1.47M | if (options & XML_PARSE_NOCDATA) { |
15127 | 679k | ctxt->sax->cdataBlock = NULL; |
15128 | 679k | options -= XML_PARSE_NOCDATA; |
15129 | 679k | ctxt->options |= XML_PARSE_NOCDATA; |
15130 | 679k | } |
15131 | 1.47M | if (options & XML_PARSE_NSCLEAN) { |
15132 | 803k | ctxt->options |= XML_PARSE_NSCLEAN; |
15133 | 803k | options -= XML_PARSE_NSCLEAN; |
15134 | 803k | } |
15135 | 1.47M | if (options & XML_PARSE_NONET) { |
15136 | 678k | ctxt->options |= XML_PARSE_NONET; |
15137 | 678k | options -= XML_PARSE_NONET; |
15138 | 678k | } |
15139 | 1.47M | if (options & XML_PARSE_COMPACT) { |
15140 | 822k | ctxt->options |= XML_PARSE_COMPACT; |
15141 | 822k | options -= XML_PARSE_COMPACT; |
15142 | 822k | } |
15143 | 1.47M | if (options & XML_PARSE_OLD10) { |
15144 | 505k | ctxt->options |= XML_PARSE_OLD10; |
15145 | 505k | options -= XML_PARSE_OLD10; |
15146 | 505k | } |
15147 | 1.47M | if (options & XML_PARSE_NOBASEFIX) { |
15148 | 550k | ctxt->options |= XML_PARSE_NOBASEFIX; |
15149 | 550k | options -= XML_PARSE_NOBASEFIX; |
15150 | 550k | } |
15151 | 1.47M | if (options & XML_PARSE_HUGE) { |
15152 | 474k | ctxt->options |= XML_PARSE_HUGE; |
15153 | 474k | options -= XML_PARSE_HUGE; |
15154 | 474k | if (ctxt->dict != NULL) |
15155 | 474k | xmlDictSetLimit(ctxt->dict, 0); |
15156 | 474k | } |
15157 | 1.47M | if (options & XML_PARSE_OLDSAX) { |
15158 | 431k | ctxt->options |= XML_PARSE_OLDSAX; |
15159 | 431k | options -= XML_PARSE_OLDSAX; |
15160 | 431k | } |
15161 | 1.47M | if (options & XML_PARSE_IGNORE_ENC) { |
15162 | 606k | ctxt->options |= XML_PARSE_IGNORE_ENC; |
15163 | 606k | options -= XML_PARSE_IGNORE_ENC; |
15164 | 606k | } |
15165 | 1.47M | if (options & XML_PARSE_BIG_LINES) { |
15166 | 535k | ctxt->options |= XML_PARSE_BIG_LINES; |
15167 | 535k | options -= XML_PARSE_BIG_LINES; |
15168 | 535k | } |
15169 | 1.47M | ctxt->linenumbers = 1; |
15170 | 1.47M | return (options); |
15171 | 1.47M | } |
15172 | | |
15173 | | /** |
15174 | | * xmlCtxtUseOptions: |
15175 | | * @ctxt: an XML parser context |
15176 | | * @options: a combination of xmlParserOption |
15177 | | * |
15178 | | * Applies the options to the parser context |
15179 | | * |
15180 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
15181 | | * in case of error. |
15182 | | */ |
15183 | | int |
15184 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
15185 | 1.07M | { |
15186 | 1.07M | return(xmlCtxtUseOptionsInternal(ctxt, options, NULL)); |
15187 | 1.07M | } |
15188 | | |
15189 | | /** |
15190 | | * xmlDoRead: |
15191 | | * @ctxt: an XML parser context |
15192 | | * @URL: the base URL to use for the document |
15193 | | * @encoding: the document encoding, or NULL |
15194 | | * @options: a combination of xmlParserOption |
15195 | | * @reuse: keep the context for reuse |
15196 | | * |
15197 | | * Common front-end for the xmlRead functions |
15198 | | * |
15199 | | * Returns the resulting document tree or NULL |
15200 | | */ |
15201 | | static xmlDocPtr |
15202 | | xmlDoRead(xmlParserCtxtPtr ctxt, const char *URL, const char *encoding, |
15203 | | int options, int reuse) |
15204 | 405k | { |
15205 | 405k | xmlDocPtr ret; |
15206 | | |
15207 | 405k | xmlCtxtUseOptionsInternal(ctxt, options, encoding); |
15208 | 405k | if (encoding != NULL) { |
15209 | 0 | xmlCharEncodingHandlerPtr hdlr; |
15210 | |
|
15211 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
15212 | 0 | if (hdlr != NULL) |
15213 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
15214 | 0 | } |
15215 | 405k | if ((URL != NULL) && (ctxt->input != NULL) && |
15216 | 405k | (ctxt->input->filename == NULL)) |
15217 | 405k | ctxt->input->filename = (char *) xmlStrdup((const xmlChar *) URL); |
15218 | 405k | xmlParseDocument(ctxt); |
15219 | 405k | if ((ctxt->wellFormed) || ctxt->recovery) |
15220 | 233k | ret = ctxt->myDoc; |
15221 | 172k | else { |
15222 | 172k | ret = NULL; |
15223 | 172k | if (ctxt->myDoc != NULL) { |
15224 | 149k | xmlFreeDoc(ctxt->myDoc); |
15225 | 149k | } |
15226 | 172k | } |
15227 | 405k | ctxt->myDoc = NULL; |
15228 | 405k | if (!reuse) { |
15229 | 405k | xmlFreeParserCtxt(ctxt); |
15230 | 405k | } |
15231 | | |
15232 | 405k | return (ret); |
15233 | 405k | } |
15234 | | |
15235 | | /** |
15236 | | * xmlReadDoc: |
15237 | | * @cur: a pointer to a zero terminated string |
15238 | | * @URL: the base URL to use for the document |
15239 | | * @encoding: the document encoding, or NULL |
15240 | | * @options: a combination of xmlParserOption |
15241 | | * |
15242 | | * parse an XML in-memory document and build a tree. |
15243 | | * |
15244 | | * Returns the resulting document tree |
15245 | | */ |
15246 | | xmlDocPtr |
15247 | | xmlReadDoc(const xmlChar * cur, const char *URL, const char *encoding, int options) |
15248 | 0 | { |
15249 | 0 | xmlParserCtxtPtr ctxt; |
15250 | |
|
15251 | 0 | if (cur == NULL) |
15252 | 0 | return (NULL); |
15253 | 0 | xmlInitParser(); |
15254 | |
|
15255 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
15256 | 0 | if (ctxt == NULL) |
15257 | 0 | return (NULL); |
15258 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15259 | 0 | } |
15260 | | |
15261 | | /** |
15262 | | * xmlReadFile: |
15263 | | * @filename: a file or URL |
15264 | | * @encoding: the document encoding, or NULL |
15265 | | * @options: a combination of xmlParserOption |
15266 | | * |
15267 | | * parse an XML file from the filesystem or the network. |
15268 | | * |
15269 | | * Returns the resulting document tree |
15270 | | */ |
15271 | | xmlDocPtr |
15272 | | xmlReadFile(const char *filename, const char *encoding, int options) |
15273 | 0 | { |
15274 | 0 | xmlParserCtxtPtr ctxt; |
15275 | |
|
15276 | 0 | xmlInitParser(); |
15277 | 0 | ctxt = xmlCreateURLParserCtxt(filename, options); |
15278 | 0 | if (ctxt == NULL) |
15279 | 0 | return (NULL); |
15280 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 0)); |
15281 | 0 | } |
15282 | | |
15283 | | /** |
15284 | | * xmlReadMemory: |
15285 | | * @buffer: a pointer to a char array |
15286 | | * @size: the size of the array |
15287 | | * @URL: the base URL to use for the document |
15288 | | * @encoding: the document encoding, or NULL |
15289 | | * @options: a combination of xmlParserOption |
15290 | | * |
15291 | | * parse an XML in-memory document and build a tree. |
15292 | | * |
15293 | | * Returns the resulting document tree |
15294 | | */ |
15295 | | xmlDocPtr |
15296 | | xmlReadMemory(const char *buffer, int size, const char *URL, const char *encoding, int options) |
15297 | 406k | { |
15298 | 406k | xmlParserCtxtPtr ctxt; |
15299 | | |
15300 | 406k | xmlInitParser(); |
15301 | 406k | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
15302 | 406k | if (ctxt == NULL) |
15303 | 645 | return (NULL); |
15304 | 405k | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15305 | 406k | } |
15306 | | |
15307 | | /** |
15308 | | * xmlReadFd: |
15309 | | * @fd: an open file descriptor |
15310 | | * @URL: the base URL to use for the document |
15311 | | * @encoding: the document encoding, or NULL |
15312 | | * @options: a combination of xmlParserOption |
15313 | | * |
15314 | | * parse an XML from a file descriptor and build a tree. |
15315 | | * NOTE that the file descriptor will not be closed when the |
15316 | | * reader is closed or reset. |
15317 | | * |
15318 | | * Returns the resulting document tree |
15319 | | */ |
15320 | | xmlDocPtr |
15321 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
15322 | 0 | { |
15323 | 0 | xmlParserCtxtPtr ctxt; |
15324 | 0 | xmlParserInputBufferPtr input; |
15325 | 0 | xmlParserInputPtr stream; |
15326 | |
|
15327 | 0 | if (fd < 0) |
15328 | 0 | return (NULL); |
15329 | 0 | xmlInitParser(); |
15330 | |
|
15331 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15332 | 0 | if (input == NULL) |
15333 | 0 | return (NULL); |
15334 | 0 | input->closecallback = NULL; |
15335 | 0 | ctxt = xmlNewParserCtxt(); |
15336 | 0 | if (ctxt == NULL) { |
15337 | 0 | xmlFreeParserInputBuffer(input); |
15338 | 0 | return (NULL); |
15339 | 0 | } |
15340 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15341 | 0 | if (stream == NULL) { |
15342 | 0 | xmlFreeParserInputBuffer(input); |
15343 | 0 | xmlFreeParserCtxt(ctxt); |
15344 | 0 | return (NULL); |
15345 | 0 | } |
15346 | 0 | inputPush(ctxt, stream); |
15347 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15348 | 0 | } |
15349 | | |
15350 | | /** |
15351 | | * xmlReadIO: |
15352 | | * @ioread: an I/O read function |
15353 | | * @ioclose: an I/O close function |
15354 | | * @ioctx: an I/O handler |
15355 | | * @URL: the base URL to use for the document |
15356 | | * @encoding: the document encoding, or NULL |
15357 | | * @options: a combination of xmlParserOption |
15358 | | * |
15359 | | * parse an XML document from I/O functions and source and build a tree. |
15360 | | * |
15361 | | * Returns the resulting document tree |
15362 | | */ |
15363 | | xmlDocPtr |
15364 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
15365 | | void *ioctx, const char *URL, const char *encoding, int options) |
15366 | 0 | { |
15367 | 0 | xmlParserCtxtPtr ctxt; |
15368 | 0 | xmlParserInputBufferPtr input; |
15369 | 0 | xmlParserInputPtr stream; |
15370 | |
|
15371 | 0 | if (ioread == NULL) |
15372 | 0 | return (NULL); |
15373 | 0 | xmlInitParser(); |
15374 | |
|
15375 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15376 | 0 | XML_CHAR_ENCODING_NONE); |
15377 | 0 | if (input == NULL) { |
15378 | 0 | if (ioclose != NULL) |
15379 | 0 | ioclose(ioctx); |
15380 | 0 | return (NULL); |
15381 | 0 | } |
15382 | 0 | ctxt = xmlNewParserCtxt(); |
15383 | 0 | if (ctxt == NULL) { |
15384 | 0 | xmlFreeParserInputBuffer(input); |
15385 | 0 | return (NULL); |
15386 | 0 | } |
15387 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15388 | 0 | if (stream == NULL) { |
15389 | 0 | xmlFreeParserInputBuffer(input); |
15390 | 0 | xmlFreeParserCtxt(ctxt); |
15391 | 0 | return (NULL); |
15392 | 0 | } |
15393 | 0 | inputPush(ctxt, stream); |
15394 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15395 | 0 | } |
15396 | | |
15397 | | /** |
15398 | | * xmlCtxtReadDoc: |
15399 | | * @ctxt: an XML parser context |
15400 | | * @cur: a pointer to a zero terminated string |
15401 | | * @URL: the base URL to use for the document |
15402 | | * @encoding: the document encoding, or NULL |
15403 | | * @options: a combination of xmlParserOption |
15404 | | * |
15405 | | * parse an XML in-memory document and build a tree. |
15406 | | * This reuses the existing @ctxt parser context |
15407 | | * |
15408 | | * Returns the resulting document tree |
15409 | | */ |
15410 | | xmlDocPtr |
15411 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar * cur, |
15412 | | const char *URL, const char *encoding, int options) |
15413 | 0 | { |
15414 | 0 | if (cur == NULL) |
15415 | 0 | return (NULL); |
15416 | 0 | return (xmlCtxtReadMemory(ctxt, (const char *) cur, xmlStrlen(cur), URL, |
15417 | 0 | encoding, options)); |
15418 | 0 | } |
15419 | | |
15420 | | /** |
15421 | | * xmlCtxtReadFile: |
15422 | | * @ctxt: an XML parser context |
15423 | | * @filename: a file or URL |
15424 | | * @encoding: the document encoding, or NULL |
15425 | | * @options: a combination of xmlParserOption |
15426 | | * |
15427 | | * parse an XML file from the filesystem or the network. |
15428 | | * This reuses the existing @ctxt parser context |
15429 | | * |
15430 | | * Returns the resulting document tree |
15431 | | */ |
15432 | | xmlDocPtr |
15433 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
15434 | | const char *encoding, int options) |
15435 | 0 | { |
15436 | 0 | xmlParserInputPtr stream; |
15437 | |
|
15438 | 0 | if (filename == NULL) |
15439 | 0 | return (NULL); |
15440 | 0 | if (ctxt == NULL) |
15441 | 0 | return (NULL); |
15442 | 0 | xmlInitParser(); |
15443 | |
|
15444 | 0 | xmlCtxtReset(ctxt); |
15445 | |
|
15446 | 0 | stream = xmlLoadExternalEntity(filename, NULL, ctxt); |
15447 | 0 | if (stream == NULL) { |
15448 | 0 | return (NULL); |
15449 | 0 | } |
15450 | 0 | inputPush(ctxt, stream); |
15451 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 1)); |
15452 | 0 | } |
15453 | | |
15454 | | /** |
15455 | | * xmlCtxtReadMemory: |
15456 | | * @ctxt: an XML parser context |
15457 | | * @buffer: a pointer to a char array |
15458 | | * @size: the size of the array |
15459 | | * @URL: the base URL to use for the document |
15460 | | * @encoding: the document encoding, or NULL |
15461 | | * @options: a combination of xmlParserOption |
15462 | | * |
15463 | | * parse an XML in-memory document and build a tree. |
15464 | | * This reuses the existing @ctxt parser context |
15465 | | * |
15466 | | * Returns the resulting document tree |
15467 | | */ |
15468 | | xmlDocPtr |
15469 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
15470 | | const char *URL, const char *encoding, int options) |
15471 | 0 | { |
15472 | 0 | xmlParserInputBufferPtr input; |
15473 | 0 | xmlParserInputPtr stream; |
15474 | |
|
15475 | 0 | if (ctxt == NULL) |
15476 | 0 | return (NULL); |
15477 | 0 | if (buffer == NULL) |
15478 | 0 | return (NULL); |
15479 | 0 | xmlInitParser(); |
15480 | |
|
15481 | 0 | xmlCtxtReset(ctxt); |
15482 | |
|
15483 | 0 | input = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
15484 | 0 | if (input == NULL) { |
15485 | 0 | return(NULL); |
15486 | 0 | } |
15487 | | |
15488 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15489 | 0 | if (stream == NULL) { |
15490 | 0 | xmlFreeParserInputBuffer(input); |
15491 | 0 | return(NULL); |
15492 | 0 | } |
15493 | | |
15494 | 0 | inputPush(ctxt, stream); |
15495 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15496 | 0 | } |
15497 | | |
15498 | | /** |
15499 | | * xmlCtxtReadFd: |
15500 | | * @ctxt: an XML parser context |
15501 | | * @fd: an open file descriptor |
15502 | | * @URL: the base URL to use for the document |
15503 | | * @encoding: the document encoding, or NULL |
15504 | | * @options: a combination of xmlParserOption |
15505 | | * |
15506 | | * parse an XML from a file descriptor and build a tree. |
15507 | | * This reuses the existing @ctxt parser context |
15508 | | * NOTE that the file descriptor will not be closed when the |
15509 | | * reader is closed or reset. |
15510 | | * |
15511 | | * Returns the resulting document tree |
15512 | | */ |
15513 | | xmlDocPtr |
15514 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
15515 | | const char *URL, const char *encoding, int options) |
15516 | 0 | { |
15517 | 0 | xmlParserInputBufferPtr input; |
15518 | 0 | xmlParserInputPtr stream; |
15519 | |
|
15520 | 0 | if (fd < 0) |
15521 | 0 | return (NULL); |
15522 | 0 | if (ctxt == NULL) |
15523 | 0 | return (NULL); |
15524 | 0 | xmlInitParser(); |
15525 | |
|
15526 | 0 | xmlCtxtReset(ctxt); |
15527 | | |
15528 | |
|
15529 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15530 | 0 | if (input == NULL) |
15531 | 0 | return (NULL); |
15532 | 0 | input->closecallback = NULL; |
15533 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15534 | 0 | if (stream == NULL) { |
15535 | 0 | xmlFreeParserInputBuffer(input); |
15536 | 0 | return (NULL); |
15537 | 0 | } |
15538 | 0 | inputPush(ctxt, stream); |
15539 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15540 | 0 | } |
15541 | | |
15542 | | /** |
15543 | | * xmlCtxtReadIO: |
15544 | | * @ctxt: an XML parser context |
15545 | | * @ioread: an I/O read function |
15546 | | * @ioclose: an I/O close function |
15547 | | * @ioctx: an I/O handler |
15548 | | * @URL: the base URL to use for the document |
15549 | | * @encoding: the document encoding, or NULL |
15550 | | * @options: a combination of xmlParserOption |
15551 | | * |
15552 | | * parse an XML document from I/O functions and source and build a tree. |
15553 | | * This reuses the existing @ctxt parser context |
15554 | | * |
15555 | | * Returns the resulting document tree |
15556 | | */ |
15557 | | xmlDocPtr |
15558 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
15559 | | xmlInputCloseCallback ioclose, void *ioctx, |
15560 | | const char *URL, |
15561 | | const char *encoding, int options) |
15562 | 0 | { |
15563 | 0 | xmlParserInputBufferPtr input; |
15564 | 0 | xmlParserInputPtr stream; |
15565 | |
|
15566 | 0 | if (ioread == NULL) |
15567 | 0 | return (NULL); |
15568 | 0 | if (ctxt == NULL) |
15569 | 0 | return (NULL); |
15570 | 0 | xmlInitParser(); |
15571 | |
|
15572 | 0 | xmlCtxtReset(ctxt); |
15573 | |
|
15574 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15575 | 0 | XML_CHAR_ENCODING_NONE); |
15576 | 0 | if (input == NULL) { |
15577 | 0 | if (ioclose != NULL) |
15578 | 0 | ioclose(ioctx); |
15579 | 0 | return (NULL); |
15580 | 0 | } |
15581 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15582 | 0 | if (stream == NULL) { |
15583 | 0 | xmlFreeParserInputBuffer(input); |
15584 | 0 | return (NULL); |
15585 | 0 | } |
15586 | 0 | inputPush(ctxt, stream); |
15587 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15588 | 0 | } |
15589 | | |