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/valid.h> |
61 | | #include <libxml/entities.h> |
62 | | #include <libxml/xmlerror.h> |
63 | | #include <libxml/encoding.h> |
64 | | #include <libxml/xmlIO.h> |
65 | | #include <libxml/uri.h> |
66 | | #ifdef LIBXML_CATALOG_ENABLED |
67 | | #include <libxml/catalog.h> |
68 | | #endif |
69 | | #ifdef LIBXML_SCHEMAS_ENABLED |
70 | | #include <libxml/xmlschemastypes.h> |
71 | | #include <libxml/relaxng.h> |
72 | | #endif |
73 | | |
74 | | #include "buf.h" |
75 | | #include "enc.h" |
76 | | |
77 | | struct _xmlStartTag { |
78 | | const xmlChar *prefix; |
79 | | const xmlChar *URI; |
80 | | int line; |
81 | | int nsNr; |
82 | | }; |
83 | | |
84 | | static void |
85 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info); |
86 | | |
87 | | static xmlParserCtxtPtr |
88 | | xmlCreateEntityParserCtxtInternal(const xmlChar *URL, const xmlChar *ID, |
89 | | const xmlChar *base, xmlParserCtxtPtr pctx); |
90 | | |
91 | | static void xmlHaltParser(xmlParserCtxtPtr ctxt); |
92 | | |
93 | | static int |
94 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
95 | | |
96 | | static void |
97 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
98 | | |
99 | | /************************************************************************ |
100 | | * * |
101 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
102 | | * * |
103 | | ************************************************************************/ |
104 | | |
105 | 0 | #define XML_PARSER_BIG_ENTITY 1000 |
106 | | #define XML_PARSER_LOT_ENTITY 5000 |
107 | | |
108 | | /* |
109 | | * XML_PARSER_NON_LINEAR is the threshold where the ratio of parsed entity |
110 | | * replacement over the size in byte of the input indicates that you have |
111 | | * and exponential behaviour. A value of 10 correspond to at least 3 entity |
112 | | * replacement per byte of input. |
113 | | */ |
114 | 0 | #define XML_PARSER_NON_LINEAR 10 |
115 | | |
116 | | /* |
117 | | * xmlParserEntityCheck |
118 | | * |
119 | | * Function to check non-linear entity expansion behaviour |
120 | | * This is here to detect and stop exponential linear entity expansion |
121 | | * This is not a limitation of the parser but a safety |
122 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
123 | | * parser option. |
124 | | */ |
125 | | static int |
126 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, size_t size, |
127 | | xmlEntityPtr ent, size_t replacement) |
128 | 8.14M | { |
129 | 8.14M | size_t consumed = 0; |
130 | 8.14M | int i; |
131 | | |
132 | 8.14M | if ((ctxt == NULL) || (ctxt->options & XML_PARSE_HUGE)) |
133 | 8.14M | return (0); |
134 | 0 | if (ctxt->lastError.code == XML_ERR_ENTITY_LOOP) |
135 | 0 | return (1); |
136 | | |
137 | | /* |
138 | | * This may look absurd but is needed to detect |
139 | | * entities problems |
140 | | */ |
141 | 0 | if ((ent != NULL) && (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
142 | 0 | (ent->content != NULL) && (ent->checked == 0) && |
143 | 0 | (ctxt->errNo != XML_ERR_ENTITY_LOOP)) { |
144 | 0 | unsigned long oldnbent = ctxt->nbentities, diff; |
145 | 0 | xmlChar *rep; |
146 | |
|
147 | 0 | ent->checked = 1; |
148 | |
|
149 | 0 | ++ctxt->depth; |
150 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
151 | 0 | XML_SUBSTITUTE_REF, 0, 0, 0); |
152 | 0 | --ctxt->depth; |
153 | 0 | if ((rep == NULL) || (ctxt->errNo == XML_ERR_ENTITY_LOOP)) { |
154 | 0 | ent->content[0] = 0; |
155 | 0 | } |
156 | |
|
157 | 0 | diff = ctxt->nbentities - oldnbent + 1; |
158 | 0 | if (diff > INT_MAX / 2) |
159 | 0 | diff = INT_MAX / 2; |
160 | 0 | ent->checked = diff * 2; |
161 | 0 | if (rep != NULL) { |
162 | 0 | if (xmlStrchr(rep, '<')) |
163 | 0 | ent->checked |= 1; |
164 | 0 | xmlFree(rep); |
165 | 0 | rep = NULL; |
166 | 0 | } |
167 | 0 | } |
168 | | |
169 | | /* |
170 | | * Prevent entity exponential check, not just replacement while |
171 | | * parsing the DTD |
172 | | * The check is potentially costly so do that only once in a thousand |
173 | | */ |
174 | 0 | if ((ctxt->instate == XML_PARSER_DTD) && (ctxt->nbentities > 10000) && |
175 | 0 | (ctxt->nbentities % 1024 == 0)) { |
176 | 0 | for (i = 0;i < ctxt->inputNr;i++) { |
177 | 0 | consumed += ctxt->inputTab[i]->consumed + |
178 | 0 | (ctxt->inputTab[i]->cur - ctxt->inputTab[i]->base); |
179 | 0 | } |
180 | 0 | if (ctxt->nbentities > consumed * XML_PARSER_NON_LINEAR) { |
181 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
182 | 0 | ctxt->instate = XML_PARSER_EOF; |
183 | 0 | return (1); |
184 | 0 | } |
185 | 0 | consumed = 0; |
186 | 0 | } |
187 | | |
188 | | |
189 | | |
190 | 0 | if (replacement != 0) { |
191 | 0 | if (replacement < XML_MAX_TEXT_LENGTH) |
192 | 0 | return(0); |
193 | | |
194 | | /* |
195 | | * If the volume of entity copy reaches 10 times the |
196 | | * amount of parsed data and over the large text threshold |
197 | | * then that's very likely to be an abuse. |
198 | | */ |
199 | 0 | if (ctxt->input != NULL) { |
200 | 0 | consumed = ctxt->input->consumed + |
201 | 0 | (ctxt->input->cur - ctxt->input->base); |
202 | 0 | } |
203 | 0 | consumed += ctxt->sizeentities; |
204 | |
|
205 | 0 | if (replacement < XML_PARSER_NON_LINEAR * consumed) |
206 | 0 | return(0); |
207 | 0 | } else if (size != 0) { |
208 | | /* |
209 | | * Do the check based on the replacement size of the entity |
210 | | */ |
211 | 0 | if (size < XML_PARSER_BIG_ENTITY) |
212 | 0 | return(0); |
213 | | |
214 | | /* |
215 | | * A limit on the amount of text data reasonably used |
216 | | */ |
217 | 0 | if (ctxt->input != NULL) { |
218 | 0 | consumed = ctxt->input->consumed + |
219 | 0 | (ctxt->input->cur - ctxt->input->base); |
220 | 0 | } |
221 | 0 | consumed += ctxt->sizeentities; |
222 | |
|
223 | 0 | if ((size < XML_PARSER_NON_LINEAR * consumed) && |
224 | 0 | (ctxt->nbentities * 3 < XML_PARSER_NON_LINEAR * consumed)) |
225 | 0 | return (0); |
226 | 0 | } else if (ent != NULL) { |
227 | | /* |
228 | | * use the number of parsed entities in the replacement |
229 | | */ |
230 | 0 | size = ent->checked / 2; |
231 | | |
232 | | /* |
233 | | * The amount of data parsed counting entities size only once |
234 | | */ |
235 | 0 | if (ctxt->input != NULL) { |
236 | 0 | consumed = ctxt->input->consumed + |
237 | 0 | (ctxt->input->cur - ctxt->input->base); |
238 | 0 | } |
239 | 0 | consumed += ctxt->sizeentities; |
240 | | |
241 | | /* |
242 | | * Check the density of entities for the amount of data |
243 | | * knowing an entity reference will take at least 3 bytes |
244 | | */ |
245 | 0 | if (size * 3 < consumed * XML_PARSER_NON_LINEAR) |
246 | 0 | return (0); |
247 | 0 | } else { |
248 | | /* |
249 | | * strange we got no data for checking |
250 | | */ |
251 | 0 | if (((ctxt->lastError.code != XML_ERR_UNDECLARED_ENTITY) && |
252 | 0 | (ctxt->lastError.code != XML_WAR_UNDECLARED_ENTITY)) || |
253 | 0 | (ctxt->nbentities <= 10000)) |
254 | 0 | return (0); |
255 | 0 | } |
256 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
257 | 0 | return (1); |
258 | 0 | } |
259 | | |
260 | | /** |
261 | | * xmlParserMaxDepth: |
262 | | * |
263 | | * arbitrary depth limit for the XML documents that we allow to |
264 | | * process. This is not a limitation of the parser but a safety |
265 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
266 | | * parser option. |
267 | | */ |
268 | | unsigned int xmlParserMaxDepth = 256; |
269 | | |
270 | | |
271 | | |
272 | | #define SAX2 1 |
273 | 235M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
274 | 2.04G | #define XML_PARSER_BUFFER_SIZE 100 |
275 | 49.4k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
276 | | |
277 | | /** |
278 | | * XML_PARSER_CHUNK_SIZE |
279 | | * |
280 | | * When calling GROW that's the minimal amount of data |
281 | | * the parser expected to have received. It is not a hard |
282 | | * limit but an optimization when reading strings like Names |
283 | | * It is not strictly needed as long as inputs available characters |
284 | | * are followed by 0, which should be provided by the I/O level |
285 | | */ |
286 | 195M | #define XML_PARSER_CHUNK_SIZE 100 |
287 | | |
288 | | /* |
289 | | * List of XML prefixed PI allowed by W3C specs |
290 | | */ |
291 | | |
292 | | static const char* const xmlW3CPIs[] = { |
293 | | "xml-stylesheet", |
294 | | "xml-model", |
295 | | NULL |
296 | | }; |
297 | | |
298 | | |
299 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
300 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
301 | | const xmlChar **str); |
302 | | |
303 | | static xmlParserErrors |
304 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
305 | | xmlSAXHandlerPtr sax, |
306 | | void *user_data, int depth, const xmlChar *URL, |
307 | | const xmlChar *ID, xmlNodePtr *list); |
308 | | |
309 | | static int |
310 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, |
311 | | const char *encoding); |
312 | | #ifdef LIBXML_LEGACY_ENABLED |
313 | | static void |
314 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
315 | | xmlNodePtr lastNode); |
316 | | #endif /* LIBXML_LEGACY_ENABLED */ |
317 | | |
318 | | static xmlParserErrors |
319 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
320 | | const xmlChar *string, void *user_data, xmlNodePtr *lst); |
321 | | |
322 | | static int |
323 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
324 | | |
325 | | /************************************************************************ |
326 | | * * |
327 | | * Some factorized error routines * |
328 | | * * |
329 | | ************************************************************************/ |
330 | | |
331 | | /** |
332 | | * xmlErrAttributeDup: |
333 | | * @ctxt: an XML parser context |
334 | | * @prefix: the attribute prefix |
335 | | * @localname: the attribute localname |
336 | | * |
337 | | * Handle a redefinition of attribute error |
338 | | */ |
339 | | static void |
340 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
341 | | const xmlChar * localname) |
342 | 10.4k | { |
343 | 10.4k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
344 | 10.4k | (ctxt->instate == XML_PARSER_EOF)) |
345 | 0 | return; |
346 | 10.4k | if (ctxt != NULL) |
347 | 10.4k | ctxt->errNo = XML_ERR_ATTRIBUTE_REDEFINED; |
348 | | |
349 | 10.4k | if (prefix == NULL) |
350 | 8.67k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
351 | 8.67k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
352 | 8.67k | (const char *) localname, NULL, NULL, 0, 0, |
353 | 8.67k | "Attribute %s redefined\n", localname); |
354 | 1.81k | else |
355 | 1.81k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
356 | 1.81k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
357 | 1.81k | (const char *) prefix, (const char *) localname, |
358 | 1.81k | NULL, 0, 0, "Attribute %s:%s redefined\n", prefix, |
359 | 1.81k | localname); |
360 | 10.4k | if (ctxt != NULL) { |
361 | 10.4k | ctxt->wellFormed = 0; |
362 | 10.4k | if (ctxt->recovery == 0) |
363 | 10.4k | ctxt->disableSAX = 1; |
364 | 10.4k | } |
365 | 10.4k | } |
366 | | |
367 | | /** |
368 | | * xmlFatalErr: |
369 | | * @ctxt: an XML parser context |
370 | | * @error: the error number |
371 | | * @extra: extra information string |
372 | | * |
373 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
374 | | */ |
375 | | static void |
376 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info) |
377 | 7.23M | { |
378 | 7.23M | const char *errmsg; |
379 | | |
380 | 7.23M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
381 | 7.23M | (ctxt->instate == XML_PARSER_EOF)) |
382 | 132k | return; |
383 | 7.10M | switch (error) { |
384 | 7.97k | case XML_ERR_INVALID_HEX_CHARREF: |
385 | 7.97k | errmsg = "CharRef: invalid hexadecimal value"; |
386 | 7.97k | break; |
387 | 29.5k | case XML_ERR_INVALID_DEC_CHARREF: |
388 | 29.5k | errmsg = "CharRef: invalid decimal value"; |
389 | 29.5k | break; |
390 | 0 | case XML_ERR_INVALID_CHARREF: |
391 | 0 | errmsg = "CharRef: invalid value"; |
392 | 0 | break; |
393 | 1.32M | case XML_ERR_INTERNAL_ERROR: |
394 | 1.32M | errmsg = "internal error"; |
395 | 1.32M | break; |
396 | 0 | case XML_ERR_PEREF_AT_EOF: |
397 | 0 | errmsg = "PEReference at end of document"; |
398 | 0 | break; |
399 | 0 | case XML_ERR_PEREF_IN_PROLOG: |
400 | 0 | errmsg = "PEReference in prolog"; |
401 | 0 | break; |
402 | 0 | case XML_ERR_PEREF_IN_EPILOG: |
403 | 0 | errmsg = "PEReference in epilog"; |
404 | 0 | break; |
405 | 0 | case XML_ERR_PEREF_NO_NAME: |
406 | 0 | errmsg = "PEReference: no name"; |
407 | 0 | break; |
408 | 4.56k | case XML_ERR_PEREF_SEMICOL_MISSING: |
409 | 4.56k | errmsg = "PEReference: expecting ';'"; |
410 | 4.56k | break; |
411 | 237k | case XML_ERR_ENTITY_LOOP: |
412 | 237k | errmsg = "Detected an entity reference loop"; |
413 | 237k | break; |
414 | 0 | case XML_ERR_ENTITY_NOT_STARTED: |
415 | 0 | errmsg = "EntityValue: \" or ' expected"; |
416 | 0 | break; |
417 | 1.02k | case XML_ERR_ENTITY_PE_INTERNAL: |
418 | 1.02k | errmsg = "PEReferences forbidden in internal subset"; |
419 | 1.02k | break; |
420 | 175 | case XML_ERR_ENTITY_NOT_FINISHED: |
421 | 175 | errmsg = "EntityValue: \" or ' expected"; |
422 | 175 | break; |
423 | 764k | case XML_ERR_ATTRIBUTE_NOT_STARTED: |
424 | 764k | errmsg = "AttValue: \" or ' expected"; |
425 | 764k | break; |
426 | 18.1k | case XML_ERR_LT_IN_ATTRIBUTE: |
427 | 18.1k | errmsg = "Unescaped '<' not allowed in attributes values"; |
428 | 18.1k | break; |
429 | 5.00k | case XML_ERR_LITERAL_NOT_STARTED: |
430 | 5.00k | errmsg = "SystemLiteral \" or ' expected"; |
431 | 5.00k | break; |
432 | 48.0k | case XML_ERR_LITERAL_NOT_FINISHED: |
433 | 48.0k | errmsg = "Unfinished System or Public ID \" or ' expected"; |
434 | 48.0k | break; |
435 | 4.59k | case XML_ERR_MISPLACED_CDATA_END: |
436 | 4.59k | errmsg = "Sequence ']]>' not allowed in content"; |
437 | 4.59k | break; |
438 | 2.86k | case XML_ERR_URI_REQUIRED: |
439 | 2.86k | errmsg = "SYSTEM or PUBLIC, the URI is missing"; |
440 | 2.86k | break; |
441 | 2.14k | case XML_ERR_PUBID_REQUIRED: |
442 | 2.14k | errmsg = "PUBLIC, the Public Identifier is missing"; |
443 | 2.14k | break; |
444 | 55.2k | case XML_ERR_HYPHEN_IN_COMMENT: |
445 | 55.2k | errmsg = "Comment must not contain '--' (double-hyphen)"; |
446 | 55.2k | break; |
447 | 746k | case XML_ERR_PI_NOT_STARTED: |
448 | 746k | errmsg = "xmlParsePI : no target name"; |
449 | 746k | break; |
450 | 2.61k | case XML_ERR_RESERVED_XML_NAME: |
451 | 2.61k | errmsg = "Invalid PI name"; |
452 | 2.61k | break; |
453 | 5.59k | case XML_ERR_NOTATION_NOT_STARTED: |
454 | 5.59k | errmsg = "NOTATION: Name expected here"; |
455 | 5.59k | break; |
456 | 66.1k | case XML_ERR_NOTATION_NOT_FINISHED: |
457 | 66.1k | errmsg = "'>' required to close NOTATION declaration"; |
458 | 66.1k | break; |
459 | 1.71M | case XML_ERR_VALUE_REQUIRED: |
460 | 1.71M | errmsg = "Entity value required"; |
461 | 1.71M | break; |
462 | 1.06k | case XML_ERR_URI_FRAGMENT: |
463 | 1.06k | errmsg = "Fragment not allowed"; |
464 | 1.06k | break; |
465 | 36.1k | case XML_ERR_ATTLIST_NOT_STARTED: |
466 | 36.1k | errmsg = "'(' required to start ATTLIST enumeration"; |
467 | 36.1k | break; |
468 | 781 | case XML_ERR_NMTOKEN_REQUIRED: |
469 | 781 | errmsg = "NmToken expected in ATTLIST enumeration"; |
470 | 781 | break; |
471 | 15.3k | case XML_ERR_ATTLIST_NOT_FINISHED: |
472 | 15.3k | errmsg = "')' required to finish ATTLIST enumeration"; |
473 | 15.3k | break; |
474 | 17.2k | case XML_ERR_MIXED_NOT_STARTED: |
475 | 17.2k | errmsg = "MixedContentDecl : '|' or ')*' expected"; |
476 | 17.2k | break; |
477 | 0 | case XML_ERR_PCDATA_REQUIRED: |
478 | 0 | errmsg = "MixedContentDecl : '#PCDATA' expected"; |
479 | 0 | break; |
480 | 15.8k | case XML_ERR_ELEMCONTENT_NOT_STARTED: |
481 | 15.8k | errmsg = "ContentDecl : Name or '(' expected"; |
482 | 15.8k | break; |
483 | 911k | case XML_ERR_ELEMCONTENT_NOT_FINISHED: |
484 | 911k | errmsg = "ContentDecl : ',' '|' or ')' expected"; |
485 | 911k | break; |
486 | 0 | case XML_ERR_PEREF_IN_INT_SUBSET: |
487 | 0 | errmsg = |
488 | 0 | "PEReference: forbidden within markup decl in internal subset"; |
489 | 0 | break; |
490 | 960k | case XML_ERR_GT_REQUIRED: |
491 | 960k | errmsg = "expected '>'"; |
492 | 960k | break; |
493 | 0 | case XML_ERR_CONDSEC_INVALID: |
494 | 0 | errmsg = "XML conditional section '[' expected"; |
495 | 0 | break; |
496 | 0 | case XML_ERR_EXT_SUBSET_NOT_FINISHED: |
497 | 0 | errmsg = "Content error in the external subset"; |
498 | 0 | break; |
499 | 0 | case XML_ERR_CONDSEC_INVALID_KEYWORD: |
500 | 0 | errmsg = |
501 | 0 | "conditional section INCLUDE or IGNORE keyword expected"; |
502 | 0 | break; |
503 | 0 | case XML_ERR_CONDSEC_NOT_FINISHED: |
504 | 0 | errmsg = "XML conditional section not closed"; |
505 | 0 | break; |
506 | 0 | case XML_ERR_XMLDECL_NOT_STARTED: |
507 | 0 | errmsg = "Text declaration '<?xml' required"; |
508 | 0 | break; |
509 | 1.73k | case XML_ERR_XMLDECL_NOT_FINISHED: |
510 | 1.73k | errmsg = "parsing XML declaration: '?>' expected"; |
511 | 1.73k | break; |
512 | 0 | case XML_ERR_EXT_ENTITY_STANDALONE: |
513 | 0 | errmsg = "external parsed entities cannot be standalone"; |
514 | 0 | break; |
515 | 38.2k | case XML_ERR_ENTITYREF_SEMICOL_MISSING: |
516 | 38.2k | errmsg = "EntityRef: expecting ';'"; |
517 | 38.2k | break; |
518 | 6.37k | case XML_ERR_DOCTYPE_NOT_FINISHED: |
519 | 6.37k | errmsg = "DOCTYPE improperly terminated"; |
520 | 6.37k | break; |
521 | 0 | case XML_ERR_LTSLASH_REQUIRED: |
522 | 0 | errmsg = "EndTag: '</' not found"; |
523 | 0 | break; |
524 | 23 | case XML_ERR_EQUAL_REQUIRED: |
525 | 23 | errmsg = "expected '='"; |
526 | 23 | break; |
527 | 190 | case XML_ERR_STRING_NOT_CLOSED: |
528 | 190 | errmsg = "String not closed expecting \" or '"; |
529 | 190 | break; |
530 | 47 | case XML_ERR_STRING_NOT_STARTED: |
531 | 47 | errmsg = "String not started expecting ' or \""; |
532 | 47 | break; |
533 | 5 | case XML_ERR_ENCODING_NAME: |
534 | 5 | errmsg = "Invalid XML encoding name"; |
535 | 5 | break; |
536 | 22 | case XML_ERR_STANDALONE_VALUE: |
537 | 22 | errmsg = "standalone accepts only 'yes' or 'no'"; |
538 | 22 | break; |
539 | 44 | case XML_ERR_DOCUMENT_EMPTY: |
540 | 44 | errmsg = "Document is empty"; |
541 | 44 | break; |
542 | 1.27k | case XML_ERR_DOCUMENT_END: |
543 | 1.27k | errmsg = "Extra content at the end of the document"; |
544 | 1.27k | break; |
545 | 48.9k | case XML_ERR_NOT_WELL_BALANCED: |
546 | 48.9k | errmsg = "chunk is not well balanced"; |
547 | 48.9k | break; |
548 | 0 | case XML_ERR_EXTRA_CONTENT: |
549 | 0 | errmsg = "extra content at the end of well balanced chunk"; |
550 | 0 | break; |
551 | 2.09k | case XML_ERR_VERSION_MISSING: |
552 | 2.09k | errmsg = "Malformed declaration expecting version"; |
553 | 2.09k | break; |
554 | 0 | case XML_ERR_NAME_TOO_LONG: |
555 | 0 | errmsg = "Name too long use XML_PARSE_HUGE option"; |
556 | 0 | break; |
557 | | #if 0 |
558 | | case: |
559 | | errmsg = ""; |
560 | | break; |
561 | | #endif |
562 | 3.44k | default: |
563 | 3.44k | errmsg = "Unregistered error message"; |
564 | 7.10M | } |
565 | 7.10M | if (ctxt != NULL) |
566 | 7.10M | ctxt->errNo = error; |
567 | 7.10M | if (info == NULL) { |
568 | 5.78M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
569 | 5.78M | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s\n", |
570 | 5.78M | errmsg); |
571 | 5.78M | } else { |
572 | 1.32M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
573 | 1.32M | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s: %s\n", |
574 | 1.32M | errmsg, info); |
575 | 1.32M | } |
576 | 7.10M | if (ctxt != NULL) { |
577 | 7.10M | ctxt->wellFormed = 0; |
578 | 7.10M | if (ctxt->recovery == 0) |
579 | 7.10M | ctxt->disableSAX = 1; |
580 | 7.10M | } |
581 | 7.10M | } |
582 | | |
583 | | /** |
584 | | * xmlFatalErrMsg: |
585 | | * @ctxt: an XML parser context |
586 | | * @error: the error number |
587 | | * @msg: the error message |
588 | | * |
589 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
590 | | */ |
591 | | static void LIBXML_ATTR_FORMAT(3,0) |
592 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
593 | | const char *msg) |
594 | 10.2M | { |
595 | 10.2M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
596 | 10.2M | (ctxt->instate == XML_PARSER_EOF)) |
597 | 0 | return; |
598 | 10.2M | if (ctxt != NULL) |
599 | 10.2M | ctxt->errNo = error; |
600 | 10.2M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
601 | 10.2M | XML_ERR_FATAL, NULL, 0, NULL, NULL, NULL, 0, 0, "%s", msg); |
602 | 10.2M | if (ctxt != NULL) { |
603 | 10.2M | ctxt->wellFormed = 0; |
604 | 10.2M | if (ctxt->recovery == 0) |
605 | 10.2M | ctxt->disableSAX = 1; |
606 | 10.2M | } |
607 | 10.2M | } |
608 | | |
609 | | /** |
610 | | * xmlWarningMsg: |
611 | | * @ctxt: an XML parser context |
612 | | * @error: the error number |
613 | | * @msg: the error message |
614 | | * @str1: extra data |
615 | | * @str2: extra data |
616 | | * |
617 | | * Handle a warning. |
618 | | */ |
619 | | static void LIBXML_ATTR_FORMAT(3,0) |
620 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
621 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
622 | 19.9k | { |
623 | 19.9k | xmlStructuredErrorFunc schannel = NULL; |
624 | | |
625 | 19.9k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
626 | 19.9k | (ctxt->instate == XML_PARSER_EOF)) |
627 | 0 | return; |
628 | 19.9k | if ((ctxt != NULL) && (ctxt->sax != NULL) && |
629 | 19.9k | (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
630 | 19.9k | schannel = ctxt->sax->serror; |
631 | 19.9k | if (ctxt != NULL) { |
632 | 19.9k | __xmlRaiseError(schannel, |
633 | 19.9k | (ctxt->sax) ? ctxt->sax->warning : NULL, |
634 | 19.9k | ctxt->userData, |
635 | 19.9k | ctxt, NULL, XML_FROM_PARSER, error, |
636 | 19.9k | XML_ERR_WARNING, NULL, 0, |
637 | 19.9k | (const char *) str1, (const char *) str2, NULL, 0, 0, |
638 | 19.9k | msg, (const char *) str1, (const char *) str2); |
639 | 19.9k | } else { |
640 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
641 | 0 | ctxt, NULL, XML_FROM_PARSER, error, |
642 | 0 | XML_ERR_WARNING, NULL, 0, |
643 | 0 | (const char *) str1, (const char *) str2, NULL, 0, 0, |
644 | 0 | msg, (const char *) str1, (const char *) str2); |
645 | 0 | } |
646 | 19.9k | } |
647 | | |
648 | | /** |
649 | | * xmlValidityError: |
650 | | * @ctxt: an XML parser context |
651 | | * @error: the error number |
652 | | * @msg: the error message |
653 | | * @str1: extra data |
654 | | * |
655 | | * Handle a validity error. |
656 | | */ |
657 | | static void LIBXML_ATTR_FORMAT(3,0) |
658 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
659 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
660 | 1.36k | { |
661 | 1.36k | xmlStructuredErrorFunc schannel = NULL; |
662 | | |
663 | 1.36k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
664 | 1.36k | (ctxt->instate == XML_PARSER_EOF)) |
665 | 0 | return; |
666 | 1.36k | if (ctxt != NULL) { |
667 | 1.36k | ctxt->errNo = error; |
668 | 1.36k | if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
669 | 1.36k | schannel = ctxt->sax->serror; |
670 | 1.36k | } |
671 | 1.36k | if (ctxt != NULL) { |
672 | 1.36k | __xmlRaiseError(schannel, |
673 | 1.36k | ctxt->vctxt.error, ctxt->vctxt.userData, |
674 | 1.36k | ctxt, NULL, XML_FROM_DTD, error, |
675 | 1.36k | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
676 | 1.36k | (const char *) str2, NULL, 0, 0, |
677 | 1.36k | msg, (const char *) str1, (const char *) str2); |
678 | 1.36k | ctxt->valid = 0; |
679 | 1.36k | } else { |
680 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
681 | 0 | ctxt, NULL, XML_FROM_DTD, error, |
682 | 0 | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
683 | 0 | (const char *) str2, NULL, 0, 0, |
684 | 0 | msg, (const char *) str1, (const char *) str2); |
685 | 0 | } |
686 | 1.36k | } |
687 | | |
688 | | /** |
689 | | * xmlFatalErrMsgInt: |
690 | | * @ctxt: an XML parser context |
691 | | * @error: the error number |
692 | | * @msg: the error message |
693 | | * @val: an integer value |
694 | | * |
695 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
696 | | */ |
697 | | static void LIBXML_ATTR_FORMAT(3,0) |
698 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
699 | | const char *msg, int val) |
700 | 145k | { |
701 | 145k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
702 | 145k | (ctxt->instate == XML_PARSER_EOF)) |
703 | 0 | return; |
704 | 145k | if (ctxt != NULL) |
705 | 145k | ctxt->errNo = error; |
706 | 145k | __xmlRaiseError(NULL, NULL, NULL, |
707 | 145k | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
708 | 145k | NULL, 0, NULL, NULL, NULL, val, 0, msg, val); |
709 | 145k | if (ctxt != NULL) { |
710 | 145k | ctxt->wellFormed = 0; |
711 | 145k | if (ctxt->recovery == 0) |
712 | 145k | ctxt->disableSAX = 1; |
713 | 145k | } |
714 | 145k | } |
715 | | |
716 | | /** |
717 | | * xmlFatalErrMsgStrIntStr: |
718 | | * @ctxt: an XML parser context |
719 | | * @error: the error number |
720 | | * @msg: the error message |
721 | | * @str1: an string info |
722 | | * @val: an integer value |
723 | | * @str2: an string info |
724 | | * |
725 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
726 | | */ |
727 | | static void LIBXML_ATTR_FORMAT(3,0) |
728 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
729 | | const char *msg, const xmlChar *str1, int val, |
730 | | const xmlChar *str2) |
731 | 278k | { |
732 | 278k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
733 | 278k | (ctxt->instate == XML_PARSER_EOF)) |
734 | 0 | return; |
735 | 278k | if (ctxt != NULL) |
736 | 278k | ctxt->errNo = error; |
737 | 278k | __xmlRaiseError(NULL, NULL, NULL, |
738 | 278k | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
739 | 278k | NULL, 0, (const char *) str1, (const char *) str2, |
740 | 278k | NULL, val, 0, msg, str1, val, str2); |
741 | 278k | if (ctxt != NULL) { |
742 | 278k | ctxt->wellFormed = 0; |
743 | 278k | if (ctxt->recovery == 0) |
744 | 278k | ctxt->disableSAX = 1; |
745 | 278k | } |
746 | 278k | } |
747 | | |
748 | | /** |
749 | | * xmlFatalErrMsgStr: |
750 | | * @ctxt: an XML parser context |
751 | | * @error: the error number |
752 | | * @msg: the error message |
753 | | * @val: a string value |
754 | | * |
755 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
756 | | */ |
757 | | static void LIBXML_ATTR_FORMAT(3,0) |
758 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
759 | | const char *msg, const xmlChar * val) |
760 | 920k | { |
761 | 920k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
762 | 920k | (ctxt->instate == XML_PARSER_EOF)) |
763 | 0 | return; |
764 | 920k | if (ctxt != NULL) |
765 | 920k | ctxt->errNo = error; |
766 | 920k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
767 | 920k | XML_FROM_PARSER, error, XML_ERR_FATAL, |
768 | 920k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
769 | 920k | val); |
770 | 920k | if (ctxt != NULL) { |
771 | 920k | ctxt->wellFormed = 0; |
772 | 920k | if (ctxt->recovery == 0) |
773 | 920k | ctxt->disableSAX = 1; |
774 | 920k | } |
775 | 920k | } |
776 | | |
777 | | /** |
778 | | * xmlErrMsgStr: |
779 | | * @ctxt: an XML parser context |
780 | | * @error: the error number |
781 | | * @msg: the error message |
782 | | * @val: a string value |
783 | | * |
784 | | * Handle a non fatal parser error |
785 | | */ |
786 | | static void LIBXML_ATTR_FORMAT(3,0) |
787 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
788 | | const char *msg, const xmlChar * val) |
789 | 23.1k | { |
790 | 23.1k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
791 | 23.1k | (ctxt->instate == XML_PARSER_EOF)) |
792 | 0 | return; |
793 | 23.1k | if (ctxt != NULL) |
794 | 23.1k | ctxt->errNo = error; |
795 | 23.1k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
796 | 23.1k | XML_FROM_PARSER, error, XML_ERR_ERROR, |
797 | 23.1k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
798 | 23.1k | val); |
799 | 23.1k | } |
800 | | |
801 | | /** |
802 | | * xmlNsErr: |
803 | | * @ctxt: an XML parser context |
804 | | * @error: the error number |
805 | | * @msg: the message |
806 | | * @info1: extra information string |
807 | | * @info2: extra information string |
808 | | * |
809 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
810 | | */ |
811 | | static void LIBXML_ATTR_FORMAT(3,0) |
812 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
813 | | const char *msg, |
814 | | const xmlChar * info1, const xmlChar * info2, |
815 | | const xmlChar * info3) |
816 | 195k | { |
817 | 195k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
818 | 195k | (ctxt->instate == XML_PARSER_EOF)) |
819 | 0 | return; |
820 | 195k | if (ctxt != NULL) |
821 | 195k | ctxt->errNo = error; |
822 | 195k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
823 | 195k | XML_ERR_ERROR, NULL, 0, (const char *) info1, |
824 | 195k | (const char *) info2, (const char *) info3, 0, 0, msg, |
825 | 195k | info1, info2, info3); |
826 | 195k | if (ctxt != NULL) |
827 | 195k | ctxt->nsWellFormed = 0; |
828 | 195k | } |
829 | | |
830 | | /** |
831 | | * xmlNsWarn |
832 | | * @ctxt: an XML parser context |
833 | | * @error: the error number |
834 | | * @msg: the message |
835 | | * @info1: extra information string |
836 | | * @info2: extra information string |
837 | | * |
838 | | * Handle a namespace warning error |
839 | | */ |
840 | | static void LIBXML_ATTR_FORMAT(3,0) |
841 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
842 | | const char *msg, |
843 | | const xmlChar * info1, const xmlChar * info2, |
844 | | const xmlChar * info3) |
845 | 20.3k | { |
846 | 20.3k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
847 | 20.3k | (ctxt->instate == XML_PARSER_EOF)) |
848 | 0 | return; |
849 | 20.3k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
850 | 20.3k | XML_ERR_WARNING, NULL, 0, (const char *) info1, |
851 | 20.3k | (const char *) info2, (const char *) info3, 0, 0, msg, |
852 | 20.3k | info1, info2, info3); |
853 | 20.3k | } |
854 | | |
855 | | /************************************************************************ |
856 | | * * |
857 | | * Library wide options * |
858 | | * * |
859 | | ************************************************************************/ |
860 | | |
861 | | /** |
862 | | * xmlHasFeature: |
863 | | * @feature: the feature to be examined |
864 | | * |
865 | | * Examines if the library has been compiled with a given feature. |
866 | | * |
867 | | * Returns a non-zero value if the feature exist, otherwise zero. |
868 | | * Returns zero (0) if the feature does not exist or an unknown |
869 | | * unknown feature is requested, non-zero otherwise. |
870 | | */ |
871 | | int |
872 | | xmlHasFeature(xmlFeature feature) |
873 | 0 | { |
874 | 0 | switch (feature) { |
875 | 0 | case XML_WITH_THREAD: |
876 | 0 | #ifdef LIBXML_THREAD_ENABLED |
877 | 0 | return(1); |
878 | | #else |
879 | | return(0); |
880 | | #endif |
881 | 0 | case XML_WITH_TREE: |
882 | 0 | #ifdef LIBXML_TREE_ENABLED |
883 | 0 | return(1); |
884 | | #else |
885 | | return(0); |
886 | | #endif |
887 | 0 | case XML_WITH_OUTPUT: |
888 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
889 | 0 | return(1); |
890 | | #else |
891 | | return(0); |
892 | | #endif |
893 | 0 | case XML_WITH_PUSH: |
894 | 0 | #ifdef LIBXML_PUSH_ENABLED |
895 | 0 | return(1); |
896 | | #else |
897 | | return(0); |
898 | | #endif |
899 | 0 | case XML_WITH_READER: |
900 | 0 | #ifdef LIBXML_READER_ENABLED |
901 | 0 | return(1); |
902 | | #else |
903 | | return(0); |
904 | | #endif |
905 | 0 | case XML_WITH_PATTERN: |
906 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
907 | 0 | return(1); |
908 | | #else |
909 | | return(0); |
910 | | #endif |
911 | 0 | case XML_WITH_WRITER: |
912 | 0 | #ifdef LIBXML_WRITER_ENABLED |
913 | 0 | return(1); |
914 | | #else |
915 | | return(0); |
916 | | #endif |
917 | 0 | case XML_WITH_SAX1: |
918 | 0 | #ifdef LIBXML_SAX1_ENABLED |
919 | 0 | return(1); |
920 | | #else |
921 | | return(0); |
922 | | #endif |
923 | 0 | case XML_WITH_FTP: |
924 | | #ifdef LIBXML_FTP_ENABLED |
925 | | return(1); |
926 | | #else |
927 | 0 | return(0); |
928 | 0 | #endif |
929 | 0 | case XML_WITH_HTTP: |
930 | 0 | #ifdef LIBXML_HTTP_ENABLED |
931 | 0 | return(1); |
932 | | #else |
933 | | return(0); |
934 | | #endif |
935 | 0 | case XML_WITH_VALID: |
936 | 0 | #ifdef LIBXML_VALID_ENABLED |
937 | 0 | return(1); |
938 | | #else |
939 | | return(0); |
940 | | #endif |
941 | 0 | case XML_WITH_HTML: |
942 | 0 | #ifdef LIBXML_HTML_ENABLED |
943 | 0 | return(1); |
944 | | #else |
945 | | return(0); |
946 | | #endif |
947 | 0 | case XML_WITH_LEGACY: |
948 | | #ifdef LIBXML_LEGACY_ENABLED |
949 | | return(1); |
950 | | #else |
951 | 0 | return(0); |
952 | 0 | #endif |
953 | 0 | case XML_WITH_C14N: |
954 | 0 | #ifdef LIBXML_C14N_ENABLED |
955 | 0 | return(1); |
956 | | #else |
957 | | return(0); |
958 | | #endif |
959 | 0 | case XML_WITH_CATALOG: |
960 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
961 | 0 | return(1); |
962 | | #else |
963 | | return(0); |
964 | | #endif |
965 | 0 | case XML_WITH_XPATH: |
966 | 0 | #ifdef LIBXML_XPATH_ENABLED |
967 | 0 | return(1); |
968 | | #else |
969 | | return(0); |
970 | | #endif |
971 | 0 | case XML_WITH_XPTR: |
972 | 0 | #ifdef LIBXML_XPTR_ENABLED |
973 | 0 | return(1); |
974 | | #else |
975 | | return(0); |
976 | | #endif |
977 | 0 | case XML_WITH_XINCLUDE: |
978 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
979 | 0 | return(1); |
980 | | #else |
981 | | return(0); |
982 | | #endif |
983 | 0 | case XML_WITH_ICONV: |
984 | 0 | #ifdef LIBXML_ICONV_ENABLED |
985 | 0 | return(1); |
986 | | #else |
987 | | return(0); |
988 | | #endif |
989 | 0 | case XML_WITH_ISO8859X: |
990 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
991 | 0 | return(1); |
992 | | #else |
993 | | return(0); |
994 | | #endif |
995 | 0 | case XML_WITH_UNICODE: |
996 | 0 | #ifdef LIBXML_UNICODE_ENABLED |
997 | 0 | return(1); |
998 | | #else |
999 | | return(0); |
1000 | | #endif |
1001 | 0 | case XML_WITH_REGEXP: |
1002 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
1003 | 0 | return(1); |
1004 | | #else |
1005 | | return(0); |
1006 | | #endif |
1007 | 0 | case XML_WITH_AUTOMATA: |
1008 | 0 | #ifdef LIBXML_AUTOMATA_ENABLED |
1009 | 0 | return(1); |
1010 | | #else |
1011 | | return(0); |
1012 | | #endif |
1013 | 0 | case XML_WITH_EXPR: |
1014 | | #ifdef LIBXML_EXPR_ENABLED |
1015 | | return(1); |
1016 | | #else |
1017 | 0 | return(0); |
1018 | 0 | #endif |
1019 | 0 | case XML_WITH_SCHEMAS: |
1020 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
1021 | 0 | return(1); |
1022 | | #else |
1023 | | return(0); |
1024 | | #endif |
1025 | 0 | case XML_WITH_SCHEMATRON: |
1026 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
1027 | 0 | return(1); |
1028 | | #else |
1029 | | return(0); |
1030 | | #endif |
1031 | 0 | case XML_WITH_MODULES: |
1032 | 0 | #ifdef LIBXML_MODULES_ENABLED |
1033 | 0 | return(1); |
1034 | | #else |
1035 | | return(0); |
1036 | | #endif |
1037 | 0 | case XML_WITH_DEBUG: |
1038 | 0 | #ifdef LIBXML_DEBUG_ENABLED |
1039 | 0 | return(1); |
1040 | | #else |
1041 | | return(0); |
1042 | | #endif |
1043 | 0 | case XML_WITH_DEBUG_MEM: |
1044 | | #ifdef DEBUG_MEMORY_LOCATION |
1045 | | return(1); |
1046 | | #else |
1047 | 0 | return(0); |
1048 | 0 | #endif |
1049 | 0 | case XML_WITH_DEBUG_RUN: |
1050 | | #ifdef LIBXML_DEBUG_RUNTIME |
1051 | | return(1); |
1052 | | #else |
1053 | 0 | return(0); |
1054 | 0 | #endif |
1055 | 0 | case XML_WITH_ZLIB: |
1056 | | #ifdef LIBXML_ZLIB_ENABLED |
1057 | | return(1); |
1058 | | #else |
1059 | 0 | return(0); |
1060 | 0 | #endif |
1061 | 0 | case XML_WITH_LZMA: |
1062 | | #ifdef LIBXML_LZMA_ENABLED |
1063 | | return(1); |
1064 | | #else |
1065 | 0 | return(0); |
1066 | 0 | #endif |
1067 | 0 | case XML_WITH_ICU: |
1068 | | #ifdef LIBXML_ICU_ENABLED |
1069 | | return(1); |
1070 | | #else |
1071 | 0 | return(0); |
1072 | 0 | #endif |
1073 | 0 | default: |
1074 | 0 | break; |
1075 | 0 | } |
1076 | 0 | return(0); |
1077 | 0 | } |
1078 | | |
1079 | | /************************************************************************ |
1080 | | * * |
1081 | | * SAX2 defaulted attributes handling * |
1082 | | * * |
1083 | | ************************************************************************/ |
1084 | | |
1085 | | /** |
1086 | | * xmlDetectSAX2: |
1087 | | * @ctxt: an XML parser context |
1088 | | * |
1089 | | * Do the SAX2 detection and specific initialization |
1090 | | */ |
1091 | | static void |
1092 | 319k | xmlDetectSAX2(xmlParserCtxtPtr ctxt) { |
1093 | 319k | xmlSAXHandlerPtr sax; |
1094 | | |
1095 | | /* Avoid unused variable warning if features are disabled. */ |
1096 | 319k | (void) sax; |
1097 | | |
1098 | 319k | if (ctxt == NULL) return; |
1099 | 319k | sax = ctxt->sax; |
1100 | 319k | #ifdef LIBXML_SAX1_ENABLED |
1101 | 319k | if ((sax) && (sax->initialized == XML_SAX2_MAGIC) && |
1102 | 319k | ((sax->startElementNs != NULL) || |
1103 | 319k | (sax->endElementNs != NULL) || |
1104 | 319k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
1105 | 319k | ctxt->sax2 = 1; |
1106 | | #else |
1107 | | ctxt->sax2 = 1; |
1108 | | #endif /* LIBXML_SAX1_ENABLED */ |
1109 | | |
1110 | 319k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
1111 | 319k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
1112 | 319k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
1113 | 319k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
1114 | 319k | (ctxt->str_xml_ns == NULL)) { |
1115 | 0 | xmlErrMemory(ctxt, NULL); |
1116 | 0 | } |
1117 | 319k | } |
1118 | | |
1119 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
1120 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
1121 | | struct _xmlDefAttrs { |
1122 | | int nbAttrs; /* number of defaulted attributes on that element */ |
1123 | | int maxAttrs; /* the size of the array */ |
1124 | | #if __STDC_VERSION__ >= 199901L |
1125 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
1126 | | const xmlChar *values[]; /* array of localname/prefix/values/external */ |
1127 | | #else |
1128 | | const xmlChar *values[5]; |
1129 | | #endif |
1130 | | }; |
1131 | | |
1132 | | /** |
1133 | | * xmlAttrNormalizeSpace: |
1134 | | * @src: the source string |
1135 | | * @dst: the target string |
1136 | | * |
1137 | | * Normalize the space in non CDATA attribute values: |
1138 | | * If the attribute type is not CDATA, then the XML processor MUST further |
1139 | | * process the normalized attribute value by discarding any leading and |
1140 | | * trailing space (#x20) characters, and by replacing sequences of space |
1141 | | * (#x20) characters by a single space (#x20) character. |
1142 | | * Note that the size of dst need to be at least src, and if one doesn't need |
1143 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
1144 | | * passing src as dst is just fine. |
1145 | | * |
1146 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1147 | | * is needed. |
1148 | | */ |
1149 | | static xmlChar * |
1150 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
1151 | 55.6k | { |
1152 | 55.6k | if ((src == NULL) || (dst == NULL)) |
1153 | 0 | return(NULL); |
1154 | | |
1155 | 58.8k | while (*src == 0x20) src++; |
1156 | 621k | while (*src != 0) { |
1157 | 565k | if (*src == 0x20) { |
1158 | 52.4k | while (*src == 0x20) src++; |
1159 | 22.6k | if (*src != 0) |
1160 | 21.2k | *dst++ = 0x20; |
1161 | 543k | } else { |
1162 | 543k | *dst++ = *src++; |
1163 | 543k | } |
1164 | 565k | } |
1165 | 55.6k | *dst = 0; |
1166 | 55.6k | if (dst == src) |
1167 | 50.7k | return(NULL); |
1168 | 4.91k | return(dst); |
1169 | 55.6k | } |
1170 | | |
1171 | | /** |
1172 | | * xmlAttrNormalizeSpace2: |
1173 | | * @src: the source string |
1174 | | * |
1175 | | * Normalize the space in non CDATA attribute values, a slightly more complex |
1176 | | * front end to avoid allocation problems when running on attribute values |
1177 | | * coming from the input. |
1178 | | * |
1179 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1180 | | * is needed. |
1181 | | */ |
1182 | | static const xmlChar * |
1183 | | xmlAttrNormalizeSpace2(xmlParserCtxtPtr ctxt, xmlChar *src, int *len) |
1184 | 4.47k | { |
1185 | 4.47k | int i; |
1186 | 4.47k | int remove_head = 0; |
1187 | 4.47k | int need_realloc = 0; |
1188 | 4.47k | const xmlChar *cur; |
1189 | | |
1190 | 4.47k | if ((ctxt == NULL) || (src == NULL) || (len == NULL)) |
1191 | 0 | return(NULL); |
1192 | 4.47k | i = *len; |
1193 | 4.47k | if (i <= 0) |
1194 | 739 | return(NULL); |
1195 | | |
1196 | 3.74k | cur = src; |
1197 | 4.51k | while (*cur == 0x20) { |
1198 | 772 | cur++; |
1199 | 772 | remove_head++; |
1200 | 772 | } |
1201 | 202M | while (*cur != 0) { |
1202 | 202M | if (*cur == 0x20) { |
1203 | 1.51k | cur++; |
1204 | 1.51k | if ((*cur == 0x20) || (*cur == 0)) { |
1205 | 441 | need_realloc = 1; |
1206 | 441 | break; |
1207 | 441 | } |
1208 | 1.51k | } else |
1209 | 202M | cur++; |
1210 | 202M | } |
1211 | 3.74k | if (need_realloc) { |
1212 | 441 | xmlChar *ret; |
1213 | | |
1214 | 441 | ret = xmlStrndup(src + remove_head, i - remove_head + 1); |
1215 | 441 | if (ret == NULL) { |
1216 | 0 | xmlErrMemory(ctxt, NULL); |
1217 | 0 | return(NULL); |
1218 | 0 | } |
1219 | 441 | xmlAttrNormalizeSpace(ret, ret); |
1220 | 441 | *len = (int) strlen((const char *)ret); |
1221 | 441 | return(ret); |
1222 | 3.29k | } else if (remove_head) { |
1223 | 565 | *len -= remove_head; |
1224 | 565 | memmove(src, src + remove_head, 1 + *len); |
1225 | 565 | return(src); |
1226 | 565 | } |
1227 | 2.73k | return(NULL); |
1228 | 3.74k | } |
1229 | | |
1230 | | /** |
1231 | | * xmlAddDefAttrs: |
1232 | | * @ctxt: an XML parser context |
1233 | | * @fullname: the element fullname |
1234 | | * @fullattr: the attribute fullname |
1235 | | * @value: the attribute value |
1236 | | * |
1237 | | * Add a defaulted attribute for an element |
1238 | | */ |
1239 | | static void |
1240 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1241 | | const xmlChar *fullname, |
1242 | | const xmlChar *fullattr, |
1243 | 55.2k | const xmlChar *value) { |
1244 | 55.2k | xmlDefAttrsPtr defaults; |
1245 | 55.2k | int len; |
1246 | 55.2k | const xmlChar *name; |
1247 | 55.2k | const xmlChar *prefix; |
1248 | | |
1249 | | /* |
1250 | | * Allows to detect attribute redefinitions |
1251 | | */ |
1252 | 55.2k | if (ctxt->attsSpecial != NULL) { |
1253 | 53.1k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1254 | 37.2k | return; |
1255 | 53.1k | } |
1256 | | |
1257 | 17.9k | if (ctxt->attsDefault == NULL) { |
1258 | 2.07k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1259 | 2.07k | if (ctxt->attsDefault == NULL) |
1260 | 0 | goto mem_error; |
1261 | 2.07k | } |
1262 | | |
1263 | | /* |
1264 | | * split the element name into prefix:localname , the string found |
1265 | | * are within the DTD and then not associated to namespace names. |
1266 | | */ |
1267 | 17.9k | name = xmlSplitQName3(fullname, &len); |
1268 | 17.9k | if (name == NULL) { |
1269 | 13.9k | name = xmlDictLookup(ctxt->dict, fullname, -1); |
1270 | 13.9k | prefix = NULL; |
1271 | 13.9k | } else { |
1272 | 4.07k | name = xmlDictLookup(ctxt->dict, name, -1); |
1273 | 4.07k | prefix = xmlDictLookup(ctxt->dict, fullname, len); |
1274 | 4.07k | } |
1275 | | |
1276 | | /* |
1277 | | * make sure there is some storage |
1278 | | */ |
1279 | 17.9k | defaults = xmlHashLookup2(ctxt->attsDefault, name, prefix); |
1280 | 17.9k | if (defaults == NULL) { |
1281 | 3.45k | defaults = (xmlDefAttrsPtr) xmlMalloc(sizeof(xmlDefAttrs) + |
1282 | 3.45k | (4 * 5) * sizeof(const xmlChar *)); |
1283 | 3.45k | if (defaults == NULL) |
1284 | 0 | goto mem_error; |
1285 | 3.45k | defaults->nbAttrs = 0; |
1286 | 3.45k | defaults->maxAttrs = 4; |
1287 | 3.45k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1288 | 3.45k | defaults, NULL) < 0) { |
1289 | 0 | xmlFree(defaults); |
1290 | 0 | goto mem_error; |
1291 | 0 | } |
1292 | 14.5k | } else if (defaults->nbAttrs >= defaults->maxAttrs) { |
1293 | 2.30k | xmlDefAttrsPtr temp; |
1294 | | |
1295 | 2.30k | temp = (xmlDefAttrsPtr) xmlRealloc(defaults, sizeof(xmlDefAttrs) + |
1296 | 2.30k | (2 * defaults->maxAttrs * 5) * sizeof(const xmlChar *)); |
1297 | 2.30k | if (temp == NULL) |
1298 | 0 | goto mem_error; |
1299 | 2.30k | defaults = temp; |
1300 | 2.30k | defaults->maxAttrs *= 2; |
1301 | 2.30k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1302 | 2.30k | defaults, NULL) < 0) { |
1303 | 0 | xmlFree(defaults); |
1304 | 0 | goto mem_error; |
1305 | 0 | } |
1306 | 2.30k | } |
1307 | | |
1308 | | /* |
1309 | | * Split the element name into prefix:localname , the string found |
1310 | | * are within the DTD and hen not associated to namespace names. |
1311 | | */ |
1312 | 17.9k | name = xmlSplitQName3(fullattr, &len); |
1313 | 17.9k | if (name == NULL) { |
1314 | 14.4k | name = xmlDictLookup(ctxt->dict, fullattr, -1); |
1315 | 14.4k | prefix = NULL; |
1316 | 14.4k | } else { |
1317 | 3.51k | name = xmlDictLookup(ctxt->dict, name, -1); |
1318 | 3.51k | prefix = xmlDictLookup(ctxt->dict, fullattr, len); |
1319 | 3.51k | } |
1320 | | |
1321 | 17.9k | defaults->values[5 * defaults->nbAttrs] = name; |
1322 | 17.9k | defaults->values[5 * defaults->nbAttrs + 1] = prefix; |
1323 | | /* intern the string and precompute the end */ |
1324 | 17.9k | len = xmlStrlen(value); |
1325 | 17.9k | value = xmlDictLookup(ctxt->dict, value, len); |
1326 | 17.9k | defaults->values[5 * defaults->nbAttrs + 2] = value; |
1327 | 17.9k | defaults->values[5 * defaults->nbAttrs + 3] = value + len; |
1328 | 17.9k | if (ctxt->external) |
1329 | 0 | defaults->values[5 * defaults->nbAttrs + 4] = BAD_CAST "external"; |
1330 | 17.9k | else |
1331 | 17.9k | defaults->values[5 * defaults->nbAttrs + 4] = NULL; |
1332 | 17.9k | defaults->nbAttrs++; |
1333 | | |
1334 | 17.9k | return; |
1335 | | |
1336 | 0 | mem_error: |
1337 | 0 | xmlErrMemory(ctxt, NULL); |
1338 | 0 | return; |
1339 | 17.9k | } |
1340 | | |
1341 | | /** |
1342 | | * xmlAddSpecialAttr: |
1343 | | * @ctxt: an XML parser context |
1344 | | * @fullname: the element fullname |
1345 | | * @fullattr: the attribute fullname |
1346 | | * @type: the attribute type |
1347 | | * |
1348 | | * Register this attribute type |
1349 | | */ |
1350 | | static void |
1351 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1352 | | const xmlChar *fullname, |
1353 | | const xmlChar *fullattr, |
1354 | | int type) |
1355 | 58.1k | { |
1356 | 58.1k | if (ctxt->attsSpecial == NULL) { |
1357 | 2.33k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1358 | 2.33k | if (ctxt->attsSpecial == NULL) |
1359 | 0 | goto mem_error; |
1360 | 2.33k | } |
1361 | | |
1362 | 58.1k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1363 | 39.8k | return; |
1364 | | |
1365 | 18.3k | xmlHashAddEntry2(ctxt->attsSpecial, fullname, fullattr, |
1366 | 18.3k | (void *) (ptrdiff_t) type); |
1367 | 18.3k | return; |
1368 | | |
1369 | 0 | mem_error: |
1370 | 0 | xmlErrMemory(ctxt, NULL); |
1371 | 0 | return; |
1372 | 58.1k | } |
1373 | | |
1374 | | /** |
1375 | | * xmlCleanSpecialAttrCallback: |
1376 | | * |
1377 | | * Removes CDATA attributes from the special attribute table |
1378 | | */ |
1379 | | static void |
1380 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1381 | | const xmlChar *fullname, const xmlChar *fullattr, |
1382 | 18.2k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1383 | 18.2k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1384 | | |
1385 | 18.2k | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1386 | 760 | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1387 | 760 | } |
1388 | 18.2k | } |
1389 | | |
1390 | | /** |
1391 | | * xmlCleanSpecialAttr: |
1392 | | * @ctxt: an XML parser context |
1393 | | * |
1394 | | * Trim the list of attributes defined to remove all those of type |
1395 | | * CDATA as they are not special. This call should be done when finishing |
1396 | | * to parse the DTD and before starting to parse the document root. |
1397 | | */ |
1398 | | static void |
1399 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1400 | 8.59k | { |
1401 | 8.59k | if (ctxt->attsSpecial == NULL) |
1402 | 6.26k | return; |
1403 | | |
1404 | 2.32k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1405 | | |
1406 | 2.32k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1407 | 54 | xmlHashFree(ctxt->attsSpecial, NULL); |
1408 | 54 | ctxt->attsSpecial = NULL; |
1409 | 54 | } |
1410 | 2.32k | return; |
1411 | 8.59k | } |
1412 | | |
1413 | | /** |
1414 | | * xmlCheckLanguageID: |
1415 | | * @lang: pointer to the string value |
1416 | | * |
1417 | | * Checks that the value conforms to the LanguageID production: |
1418 | | * |
1419 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1420 | | * the XML Second edition. |
1421 | | * |
1422 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1423 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1424 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1425 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1426 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1427 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1428 | | * |
1429 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1430 | | * |
1431 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1432 | | * langtag = language |
1433 | | * ["-" script] |
1434 | | * ["-" region] |
1435 | | * *("-" variant) |
1436 | | * *("-" extension) |
1437 | | * ["-" privateuse] |
1438 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1439 | | * ["-" extlang] ; sometimes followed by |
1440 | | * ; extended language subtags |
1441 | | * / 4ALPHA ; or reserved for future use |
1442 | | * / 5*8ALPHA ; or registered language subtag |
1443 | | * |
1444 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1445 | | * *2("-" 3ALPHA) ; permanently reserved |
1446 | | * |
1447 | | * script = 4ALPHA ; ISO 15924 code |
1448 | | * |
1449 | | * region = 2ALPHA ; ISO 3166-1 code |
1450 | | * / 3DIGIT ; UN M.49 code |
1451 | | * |
1452 | | * variant = 5*8alphanum ; registered variants |
1453 | | * / (DIGIT 3alphanum) |
1454 | | * |
1455 | | * extension = singleton 1*("-" (2*8alphanum)) |
1456 | | * |
1457 | | * ; Single alphanumerics |
1458 | | * ; "x" reserved for private use |
1459 | | * singleton = DIGIT ; 0 - 9 |
1460 | | * / %x41-57 ; A - W |
1461 | | * / %x59-5A ; Y - Z |
1462 | | * / %x61-77 ; a - w |
1463 | | * / %x79-7A ; y - z |
1464 | | * |
1465 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1466 | | * The parser below doesn't try to cope with extension or privateuse |
1467 | | * that could be added but that's not interoperable anyway |
1468 | | * |
1469 | | * Returns 1 if correct 0 otherwise |
1470 | | **/ |
1471 | | int |
1472 | | xmlCheckLanguageID(const xmlChar * lang) |
1473 | 0 | { |
1474 | 0 | const xmlChar *cur = lang, *nxt; |
1475 | |
|
1476 | 0 | if (cur == NULL) |
1477 | 0 | return (0); |
1478 | 0 | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1479 | 0 | ((cur[0] == 'I') && (cur[1] == '-')) || |
1480 | 0 | ((cur[0] == 'x') && (cur[1] == '-')) || |
1481 | 0 | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1482 | | /* |
1483 | | * Still allow IANA code and user code which were coming |
1484 | | * from the previous version of the XML-1.0 specification |
1485 | | * it's deprecated but we should not fail |
1486 | | */ |
1487 | 0 | cur += 2; |
1488 | 0 | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1489 | 0 | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1490 | 0 | cur++; |
1491 | 0 | return(cur[0] == 0); |
1492 | 0 | } |
1493 | 0 | nxt = cur; |
1494 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1495 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1496 | 0 | nxt++; |
1497 | 0 | if (nxt - cur >= 4) { |
1498 | | /* |
1499 | | * Reserved |
1500 | | */ |
1501 | 0 | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1502 | 0 | return(0); |
1503 | 0 | return(1); |
1504 | 0 | } |
1505 | 0 | if (nxt - cur < 2) |
1506 | 0 | return(0); |
1507 | | /* we got an ISO 639 code */ |
1508 | 0 | if (nxt[0] == 0) |
1509 | 0 | return(1); |
1510 | 0 | if (nxt[0] != '-') |
1511 | 0 | return(0); |
1512 | | |
1513 | 0 | nxt++; |
1514 | 0 | cur = nxt; |
1515 | | /* now we can have extlang or script or region or variant */ |
1516 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1517 | 0 | goto region_m49; |
1518 | | |
1519 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1520 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1521 | 0 | nxt++; |
1522 | 0 | if (nxt - cur == 4) |
1523 | 0 | goto script; |
1524 | 0 | if (nxt - cur == 2) |
1525 | 0 | goto region; |
1526 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1527 | 0 | goto variant; |
1528 | 0 | if (nxt - cur != 3) |
1529 | 0 | return(0); |
1530 | | /* we parsed an extlang */ |
1531 | 0 | if (nxt[0] == 0) |
1532 | 0 | return(1); |
1533 | 0 | if (nxt[0] != '-') |
1534 | 0 | return(0); |
1535 | | |
1536 | 0 | nxt++; |
1537 | 0 | cur = nxt; |
1538 | | /* now we can have script or region or variant */ |
1539 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1540 | 0 | goto region_m49; |
1541 | | |
1542 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1543 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1544 | 0 | nxt++; |
1545 | 0 | if (nxt - cur == 2) |
1546 | 0 | goto region; |
1547 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1548 | 0 | goto variant; |
1549 | 0 | if (nxt - cur != 4) |
1550 | 0 | return(0); |
1551 | | /* we parsed a script */ |
1552 | 0 | script: |
1553 | 0 | if (nxt[0] == 0) |
1554 | 0 | return(1); |
1555 | 0 | if (nxt[0] != '-') |
1556 | 0 | return(0); |
1557 | | |
1558 | 0 | nxt++; |
1559 | 0 | cur = nxt; |
1560 | | /* now we can have region or variant */ |
1561 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1562 | 0 | goto region_m49; |
1563 | | |
1564 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1565 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1566 | 0 | nxt++; |
1567 | |
|
1568 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1569 | 0 | goto variant; |
1570 | 0 | if (nxt - cur != 2) |
1571 | 0 | return(0); |
1572 | | /* we parsed a region */ |
1573 | 0 | region: |
1574 | 0 | if (nxt[0] == 0) |
1575 | 0 | return(1); |
1576 | 0 | if (nxt[0] != '-') |
1577 | 0 | return(0); |
1578 | | |
1579 | 0 | nxt++; |
1580 | 0 | cur = nxt; |
1581 | | /* now we can just have a variant */ |
1582 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1583 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1584 | 0 | nxt++; |
1585 | |
|
1586 | 0 | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1587 | 0 | return(0); |
1588 | | |
1589 | | /* we parsed a variant */ |
1590 | 0 | variant: |
1591 | 0 | if (nxt[0] == 0) |
1592 | 0 | return(1); |
1593 | 0 | if (nxt[0] != '-') |
1594 | 0 | return(0); |
1595 | | /* extensions and private use subtags not checked */ |
1596 | 0 | return (1); |
1597 | | |
1598 | 0 | region_m49: |
1599 | 0 | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1600 | 0 | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1601 | 0 | nxt += 3; |
1602 | 0 | goto region; |
1603 | 0 | } |
1604 | 0 | return(0); |
1605 | 0 | } |
1606 | | |
1607 | | /************************************************************************ |
1608 | | * * |
1609 | | * Parser stacks related functions and macros * |
1610 | | * * |
1611 | | ************************************************************************/ |
1612 | | |
1613 | | static xmlEntityPtr xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, |
1614 | | const xmlChar ** str); |
1615 | | |
1616 | | #ifdef SAX2 |
1617 | | /** |
1618 | | * nsPush: |
1619 | | * @ctxt: an XML parser context |
1620 | | * @prefix: the namespace prefix or NULL |
1621 | | * @URL: the namespace name |
1622 | | * |
1623 | | * Pushes a new parser namespace on top of the ns stack |
1624 | | * |
1625 | | * Returns -1 in case of error, -2 if the namespace should be discarded |
1626 | | * and the index in the stack otherwise. |
1627 | | */ |
1628 | | static int |
1629 | | nsPush(xmlParserCtxtPtr ctxt, const xmlChar *prefix, const xmlChar *URL) |
1630 | 53.2M | { |
1631 | 53.2M | if (ctxt->options & XML_PARSE_NSCLEAN) { |
1632 | 0 | int i; |
1633 | 0 | for (i = ctxt->nsNr - 2;i >= 0;i -= 2) { |
1634 | 0 | if (ctxt->nsTab[i] == prefix) { |
1635 | | /* in scope */ |
1636 | 0 | if (ctxt->nsTab[i + 1] == URL) |
1637 | 0 | return(-2); |
1638 | | /* out of scope keep it */ |
1639 | 0 | break; |
1640 | 0 | } |
1641 | 0 | } |
1642 | 0 | } |
1643 | 53.2M | if ((ctxt->nsMax == 0) || (ctxt->nsTab == NULL)) { |
1644 | 153k | ctxt->nsMax = 10; |
1645 | 153k | ctxt->nsNr = 0; |
1646 | 153k | ctxt->nsTab = (const xmlChar **) |
1647 | 153k | xmlMalloc(ctxt->nsMax * sizeof(xmlChar *)); |
1648 | 153k | if (ctxt->nsTab == NULL) { |
1649 | 0 | xmlErrMemory(ctxt, NULL); |
1650 | 0 | ctxt->nsMax = 0; |
1651 | 0 | return (-1); |
1652 | 0 | } |
1653 | 53.0M | } else if (ctxt->nsNr >= ctxt->nsMax) { |
1654 | 633k | const xmlChar ** tmp; |
1655 | 633k | ctxt->nsMax *= 2; |
1656 | 633k | tmp = (const xmlChar **) xmlRealloc((char *) ctxt->nsTab, |
1657 | 633k | ctxt->nsMax * sizeof(ctxt->nsTab[0])); |
1658 | 633k | if (tmp == NULL) { |
1659 | 0 | xmlErrMemory(ctxt, NULL); |
1660 | 0 | ctxt->nsMax /= 2; |
1661 | 0 | return (-1); |
1662 | 0 | } |
1663 | 633k | ctxt->nsTab = tmp; |
1664 | 633k | } |
1665 | 53.2M | ctxt->nsTab[ctxt->nsNr++] = prefix; |
1666 | 53.2M | ctxt->nsTab[ctxt->nsNr++] = URL; |
1667 | 53.2M | return (ctxt->nsNr); |
1668 | 53.2M | } |
1669 | | /** |
1670 | | * nsPop: |
1671 | | * @ctxt: an XML parser context |
1672 | | * @nr: the number to pop |
1673 | | * |
1674 | | * Pops the top @nr parser prefix/namespace from the ns stack |
1675 | | * |
1676 | | * Returns the number of namespaces removed |
1677 | | */ |
1678 | | static int |
1679 | | nsPop(xmlParserCtxtPtr ctxt, int nr) |
1680 | 117k | { |
1681 | 117k | int i; |
1682 | | |
1683 | 117k | if (ctxt->nsTab == NULL) return(0); |
1684 | 117k | if (ctxt->nsNr < nr) { |
1685 | 0 | xmlGenericError(xmlGenericErrorContext, "Pbm popping %d NS\n", nr); |
1686 | 0 | nr = ctxt->nsNr; |
1687 | 0 | } |
1688 | 117k | if (ctxt->nsNr <= 0) |
1689 | 0 | return (0); |
1690 | | |
1691 | 1.20M | for (i = 0;i < nr;i++) { |
1692 | 1.08M | ctxt->nsNr--; |
1693 | 1.08M | ctxt->nsTab[ctxt->nsNr] = NULL; |
1694 | 1.08M | } |
1695 | 117k | return(nr); |
1696 | 117k | } |
1697 | | #endif |
1698 | | |
1699 | | static int |
1700 | 56.2k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1701 | 56.2k | const xmlChar **atts; |
1702 | 56.2k | int *attallocs; |
1703 | 56.2k | int maxatts; |
1704 | | |
1705 | 56.2k | if (ctxt->atts == NULL) { |
1706 | 56.1k | maxatts = 55; /* allow for 10 attrs by default */ |
1707 | 56.1k | atts = (const xmlChar **) |
1708 | 56.1k | xmlMalloc(maxatts * sizeof(xmlChar *)); |
1709 | 56.1k | if (atts == NULL) goto mem_error; |
1710 | 56.1k | ctxt->atts = atts; |
1711 | 56.1k | attallocs = (int *) xmlMalloc((maxatts / 5) * sizeof(int)); |
1712 | 56.1k | if (attallocs == NULL) goto mem_error; |
1713 | 56.1k | ctxt->attallocs = attallocs; |
1714 | 56.1k | ctxt->maxatts = maxatts; |
1715 | 56.1k | } else if (nr + 5 > ctxt->maxatts) { |
1716 | 64 | maxatts = (nr + 5) * 2; |
1717 | 64 | atts = (const xmlChar **) xmlRealloc((void *) ctxt->atts, |
1718 | 64 | maxatts * sizeof(const xmlChar *)); |
1719 | 64 | if (atts == NULL) goto mem_error; |
1720 | 64 | ctxt->atts = atts; |
1721 | 64 | attallocs = (int *) xmlRealloc((void *) ctxt->attallocs, |
1722 | 64 | (maxatts / 5) * sizeof(int)); |
1723 | 64 | if (attallocs == NULL) goto mem_error; |
1724 | 64 | ctxt->attallocs = attallocs; |
1725 | 64 | ctxt->maxatts = maxatts; |
1726 | 64 | } |
1727 | 56.2k | return(ctxt->maxatts); |
1728 | 0 | mem_error: |
1729 | 0 | xmlErrMemory(ctxt, NULL); |
1730 | 0 | return(-1); |
1731 | 56.2k | } |
1732 | | |
1733 | | /** |
1734 | | * inputPush: |
1735 | | * @ctxt: an XML parser context |
1736 | | * @value: the parser input |
1737 | | * |
1738 | | * Pushes a new parser input on top of the input stack |
1739 | | * |
1740 | | * Returns -1 in case of error, the index in the stack otherwise |
1741 | | */ |
1742 | | int |
1743 | | inputPush(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1744 | 5.95M | { |
1745 | 5.95M | if ((ctxt == NULL) || (value == NULL)) |
1746 | 0 | return(-1); |
1747 | 5.95M | if (ctxt->inputNr >= ctxt->inputMax) { |
1748 | 4.32k | ctxt->inputMax *= 2; |
1749 | 4.32k | ctxt->inputTab = |
1750 | 4.32k | (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1751 | 4.32k | ctxt->inputMax * |
1752 | 4.32k | sizeof(ctxt->inputTab[0])); |
1753 | 4.32k | if (ctxt->inputTab == NULL) { |
1754 | 0 | xmlErrMemory(ctxt, NULL); |
1755 | 0 | xmlFreeInputStream(value); |
1756 | 0 | ctxt->inputMax /= 2; |
1757 | 0 | value = NULL; |
1758 | 0 | return (-1); |
1759 | 0 | } |
1760 | 4.32k | } |
1761 | 5.95M | ctxt->inputTab[ctxt->inputNr] = value; |
1762 | 5.95M | ctxt->input = value; |
1763 | 5.95M | return (ctxt->inputNr++); |
1764 | 5.95M | } |
1765 | | /** |
1766 | | * inputPop: |
1767 | | * @ctxt: an XML parser context |
1768 | | * |
1769 | | * Pops the top parser input from the input stack |
1770 | | * |
1771 | | * Returns the input just removed |
1772 | | */ |
1773 | | xmlParserInputPtr |
1774 | | inputPop(xmlParserCtxtPtr ctxt) |
1775 | 6.59M | { |
1776 | 6.59M | xmlParserInputPtr ret; |
1777 | | |
1778 | 6.59M | if (ctxt == NULL) |
1779 | 0 | return(NULL); |
1780 | 6.59M | if (ctxt->inputNr <= 0) |
1781 | 639k | return (NULL); |
1782 | 5.95M | ctxt->inputNr--; |
1783 | 5.95M | if (ctxt->inputNr > 0) |
1784 | 5.63M | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1785 | 319k | else |
1786 | 319k | ctxt->input = NULL; |
1787 | 5.95M | ret = ctxt->inputTab[ctxt->inputNr]; |
1788 | 5.95M | ctxt->inputTab[ctxt->inputNr] = NULL; |
1789 | 5.95M | return (ret); |
1790 | 6.59M | } |
1791 | | /** |
1792 | | * nodePush: |
1793 | | * @ctxt: an XML parser context |
1794 | | * @value: the element node |
1795 | | * |
1796 | | * Pushes a new element node on top of the node stack |
1797 | | * |
1798 | | * Returns -1 in case of error, the index in the stack otherwise |
1799 | | */ |
1800 | | int |
1801 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
1802 | 834k | { |
1803 | 834k | if (ctxt == NULL) return(0); |
1804 | 834k | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1805 | 3.50k | xmlNodePtr *tmp; |
1806 | | |
1807 | 3.50k | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
1808 | 3.50k | ctxt->nodeMax * 2 * |
1809 | 3.50k | sizeof(ctxt->nodeTab[0])); |
1810 | 3.50k | if (tmp == NULL) { |
1811 | 0 | xmlErrMemory(ctxt, NULL); |
1812 | 0 | return (-1); |
1813 | 0 | } |
1814 | 3.50k | ctxt->nodeTab = tmp; |
1815 | 3.50k | ctxt->nodeMax *= 2; |
1816 | 3.50k | } |
1817 | 834k | if ((((unsigned int) ctxt->nodeNr) > xmlParserMaxDepth) && |
1818 | 834k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
1819 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
1820 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
1821 | 0 | xmlParserMaxDepth); |
1822 | 0 | xmlHaltParser(ctxt); |
1823 | 0 | return(-1); |
1824 | 0 | } |
1825 | 834k | ctxt->nodeTab[ctxt->nodeNr] = value; |
1826 | 834k | ctxt->node = value; |
1827 | 834k | return (ctxt->nodeNr++); |
1828 | 834k | } |
1829 | | |
1830 | | /** |
1831 | | * nodePop: |
1832 | | * @ctxt: an XML parser context |
1833 | | * |
1834 | | * Pops the top element node from the node stack |
1835 | | * |
1836 | | * Returns the node just removed |
1837 | | */ |
1838 | | xmlNodePtr |
1839 | | nodePop(xmlParserCtxtPtr ctxt) |
1840 | 401k | { |
1841 | 401k | xmlNodePtr ret; |
1842 | | |
1843 | 401k | if (ctxt == NULL) return(NULL); |
1844 | 401k | if (ctxt->nodeNr <= 0) |
1845 | 171k | return (NULL); |
1846 | 229k | ctxt->nodeNr--; |
1847 | 229k | if (ctxt->nodeNr > 0) |
1848 | 202k | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
1849 | 27.1k | else |
1850 | 27.1k | ctxt->node = NULL; |
1851 | 229k | ret = ctxt->nodeTab[ctxt->nodeNr]; |
1852 | 229k | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
1853 | 229k | return (ret); |
1854 | 401k | } |
1855 | | |
1856 | | /** |
1857 | | * nameNsPush: |
1858 | | * @ctxt: an XML parser context |
1859 | | * @value: the element name |
1860 | | * @prefix: the element prefix |
1861 | | * @URI: the element namespace name |
1862 | | * @line: the current line number for error messages |
1863 | | * @nsNr: the number of namespaces pushed on the namespace table |
1864 | | * |
1865 | | * Pushes a new element name/prefix/URL on top of the name stack |
1866 | | * |
1867 | | * Returns -1 in case of error, the index in the stack otherwise |
1868 | | */ |
1869 | | static int |
1870 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
1871 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
1872 | 5.66M | { |
1873 | 5.66M | xmlStartTag *tag; |
1874 | | |
1875 | 5.66M | if (ctxt->nameNr >= ctxt->nameMax) { |
1876 | 5.64k | const xmlChar * *tmp; |
1877 | 5.64k | xmlStartTag *tmp2; |
1878 | 5.64k | ctxt->nameMax *= 2; |
1879 | 5.64k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1880 | 5.64k | ctxt->nameMax * |
1881 | 5.64k | sizeof(ctxt->nameTab[0])); |
1882 | 5.64k | if (tmp == NULL) { |
1883 | 0 | ctxt->nameMax /= 2; |
1884 | 0 | goto mem_error; |
1885 | 0 | } |
1886 | 5.64k | ctxt->nameTab = tmp; |
1887 | 5.64k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
1888 | 5.64k | ctxt->nameMax * |
1889 | 5.64k | sizeof(ctxt->pushTab[0])); |
1890 | 5.64k | if (tmp2 == NULL) { |
1891 | 0 | ctxt->nameMax /= 2; |
1892 | 0 | goto mem_error; |
1893 | 0 | } |
1894 | 5.64k | ctxt->pushTab = tmp2; |
1895 | 5.65M | } else if (ctxt->pushTab == NULL) { |
1896 | 157k | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
1897 | 157k | sizeof(ctxt->pushTab[0])); |
1898 | 157k | if (ctxt->pushTab == NULL) |
1899 | 0 | goto mem_error; |
1900 | 157k | } |
1901 | 5.66M | ctxt->nameTab[ctxt->nameNr] = value; |
1902 | 5.66M | ctxt->name = value; |
1903 | 5.66M | tag = &ctxt->pushTab[ctxt->nameNr]; |
1904 | 5.66M | tag->prefix = prefix; |
1905 | 5.66M | tag->URI = URI; |
1906 | 5.66M | tag->line = line; |
1907 | 5.66M | tag->nsNr = nsNr; |
1908 | 5.66M | return (ctxt->nameNr++); |
1909 | 0 | mem_error: |
1910 | 0 | xmlErrMemory(ctxt, NULL); |
1911 | 0 | return (-1); |
1912 | 5.66M | } |
1913 | | #ifdef LIBXML_PUSH_ENABLED |
1914 | | /** |
1915 | | * nameNsPop: |
1916 | | * @ctxt: an XML parser context |
1917 | | * |
1918 | | * Pops the top element/prefix/URI name from the name stack |
1919 | | * |
1920 | | * Returns the name just removed |
1921 | | */ |
1922 | | static const xmlChar * |
1923 | | nameNsPop(xmlParserCtxtPtr ctxt) |
1924 | 0 | { |
1925 | 0 | const xmlChar *ret; |
1926 | |
|
1927 | 0 | if (ctxt->nameNr <= 0) |
1928 | 0 | return (NULL); |
1929 | 0 | ctxt->nameNr--; |
1930 | 0 | if (ctxt->nameNr > 0) |
1931 | 0 | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1932 | 0 | else |
1933 | 0 | ctxt->name = NULL; |
1934 | 0 | ret = ctxt->nameTab[ctxt->nameNr]; |
1935 | 0 | ctxt->nameTab[ctxt->nameNr] = NULL; |
1936 | 0 | return (ret); |
1937 | 0 | } |
1938 | | #endif /* LIBXML_PUSH_ENABLED */ |
1939 | | |
1940 | | /** |
1941 | | * namePush: |
1942 | | * @ctxt: an XML parser context |
1943 | | * @value: the element name |
1944 | | * |
1945 | | * Pushes a new element name on top of the name stack |
1946 | | * |
1947 | | * Returns -1 in case of error, the index in the stack otherwise |
1948 | | */ |
1949 | | int |
1950 | | namePush(xmlParserCtxtPtr ctxt, const xmlChar * value) |
1951 | 0 | { |
1952 | 0 | if (ctxt == NULL) return (-1); |
1953 | | |
1954 | 0 | if (ctxt->nameNr >= ctxt->nameMax) { |
1955 | 0 | const xmlChar * *tmp; |
1956 | 0 | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1957 | 0 | ctxt->nameMax * 2 * |
1958 | 0 | sizeof(ctxt->nameTab[0])); |
1959 | 0 | if (tmp == NULL) { |
1960 | 0 | goto mem_error; |
1961 | 0 | } |
1962 | 0 | ctxt->nameTab = tmp; |
1963 | 0 | ctxt->nameMax *= 2; |
1964 | 0 | } |
1965 | 0 | ctxt->nameTab[ctxt->nameNr] = value; |
1966 | 0 | ctxt->name = value; |
1967 | 0 | return (ctxt->nameNr++); |
1968 | 0 | mem_error: |
1969 | 0 | xmlErrMemory(ctxt, NULL); |
1970 | 0 | return (-1); |
1971 | 0 | } |
1972 | | /** |
1973 | | * namePop: |
1974 | | * @ctxt: an XML parser context |
1975 | | * |
1976 | | * Pops the top element name from the name stack |
1977 | | * |
1978 | | * Returns the name just removed |
1979 | | */ |
1980 | | const xmlChar * |
1981 | | namePop(xmlParserCtxtPtr ctxt) |
1982 | 476k | { |
1983 | 476k | const xmlChar *ret; |
1984 | | |
1985 | 476k | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
1986 | 0 | return (NULL); |
1987 | 476k | ctxt->nameNr--; |
1988 | 476k | if (ctxt->nameNr > 0) |
1989 | 420k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1990 | 55.9k | else |
1991 | 55.9k | ctxt->name = NULL; |
1992 | 476k | ret = ctxt->nameTab[ctxt->nameNr]; |
1993 | 476k | ctxt->nameTab[ctxt->nameNr] = NULL; |
1994 | 476k | return (ret); |
1995 | 476k | } |
1996 | | |
1997 | 5.79M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
1998 | 5.79M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
1999 | 5.73k | int *tmp; |
2000 | | |
2001 | 5.73k | ctxt->spaceMax *= 2; |
2002 | 5.73k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
2003 | 5.73k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
2004 | 5.73k | if (tmp == NULL) { |
2005 | 0 | xmlErrMemory(ctxt, NULL); |
2006 | 0 | ctxt->spaceMax /=2; |
2007 | 0 | return(-1); |
2008 | 0 | } |
2009 | 5.73k | ctxt->spaceTab = tmp; |
2010 | 5.73k | } |
2011 | 5.79M | ctxt->spaceTab[ctxt->spaceNr] = val; |
2012 | 5.79M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2013 | 5.79M | return(ctxt->spaceNr++); |
2014 | 5.79M | } |
2015 | | |
2016 | 606k | static int spacePop(xmlParserCtxtPtr ctxt) { |
2017 | 606k | int ret; |
2018 | 606k | if (ctxt->spaceNr <= 0) return(0); |
2019 | 606k | ctxt->spaceNr--; |
2020 | 606k | if (ctxt->spaceNr > 0) |
2021 | 606k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2022 | 0 | else |
2023 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
2024 | 606k | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2025 | 606k | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2026 | 606k | return(ret); |
2027 | 606k | } |
2028 | | |
2029 | | /* |
2030 | | * Macros for accessing the content. Those should be used only by the parser, |
2031 | | * and not exported. |
2032 | | * |
2033 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2034 | | * use them |
2035 | | * |
2036 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2037 | | * To be used with extreme caution since operations consuming |
2038 | | * characters may move the input buffer to a different location ! |
2039 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2040 | | * This should be used internally by the parser |
2041 | | * only to compare to ASCII values otherwise it would break when |
2042 | | * running with UTF-8 encoding. |
2043 | | * RAW same as CUR but in the input buffer, bypass any token |
2044 | | * extraction that may have been done |
2045 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2046 | | * to compare on ASCII based substring. |
2047 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2048 | | * strings without newlines within the parser. |
2049 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2050 | | * defined char within the parser. |
2051 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2052 | | * |
2053 | | * NEXT Skip to the next character, this does the proper decoding |
2054 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2055 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2056 | | * CUR_CHAR(l) returns the current unicode character (int), set l |
2057 | | * to the number of xmlChars used for the encoding [0-5]. |
2058 | | * CUR_SCHAR same but operate on a string instead of the context |
2059 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2060 | | * the index |
2061 | | * GROW, SHRINK handling of input buffers |
2062 | | */ |
2063 | | |
2064 | 104M | #define RAW (*ctxt->input->cur) |
2065 | 123M | #define CUR (*ctxt->input->cur) |
2066 | 351M | #define NXT(val) ctxt->input->cur[(val)] |
2067 | 1.67M | #define CUR_PTR ctxt->input->cur |
2068 | 447k | #define BASE_PTR ctxt->input->base |
2069 | | |
2070 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2071 | 50.7M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2072 | 25.4M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2073 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2074 | 46.8M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2075 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2076 | 38.3M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2077 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2078 | 32.5M | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2079 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2080 | 26.1M | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2081 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2082 | 10.6M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2083 | 10.6M | ((unsigned char *) s)[ 8 ] == c9 ) |
2084 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2085 | 195k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2086 | 195k | ((unsigned char *) s)[ 9 ] == c10 ) |
2087 | | |
2088 | 8.94M | #define SKIP(val) do { \ |
2089 | 8.94M | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2090 | 8.94M | if (*ctxt->input->cur == 0) \ |
2091 | 8.94M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2092 | 8.94M | } while (0) |
2093 | | |
2094 | 0 | #define SKIPL(val) do { \ |
2095 | 0 | int skipl; \ |
2096 | 0 | for(skipl=0; skipl<val; skipl++) { \ |
2097 | 0 | if (*(ctxt->input->cur) == '\n') { \ |
2098 | 0 | ctxt->input->line++; ctxt->input->col = 1; \ |
2099 | 0 | } else ctxt->input->col++; \ |
2100 | 0 | ctxt->input->cur++; \ |
2101 | 0 | } \ |
2102 | 0 | if (*ctxt->input->cur == 0) \ |
2103 | 0 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2104 | 0 | } while (0) |
2105 | | |
2106 | 36.7M | #define SHRINK if ((ctxt->progressive == 0) && \ |
2107 | 36.7M | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2108 | 36.7M | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2109 | 36.7M | xmlSHRINK (ctxt); |
2110 | | |
2111 | 96.7k | static void xmlSHRINK (xmlParserCtxtPtr ctxt) { |
2112 | 96.7k | xmlParserInputShrink(ctxt->input); |
2113 | 96.7k | if (*ctxt->input->cur == 0) |
2114 | 18.6k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2115 | 96.7k | } |
2116 | | |
2117 | 970M | #define GROW if ((ctxt->progressive == 0) && \ |
2118 | 970M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2119 | 970M | xmlGROW (ctxt); |
2120 | | |
2121 | 55.3M | static void xmlGROW (xmlParserCtxtPtr ctxt) { |
2122 | 55.3M | ptrdiff_t curEnd = ctxt->input->end - ctxt->input->cur; |
2123 | 55.3M | ptrdiff_t curBase = ctxt->input->cur - ctxt->input->base; |
2124 | | |
2125 | 55.3M | if (((curEnd > XML_MAX_LOOKUP_LIMIT) || |
2126 | 55.3M | (curBase > XML_MAX_LOOKUP_LIMIT)) && |
2127 | 55.3M | ((ctxt->input->buf) && |
2128 | 2.83k | (ctxt->input->buf->readcallback != xmlInputReadCallbackNop)) && |
2129 | 55.3M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
2130 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
2131 | 0 | xmlHaltParser(ctxt); |
2132 | 0 | return; |
2133 | 0 | } |
2134 | 55.3M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2135 | 55.3M | if ((ctxt->input->cur > ctxt->input->end) || |
2136 | 55.3M | (ctxt->input->cur < ctxt->input->base)) { |
2137 | 0 | xmlHaltParser(ctxt); |
2138 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "cur index out of bound"); |
2139 | 0 | return; |
2140 | 0 | } |
2141 | 55.3M | if ((ctxt->input->cur != NULL) && (*ctxt->input->cur == 0)) |
2142 | 509k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2143 | 55.3M | } |
2144 | | |
2145 | 50.9M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2146 | | |
2147 | 42.8M | #define NEXT xmlNextChar(ctxt) |
2148 | | |
2149 | 11.3M | #define NEXT1 { \ |
2150 | 11.3M | ctxt->input->col++; \ |
2151 | 11.3M | ctxt->input->cur++; \ |
2152 | 11.3M | if (*ctxt->input->cur == 0) \ |
2153 | 11.3M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2154 | 11.3M | } |
2155 | | |
2156 | 1.72G | #define NEXTL(l) do { \ |
2157 | 1.72G | if (*(ctxt->input->cur) == '\n') { \ |
2158 | 7.75M | ctxt->input->line++; ctxt->input->col = 1; \ |
2159 | 1.72G | } else ctxt->input->col++; \ |
2160 | 1.72G | ctxt->input->cur += l; \ |
2161 | 1.72G | } while (0) |
2162 | | |
2163 | 1.73G | #define CUR_CHAR(l) xmlCurrentChar(ctxt, &l) |
2164 | 931M | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2165 | | |
2166 | | #define COPY_BUF(l,b,i,v) \ |
2167 | 2.54G | if (l == 1) b[i++] = (xmlChar) v; \ |
2168 | 2.54G | else i += xmlCopyCharMultiByte(&b[i],v) |
2169 | | |
2170 | | #define CUR_CONSUMED \ |
2171 | 27.9M | (ctxt->input->consumed + (ctxt->input->cur - ctxt->input->base)) |
2172 | | |
2173 | | /** |
2174 | | * xmlSkipBlankChars: |
2175 | | * @ctxt: the XML parser context |
2176 | | * |
2177 | | * skip all blanks character found at that point in the input streams. |
2178 | | * It pops up finished entities in the process if allowable at that point. |
2179 | | * |
2180 | | * Returns the number of space chars skipped |
2181 | | */ |
2182 | | |
2183 | | int |
2184 | 50.9M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2185 | 50.9M | int res = 0; |
2186 | | |
2187 | | /* |
2188 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2189 | | * the ASCII range. |
2190 | | */ |
2191 | 50.9M | if (((ctxt->inputNr == 1) && (ctxt->instate != XML_PARSER_DTD)) || |
2192 | 50.9M | (ctxt->instate == XML_PARSER_START)) { |
2193 | 6.80M | const xmlChar *cur; |
2194 | | /* |
2195 | | * if we are in the document content, go really fast |
2196 | | */ |
2197 | 6.80M | cur = ctxt->input->cur; |
2198 | 6.80M | while (IS_BLANK_CH(*cur)) { |
2199 | 389k | if (*cur == '\n') { |
2200 | 27.2k | ctxt->input->line++; ctxt->input->col = 1; |
2201 | 361k | } else { |
2202 | 361k | ctxt->input->col++; |
2203 | 361k | } |
2204 | 389k | cur++; |
2205 | 389k | if (res < INT_MAX) |
2206 | 389k | res++; |
2207 | 389k | if (*cur == 0) { |
2208 | 1.08k | ctxt->input->cur = cur; |
2209 | 1.08k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2210 | 1.08k | cur = ctxt->input->cur; |
2211 | 1.08k | } |
2212 | 389k | } |
2213 | 6.80M | ctxt->input->cur = cur; |
2214 | 44.0M | } else { |
2215 | 44.0M | int expandPE = ((ctxt->external != 0) || (ctxt->inputNr != 1)); |
2216 | | |
2217 | 54.6M | while (1) { |
2218 | 54.6M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2219 | 4.76M | NEXT; |
2220 | 49.8M | } else if (CUR == '%') { |
2221 | | /* |
2222 | | * Need to handle support of entities branching here |
2223 | | */ |
2224 | 5.66M | if ((expandPE == 0) || (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2225 | 34.3k | break; |
2226 | 5.63M | xmlParsePEReference(ctxt); |
2227 | 44.2M | } else if (CUR == 0) { |
2228 | 154k | if (ctxt->inputNr <= 1) |
2229 | 4.92k | break; |
2230 | 149k | xmlPopInput(ctxt); |
2231 | 44.0M | } else { |
2232 | 44.0M | break; |
2233 | 44.0M | } |
2234 | | |
2235 | | /* |
2236 | | * Also increase the counter when entering or exiting a PERef. |
2237 | | * The spec says: "When a parameter-entity reference is recognized |
2238 | | * in the DTD and included, its replacement text MUST be enlarged |
2239 | | * by the attachment of one leading and one following space (#x20) |
2240 | | * character." |
2241 | | */ |
2242 | 10.5M | if (res < INT_MAX) |
2243 | 10.5M | res++; |
2244 | 10.5M | } |
2245 | 44.0M | } |
2246 | 50.9M | return(res); |
2247 | 50.9M | } |
2248 | | |
2249 | | /************************************************************************ |
2250 | | * * |
2251 | | * Commodity functions to handle entities * |
2252 | | * * |
2253 | | ************************************************************************/ |
2254 | | |
2255 | | /** |
2256 | | * xmlPopInput: |
2257 | | * @ctxt: an XML parser context |
2258 | | * |
2259 | | * xmlPopInput: the current input pointed by ctxt->input came to an end |
2260 | | * pop it and return the next char. |
2261 | | * |
2262 | | * Returns the current xmlChar in the parser context |
2263 | | */ |
2264 | | xmlChar |
2265 | 1.46M | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2266 | 1.46M | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2267 | 1.46M | if (xmlParserDebugEntities) |
2268 | 0 | xmlGenericError(xmlGenericErrorContext, |
2269 | 0 | "Popping input %d\n", ctxt->inputNr); |
2270 | 1.46M | if ((ctxt->inputNr > 1) && (ctxt->inSubset == 0) && |
2271 | 1.46M | (ctxt->instate != XML_PARSER_EOF)) |
2272 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
2273 | 0 | "Unfinished entity outside the DTD"); |
2274 | 1.46M | xmlFreeInputStream(inputPop(ctxt)); |
2275 | 1.46M | if (*ctxt->input->cur == 0) |
2276 | 5.72k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2277 | 1.46M | return(CUR); |
2278 | 1.46M | } |
2279 | | |
2280 | | /** |
2281 | | * xmlPushInput: |
2282 | | * @ctxt: an XML parser context |
2283 | | * @input: an XML parser input fragment (entity, XML fragment ...). |
2284 | | * |
2285 | | * xmlPushInput: switch to a new input stream which is stacked on top |
2286 | | * of the previous one(s). |
2287 | | * Returns -1 in case of error or the index in the input stack |
2288 | | */ |
2289 | | int |
2290 | 5.64M | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2291 | 5.64M | int ret; |
2292 | 5.64M | if (input == NULL) return(-1); |
2293 | | |
2294 | 5.64M | if (xmlParserDebugEntities) { |
2295 | 0 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
2296 | 0 | xmlGenericError(xmlGenericErrorContext, |
2297 | 0 | "%s(%d): ", ctxt->input->filename, |
2298 | 0 | ctxt->input->line); |
2299 | 0 | xmlGenericError(xmlGenericErrorContext, |
2300 | 0 | "Pushing input %d : %.30s\n", ctxt->inputNr+1, input->cur); |
2301 | 0 | } |
2302 | 5.64M | if (((ctxt->inputNr > 40) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2303 | 5.64M | (ctxt->inputNr > 1024)) { |
2304 | 4.07k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2305 | 4.17M | while (ctxt->inputNr > 1) |
2306 | 4.17M | xmlFreeInputStream(inputPop(ctxt)); |
2307 | 4.07k | return(-1); |
2308 | 4.07k | } |
2309 | 5.63M | ret = inputPush(ctxt, input); |
2310 | 5.63M | if (ctxt->instate == XML_PARSER_EOF) |
2311 | 0 | return(-1); |
2312 | 5.63M | GROW; |
2313 | 5.63M | return(ret); |
2314 | 5.63M | } |
2315 | | |
2316 | | /** |
2317 | | * xmlParseCharRef: |
2318 | | * @ctxt: an XML parser context |
2319 | | * |
2320 | | * parse Reference declarations |
2321 | | * |
2322 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2323 | | * '&#x' [0-9a-fA-F]+ ';' |
2324 | | * |
2325 | | * [ WFC: Legal Character ] |
2326 | | * Characters referred to using character references must match the |
2327 | | * production for Char. |
2328 | | * |
2329 | | * Returns the value parsed (as an int), 0 in case of error |
2330 | | */ |
2331 | | int |
2332 | 532k | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2333 | 532k | int val = 0; |
2334 | 532k | int count = 0; |
2335 | | |
2336 | | /* |
2337 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2338 | | */ |
2339 | 532k | if ((RAW == '&') && (NXT(1) == '#') && |
2340 | 532k | (NXT(2) == 'x')) { |
2341 | 45.1k | SKIP(3); |
2342 | 45.1k | GROW; |
2343 | 165k | while (RAW != ';') { /* loop blocked by count */ |
2344 | 126k | if (count++ > 20) { |
2345 | 2.73k | count = 0; |
2346 | 2.73k | GROW; |
2347 | 2.73k | if (ctxt->instate == XML_PARSER_EOF) |
2348 | 0 | return(0); |
2349 | 2.73k | } |
2350 | 126k | if ((RAW >= '0') && (RAW <= '9')) |
2351 | 67.7k | val = val * 16 + (CUR - '0'); |
2352 | 58.9k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2353 | 37.6k | val = val * 16 + (CUR - 'a') + 10; |
2354 | 21.3k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2355 | 15.2k | val = val * 16 + (CUR - 'A') + 10; |
2356 | 6.02k | else { |
2357 | 6.02k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2358 | 6.02k | val = 0; |
2359 | 6.02k | break; |
2360 | 6.02k | } |
2361 | 120k | if (val > 0x110000) |
2362 | 42.6k | val = 0x110000; |
2363 | | |
2364 | 120k | NEXT; |
2365 | 120k | count++; |
2366 | 120k | } |
2367 | 45.1k | if (RAW == ';') { |
2368 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2369 | 39.1k | ctxt->input->col++; |
2370 | 39.1k | ctxt->input->cur++; |
2371 | 39.1k | } |
2372 | 487k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2373 | 487k | SKIP(2); |
2374 | 487k | GROW; |
2375 | 1.78M | while (RAW != ';') { /* loop blocked by count */ |
2376 | 1.32M | if (count++ > 20) { |
2377 | 25.9k | count = 0; |
2378 | 25.9k | GROW; |
2379 | 25.9k | if (ctxt->instate == XML_PARSER_EOF) |
2380 | 0 | return(0); |
2381 | 25.9k | } |
2382 | 1.32M | if ((RAW >= '0') && (RAW <= '9')) |
2383 | 1.29M | val = val * 10 + (CUR - '0'); |
2384 | 25.2k | else { |
2385 | 25.2k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2386 | 25.2k | val = 0; |
2387 | 25.2k | break; |
2388 | 25.2k | } |
2389 | 1.29M | if (val > 0x110000) |
2390 | 278k | val = 0x110000; |
2391 | | |
2392 | 1.29M | NEXT; |
2393 | 1.29M | count++; |
2394 | 1.29M | } |
2395 | 487k | if (RAW == ';') { |
2396 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2397 | 461k | ctxt->input->col++; |
2398 | 461k | ctxt->input->cur++; |
2399 | 461k | } |
2400 | 487k | } else { |
2401 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2402 | 0 | } |
2403 | | |
2404 | | /* |
2405 | | * [ WFC: Legal Character ] |
2406 | | * Characters referred to using character references must match the |
2407 | | * production for Char. |
2408 | | */ |
2409 | 532k | if (val >= 0x110000) { |
2410 | 14.0k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2411 | 14.0k | "xmlParseCharRef: character reference out of bounds\n", |
2412 | 14.0k | val); |
2413 | 518k | } else if (IS_CHAR(val)) { |
2414 | 477k | return(val); |
2415 | 477k | } else { |
2416 | 40.2k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2417 | 40.2k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2418 | 40.2k | val); |
2419 | 40.2k | } |
2420 | 54.3k | return(0); |
2421 | 532k | } |
2422 | | |
2423 | | /** |
2424 | | * xmlParseStringCharRef: |
2425 | | * @ctxt: an XML parser context |
2426 | | * @str: a pointer to an index in the string |
2427 | | * |
2428 | | * parse Reference declarations, variant parsing from a string rather |
2429 | | * than an an input flow. |
2430 | | * |
2431 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2432 | | * '&#x' [0-9a-fA-F]+ ';' |
2433 | | * |
2434 | | * [ WFC: Legal Character ] |
2435 | | * Characters referred to using character references must match the |
2436 | | * production for Char. |
2437 | | * |
2438 | | * Returns the value parsed (as an int), 0 in case of error, str will be |
2439 | | * updated to the current value of the index |
2440 | | */ |
2441 | | static int |
2442 | 34.2k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2443 | 34.2k | const xmlChar *ptr; |
2444 | 34.2k | xmlChar cur; |
2445 | 34.2k | int val = 0; |
2446 | | |
2447 | 34.2k | if ((str == NULL) || (*str == NULL)) return(0); |
2448 | 34.2k | ptr = *str; |
2449 | 34.2k | cur = *ptr; |
2450 | 34.2k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2451 | 15.3k | ptr += 3; |
2452 | 15.3k | cur = *ptr; |
2453 | 79.6k | while (cur != ';') { /* Non input consuming loop */ |
2454 | 66.1k | if ((cur >= '0') && (cur <= '9')) |
2455 | 35.6k | val = val * 16 + (cur - '0'); |
2456 | 30.5k | else if ((cur >= 'a') && (cur <= 'f')) |
2457 | 17.7k | val = val * 16 + (cur - 'a') + 10; |
2458 | 12.7k | else if ((cur >= 'A') && (cur <= 'F')) |
2459 | 10.8k | val = val * 16 + (cur - 'A') + 10; |
2460 | 1.94k | else { |
2461 | 1.94k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2462 | 1.94k | val = 0; |
2463 | 1.94k | break; |
2464 | 1.94k | } |
2465 | 64.2k | if (val > 0x110000) |
2466 | 3.23k | val = 0x110000; |
2467 | | |
2468 | 64.2k | ptr++; |
2469 | 64.2k | cur = *ptr; |
2470 | 64.2k | } |
2471 | 15.3k | if (cur == ';') |
2472 | 13.4k | ptr++; |
2473 | 18.8k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2474 | 18.8k | ptr += 2; |
2475 | 18.8k | cur = *ptr; |
2476 | 48.1k | while (cur != ';') { /* Non input consuming loops */ |
2477 | 33.5k | if ((cur >= '0') && (cur <= '9')) |
2478 | 29.3k | val = val * 10 + (cur - '0'); |
2479 | 4.25k | else { |
2480 | 4.25k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2481 | 4.25k | val = 0; |
2482 | 4.25k | break; |
2483 | 4.25k | } |
2484 | 29.3k | if (val > 0x110000) |
2485 | 724 | val = 0x110000; |
2486 | | |
2487 | 29.3k | ptr++; |
2488 | 29.3k | cur = *ptr; |
2489 | 29.3k | } |
2490 | 18.8k | if (cur == ';') |
2491 | 14.5k | ptr++; |
2492 | 18.8k | } else { |
2493 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2494 | 0 | return(0); |
2495 | 0 | } |
2496 | 34.2k | *str = ptr; |
2497 | | |
2498 | | /* |
2499 | | * [ WFC: Legal Character ] |
2500 | | * Characters referred to using character references must match the |
2501 | | * production for Char. |
2502 | | */ |
2503 | 34.2k | if (val >= 0x110000) { |
2504 | 291 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2505 | 291 | "xmlParseStringCharRef: character reference out of bounds\n", |
2506 | 291 | val); |
2507 | 33.9k | } else if (IS_CHAR(val)) { |
2508 | 25.8k | return(val); |
2509 | 25.8k | } else { |
2510 | 8.06k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2511 | 8.06k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2512 | 8.06k | val); |
2513 | 8.06k | } |
2514 | 8.35k | return(0); |
2515 | 34.2k | } |
2516 | | |
2517 | | /** |
2518 | | * xmlParserHandlePEReference: |
2519 | | * @ctxt: the parser context |
2520 | | * |
2521 | | * [69] PEReference ::= '%' Name ';' |
2522 | | * |
2523 | | * [ WFC: No Recursion ] |
2524 | | * A parsed entity must not contain a recursive |
2525 | | * reference to itself, either directly or indirectly. |
2526 | | * |
2527 | | * [ WFC: Entity Declared ] |
2528 | | * In a document without any DTD, a document with only an internal DTD |
2529 | | * subset which contains no parameter entity references, or a document |
2530 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2531 | | * entity must precede any reference to it... |
2532 | | * |
2533 | | * [ VC: Entity Declared ] |
2534 | | * In a document with an external subset or external parameter entities |
2535 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2536 | | * must precede any reference to it... |
2537 | | * |
2538 | | * [ WFC: In DTD ] |
2539 | | * Parameter-entity references may only appear in the DTD. |
2540 | | * NOTE: misleading but this is handled. |
2541 | | * |
2542 | | * A PEReference may have been detected in the current input stream |
2543 | | * the handling is done accordingly to |
2544 | | * http://www.w3.org/TR/REC-xml#entproc |
2545 | | * i.e. |
2546 | | * - Included in literal in entity values |
2547 | | * - Included as Parameter Entity reference within DTDs |
2548 | | */ |
2549 | | void |
2550 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2551 | 0 | switch(ctxt->instate) { |
2552 | 0 | case XML_PARSER_CDATA_SECTION: |
2553 | 0 | return; |
2554 | 0 | case XML_PARSER_COMMENT: |
2555 | 0 | return; |
2556 | 0 | case XML_PARSER_START_TAG: |
2557 | 0 | return; |
2558 | 0 | case XML_PARSER_END_TAG: |
2559 | 0 | return; |
2560 | 0 | case XML_PARSER_EOF: |
2561 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_AT_EOF, NULL); |
2562 | 0 | return; |
2563 | 0 | case XML_PARSER_PROLOG: |
2564 | 0 | case XML_PARSER_START: |
2565 | 0 | case XML_PARSER_MISC: |
2566 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_PROLOG, NULL); |
2567 | 0 | return; |
2568 | 0 | case XML_PARSER_ENTITY_DECL: |
2569 | 0 | case XML_PARSER_CONTENT: |
2570 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
2571 | 0 | case XML_PARSER_PI: |
2572 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
2573 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
2574 | | /* we just ignore it there */ |
2575 | 0 | return; |
2576 | 0 | case XML_PARSER_EPILOG: |
2577 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_EPILOG, NULL); |
2578 | 0 | return; |
2579 | 0 | case XML_PARSER_ENTITY_VALUE: |
2580 | | /* |
2581 | | * NOTE: in the case of entity values, we don't do the |
2582 | | * substitution here since we need the literal |
2583 | | * entity value to be able to save the internal |
2584 | | * subset of the document. |
2585 | | * This will be handled by xmlStringDecodeEntities |
2586 | | */ |
2587 | 0 | return; |
2588 | 0 | case XML_PARSER_DTD: |
2589 | | /* |
2590 | | * [WFC: Well-Formedness Constraint: PEs in Internal Subset] |
2591 | | * In the internal DTD subset, parameter-entity references |
2592 | | * can occur only where markup declarations can occur, not |
2593 | | * within markup declarations. |
2594 | | * In that case this is handled in xmlParseMarkupDecl |
2595 | | */ |
2596 | 0 | if ((ctxt->external == 0) && (ctxt->inputNr == 1)) |
2597 | 0 | return; |
2598 | 0 | if (IS_BLANK_CH(NXT(1)) || NXT(1) == 0) |
2599 | 0 | return; |
2600 | 0 | break; |
2601 | 0 | case XML_PARSER_IGNORE: |
2602 | 0 | return; |
2603 | 0 | } |
2604 | | |
2605 | 0 | xmlParsePEReference(ctxt); |
2606 | 0 | } |
2607 | | |
2608 | | /* |
2609 | | * Macro used to grow the current buffer. |
2610 | | * buffer##_size is expected to be a size_t |
2611 | | * mem_error: is expected to handle memory allocation failures |
2612 | | */ |
2613 | 175k | #define growBuffer(buffer, n) { \ |
2614 | 175k | xmlChar *tmp; \ |
2615 | 175k | size_t new_size = buffer##_size * 2 + n; \ |
2616 | 175k | if (new_size < buffer##_size) goto mem_error; \ |
2617 | 175k | tmp = (xmlChar *) xmlRealloc(buffer, new_size); \ |
2618 | 175k | if (tmp == NULL) goto mem_error; \ |
2619 | 175k | buffer = tmp; \ |
2620 | 175k | buffer##_size = new_size; \ |
2621 | 175k | } |
2622 | | |
2623 | | /** |
2624 | | * xmlStringLenDecodeEntities: |
2625 | | * @ctxt: the parser context |
2626 | | * @str: the input string |
2627 | | * @len: the string length |
2628 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2629 | | * @end: an end marker xmlChar, 0 if none |
2630 | | * @end2: an end marker xmlChar, 0 if none |
2631 | | * @end3: an end marker xmlChar, 0 if none |
2632 | | * |
2633 | | * Takes a entity string content and process to do the adequate substitutions. |
2634 | | * |
2635 | | * [67] Reference ::= EntityRef | CharRef |
2636 | | * |
2637 | | * [69] PEReference ::= '%' Name ';' |
2638 | | * |
2639 | | * Returns A newly allocated string with the substitution done. The caller |
2640 | | * must deallocate it ! |
2641 | | */ |
2642 | | xmlChar * |
2643 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2644 | 1.80M | int what, xmlChar end, xmlChar end2, xmlChar end3) { |
2645 | 1.80M | xmlChar *buffer = NULL; |
2646 | 1.80M | size_t buffer_size = 0; |
2647 | 1.80M | size_t nbchars = 0; |
2648 | | |
2649 | 1.80M | xmlChar *current = NULL; |
2650 | 1.80M | xmlChar *rep = NULL; |
2651 | 1.80M | const xmlChar *last; |
2652 | 1.80M | xmlEntityPtr ent; |
2653 | 1.80M | int c,l; |
2654 | | |
2655 | 1.80M | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2656 | 0 | return(NULL); |
2657 | 1.80M | last = str + len; |
2658 | | |
2659 | 1.80M | if (((ctxt->depth > 40) && |
2660 | 1.80M | ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2661 | 1.80M | (ctxt->depth > 1024)) { |
2662 | 161 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2663 | 161 | return(NULL); |
2664 | 161 | } |
2665 | | |
2666 | | /* |
2667 | | * allocate a translation buffer. |
2668 | | */ |
2669 | 1.80M | buffer_size = XML_PARSER_BIG_BUFFER_SIZE; |
2670 | 1.80M | buffer = (xmlChar *) xmlMallocAtomic(buffer_size); |
2671 | 1.80M | if (buffer == NULL) goto mem_error; |
2672 | | |
2673 | | /* |
2674 | | * OK loop until we reach one of the ending char or a size limit. |
2675 | | * we are operating on already parsed values. |
2676 | | */ |
2677 | 1.80M | if (str < last) |
2678 | 302k | c = CUR_SCHAR(str, l); |
2679 | 1.50M | else |
2680 | 1.50M | c = 0; |
2681 | 923M | while ((c != 0) && (c != end) && /* non input consuming loop */ |
2682 | 923M | (c != end2) && (c != end3) && |
2683 | 923M | (ctxt->instate != XML_PARSER_EOF)) { |
2684 | | |
2685 | 922M | if (c == 0) break; |
2686 | 922M | if ((c == '&') && (str[1] == '#')) { |
2687 | 34.2k | int val = xmlParseStringCharRef(ctxt, &str); |
2688 | 34.2k | if (val == 0) |
2689 | 8.35k | goto int_error; |
2690 | 25.8k | COPY_BUF(0,buffer,nbchars,val); |
2691 | 25.8k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2692 | 2.98k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2693 | 2.98k | } |
2694 | 922M | } else if ((c == '&') && (what & XML_SUBSTITUTE_REF)) { |
2695 | 2.01M | if (xmlParserDebugEntities) |
2696 | 0 | xmlGenericError(xmlGenericErrorContext, |
2697 | 0 | "String decoding Entity Reference: %.30s\n", |
2698 | 0 | str); |
2699 | 2.01M | ent = xmlParseStringEntityRef(ctxt, &str); |
2700 | 2.01M | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2701 | 2.01M | if (ent != NULL) |
2702 | 1.77M | ctxt->nbentities += ent->checked / 2; |
2703 | 2.01M | if ((ent != NULL) && |
2704 | 2.01M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
2705 | 15.2k | if (ent->content != NULL) { |
2706 | 15.2k | COPY_BUF(0,buffer,nbchars,ent->content[0]); |
2707 | 15.2k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2708 | 2.18k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2709 | 2.18k | } |
2710 | 15.2k | } else { |
2711 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
2712 | 0 | "predefined entity has no content\n"); |
2713 | 0 | goto int_error; |
2714 | 0 | } |
2715 | 1.99M | } else if ((ent != NULL) && (ent->content != NULL)) { |
2716 | 1.75M | ctxt->depth++; |
2717 | 1.75M | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2718 | 1.75M | 0, 0, 0); |
2719 | 1.75M | ctxt->depth--; |
2720 | 1.75M | if (rep == NULL) { |
2721 | 147k | ent->content[0] = 0; |
2722 | 147k | goto int_error; |
2723 | 147k | } |
2724 | | |
2725 | 1.61M | current = rep; |
2726 | 1.11G | while (*current != 0) { /* non input consuming loop */ |
2727 | 1.11G | buffer[nbchars++] = *current++; |
2728 | 1.11G | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2729 | 32.4k | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2730 | 0 | goto int_error; |
2731 | 97.3k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2732 | 97.3k | } |
2733 | 1.11G | } |
2734 | 1.61M | xmlFree(rep); |
2735 | 1.61M | rep = NULL; |
2736 | 1.61M | } else if (ent != NULL) { |
2737 | 3.51k | int i = xmlStrlen(ent->name); |
2738 | 3.51k | const xmlChar *cur = ent->name; |
2739 | | |
2740 | 3.51k | buffer[nbchars++] = '&'; |
2741 | 3.51k | if (nbchars + i + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2742 | 2.07k | growBuffer(buffer, i + XML_PARSER_BUFFER_SIZE); |
2743 | 2.07k | } |
2744 | 10.1k | for (;i > 0;i--) |
2745 | 6.65k | buffer[nbchars++] = *cur++; |
2746 | 3.51k | buffer[nbchars++] = ';'; |
2747 | 3.51k | } |
2748 | 920M | } else if (c == '%' && (what & XML_SUBSTITUTE_PEREF)) { |
2749 | 37.8k | if (xmlParserDebugEntities) |
2750 | 0 | xmlGenericError(xmlGenericErrorContext, |
2751 | 0 | "String decoding PE Reference: %.30s\n", str); |
2752 | 37.8k | ent = xmlParseStringPEReference(ctxt, &str); |
2753 | 37.8k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2754 | 37.8k | if (ent != NULL) |
2755 | 14.8k | ctxt->nbentities += ent->checked / 2; |
2756 | 37.8k | if (ent != NULL) { |
2757 | 14.8k | if (ent->content == NULL) { |
2758 | | /* |
2759 | | * Note: external parsed entities will not be loaded, |
2760 | | * it is not required for a non-validating parser to |
2761 | | * complete external PEReferences coming from the |
2762 | | * internal subset |
2763 | | */ |
2764 | 10 | if (((ctxt->options & XML_PARSE_NOENT) != 0) || |
2765 | 10 | ((ctxt->options & XML_PARSE_DTDVALID) != 0) || |
2766 | 10 | (ctxt->validate != 0)) { |
2767 | 0 | xmlLoadEntityContent(ctxt, ent); |
2768 | 10 | } else { |
2769 | 10 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
2770 | 10 | "not validating will not read content for PE entity %s\n", |
2771 | 10 | ent->name, NULL); |
2772 | 10 | } |
2773 | 10 | } |
2774 | 14.8k | ctxt->depth++; |
2775 | 14.8k | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2776 | 14.8k | 0, 0, 0); |
2777 | 14.8k | ctxt->depth--; |
2778 | 14.8k | if (rep == NULL) { |
2779 | 11.2k | if (ent->content != NULL) |
2780 | 11.2k | ent->content[0] = 0; |
2781 | 11.2k | goto int_error; |
2782 | 11.2k | } |
2783 | 3.53k | current = rep; |
2784 | 7.23M | while (*current != 0) { /* non input consuming loop */ |
2785 | 7.22M | buffer[nbchars++] = *current++; |
2786 | 7.22M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2787 | 1.23k | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2788 | 0 | goto int_error; |
2789 | 3.70k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2790 | 3.70k | } |
2791 | 7.22M | } |
2792 | 3.53k | xmlFree(rep); |
2793 | 3.53k | rep = NULL; |
2794 | 3.53k | } |
2795 | 920M | } else { |
2796 | 920M | COPY_BUF(l,buffer,nbchars,c); |
2797 | 920M | str += l; |
2798 | 920M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2799 | 213k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2800 | 213k | } |
2801 | 920M | } |
2802 | 922M | if (str < last) |
2803 | 921M | c = CUR_SCHAR(str, l); |
2804 | 135k | else |
2805 | 135k | c = 0; |
2806 | 922M | } |
2807 | 1.63M | buffer[nbchars] = 0; |
2808 | 1.63M | return(buffer); |
2809 | | |
2810 | 0 | mem_error: |
2811 | 0 | xmlErrMemory(ctxt, NULL); |
2812 | 166k | int_error: |
2813 | 166k | if (rep != NULL) |
2814 | 0 | xmlFree(rep); |
2815 | 166k | if (buffer != NULL) |
2816 | 166k | xmlFree(buffer); |
2817 | 166k | return(NULL); |
2818 | 0 | } |
2819 | | |
2820 | | /** |
2821 | | * xmlStringDecodeEntities: |
2822 | | * @ctxt: the parser context |
2823 | | * @str: the input string |
2824 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2825 | | * @end: an end marker xmlChar, 0 if none |
2826 | | * @end2: an end marker xmlChar, 0 if none |
2827 | | * @end3: an end marker xmlChar, 0 if none |
2828 | | * |
2829 | | * Takes a entity string content and process to do the adequate substitutions. |
2830 | | * |
2831 | | * [67] Reference ::= EntityRef | CharRef |
2832 | | * |
2833 | | * [69] PEReference ::= '%' Name ';' |
2834 | | * |
2835 | | * Returns A newly allocated string with the substitution done. The caller |
2836 | | * must deallocate it ! |
2837 | | */ |
2838 | | xmlChar * |
2839 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int what, |
2840 | 1.80M | xmlChar end, xmlChar end2, xmlChar end3) { |
2841 | 1.80M | if ((ctxt == NULL) || (str == NULL)) return(NULL); |
2842 | 1.80M | return(xmlStringLenDecodeEntities(ctxt, str, xmlStrlen(str), what, |
2843 | 1.80M | end, end2, end3)); |
2844 | 1.80M | } |
2845 | | |
2846 | | /************************************************************************ |
2847 | | * * |
2848 | | * Commodity functions, cleanup needed ? * |
2849 | | * * |
2850 | | ************************************************************************/ |
2851 | | |
2852 | | /** |
2853 | | * areBlanks: |
2854 | | * @ctxt: an XML parser context |
2855 | | * @str: a xmlChar * |
2856 | | * @len: the size of @str |
2857 | | * @blank_chars: we know the chars are blanks |
2858 | | * |
2859 | | * Is this a sequence of blank chars that one can ignore ? |
2860 | | * |
2861 | | * Returns 1 if ignorable 0 otherwise. |
2862 | | */ |
2863 | | |
2864 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2865 | 1.84M | int blank_chars) { |
2866 | 1.84M | int i, ret; |
2867 | 1.84M | xmlNodePtr lastChild; |
2868 | | |
2869 | | /* |
2870 | | * Don't spend time trying to differentiate them, the same callback is |
2871 | | * used ! |
2872 | | */ |
2873 | 1.84M | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2874 | 1.84M | return(0); |
2875 | | |
2876 | | /* |
2877 | | * Check for xml:space value. |
2878 | | */ |
2879 | 0 | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2880 | 0 | (*(ctxt->space) == -2)) |
2881 | 0 | return(0); |
2882 | | |
2883 | | /* |
2884 | | * Check that the string is made of blanks |
2885 | | */ |
2886 | 0 | if (blank_chars == 0) { |
2887 | 0 | for (i = 0;i < len;i++) |
2888 | 0 | if (!(IS_BLANK_CH(str[i]))) return(0); |
2889 | 0 | } |
2890 | | |
2891 | | /* |
2892 | | * Look if the element is mixed content in the DTD if available |
2893 | | */ |
2894 | 0 | if (ctxt->node == NULL) return(0); |
2895 | 0 | if (ctxt->myDoc != NULL) { |
2896 | 0 | ret = xmlIsMixedElement(ctxt->myDoc, ctxt->node->name); |
2897 | 0 | if (ret == 0) return(1); |
2898 | 0 | if (ret == 1) return(0); |
2899 | 0 | } |
2900 | | |
2901 | | /* |
2902 | | * Otherwise, heuristic :-\ |
2903 | | */ |
2904 | 0 | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2905 | 0 | if ((ctxt->node->children == NULL) && |
2906 | 0 | (RAW == '<') && (NXT(1) == '/')) return(0); |
2907 | | |
2908 | 0 | lastChild = xmlGetLastChild(ctxt->node); |
2909 | 0 | if (lastChild == NULL) { |
2910 | 0 | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2911 | 0 | (ctxt->node->content != NULL)) return(0); |
2912 | 0 | } else if (xmlNodeIsText(lastChild)) |
2913 | 0 | return(0); |
2914 | 0 | else if ((ctxt->node->children != NULL) && |
2915 | 0 | (xmlNodeIsText(ctxt->node->children))) |
2916 | 0 | return(0); |
2917 | 0 | return(1); |
2918 | 0 | } |
2919 | | |
2920 | | /************************************************************************ |
2921 | | * * |
2922 | | * Extra stuff for namespace support * |
2923 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2924 | | * * |
2925 | | ************************************************************************/ |
2926 | | |
2927 | | /** |
2928 | | * xmlSplitQName: |
2929 | | * @ctxt: an XML parser context |
2930 | | * @name: an XML parser context |
2931 | | * @prefix: a xmlChar ** |
2932 | | * |
2933 | | * parse an UTF8 encoded XML qualified name string |
2934 | | * |
2935 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2936 | | * |
2937 | | * [NS 6] Prefix ::= NCName |
2938 | | * |
2939 | | * [NS 7] LocalPart ::= NCName |
2940 | | * |
2941 | | * Returns the local part, and prefix is updated |
2942 | | * to get the Prefix if any. |
2943 | | */ |
2944 | | |
2945 | | xmlChar * |
2946 | 29.7k | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefix) { |
2947 | 29.7k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2948 | 29.7k | xmlChar *buffer = NULL; |
2949 | 29.7k | int len = 0; |
2950 | 29.7k | int max = XML_MAX_NAMELEN; |
2951 | 29.7k | xmlChar *ret = NULL; |
2952 | 29.7k | const xmlChar *cur = name; |
2953 | 29.7k | int c; |
2954 | | |
2955 | 29.7k | if (prefix == NULL) return(NULL); |
2956 | 29.7k | *prefix = NULL; |
2957 | | |
2958 | 29.7k | if (cur == NULL) return(NULL); |
2959 | | |
2960 | | #ifndef XML_XML_NAMESPACE |
2961 | | /* xml: prefix is not really a namespace */ |
2962 | | if ((cur[0] == 'x') && (cur[1] == 'm') && |
2963 | | (cur[2] == 'l') && (cur[3] == ':')) |
2964 | | return(xmlStrdup(name)); |
2965 | | #endif |
2966 | | |
2967 | | /* nasty but well=formed */ |
2968 | 29.7k | if (cur[0] == ':') |
2969 | 690 | return(xmlStrdup(name)); |
2970 | | |
2971 | 29.0k | c = *cur++; |
2972 | 759k | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
2973 | 730k | buf[len++] = c; |
2974 | 730k | c = *cur++; |
2975 | 730k | } |
2976 | 29.0k | if (len >= max) { |
2977 | | /* |
2978 | | * Okay someone managed to make a huge name, so he's ready to pay |
2979 | | * for the processing speed. |
2980 | | */ |
2981 | 4.59k | max = len * 2; |
2982 | | |
2983 | 4.59k | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
2984 | 4.59k | if (buffer == NULL) { |
2985 | 0 | xmlErrMemory(ctxt, NULL); |
2986 | 0 | return(NULL); |
2987 | 0 | } |
2988 | 4.59k | memcpy(buffer, buf, len); |
2989 | 1.47M | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
2990 | 1.47M | if (len + 10 > max) { |
2991 | 4.64k | xmlChar *tmp; |
2992 | | |
2993 | 4.64k | max *= 2; |
2994 | 4.64k | tmp = (xmlChar *) xmlRealloc(buffer, |
2995 | 4.64k | max * sizeof(xmlChar)); |
2996 | 4.64k | if (tmp == NULL) { |
2997 | 0 | xmlFree(buffer); |
2998 | 0 | xmlErrMemory(ctxt, NULL); |
2999 | 0 | return(NULL); |
3000 | 0 | } |
3001 | 4.64k | buffer = tmp; |
3002 | 4.64k | } |
3003 | 1.47M | buffer[len++] = c; |
3004 | 1.47M | c = *cur++; |
3005 | 1.47M | } |
3006 | 4.59k | buffer[len] = 0; |
3007 | 4.59k | } |
3008 | | |
3009 | 29.0k | if ((c == ':') && (*cur == 0)) { |
3010 | 1.10k | if (buffer != NULL) |
3011 | 278 | xmlFree(buffer); |
3012 | 1.10k | *prefix = NULL; |
3013 | 1.10k | return(xmlStrdup(name)); |
3014 | 1.10k | } |
3015 | | |
3016 | 27.9k | if (buffer == NULL) |
3017 | 23.6k | ret = xmlStrndup(buf, len); |
3018 | 4.31k | else { |
3019 | 4.31k | ret = buffer; |
3020 | 4.31k | buffer = NULL; |
3021 | 4.31k | max = XML_MAX_NAMELEN; |
3022 | 4.31k | } |
3023 | | |
3024 | | |
3025 | 27.9k | if (c == ':') { |
3026 | 8.83k | c = *cur; |
3027 | 8.83k | *prefix = ret; |
3028 | 8.83k | if (c == 0) { |
3029 | 0 | return(xmlStrndup(BAD_CAST "", 0)); |
3030 | 0 | } |
3031 | 8.83k | len = 0; |
3032 | | |
3033 | | /* |
3034 | | * Check that the first character is proper to start |
3035 | | * a new name |
3036 | | */ |
3037 | 8.83k | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3038 | 8.83k | ((c >= 0x41) && (c <= 0x5A)) || |
3039 | 8.83k | (c == '_') || (c == ':'))) { |
3040 | 1.95k | int l; |
3041 | 1.95k | int first = CUR_SCHAR(cur, l); |
3042 | | |
3043 | 1.95k | if (!IS_LETTER(first) && (first != '_')) { |
3044 | 31 | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3045 | 31 | "Name %s is not XML Namespace compliant\n", |
3046 | 31 | name); |
3047 | 31 | } |
3048 | 1.95k | } |
3049 | 8.83k | cur++; |
3050 | | |
3051 | 502k | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3052 | 493k | buf[len++] = c; |
3053 | 493k | c = *cur++; |
3054 | 493k | } |
3055 | 8.83k | if (len >= max) { |
3056 | | /* |
3057 | | * Okay someone managed to make a huge name, so he's ready to pay |
3058 | | * for the processing speed. |
3059 | | */ |
3060 | 4.40k | max = len * 2; |
3061 | | |
3062 | 4.40k | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3063 | 4.40k | if (buffer == NULL) { |
3064 | 0 | xmlErrMemory(ctxt, NULL); |
3065 | 0 | return(NULL); |
3066 | 0 | } |
3067 | 4.40k | memcpy(buffer, buf, len); |
3068 | 25.7M | while (c != 0) { /* tested bigname2.xml */ |
3069 | 25.7M | if (len + 10 > max) { |
3070 | 5.09k | xmlChar *tmp; |
3071 | | |
3072 | 5.09k | max *= 2; |
3073 | 5.09k | tmp = (xmlChar *) xmlRealloc(buffer, |
3074 | 5.09k | max * sizeof(xmlChar)); |
3075 | 5.09k | if (tmp == NULL) { |
3076 | 0 | xmlErrMemory(ctxt, NULL); |
3077 | 0 | xmlFree(buffer); |
3078 | 0 | return(NULL); |
3079 | 0 | } |
3080 | 5.09k | buffer = tmp; |
3081 | 5.09k | } |
3082 | 25.7M | buffer[len++] = c; |
3083 | 25.7M | c = *cur++; |
3084 | 25.7M | } |
3085 | 4.40k | buffer[len] = 0; |
3086 | 4.40k | } |
3087 | | |
3088 | 8.83k | if (buffer == NULL) |
3089 | 4.42k | ret = xmlStrndup(buf, len); |
3090 | 4.40k | else { |
3091 | 4.40k | ret = buffer; |
3092 | 4.40k | } |
3093 | 8.83k | } |
3094 | | |
3095 | 27.9k | return(ret); |
3096 | 27.9k | } |
3097 | | |
3098 | | /************************************************************************ |
3099 | | * * |
3100 | | * The parser itself * |
3101 | | * Relates to http://www.w3.org/TR/REC-xml * |
3102 | | * * |
3103 | | ************************************************************************/ |
3104 | | |
3105 | | /************************************************************************ |
3106 | | * * |
3107 | | * Routines to parse Name, NCName and NmToken * |
3108 | | * * |
3109 | | ************************************************************************/ |
3110 | | #ifdef DEBUG |
3111 | | static unsigned long nbParseName = 0; |
3112 | | static unsigned long nbParseNmToken = 0; |
3113 | | static unsigned long nbParseNCName = 0; |
3114 | | static unsigned long nbParseNCNameComplex = 0; |
3115 | | static unsigned long nbParseNameComplex = 0; |
3116 | | static unsigned long nbParseStringName = 0; |
3117 | | #endif |
3118 | | |
3119 | | /* |
3120 | | * The two following functions are related to the change of accepted |
3121 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3122 | | * They correspond to the modified production [4] and the new production [4a] |
3123 | | * changes in that revision. Also note that the macros used for the |
3124 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3125 | | * anymore. |
3126 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3127 | | * new XML_PARSE_OLD10 option is given to the parser. |
3128 | | */ |
3129 | | static int |
3130 | 2.43M | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3131 | 2.43M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3132 | | /* |
3133 | | * Use the new checks of production [4] [4a] amd [5] of the |
3134 | | * Update 5 of XML-1.0 |
3135 | | */ |
3136 | 2.43M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3137 | 2.43M | (((c >= 'a') && (c <= 'z')) || |
3138 | 2.43M | ((c >= 'A') && (c <= 'Z')) || |
3139 | 2.43M | (c == '_') || (c == ':') || |
3140 | 2.43M | ((c >= 0xC0) && (c <= 0xD6)) || |
3141 | 2.43M | ((c >= 0xD8) && (c <= 0xF6)) || |
3142 | 2.43M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3143 | 2.43M | ((c >= 0x370) && (c <= 0x37D)) || |
3144 | 2.43M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3145 | 2.43M | ((c >= 0x200C) && (c <= 0x200D)) || |
3146 | 2.43M | ((c >= 0x2070) && (c <= 0x218F)) || |
3147 | 2.43M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3148 | 2.43M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3149 | 2.43M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3150 | 2.43M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3151 | 2.43M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3152 | 2.14M | return(1); |
3153 | 2.43M | } else { |
3154 | 0 | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3155 | 0 | return(1); |
3156 | 0 | } |
3157 | 296k | return(0); |
3158 | 2.43M | } |
3159 | | |
3160 | | static int |
3161 | 139M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3162 | 139M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3163 | | /* |
3164 | | * Use the new checks of production [4] [4a] amd [5] of the |
3165 | | * Update 5 of XML-1.0 |
3166 | | */ |
3167 | 139M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3168 | 139M | (((c >= 'a') && (c <= 'z')) || |
3169 | 139M | ((c >= 'A') && (c <= 'Z')) || |
3170 | 139M | ((c >= '0') && (c <= '9')) || /* !start */ |
3171 | 139M | (c == '_') || (c == ':') || |
3172 | 139M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3173 | 139M | ((c >= 0xC0) && (c <= 0xD6)) || |
3174 | 139M | ((c >= 0xD8) && (c <= 0xF6)) || |
3175 | 139M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3176 | 139M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3177 | 139M | ((c >= 0x370) && (c <= 0x37D)) || |
3178 | 139M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3179 | 139M | ((c >= 0x200C) && (c <= 0x200D)) || |
3180 | 139M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3181 | 139M | ((c >= 0x2070) && (c <= 0x218F)) || |
3182 | 139M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3183 | 139M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3184 | 139M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3185 | 139M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3186 | 139M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3187 | 137M | return(1); |
3188 | 139M | } else { |
3189 | 0 | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3190 | 0 | (c == '.') || (c == '-') || |
3191 | 0 | (c == '_') || (c == ':') || |
3192 | 0 | (IS_COMBINING(c)) || |
3193 | 0 | (IS_EXTENDER(c))) |
3194 | 0 | return(1); |
3195 | 0 | } |
3196 | 2.16M | return(0); |
3197 | 139M | } |
3198 | | |
3199 | | static xmlChar * xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, |
3200 | | int *len, int *alloc, int normalize); |
3201 | | |
3202 | | static const xmlChar * |
3203 | 3.76M | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3204 | 3.76M | int len = 0, l; |
3205 | 3.76M | int c; |
3206 | 3.76M | int count = 0; |
3207 | | |
3208 | | #ifdef DEBUG |
3209 | | nbParseNameComplex++; |
3210 | | #endif |
3211 | | |
3212 | | /* |
3213 | | * Handler for more complex cases |
3214 | | */ |
3215 | 3.76M | GROW; |
3216 | 3.76M | if (ctxt->instate == XML_PARSER_EOF) |
3217 | 0 | return(NULL); |
3218 | 3.76M | c = CUR_CHAR(l); |
3219 | 3.76M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3220 | | /* |
3221 | | * Use the new checks of production [4] [4a] amd [5] of the |
3222 | | * Update 5 of XML-1.0 |
3223 | | */ |
3224 | 3.76M | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3225 | 3.76M | (!(((c >= 'a') && (c <= 'z')) || |
3226 | 3.76M | ((c >= 'A') && (c <= 'Z')) || |
3227 | 3.76M | (c == '_') || (c == ':') || |
3228 | 3.76M | ((c >= 0xC0) && (c <= 0xD6)) || |
3229 | 3.76M | ((c >= 0xD8) && (c <= 0xF6)) || |
3230 | 3.76M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3231 | 3.76M | ((c >= 0x370) && (c <= 0x37D)) || |
3232 | 3.76M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3233 | 3.76M | ((c >= 0x200C) && (c <= 0x200D)) || |
3234 | 3.76M | ((c >= 0x2070) && (c <= 0x218F)) || |
3235 | 3.76M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3236 | 3.76M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3237 | 3.76M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3238 | 3.76M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3239 | 3.76M | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3240 | 3.32M | return(NULL); |
3241 | 3.32M | } |
3242 | 449k | len += l; |
3243 | 449k | NEXTL(l); |
3244 | 449k | c = CUR_CHAR(l); |
3245 | 63.7M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3246 | 63.7M | (((c >= 'a') && (c <= 'z')) || |
3247 | 63.6M | ((c >= 'A') && (c <= 'Z')) || |
3248 | 63.6M | ((c >= '0') && (c <= '9')) || /* !start */ |
3249 | 63.6M | (c == '_') || (c == ':') || |
3250 | 63.6M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3251 | 63.6M | ((c >= 0xC0) && (c <= 0xD6)) || |
3252 | 63.6M | ((c >= 0xD8) && (c <= 0xF6)) || |
3253 | 63.6M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3254 | 63.6M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3255 | 63.6M | ((c >= 0x370) && (c <= 0x37D)) || |
3256 | 63.6M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3257 | 63.6M | ((c >= 0x200C) && (c <= 0x200D)) || |
3258 | 63.6M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3259 | 63.6M | ((c >= 0x2070) && (c <= 0x218F)) || |
3260 | 63.6M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3261 | 63.6M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3262 | 63.6M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3263 | 63.6M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3264 | 63.6M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3265 | 63.6M | )) { |
3266 | 63.2M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3267 | 552k | count = 0; |
3268 | 552k | GROW; |
3269 | 552k | if (ctxt->instate == XML_PARSER_EOF) |
3270 | 0 | return(NULL); |
3271 | 552k | } |
3272 | 63.2M | len += l; |
3273 | 63.2M | NEXTL(l); |
3274 | 63.2M | c = CUR_CHAR(l); |
3275 | 63.2M | } |
3276 | 449k | } else { |
3277 | 0 | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3278 | 0 | (!IS_LETTER(c) && (c != '_') && |
3279 | 0 | (c != ':'))) { |
3280 | 0 | return(NULL); |
3281 | 0 | } |
3282 | 0 | len += l; |
3283 | 0 | NEXTL(l); |
3284 | 0 | c = CUR_CHAR(l); |
3285 | |
|
3286 | 0 | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3287 | 0 | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3288 | 0 | (c == '.') || (c == '-') || |
3289 | 0 | (c == '_') || (c == ':') || |
3290 | 0 | (IS_COMBINING(c)) || |
3291 | 0 | (IS_EXTENDER(c)))) { |
3292 | 0 | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3293 | 0 | count = 0; |
3294 | 0 | GROW; |
3295 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
3296 | 0 | return(NULL); |
3297 | 0 | } |
3298 | 0 | len += l; |
3299 | 0 | NEXTL(l); |
3300 | 0 | c = CUR_CHAR(l); |
3301 | 0 | } |
3302 | 0 | } |
3303 | 449k | if ((len > XML_MAX_NAME_LENGTH) && |
3304 | 449k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3305 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3306 | 0 | return(NULL); |
3307 | 0 | } |
3308 | 449k | if (ctxt->input->cur - ctxt->input->base < len) { |
3309 | | /* |
3310 | | * There were a couple of bugs where PERefs lead to to a change |
3311 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3312 | | * pointer to xmlDictLookup. |
3313 | | */ |
3314 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3315 | 0 | "unexpected change of input buffer"); |
3316 | 0 | return (NULL); |
3317 | 0 | } |
3318 | 449k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3319 | 519 | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len)); |
3320 | 448k | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len)); |
3321 | 449k | } |
3322 | | |
3323 | | /** |
3324 | | * xmlParseName: |
3325 | | * @ctxt: an XML parser context |
3326 | | * |
3327 | | * parse an XML name. |
3328 | | * |
3329 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3330 | | * CombiningChar | Extender |
3331 | | * |
3332 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3333 | | * |
3334 | | * [6] Names ::= Name (#x20 Name)* |
3335 | | * |
3336 | | * Returns the Name parsed or NULL |
3337 | | */ |
3338 | | |
3339 | | const xmlChar * |
3340 | 15.7M | xmlParseName(xmlParserCtxtPtr ctxt) { |
3341 | 15.7M | const xmlChar *in; |
3342 | 15.7M | const xmlChar *ret; |
3343 | 15.7M | int count = 0; |
3344 | | |
3345 | 15.7M | GROW; |
3346 | | |
3347 | | #ifdef DEBUG |
3348 | | nbParseName++; |
3349 | | #endif |
3350 | | |
3351 | | /* |
3352 | | * Accelerator for simple ASCII names |
3353 | | */ |
3354 | 15.7M | in = ctxt->input->cur; |
3355 | 15.7M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3356 | 15.7M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3357 | 15.7M | (*in == '_') || (*in == ':')) { |
3358 | 12.1M | in++; |
3359 | 29.4M | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3360 | 29.4M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3361 | 29.4M | ((*in >= 0x30) && (*in <= 0x39)) || |
3362 | 29.4M | (*in == '_') || (*in == '-') || |
3363 | 29.4M | (*in == ':') || (*in == '.')) |
3364 | 17.3M | in++; |
3365 | 12.1M | if ((*in > 0) && (*in < 0x80)) { |
3366 | 11.9M | count = in - ctxt->input->cur; |
3367 | 11.9M | if ((count > XML_MAX_NAME_LENGTH) && |
3368 | 11.9M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3369 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3370 | 0 | return(NULL); |
3371 | 0 | } |
3372 | 11.9M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3373 | 11.9M | ctxt->input->cur = in; |
3374 | 11.9M | ctxt->input->col += count; |
3375 | 11.9M | if (ret == NULL) |
3376 | 0 | xmlErrMemory(ctxt, NULL); |
3377 | 11.9M | return(ret); |
3378 | 11.9M | } |
3379 | 12.1M | } |
3380 | | /* accelerator for special cases */ |
3381 | 3.76M | return(xmlParseNameComplex(ctxt)); |
3382 | 15.7M | } |
3383 | | |
3384 | | static const xmlChar * |
3385 | 379k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3386 | 379k | int len = 0, l; |
3387 | 379k | int c; |
3388 | 379k | int count = 0; |
3389 | 379k | size_t startPosition = 0; |
3390 | | |
3391 | | #ifdef DEBUG |
3392 | | nbParseNCNameComplex++; |
3393 | | #endif |
3394 | | |
3395 | | /* |
3396 | | * Handler for more complex cases |
3397 | | */ |
3398 | 379k | GROW; |
3399 | 379k | startPosition = CUR_PTR - BASE_PTR; |
3400 | 379k | c = CUR_CHAR(l); |
3401 | 379k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3402 | 379k | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3403 | 312k | return(NULL); |
3404 | 312k | } |
3405 | | |
3406 | 45.8M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3407 | 45.8M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3408 | 45.8M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3409 | 442k | if ((len > XML_MAX_NAME_LENGTH) && |
3410 | 442k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3411 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3412 | 0 | return(NULL); |
3413 | 0 | } |
3414 | 442k | count = 0; |
3415 | 442k | GROW; |
3416 | 442k | if (ctxt->instate == XML_PARSER_EOF) |
3417 | 0 | return(NULL); |
3418 | 442k | } |
3419 | 45.8M | len += l; |
3420 | 45.8M | NEXTL(l); |
3421 | 45.8M | c = CUR_CHAR(l); |
3422 | 45.8M | if (c == 0) { |
3423 | 3.03k | count = 0; |
3424 | | /* |
3425 | | * when shrinking to extend the buffer we really need to preserve |
3426 | | * the part of the name we already parsed. Hence rolling back |
3427 | | * by current length. |
3428 | | */ |
3429 | 3.03k | ctxt->input->cur -= l; |
3430 | 3.03k | GROW; |
3431 | 3.03k | if (ctxt->instate == XML_PARSER_EOF) |
3432 | 0 | return(NULL); |
3433 | 3.03k | ctxt->input->cur += l; |
3434 | 3.03k | c = CUR_CHAR(l); |
3435 | 3.03k | } |
3436 | 45.8M | } |
3437 | 67.5k | if ((len > XML_MAX_NAME_LENGTH) && |
3438 | 67.5k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3439 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3440 | 0 | return(NULL); |
3441 | 0 | } |
3442 | 67.5k | return(xmlDictLookup(ctxt->dict, (BASE_PTR + startPosition), len)); |
3443 | 67.5k | } |
3444 | | |
3445 | | /** |
3446 | | * xmlParseNCName: |
3447 | | * @ctxt: an XML parser context |
3448 | | * @len: length of the string parsed |
3449 | | * |
3450 | | * parse an XML name. |
3451 | | * |
3452 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3453 | | * CombiningChar | Extender |
3454 | | * |
3455 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3456 | | * |
3457 | | * Returns the Name parsed or NULL |
3458 | | */ |
3459 | | |
3460 | | static const xmlChar * |
3461 | 6.40M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3462 | 6.40M | const xmlChar *in, *e; |
3463 | 6.40M | const xmlChar *ret; |
3464 | 6.40M | int count = 0; |
3465 | | |
3466 | | #ifdef DEBUG |
3467 | | nbParseNCName++; |
3468 | | #endif |
3469 | | |
3470 | | /* |
3471 | | * Accelerator for simple ASCII names |
3472 | | */ |
3473 | 6.40M | in = ctxt->input->cur; |
3474 | 6.40M | e = ctxt->input->end; |
3475 | 6.40M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3476 | 6.40M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3477 | 6.40M | (*in == '_')) && (in < e)) { |
3478 | 6.06M | in++; |
3479 | 43.8M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3480 | 43.8M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3481 | 43.8M | ((*in >= 0x30) && (*in <= 0x39)) || |
3482 | 43.8M | (*in == '_') || (*in == '-') || |
3483 | 43.8M | (*in == '.')) && (in < e)) |
3484 | 37.8M | in++; |
3485 | 6.06M | if (in >= e) |
3486 | 3.33k | goto complex; |
3487 | 6.05M | if ((*in > 0) && (*in < 0x80)) { |
3488 | 6.02M | count = in - ctxt->input->cur; |
3489 | 6.02M | if ((count > XML_MAX_NAME_LENGTH) && |
3490 | 6.02M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3491 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3492 | 0 | return(NULL); |
3493 | 0 | } |
3494 | 6.02M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3495 | 6.02M | ctxt->input->cur = in; |
3496 | 6.02M | ctxt->input->col += count; |
3497 | 6.02M | if (ret == NULL) { |
3498 | 0 | xmlErrMemory(ctxt, NULL); |
3499 | 0 | } |
3500 | 6.02M | return(ret); |
3501 | 6.02M | } |
3502 | 6.05M | } |
3503 | 379k | complex: |
3504 | 379k | return(xmlParseNCNameComplex(ctxt)); |
3505 | 6.40M | } |
3506 | | |
3507 | | /** |
3508 | | * xmlParseNameAndCompare: |
3509 | | * @ctxt: an XML parser context |
3510 | | * |
3511 | | * parse an XML name and compares for match |
3512 | | * (specialized for endtag parsing) |
3513 | | * |
3514 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3515 | | * and the name for mismatch |
3516 | | */ |
3517 | | |
3518 | | static const xmlChar * |
3519 | 191k | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3520 | 191k | register const xmlChar *cmp = other; |
3521 | 191k | register const xmlChar *in; |
3522 | 191k | const xmlChar *ret; |
3523 | | |
3524 | 191k | GROW; |
3525 | 191k | if (ctxt->instate == XML_PARSER_EOF) |
3526 | 0 | return(NULL); |
3527 | | |
3528 | 191k | in = ctxt->input->cur; |
3529 | 935k | while (*in != 0 && *in == *cmp) { |
3530 | 744k | ++in; |
3531 | 744k | ++cmp; |
3532 | 744k | } |
3533 | 191k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3534 | | /* success */ |
3535 | 174k | ctxt->input->col += in - ctxt->input->cur; |
3536 | 174k | ctxt->input->cur = in; |
3537 | 174k | return (const xmlChar*) 1; |
3538 | 174k | } |
3539 | | /* failure (or end of input buffer), check with full function */ |
3540 | 16.2k | ret = xmlParseName (ctxt); |
3541 | | /* strings coming from the dictionary direct compare possible */ |
3542 | 16.2k | if (ret == other) { |
3543 | 740 | return (const xmlChar*) 1; |
3544 | 740 | } |
3545 | 15.4k | return ret; |
3546 | 16.2k | } |
3547 | | |
3548 | | /** |
3549 | | * xmlParseStringName: |
3550 | | * @ctxt: an XML parser context |
3551 | | * @str: a pointer to the string pointer (IN/OUT) |
3552 | | * |
3553 | | * parse an XML name. |
3554 | | * |
3555 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3556 | | * CombiningChar | Extender |
3557 | | * |
3558 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3559 | | * |
3560 | | * [6] Names ::= Name (#x20 Name)* |
3561 | | * |
3562 | | * Returns the Name parsed or NULL. The @str pointer |
3563 | | * is updated to the current location in the string. |
3564 | | */ |
3565 | | |
3566 | | static xmlChar * |
3567 | 2.08M | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3568 | 2.08M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3569 | 2.08M | const xmlChar *cur = *str; |
3570 | 2.08M | int len = 0, l; |
3571 | 2.08M | int c; |
3572 | | |
3573 | | #ifdef DEBUG |
3574 | | nbParseStringName++; |
3575 | | #endif |
3576 | | |
3577 | 2.08M | c = CUR_SCHAR(cur, l); |
3578 | 2.08M | if (!xmlIsNameStartChar(ctxt, c)) { |
3579 | 28.3k | return(NULL); |
3580 | 28.3k | } |
3581 | | |
3582 | 2.06M | COPY_BUF(l,buf,len,c); |
3583 | 2.06M | cur += l; |
3584 | 2.06M | c = CUR_SCHAR(cur, l); |
3585 | 2.69M | while (xmlIsNameChar(ctxt, c)) { |
3586 | 642k | COPY_BUF(l,buf,len,c); |
3587 | 642k | cur += l; |
3588 | 642k | c = CUR_SCHAR(cur, l); |
3589 | 642k | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3590 | | /* |
3591 | | * Okay someone managed to make a huge name, so he's ready to pay |
3592 | | * for the processing speed. |
3593 | | */ |
3594 | 5.76k | xmlChar *buffer; |
3595 | 5.76k | int max = len * 2; |
3596 | | |
3597 | 5.76k | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3598 | 5.76k | if (buffer == NULL) { |
3599 | 0 | xmlErrMemory(ctxt, NULL); |
3600 | 0 | return(NULL); |
3601 | 0 | } |
3602 | 5.76k | memcpy(buffer, buf, len); |
3603 | 4.26M | while (xmlIsNameChar(ctxt, c)) { |
3604 | 4.25M | if (len + 10 > max) { |
3605 | 5.13k | xmlChar *tmp; |
3606 | | |
3607 | 5.13k | if ((len > XML_MAX_NAME_LENGTH) && |
3608 | 5.13k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3609 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3610 | 0 | xmlFree(buffer); |
3611 | 0 | return(NULL); |
3612 | 0 | } |
3613 | 5.13k | max *= 2; |
3614 | 5.13k | tmp = (xmlChar *) xmlRealloc(buffer, |
3615 | 5.13k | max * sizeof(xmlChar)); |
3616 | 5.13k | if (tmp == NULL) { |
3617 | 0 | xmlErrMemory(ctxt, NULL); |
3618 | 0 | xmlFree(buffer); |
3619 | 0 | return(NULL); |
3620 | 0 | } |
3621 | 5.13k | buffer = tmp; |
3622 | 5.13k | } |
3623 | 4.25M | COPY_BUF(l,buffer,len,c); |
3624 | 4.25M | cur += l; |
3625 | 4.25M | c = CUR_SCHAR(cur, l); |
3626 | 4.25M | } |
3627 | 5.76k | buffer[len] = 0; |
3628 | 5.76k | *str = cur; |
3629 | 5.76k | return(buffer); |
3630 | 5.76k | } |
3631 | 642k | } |
3632 | 2.05M | if ((len > XML_MAX_NAME_LENGTH) && |
3633 | 2.05M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3634 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3635 | 0 | return(NULL); |
3636 | 0 | } |
3637 | 2.05M | *str = cur; |
3638 | 2.05M | return(xmlStrndup(buf, len)); |
3639 | 2.05M | } |
3640 | | |
3641 | | /** |
3642 | | * xmlParseNmtoken: |
3643 | | * @ctxt: an XML parser context |
3644 | | * |
3645 | | * parse an XML Nmtoken. |
3646 | | * |
3647 | | * [7] Nmtoken ::= (NameChar)+ |
3648 | | * |
3649 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3650 | | * |
3651 | | * Returns the Nmtoken parsed or NULL |
3652 | | */ |
3653 | | |
3654 | | xmlChar * |
3655 | 67.5k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3656 | 67.5k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3657 | 67.5k | int len = 0, l; |
3658 | 67.5k | int c; |
3659 | 67.5k | int count = 0; |
3660 | | |
3661 | | #ifdef DEBUG |
3662 | | nbParseNmToken++; |
3663 | | #endif |
3664 | | |
3665 | 67.5k | GROW; |
3666 | 67.5k | if (ctxt->instate == XML_PARSER_EOF) |
3667 | 0 | return(NULL); |
3668 | 67.5k | c = CUR_CHAR(l); |
3669 | | |
3670 | 1.21M | while (xmlIsNameChar(ctxt, c)) { |
3671 | 1.17M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3672 | 0 | count = 0; |
3673 | 0 | GROW; |
3674 | 0 | } |
3675 | 1.17M | COPY_BUF(l,buf,len,c); |
3676 | 1.17M | NEXTL(l); |
3677 | 1.17M | c = CUR_CHAR(l); |
3678 | 1.17M | if (c == 0) { |
3679 | 1.85k | count = 0; |
3680 | 1.85k | GROW; |
3681 | 1.85k | if (ctxt->instate == XML_PARSER_EOF) |
3682 | 0 | return(NULL); |
3683 | 1.85k | c = CUR_CHAR(l); |
3684 | 1.85k | } |
3685 | 1.17M | if (len >= XML_MAX_NAMELEN) { |
3686 | | /* |
3687 | | * Okay someone managed to make a huge token, so he's ready to pay |
3688 | | * for the processing speed. |
3689 | | */ |
3690 | 27.5k | xmlChar *buffer; |
3691 | 27.5k | int max = len * 2; |
3692 | | |
3693 | 27.5k | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3694 | 27.5k | if (buffer == NULL) { |
3695 | 0 | xmlErrMemory(ctxt, NULL); |
3696 | 0 | return(NULL); |
3697 | 0 | } |
3698 | 27.5k | memcpy(buffer, buf, len); |
3699 | 85.2M | while (xmlIsNameChar(ctxt, c)) { |
3700 | 85.2M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3701 | 830k | count = 0; |
3702 | 830k | GROW; |
3703 | 830k | if (ctxt->instate == XML_PARSER_EOF) { |
3704 | 0 | xmlFree(buffer); |
3705 | 0 | return(NULL); |
3706 | 0 | } |
3707 | 830k | } |
3708 | 85.2M | if (len + 10 > max) { |
3709 | 34.1k | xmlChar *tmp; |
3710 | | |
3711 | 34.1k | if ((max > XML_MAX_NAME_LENGTH) && |
3712 | 34.1k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3713 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3714 | 0 | xmlFree(buffer); |
3715 | 0 | return(NULL); |
3716 | 0 | } |
3717 | 34.1k | max *= 2; |
3718 | 34.1k | tmp = (xmlChar *) xmlRealloc(buffer, |
3719 | 34.1k | max * sizeof(xmlChar)); |
3720 | 34.1k | if (tmp == NULL) { |
3721 | 0 | xmlErrMemory(ctxt, NULL); |
3722 | 0 | xmlFree(buffer); |
3723 | 0 | return(NULL); |
3724 | 0 | } |
3725 | 34.1k | buffer = tmp; |
3726 | 34.1k | } |
3727 | 85.2M | COPY_BUF(l,buffer,len,c); |
3728 | 85.2M | NEXTL(l); |
3729 | 85.2M | c = CUR_CHAR(l); |
3730 | 85.2M | } |
3731 | 27.5k | buffer[len] = 0; |
3732 | 27.5k | return(buffer); |
3733 | 27.5k | } |
3734 | 1.17M | } |
3735 | 39.9k | if (len == 0) |
3736 | 15.1k | return(NULL); |
3737 | 24.8k | if ((len > XML_MAX_NAME_LENGTH) && |
3738 | 24.8k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3739 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3740 | 0 | return(NULL); |
3741 | 0 | } |
3742 | 24.8k | return(xmlStrndup(buf, len)); |
3743 | 24.8k | } |
3744 | | |
3745 | | /** |
3746 | | * xmlParseEntityValue: |
3747 | | * @ctxt: an XML parser context |
3748 | | * @orig: if non-NULL store a copy of the original entity value |
3749 | | * |
3750 | | * parse a value for ENTITY declarations |
3751 | | * |
3752 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3753 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3754 | | * |
3755 | | * Returns the EntityValue parsed with reference substituted or NULL |
3756 | | */ |
3757 | | |
3758 | | xmlChar * |
3759 | 39.3k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3760 | 39.3k | xmlChar *buf = NULL; |
3761 | 39.3k | int len = 0; |
3762 | 39.3k | int size = XML_PARSER_BUFFER_SIZE; |
3763 | 39.3k | int c, l; |
3764 | 39.3k | xmlChar stop; |
3765 | 39.3k | xmlChar *ret = NULL; |
3766 | 39.3k | const xmlChar *cur = NULL; |
3767 | 39.3k | xmlParserInputPtr input; |
3768 | | |
3769 | 39.3k | if (RAW == '"') stop = '"'; |
3770 | 12.0k | else if (RAW == '\'') stop = '\''; |
3771 | 0 | else { |
3772 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_STARTED, NULL); |
3773 | 0 | return(NULL); |
3774 | 0 | } |
3775 | 39.3k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
3776 | 39.3k | if (buf == NULL) { |
3777 | 0 | xmlErrMemory(ctxt, NULL); |
3778 | 0 | return(NULL); |
3779 | 0 | } |
3780 | | |
3781 | | /* |
3782 | | * The content of the entity definition is copied in a buffer. |
3783 | | */ |
3784 | | |
3785 | 39.3k | ctxt->instate = XML_PARSER_ENTITY_VALUE; |
3786 | 39.3k | input = ctxt->input; |
3787 | 39.3k | GROW; |
3788 | 39.3k | if (ctxt->instate == XML_PARSER_EOF) |
3789 | 0 | goto error; |
3790 | 39.3k | NEXT; |
3791 | 39.3k | c = CUR_CHAR(l); |
3792 | | /* |
3793 | | * NOTE: 4.4.5 Included in Literal |
3794 | | * When a parameter entity reference appears in a literal entity |
3795 | | * value, ... a single or double quote character in the replacement |
3796 | | * text is always treated as a normal data character and will not |
3797 | | * terminate the literal. |
3798 | | * In practice it means we stop the loop only when back at parsing |
3799 | | * the initial entity and the quote is found |
3800 | | */ |
3801 | 574M | while (((IS_CHAR(c)) && ((c != stop) || /* checked */ |
3802 | 574M | (ctxt->input != input))) && (ctxt->instate != XML_PARSER_EOF)) { |
3803 | 574M | if (len + 5 >= size) { |
3804 | 17.2k | xmlChar *tmp; |
3805 | | |
3806 | 17.2k | size *= 2; |
3807 | 17.2k | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
3808 | 17.2k | if (tmp == NULL) { |
3809 | 0 | xmlErrMemory(ctxt, NULL); |
3810 | 0 | goto error; |
3811 | 0 | } |
3812 | 17.2k | buf = tmp; |
3813 | 17.2k | } |
3814 | 574M | COPY_BUF(l,buf,len,c); |
3815 | 574M | NEXTL(l); |
3816 | | |
3817 | 574M | GROW; |
3818 | 574M | c = CUR_CHAR(l); |
3819 | 574M | if (c == 0) { |
3820 | 140 | GROW; |
3821 | 140 | c = CUR_CHAR(l); |
3822 | 140 | } |
3823 | 574M | } |
3824 | 39.3k | buf[len] = 0; |
3825 | 39.3k | if (ctxt->instate == XML_PARSER_EOF) |
3826 | 0 | goto error; |
3827 | 39.3k | if (c != stop) { |
3828 | 175 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3829 | 175 | goto error; |
3830 | 175 | } |
3831 | 39.1k | NEXT; |
3832 | | |
3833 | | /* |
3834 | | * Raise problem w.r.t. '&' and '%' being used in non-entities |
3835 | | * reference constructs. Note Charref will be handled in |
3836 | | * xmlStringDecodeEntities() |
3837 | | */ |
3838 | 39.1k | cur = buf; |
3839 | 1.64G | while (*cur != 0) { /* non input consuming */ |
3840 | 1.64G | if ((*cur == '%') || ((*cur == '&') && (cur[1] != '#'))) { |
3841 | 37.6k | xmlChar *name; |
3842 | 37.6k | xmlChar tmp = *cur; |
3843 | 37.6k | int nameOk = 0; |
3844 | | |
3845 | 37.6k | cur++; |
3846 | 37.6k | name = xmlParseStringName(ctxt, &cur); |
3847 | 37.6k | if (name != NULL) { |
3848 | 35.2k | nameOk = 1; |
3849 | 35.2k | xmlFree(name); |
3850 | 35.2k | } |
3851 | 37.6k | if ((nameOk == 0) || (*cur != ';')) { |
3852 | 6.14k | xmlFatalErrMsgInt(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3853 | 6.14k | "EntityValue: '%c' forbidden except for entities references\n", |
3854 | 6.14k | tmp); |
3855 | 6.14k | goto error; |
3856 | 6.14k | } |
3857 | 31.5k | if ((tmp == '%') && (ctxt->inSubset == 1) && |
3858 | 31.5k | (ctxt->inputNr == 1)) { |
3859 | 1.02k | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3860 | 1.02k | goto error; |
3861 | 1.02k | } |
3862 | 30.4k | if (*cur == 0) |
3863 | 0 | break; |
3864 | 30.4k | } |
3865 | 1.64G | cur++; |
3866 | 1.64G | } |
3867 | | |
3868 | | /* |
3869 | | * Then PEReference entities are substituted. |
3870 | | * |
3871 | | * NOTE: 4.4.7 Bypassed |
3872 | | * When a general entity reference appears in the EntityValue in |
3873 | | * an entity declaration, it is bypassed and left as is. |
3874 | | * so XML_SUBSTITUTE_REF is not set here. |
3875 | | */ |
3876 | 31.9k | ++ctxt->depth; |
3877 | 31.9k | ret = xmlStringDecodeEntities(ctxt, buf, XML_SUBSTITUTE_PEREF, |
3878 | 31.9k | 0, 0, 0); |
3879 | 31.9k | --ctxt->depth; |
3880 | 31.9k | if (orig != NULL) { |
3881 | 31.9k | *orig = buf; |
3882 | 31.9k | buf = NULL; |
3883 | 31.9k | } |
3884 | | |
3885 | 39.3k | error: |
3886 | 39.3k | if (buf != NULL) |
3887 | 7.35k | xmlFree(buf); |
3888 | 39.3k | return(ret); |
3889 | 31.9k | } |
3890 | | |
3891 | | /** |
3892 | | * xmlParseAttValueComplex: |
3893 | | * @ctxt: an XML parser context |
3894 | | * @len: the resulting attribute len |
3895 | | * @normalize: whether to apply the inner normalization |
3896 | | * |
3897 | | * parse a value for an attribute, this is the fallback function |
3898 | | * of xmlParseAttValue() when the attribute parsing requires handling |
3899 | | * of non-ASCII characters, or normalization compaction. |
3900 | | * |
3901 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
3902 | | */ |
3903 | | static xmlChar * |
3904 | 98.8k | xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) { |
3905 | 98.8k | xmlChar limit = 0; |
3906 | 98.8k | xmlChar *buf = NULL; |
3907 | 98.8k | xmlChar *rep = NULL; |
3908 | 98.8k | size_t len = 0; |
3909 | 98.8k | size_t buf_size = 0; |
3910 | 98.8k | int c, l, in_space = 0; |
3911 | 98.8k | xmlChar *current = NULL; |
3912 | 98.8k | xmlEntityPtr ent; |
3913 | | |
3914 | 98.8k | if (NXT(0) == '"') { |
3915 | 96.7k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3916 | 96.7k | limit = '"'; |
3917 | 96.7k | NEXT; |
3918 | 96.7k | } else if (NXT(0) == '\'') { |
3919 | 2.01k | limit = '\''; |
3920 | 2.01k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3921 | 2.01k | NEXT; |
3922 | 2.01k | } else { |
3923 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3924 | 0 | return(NULL); |
3925 | 0 | } |
3926 | | |
3927 | | /* |
3928 | | * allocate a translation buffer. |
3929 | | */ |
3930 | 98.8k | buf_size = XML_PARSER_BUFFER_SIZE; |
3931 | 98.8k | buf = (xmlChar *) xmlMallocAtomic(buf_size); |
3932 | 98.8k | if (buf == NULL) goto mem_error; |
3933 | | |
3934 | | /* |
3935 | | * OK loop until we reach one of the ending char or a size limit. |
3936 | | */ |
3937 | 98.8k | c = CUR_CHAR(l); |
3938 | 305M | while (((NXT(0) != limit) && /* checked */ |
3939 | 305M | (IS_CHAR(c)) && (c != '<')) && |
3940 | 305M | (ctxt->instate != XML_PARSER_EOF)) { |
3941 | | /* |
3942 | | * Impose a reasonable limit on attribute size, unless XML_PARSE_HUGE |
3943 | | * special option is given |
3944 | | */ |
3945 | 305M | if ((len > XML_MAX_TEXT_LENGTH) && |
3946 | 305M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
3947 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
3948 | 0 | "AttValue length too long\n"); |
3949 | 0 | goto mem_error; |
3950 | 0 | } |
3951 | 305M | if (c == '&') { |
3952 | 972k | in_space = 0; |
3953 | 972k | if (NXT(1) == '#') { |
3954 | 503k | int val = xmlParseCharRef(ctxt); |
3955 | | |
3956 | 503k | if (val == '&') { |
3957 | 413k | if (ctxt->replaceEntities) { |
3958 | 0 | if (len + 10 > buf_size) { |
3959 | 0 | growBuffer(buf, 10); |
3960 | 0 | } |
3961 | 0 | buf[len++] = '&'; |
3962 | 413k | } else { |
3963 | | /* |
3964 | | * The reparsing will be done in xmlStringGetNodeList() |
3965 | | * called by the attribute() function in SAX.c |
3966 | | */ |
3967 | 413k | if (len + 10 > buf_size) { |
3968 | 870 | growBuffer(buf, 10); |
3969 | 870 | } |
3970 | 413k | buf[len++] = '&'; |
3971 | 413k | buf[len++] = '#'; |
3972 | 413k | buf[len++] = '3'; |
3973 | 413k | buf[len++] = '8'; |
3974 | 413k | buf[len++] = ';'; |
3975 | 413k | } |
3976 | 413k | } else if (val != 0) { |
3977 | 48.2k | if (len + 10 > buf_size) { |
3978 | 428 | growBuffer(buf, 10); |
3979 | 428 | } |
3980 | 48.2k | len += xmlCopyChar(0, &buf[len], val); |
3981 | 48.2k | } |
3982 | 503k | } else { |
3983 | 469k | ent = xmlParseEntityRef(ctxt); |
3984 | 469k | ctxt->nbentities++; |
3985 | 469k | if (ent != NULL) |
3986 | 335k | ctxt->nbentities += ent->owner; |
3987 | 469k | if ((ent != NULL) && |
3988 | 469k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
3989 | 3.43k | if (len + 10 > buf_size) { |
3990 | 478 | growBuffer(buf, 10); |
3991 | 478 | } |
3992 | 3.43k | if ((ctxt->replaceEntities == 0) && |
3993 | 3.43k | (ent->content[0] == '&')) { |
3994 | 1.45k | buf[len++] = '&'; |
3995 | 1.45k | buf[len++] = '#'; |
3996 | 1.45k | buf[len++] = '3'; |
3997 | 1.45k | buf[len++] = '8'; |
3998 | 1.45k | buf[len++] = ';'; |
3999 | 1.98k | } else { |
4000 | 1.98k | buf[len++] = ent->content[0]; |
4001 | 1.98k | } |
4002 | 465k | } else if ((ent != NULL) && |
4003 | 465k | (ctxt->replaceEntities != 0)) { |
4004 | 0 | if (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) { |
4005 | 0 | ++ctxt->depth; |
4006 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4007 | 0 | XML_SUBSTITUTE_REF, |
4008 | 0 | 0, 0, 0); |
4009 | 0 | --ctxt->depth; |
4010 | 0 | if (rep != NULL) { |
4011 | 0 | current = rep; |
4012 | 0 | while (*current != 0) { /* non input consuming */ |
4013 | 0 | if ((*current == 0xD) || (*current == 0xA) || |
4014 | 0 | (*current == 0x9)) { |
4015 | 0 | buf[len++] = 0x20; |
4016 | 0 | current++; |
4017 | 0 | } else |
4018 | 0 | buf[len++] = *current++; |
4019 | 0 | if (len + 10 > buf_size) { |
4020 | 0 | growBuffer(buf, 10); |
4021 | 0 | } |
4022 | 0 | } |
4023 | 0 | xmlFree(rep); |
4024 | 0 | rep = NULL; |
4025 | 0 | } |
4026 | 0 | } else { |
4027 | 0 | if (len + 10 > buf_size) { |
4028 | 0 | growBuffer(buf, 10); |
4029 | 0 | } |
4030 | 0 | if (ent->content != NULL) |
4031 | 0 | buf[len++] = ent->content[0]; |
4032 | 0 | } |
4033 | 465k | } else if (ent != NULL) { |
4034 | 331k | int i = xmlStrlen(ent->name); |
4035 | 331k | const xmlChar *cur = ent->name; |
4036 | | |
4037 | | /* |
4038 | | * This may look absurd but is needed to detect |
4039 | | * entities problems |
4040 | | */ |
4041 | 331k | if ((ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
4042 | 331k | (ent->content != NULL) && (ent->checked == 0)) { |
4043 | 746 | unsigned long oldnbent = ctxt->nbentities, diff; |
4044 | | |
4045 | 746 | ++ctxt->depth; |
4046 | 746 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4047 | 746 | XML_SUBSTITUTE_REF, 0, 0, 0); |
4048 | 746 | --ctxt->depth; |
4049 | | |
4050 | 746 | diff = ctxt->nbentities - oldnbent + 1; |
4051 | 746 | if (diff > INT_MAX / 2) |
4052 | 4 | diff = INT_MAX / 2; |
4053 | 746 | ent->checked = diff * 2; |
4054 | 746 | if (rep != NULL) { |
4055 | 590 | if (xmlStrchr(rep, '<')) |
4056 | 222 | ent->checked |= 1; |
4057 | 590 | xmlFree(rep); |
4058 | 590 | rep = NULL; |
4059 | 590 | } else { |
4060 | 156 | ent->content[0] = 0; |
4061 | 156 | } |
4062 | 746 | } |
4063 | | |
4064 | | /* |
4065 | | * Just output the reference |
4066 | | */ |
4067 | 331k | buf[len++] = '&'; |
4068 | 332k | while (len + i + 10 > buf_size) { |
4069 | 1.36k | growBuffer(buf, i + 10); |
4070 | 1.36k | } |
4071 | 663k | for (;i > 0;i--) |
4072 | 332k | buf[len++] = *cur++; |
4073 | 331k | buf[len++] = ';'; |
4074 | 331k | } |
4075 | 469k | } |
4076 | 304M | } else { |
4077 | 304M | if ((c == 0x20) || (c == 0xD) || (c == 0xA) || (c == 0x9)) { |
4078 | 3.65M | if ((len != 0) || (!normalize)) { |
4079 | 3.65M | if ((!normalize) || (!in_space)) { |
4080 | 3.65M | COPY_BUF(l,buf,len,0x20); |
4081 | 3.65M | while (len + 10 > buf_size) { |
4082 | 9.98k | growBuffer(buf, 10); |
4083 | 9.98k | } |
4084 | 3.65M | } |
4085 | 3.65M | in_space = 1; |
4086 | 3.65M | } |
4087 | 300M | } else { |
4088 | 300M | in_space = 0; |
4089 | 300M | COPY_BUF(l,buf,len,c); |
4090 | 300M | if (len + 10 > buf_size) { |
4091 | 50.2k | growBuffer(buf, 10); |
4092 | 50.2k | } |
4093 | 300M | } |
4094 | 304M | NEXTL(l); |
4095 | 304M | } |
4096 | 305M | GROW; |
4097 | 305M | c = CUR_CHAR(l); |
4098 | 305M | } |
4099 | 98.8k | if (ctxt->instate == XML_PARSER_EOF) |
4100 | 0 | goto error; |
4101 | | |
4102 | 98.8k | if ((in_space) && (normalize)) { |
4103 | 1.90k | while ((len > 0) && (buf[len - 1] == 0x20)) len--; |
4104 | 758 | } |
4105 | 98.8k | buf[len] = 0; |
4106 | 98.8k | if (RAW == '<') { |
4107 | 18.1k | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4108 | 80.6k | } else if (RAW != limit) { |
4109 | 9.57k | if ((c != 0) && (!IS_CHAR(c))) { |
4110 | 1.18k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4111 | 1.18k | "invalid character in attribute value\n"); |
4112 | 8.39k | } else { |
4113 | 8.39k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4114 | 8.39k | "AttValue: ' expected\n"); |
4115 | 8.39k | } |
4116 | 9.57k | } else |
4117 | 71.0k | NEXT; |
4118 | | |
4119 | | /* |
4120 | | * There we potentially risk an overflow, don't allow attribute value of |
4121 | | * length more than INT_MAX it is a very reasonable assumption ! |
4122 | | */ |
4123 | 98.8k | if (len >= INT_MAX) { |
4124 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4125 | 0 | "AttValue length too long\n"); |
4126 | 0 | goto mem_error; |
4127 | 0 | } |
4128 | | |
4129 | 98.8k | if (attlen != NULL) *attlen = (int) len; |
4130 | 98.8k | return(buf); |
4131 | | |
4132 | 0 | mem_error: |
4133 | 0 | xmlErrMemory(ctxt, NULL); |
4134 | 0 | error: |
4135 | 0 | if (buf != NULL) |
4136 | 0 | xmlFree(buf); |
4137 | 0 | if (rep != NULL) |
4138 | 0 | xmlFree(rep); |
4139 | 0 | return(NULL); |
4140 | 0 | } |
4141 | | |
4142 | | /** |
4143 | | * xmlParseAttValue: |
4144 | | * @ctxt: an XML parser context |
4145 | | * |
4146 | | * parse a value for an attribute |
4147 | | * Note: the parser won't do substitution of entities here, this |
4148 | | * will be handled later in xmlStringGetNodeList |
4149 | | * |
4150 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4151 | | * "'" ([^<&'] | Reference)* "'" |
4152 | | * |
4153 | | * 3.3.3 Attribute-Value Normalization: |
4154 | | * Before the value of an attribute is passed to the application or |
4155 | | * checked for validity, the XML processor must normalize it as follows: |
4156 | | * - a character reference is processed by appending the referenced |
4157 | | * character to the attribute value |
4158 | | * - an entity reference is processed by recursively processing the |
4159 | | * replacement text of the entity |
4160 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4161 | | * appending #x20 to the normalized value, except that only a single |
4162 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4163 | | * parsed entity or the literal entity value of an internal parsed entity |
4164 | | * - other characters are processed by appending them to the normalized value |
4165 | | * If the declared value is not CDATA, then the XML processor must further |
4166 | | * process the normalized attribute value by discarding any leading and |
4167 | | * trailing space (#x20) characters, and by replacing sequences of space |
4168 | | * (#x20) characters by a single space (#x20) character. |
4169 | | * All attributes for which no declaration has been read should be treated |
4170 | | * by a non-validating parser as if declared CDATA. |
4171 | | * |
4172 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4173 | | */ |
4174 | | |
4175 | | |
4176 | | xmlChar * |
4177 | 817k | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4178 | 817k | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4179 | 817k | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0)); |
4180 | 817k | } |
4181 | | |
4182 | | /** |
4183 | | * xmlParseSystemLiteral: |
4184 | | * @ctxt: an XML parser context |
4185 | | * |
4186 | | * parse an XML Literal |
4187 | | * |
4188 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4189 | | * |
4190 | | * Returns the SystemLiteral parsed or NULL |
4191 | | */ |
4192 | | |
4193 | | xmlChar * |
4194 | 23.7k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4195 | 23.7k | xmlChar *buf = NULL; |
4196 | 23.7k | int len = 0; |
4197 | 23.7k | int size = XML_PARSER_BUFFER_SIZE; |
4198 | 23.7k | int cur, l; |
4199 | 23.7k | xmlChar stop; |
4200 | 23.7k | int state = ctxt->instate; |
4201 | 23.7k | int count = 0; |
4202 | | |
4203 | 23.7k | SHRINK; |
4204 | 23.7k | if (RAW == '"') { |
4205 | 16.2k | NEXT; |
4206 | 16.2k | stop = '"'; |
4207 | 16.2k | } else if (RAW == '\'') { |
4208 | 4.62k | NEXT; |
4209 | 4.62k | stop = '\''; |
4210 | 4.62k | } else { |
4211 | 2.86k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4212 | 2.86k | return(NULL); |
4213 | 2.86k | } |
4214 | | |
4215 | 20.8k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4216 | 20.8k | if (buf == NULL) { |
4217 | 0 | xmlErrMemory(ctxt, NULL); |
4218 | 0 | return(NULL); |
4219 | 0 | } |
4220 | 20.8k | ctxt->instate = XML_PARSER_SYSTEM_LITERAL; |
4221 | 20.8k | cur = CUR_CHAR(l); |
4222 | 30.1M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4223 | 30.1M | if (len + 5 >= size) { |
4224 | 16.3k | xmlChar *tmp; |
4225 | | |
4226 | 16.3k | if ((size > XML_MAX_NAME_LENGTH) && |
4227 | 16.3k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
4228 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4229 | 0 | xmlFree(buf); |
4230 | 0 | ctxt->instate = (xmlParserInputState) state; |
4231 | 0 | return(NULL); |
4232 | 0 | } |
4233 | 16.3k | size *= 2; |
4234 | 16.3k | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
4235 | 16.3k | if (tmp == NULL) { |
4236 | 0 | xmlFree(buf); |
4237 | 0 | xmlErrMemory(ctxt, NULL); |
4238 | 0 | ctxt->instate = (xmlParserInputState) state; |
4239 | 0 | return(NULL); |
4240 | 0 | } |
4241 | 16.3k | buf = tmp; |
4242 | 16.3k | } |
4243 | 30.1M | count++; |
4244 | 30.1M | if (count > 50) { |
4245 | 587k | SHRINK; |
4246 | 587k | GROW; |
4247 | 587k | count = 0; |
4248 | 587k | if (ctxt->instate == XML_PARSER_EOF) { |
4249 | 0 | xmlFree(buf); |
4250 | 0 | return(NULL); |
4251 | 0 | } |
4252 | 587k | } |
4253 | 30.1M | COPY_BUF(l,buf,len,cur); |
4254 | 30.1M | NEXTL(l); |
4255 | 30.1M | cur = CUR_CHAR(l); |
4256 | 30.1M | if (cur == 0) { |
4257 | 40.7k | GROW; |
4258 | 40.7k | SHRINK; |
4259 | 40.7k | cur = CUR_CHAR(l); |
4260 | 40.7k | } |
4261 | 30.1M | } |
4262 | 20.8k | buf[len] = 0; |
4263 | 20.8k | ctxt->instate = (xmlParserInputState) state; |
4264 | 20.8k | if (!IS_CHAR(cur)) { |
4265 | 7.12k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4266 | 13.7k | } else { |
4267 | 13.7k | NEXT; |
4268 | 13.7k | } |
4269 | 20.8k | return(buf); |
4270 | 20.8k | } |
4271 | | |
4272 | | /** |
4273 | | * xmlParsePubidLiteral: |
4274 | | * @ctxt: an XML parser context |
4275 | | * |
4276 | | * parse an XML public literal |
4277 | | * |
4278 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4279 | | * |
4280 | | * Returns the PubidLiteral parsed or NULL. |
4281 | | */ |
4282 | | |
4283 | | xmlChar * |
4284 | 49.0k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4285 | 49.0k | xmlChar *buf = NULL; |
4286 | 49.0k | int len = 0; |
4287 | 49.0k | int size = XML_PARSER_BUFFER_SIZE; |
4288 | 49.0k | xmlChar cur; |
4289 | 49.0k | xmlChar stop; |
4290 | 49.0k | int count = 0; |
4291 | 49.0k | xmlParserInputState oldstate = ctxt->instate; |
4292 | | |
4293 | 49.0k | SHRINK; |
4294 | 49.0k | if (RAW == '"') { |
4295 | 42.4k | NEXT; |
4296 | 42.4k | stop = '"'; |
4297 | 42.4k | } else if (RAW == '\'') { |
4298 | 4.48k | NEXT; |
4299 | 4.48k | stop = '\''; |
4300 | 4.48k | } else { |
4301 | 2.14k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4302 | 2.14k | return(NULL); |
4303 | 2.14k | } |
4304 | 46.9k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4305 | 46.9k | if (buf == NULL) { |
4306 | 0 | xmlErrMemory(ctxt, NULL); |
4307 | 0 | return(NULL); |
4308 | 0 | } |
4309 | 46.9k | ctxt->instate = XML_PARSER_PUBLIC_LITERAL; |
4310 | 46.9k | cur = CUR; |
4311 | 10.8M | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop)) { /* checked */ |
4312 | 10.7M | if (len + 1 >= size) { |
4313 | 4.78k | xmlChar *tmp; |
4314 | | |
4315 | 4.78k | if ((size > XML_MAX_NAME_LENGTH) && |
4316 | 4.78k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
4317 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4318 | 0 | xmlFree(buf); |
4319 | 0 | return(NULL); |
4320 | 0 | } |
4321 | 4.78k | size *= 2; |
4322 | 4.78k | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
4323 | 4.78k | if (tmp == NULL) { |
4324 | 0 | xmlErrMemory(ctxt, NULL); |
4325 | 0 | xmlFree(buf); |
4326 | 0 | return(NULL); |
4327 | 0 | } |
4328 | 4.78k | buf = tmp; |
4329 | 4.78k | } |
4330 | 10.7M | buf[len++] = cur; |
4331 | 10.7M | count++; |
4332 | 10.7M | if (count > 50) { |
4333 | 199k | SHRINK; |
4334 | 199k | GROW; |
4335 | 199k | count = 0; |
4336 | 199k | if (ctxt->instate == XML_PARSER_EOF) { |
4337 | 0 | xmlFree(buf); |
4338 | 0 | return(NULL); |
4339 | 0 | } |
4340 | 199k | } |
4341 | 10.7M | NEXT; |
4342 | 10.7M | cur = CUR; |
4343 | 10.7M | if (cur == 0) { |
4344 | 8.27k | GROW; |
4345 | 8.27k | SHRINK; |
4346 | 8.27k | cur = CUR; |
4347 | 8.27k | } |
4348 | 10.7M | } |
4349 | 46.9k | buf[len] = 0; |
4350 | 46.9k | if (cur != stop) { |
4351 | 40.9k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4352 | 40.9k | } else { |
4353 | 6.02k | NEXT; |
4354 | 6.02k | } |
4355 | 46.9k | ctxt->instate = oldstate; |
4356 | 46.9k | return(buf); |
4357 | 46.9k | } |
4358 | | |
4359 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata); |
4360 | | |
4361 | | /* |
4362 | | * used for the test in the inner loop of the char data testing |
4363 | | */ |
4364 | | static const unsigned char test_char_data[256] = { |
4365 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4366 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4367 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4368 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4369 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4370 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4371 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4372 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4373 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4374 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4375 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4376 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4377 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4378 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4379 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4380 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4381 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4382 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4383 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4384 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4385 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4386 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4387 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4388 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4389 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4390 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4391 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4392 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4393 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4394 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4395 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4396 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4397 | | }; |
4398 | | |
4399 | | /** |
4400 | | * xmlParseCharData: |
4401 | | * @ctxt: an XML parser context |
4402 | | * @cdata: int indicating whether we are within a CDATA section |
4403 | | * |
4404 | | * parse a CharData section. |
4405 | | * if we are within a CDATA section ']]>' marks an end of section. |
4406 | | * |
4407 | | * The right angle bracket (>) may be represented using the string ">", |
4408 | | * and must, for compatibility, be escaped using ">" or a character |
4409 | | * reference when it appears in the string "]]>" in content, when that |
4410 | | * string is not marking the end of a CDATA section. |
4411 | | * |
4412 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4413 | | */ |
4414 | | |
4415 | | void |
4416 | 830k | xmlParseCharData(xmlParserCtxtPtr ctxt, int cdata) { |
4417 | 830k | const xmlChar *in; |
4418 | 830k | int nbchar = 0; |
4419 | 830k | int line = ctxt->input->line; |
4420 | 830k | int col = ctxt->input->col; |
4421 | 830k | int ccol; |
4422 | | |
4423 | 830k | SHRINK; |
4424 | 830k | GROW; |
4425 | | /* |
4426 | | * Accelerated common case where input don't need to be |
4427 | | * modified before passing it to the handler. |
4428 | | */ |
4429 | 830k | if (!cdata) { |
4430 | 830k | in = ctxt->input->cur; |
4431 | 834k | do { |
4432 | 900k | get_more_space: |
4433 | 1.04M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4434 | 900k | if (*in == 0xA) { |
4435 | 258k | do { |
4436 | 258k | ctxt->input->line++; ctxt->input->col = 1; |
4437 | 258k | in++; |
4438 | 258k | } while (*in == 0xA); |
4439 | 65.7k | goto get_more_space; |
4440 | 65.7k | } |
4441 | 834k | if (*in == '<') { |
4442 | 60.6k | nbchar = in - ctxt->input->cur; |
4443 | 60.6k | if (nbchar > 0) { |
4444 | 59.9k | const xmlChar *tmp = ctxt->input->cur; |
4445 | 59.9k | ctxt->input->cur = in; |
4446 | | |
4447 | 59.9k | if ((ctxt->sax != NULL) && |
4448 | 59.9k | (ctxt->sax->ignorableWhitespace != |
4449 | 59.9k | ctxt->sax->characters)) { |
4450 | 0 | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4451 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4452 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4453 | 0 | tmp, nbchar); |
4454 | 0 | } else { |
4455 | 0 | if (ctxt->sax->characters != NULL) |
4456 | 0 | ctxt->sax->characters(ctxt->userData, |
4457 | 0 | tmp, nbchar); |
4458 | 0 | if (*ctxt->space == -1) |
4459 | 0 | *ctxt->space = -2; |
4460 | 0 | } |
4461 | 59.9k | } else if ((ctxt->sax != NULL) && |
4462 | 59.9k | (ctxt->sax->characters != NULL)) { |
4463 | 59.9k | ctxt->sax->characters(ctxt->userData, |
4464 | 59.9k | tmp, nbchar); |
4465 | 59.9k | } |
4466 | 59.9k | } |
4467 | 60.6k | return; |
4468 | 60.6k | } |
4469 | | |
4470 | 849k | get_more: |
4471 | 849k | ccol = ctxt->input->col; |
4472 | 4.88M | while (test_char_data[*in]) { |
4473 | 4.03M | in++; |
4474 | 4.03M | ccol++; |
4475 | 4.03M | } |
4476 | 849k | ctxt->input->col = ccol; |
4477 | 849k | if (*in == 0xA) { |
4478 | 192k | do { |
4479 | 192k | ctxt->input->line++; ctxt->input->col = 1; |
4480 | 192k | in++; |
4481 | 192k | } while (*in == 0xA); |
4482 | 29.4k | goto get_more; |
4483 | 29.4k | } |
4484 | 820k | if (*in == ']') { |
4485 | 48.6k | if ((in[1] == ']') && (in[2] == '>')) { |
4486 | 2.55k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4487 | 2.55k | ctxt->input->cur = in + 1; |
4488 | 2.55k | return; |
4489 | 2.55k | } |
4490 | 46.0k | in++; |
4491 | 46.0k | ctxt->input->col++; |
4492 | 46.0k | goto get_more; |
4493 | 48.6k | } |
4494 | 771k | nbchar = in - ctxt->input->cur; |
4495 | 771k | if (nbchar > 0) { |
4496 | 623k | if ((ctxt->sax != NULL) && |
4497 | 623k | (ctxt->sax->ignorableWhitespace != |
4498 | 623k | ctxt->sax->characters) && |
4499 | 623k | (IS_BLANK_CH(*ctxt->input->cur))) { |
4500 | 0 | const xmlChar *tmp = ctxt->input->cur; |
4501 | 0 | ctxt->input->cur = in; |
4502 | |
|
4503 | 0 | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4504 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4505 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4506 | 0 | tmp, nbchar); |
4507 | 0 | } else { |
4508 | 0 | if (ctxt->sax->characters != NULL) |
4509 | 0 | ctxt->sax->characters(ctxt->userData, |
4510 | 0 | tmp, nbchar); |
4511 | 0 | if (*ctxt->space == -1) |
4512 | 0 | *ctxt->space = -2; |
4513 | 0 | } |
4514 | 0 | line = ctxt->input->line; |
4515 | 0 | col = ctxt->input->col; |
4516 | 623k | } else if (ctxt->sax != NULL) { |
4517 | 623k | if (ctxt->sax->characters != NULL) |
4518 | 623k | ctxt->sax->characters(ctxt->userData, |
4519 | 623k | ctxt->input->cur, nbchar); |
4520 | 623k | line = ctxt->input->line; |
4521 | 623k | col = ctxt->input->col; |
4522 | 623k | } |
4523 | | /* something really bad happened in the SAX callback */ |
4524 | 623k | if (ctxt->instate != XML_PARSER_CONTENT) |
4525 | 0 | return; |
4526 | 623k | } |
4527 | 771k | ctxt->input->cur = in; |
4528 | 771k | if (*in == 0xD) { |
4529 | 9.27k | in++; |
4530 | 9.27k | if (*in == 0xA) { |
4531 | 2.16k | ctxt->input->cur = in; |
4532 | 2.16k | in++; |
4533 | 2.16k | ctxt->input->line++; ctxt->input->col = 1; |
4534 | 2.16k | continue; /* while */ |
4535 | 2.16k | } |
4536 | 7.10k | in--; |
4537 | 7.10k | } |
4538 | 769k | if (*in == '<') { |
4539 | 236k | return; |
4540 | 236k | } |
4541 | 532k | if (*in == '&') { |
4542 | 186k | return; |
4543 | 186k | } |
4544 | 345k | SHRINK; |
4545 | 345k | GROW; |
4546 | 345k | if (ctxt->instate == XML_PARSER_EOF) |
4547 | 0 | return; |
4548 | 345k | in = ctxt->input->cur; |
4549 | 348k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
4550 | 343k | nbchar = 0; |
4551 | 343k | } |
4552 | 343k | ctxt->input->line = line; |
4553 | 343k | ctxt->input->col = col; |
4554 | 343k | xmlParseCharDataComplex(ctxt, cdata); |
4555 | 343k | } |
4556 | | |
4557 | | /** |
4558 | | * xmlParseCharDataComplex: |
4559 | | * @ctxt: an XML parser context |
4560 | | * @cdata: int indicating whether we are within a CDATA section |
4561 | | * |
4562 | | * parse a CharData section.this is the fallback function |
4563 | | * of xmlParseCharData() when the parsing requires handling |
4564 | | * of non-ASCII characters. |
4565 | | */ |
4566 | | static void |
4567 | 343k | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata) { |
4568 | 343k | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4569 | 343k | int nbchar = 0; |
4570 | 343k | int cur, l; |
4571 | 343k | int count = 0; |
4572 | | |
4573 | 343k | SHRINK; |
4574 | 343k | GROW; |
4575 | 343k | cur = CUR_CHAR(l); |
4576 | 234M | while ((cur != '<') && /* checked */ |
4577 | 234M | (cur != '&') && |
4578 | 234M | (IS_CHAR(cur))) /* test also done in xmlCurrentChar() */ { |
4579 | 233M | if ((cur == ']') && (NXT(1) == ']') && |
4580 | 233M | (NXT(2) == '>')) { |
4581 | 2.03k | if (cdata) break; |
4582 | 2.03k | else { |
4583 | 2.03k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4584 | 2.03k | } |
4585 | 2.03k | } |
4586 | 233M | COPY_BUF(l,buf,nbchar,cur); |
4587 | | /* move current position before possible calling of ctxt->sax->characters */ |
4588 | 233M | NEXTL(l); |
4589 | 233M | cur = CUR_CHAR(l); |
4590 | 233M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4591 | 2.16M | buf[nbchar] = 0; |
4592 | | |
4593 | | /* |
4594 | | * OK the segment is to be consumed as chars. |
4595 | | */ |
4596 | 2.16M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4597 | 1.71M | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4598 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4599 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4600 | 0 | buf, nbchar); |
4601 | 1.71M | } else { |
4602 | 1.71M | if (ctxt->sax->characters != NULL) |
4603 | 1.71M | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4604 | 1.71M | if ((ctxt->sax->characters != |
4605 | 1.71M | ctxt->sax->ignorableWhitespace) && |
4606 | 1.71M | (*ctxt->space == -1)) |
4607 | 0 | *ctxt->space = -2; |
4608 | 1.71M | } |
4609 | 1.71M | } |
4610 | 2.16M | nbchar = 0; |
4611 | | /* something really bad happened in the SAX callback */ |
4612 | 2.16M | if (ctxt->instate != XML_PARSER_CONTENT) |
4613 | 0 | return; |
4614 | 2.16M | } |
4615 | 233M | count++; |
4616 | 233M | if (count > 50) { |
4617 | 4.52M | SHRINK; |
4618 | 4.52M | GROW; |
4619 | 4.52M | count = 0; |
4620 | 4.52M | if (ctxt->instate == XML_PARSER_EOF) |
4621 | 0 | return; |
4622 | 4.52M | } |
4623 | 233M | } |
4624 | 343k | if (nbchar != 0) { |
4625 | 254k | buf[nbchar] = 0; |
4626 | | /* |
4627 | | * OK the segment is to be consumed as chars. |
4628 | | */ |
4629 | 254k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4630 | 131k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4631 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4632 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4633 | 131k | } else { |
4634 | 131k | if (ctxt->sax->characters != NULL) |
4635 | 131k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4636 | 131k | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4637 | 131k | (*ctxt->space == -1)) |
4638 | 0 | *ctxt->space = -2; |
4639 | 131k | } |
4640 | 131k | } |
4641 | 254k | } |
4642 | 343k | if ((cur != 0) && (!IS_CHAR(cur))) { |
4643 | | /* Generate the error and skip the offending character */ |
4644 | 70.7k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4645 | 70.7k | "PCDATA invalid Char value %d\n", |
4646 | 70.7k | cur); |
4647 | 70.7k | NEXTL(l); |
4648 | 70.7k | } |
4649 | 343k | } |
4650 | | |
4651 | | /** |
4652 | | * xmlParseExternalID: |
4653 | | * @ctxt: an XML parser context |
4654 | | * @publicID: a xmlChar** receiving PubidLiteral |
4655 | | * @strict: indicate whether we should restrict parsing to only |
4656 | | * production [75], see NOTE below |
4657 | | * |
4658 | | * Parse an External ID or a Public ID |
4659 | | * |
4660 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4661 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
4662 | | * |
4663 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4664 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4665 | | * |
4666 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4667 | | * |
4668 | | * Returns the function returns SystemLiteral and in the second |
4669 | | * case publicID receives PubidLiteral, is strict is off |
4670 | | * it is possible to return NULL and have publicID set. |
4671 | | */ |
4672 | | |
4673 | | xmlChar * |
4674 | 1.80M | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
4675 | 1.80M | xmlChar *URI = NULL; |
4676 | | |
4677 | 1.80M | SHRINK; |
4678 | | |
4679 | 1.80M | *publicID = NULL; |
4680 | 1.80M | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4681 | 10.4k | SKIP(6); |
4682 | 10.4k | if (SKIP_BLANKS == 0) { |
4683 | 2.88k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4684 | 2.88k | "Space required after 'SYSTEM'\n"); |
4685 | 2.88k | } |
4686 | 10.4k | URI = xmlParseSystemLiteral(ctxt); |
4687 | 10.4k | if (URI == NULL) { |
4688 | 1.46k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4689 | 1.46k | } |
4690 | 1.79M | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4691 | 49.0k | SKIP(6); |
4692 | 49.0k | if (SKIP_BLANKS == 0) { |
4693 | 47.7k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4694 | 47.7k | "Space required after 'PUBLIC'\n"); |
4695 | 47.7k | } |
4696 | 49.0k | *publicID = xmlParsePubidLiteral(ctxt); |
4697 | 49.0k | if (*publicID == NULL) { |
4698 | 2.14k | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4699 | 2.14k | } |
4700 | 49.0k | if (strict) { |
4701 | | /* |
4702 | | * We don't handle [83] so "S SystemLiteral" is required. |
4703 | | */ |
4704 | 4.50k | if (SKIP_BLANKS == 0) { |
4705 | 4.36k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4706 | 4.36k | "Space required after the Public Identifier\n"); |
4707 | 4.36k | } |
4708 | 44.5k | } else { |
4709 | | /* |
4710 | | * We handle [83] so we return immediately, if |
4711 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4712 | | * system literal was found, but this is harmless since we must |
4713 | | * be at the end of a NotationDecl. |
4714 | | */ |
4715 | 44.5k | if (SKIP_BLANKS == 0) return(NULL); |
4716 | 17.1k | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4717 | 17.1k | } |
4718 | 13.2k | URI = xmlParseSystemLiteral(ctxt); |
4719 | 13.2k | if (URI == NULL) { |
4720 | 1.39k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4721 | 1.39k | } |
4722 | 13.2k | } |
4723 | 1.76M | return(URI); |
4724 | 1.80M | } |
4725 | | |
4726 | | /** |
4727 | | * xmlParseCommentComplex: |
4728 | | * @ctxt: an XML parser context |
4729 | | * @buf: the already parsed part of the buffer |
4730 | | * @len: number of bytes in the buffer |
4731 | | * @size: allocated size of the buffer |
4732 | | * |
4733 | | * Skip an XML (SGML) comment <!-- .... --> |
4734 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4735 | | * must not occur within comments. " |
4736 | | * This is the slow routine in case the accelerator for ascii didn't work |
4737 | | * |
4738 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4739 | | */ |
4740 | | static void |
4741 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4742 | 42.6k | size_t len, size_t size) { |
4743 | 42.6k | int q, ql; |
4744 | 42.6k | int r, rl; |
4745 | 42.6k | int cur, l; |
4746 | 42.6k | size_t count = 0; |
4747 | 42.6k | int inputid; |
4748 | | |
4749 | 42.6k | inputid = ctxt->input->id; |
4750 | | |
4751 | 42.6k | if (buf == NULL) { |
4752 | 8.79k | len = 0; |
4753 | 8.79k | size = XML_PARSER_BUFFER_SIZE; |
4754 | 8.79k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4755 | 8.79k | if (buf == NULL) { |
4756 | 0 | xmlErrMemory(ctxt, NULL); |
4757 | 0 | return; |
4758 | 0 | } |
4759 | 8.79k | } |
4760 | 42.6k | GROW; /* Assure there's enough input data */ |
4761 | 42.6k | q = CUR_CHAR(ql); |
4762 | 42.6k | if (q == 0) |
4763 | 7.52k | goto not_terminated; |
4764 | 35.1k | if (!IS_CHAR(q)) { |
4765 | 563 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4766 | 563 | "xmlParseComment: invalid xmlChar value %d\n", |
4767 | 563 | q); |
4768 | 563 | xmlFree (buf); |
4769 | 563 | return; |
4770 | 563 | } |
4771 | 34.6k | NEXTL(ql); |
4772 | 34.6k | r = CUR_CHAR(rl); |
4773 | 34.6k | if (r == 0) |
4774 | 2.11k | goto not_terminated; |
4775 | 32.4k | if (!IS_CHAR(r)) { |
4776 | 616 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4777 | 616 | "xmlParseComment: invalid xmlChar value %d\n", |
4778 | 616 | q); |
4779 | 616 | xmlFree (buf); |
4780 | 616 | return; |
4781 | 616 | } |
4782 | 31.8k | NEXTL(rl); |
4783 | 31.8k | cur = CUR_CHAR(l); |
4784 | 31.8k | if (cur == 0) |
4785 | 787 | goto not_terminated; |
4786 | 147M | while (IS_CHAR(cur) && /* checked */ |
4787 | 147M | ((cur != '>') || |
4788 | 147M | (r != '-') || (q != '-'))) { |
4789 | 147M | if ((r == '-') && (q == '-')) { |
4790 | 55.2k | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
4791 | 55.2k | } |
4792 | 147M | if ((len > XML_MAX_TEXT_LENGTH) && |
4793 | 147M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
4794 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4795 | 0 | "Comment too big found", NULL); |
4796 | 0 | xmlFree (buf); |
4797 | 0 | return; |
4798 | 0 | } |
4799 | 147M | if (len + 5 >= size) { |
4800 | 30.6k | xmlChar *new_buf; |
4801 | 30.6k | size_t new_size; |
4802 | | |
4803 | 30.6k | new_size = size * 2; |
4804 | 30.6k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
4805 | 30.6k | if (new_buf == NULL) { |
4806 | 0 | xmlFree (buf); |
4807 | 0 | xmlErrMemory(ctxt, NULL); |
4808 | 0 | return; |
4809 | 0 | } |
4810 | 30.6k | buf = new_buf; |
4811 | 30.6k | size = new_size; |
4812 | 30.6k | } |
4813 | 147M | COPY_BUF(ql,buf,len,q); |
4814 | 147M | q = r; |
4815 | 147M | ql = rl; |
4816 | 147M | r = cur; |
4817 | 147M | rl = l; |
4818 | | |
4819 | 147M | count++; |
4820 | 147M | if (count > 50) { |
4821 | 2.88M | SHRINK; |
4822 | 2.88M | GROW; |
4823 | 2.88M | count = 0; |
4824 | 2.88M | if (ctxt->instate == XML_PARSER_EOF) { |
4825 | 0 | xmlFree(buf); |
4826 | 0 | return; |
4827 | 0 | } |
4828 | 2.88M | } |
4829 | 147M | NEXTL(l); |
4830 | 147M | cur = CUR_CHAR(l); |
4831 | 147M | if (cur == 0) { |
4832 | 25.4k | SHRINK; |
4833 | 25.4k | GROW; |
4834 | 25.4k | cur = CUR_CHAR(l); |
4835 | 25.4k | } |
4836 | 147M | } |
4837 | 31.0k | buf[len] = 0; |
4838 | 31.0k | if (cur == 0) { |
4839 | 14.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4840 | 14.5k | "Comment not terminated \n<!--%.50s\n", buf); |
4841 | 16.5k | } else if (!IS_CHAR(cur)) { |
4842 | 1.44k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4843 | 1.44k | "xmlParseComment: invalid xmlChar value %d\n", |
4844 | 1.44k | cur); |
4845 | 15.1k | } else { |
4846 | 15.1k | if (inputid != ctxt->input->id) { |
4847 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
4848 | 0 | "Comment doesn't start and stop in the same" |
4849 | 0 | " entity\n"); |
4850 | 0 | } |
4851 | 15.1k | NEXT; |
4852 | 15.1k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
4853 | 15.1k | (!ctxt->disableSAX)) |
4854 | 1.17k | ctxt->sax->comment(ctxt->userData, buf); |
4855 | 15.1k | } |
4856 | 31.0k | xmlFree(buf); |
4857 | 31.0k | return; |
4858 | 10.4k | not_terminated: |
4859 | 10.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4860 | 10.4k | "Comment not terminated\n", NULL); |
4861 | 10.4k | xmlFree(buf); |
4862 | 10.4k | return; |
4863 | 31.0k | } |
4864 | | |
4865 | | /** |
4866 | | * xmlParseComment: |
4867 | | * @ctxt: an XML parser context |
4868 | | * |
4869 | | * Skip an XML (SGML) comment <!-- .... --> |
4870 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4871 | | * must not occur within comments. " |
4872 | | * |
4873 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4874 | | */ |
4875 | | void |
4876 | 55.7k | xmlParseComment(xmlParserCtxtPtr ctxt) { |
4877 | 55.7k | xmlChar *buf = NULL; |
4878 | 55.7k | size_t size = XML_PARSER_BUFFER_SIZE; |
4879 | 55.7k | size_t len = 0; |
4880 | 55.7k | xmlParserInputState state; |
4881 | 55.7k | const xmlChar *in; |
4882 | 55.7k | size_t nbchar = 0; |
4883 | 55.7k | int ccol; |
4884 | 55.7k | int inputid; |
4885 | | |
4886 | | /* |
4887 | | * Check that there is a comment right here. |
4888 | | */ |
4889 | 55.7k | if ((RAW != '<') || (NXT(1) != '!') || |
4890 | 55.7k | (NXT(2) != '-') || (NXT(3) != '-')) return; |
4891 | 54.0k | state = ctxt->instate; |
4892 | 54.0k | ctxt->instate = XML_PARSER_COMMENT; |
4893 | 54.0k | inputid = ctxt->input->id; |
4894 | 54.0k | SKIP(4); |
4895 | 54.0k | SHRINK; |
4896 | 54.0k | GROW; |
4897 | | |
4898 | | /* |
4899 | | * Accelerated common case where input don't need to be |
4900 | | * modified before passing it to the handler. |
4901 | | */ |
4902 | 54.0k | in = ctxt->input->cur; |
4903 | 54.3k | do { |
4904 | 54.3k | if (*in == 0xA) { |
4905 | 24.5k | do { |
4906 | 24.5k | ctxt->input->line++; ctxt->input->col = 1; |
4907 | 24.5k | in++; |
4908 | 24.5k | } while (*in == 0xA); |
4909 | 18.7k | } |
4910 | 95.6k | get_more: |
4911 | 95.6k | ccol = ctxt->input->col; |
4912 | 346k | while (((*in > '-') && (*in <= 0x7F)) || |
4913 | 346k | ((*in >= 0x20) && (*in < '-')) || |
4914 | 346k | (*in == 0x09)) { |
4915 | 250k | in++; |
4916 | 250k | ccol++; |
4917 | 250k | } |
4918 | 95.6k | ctxt->input->col = ccol; |
4919 | 95.6k | if (*in == 0xA) { |
4920 | 7.74k | do { |
4921 | 7.74k | ctxt->input->line++; ctxt->input->col = 1; |
4922 | 7.74k | in++; |
4923 | 7.74k | } while (*in == 0xA); |
4924 | 4.88k | goto get_more; |
4925 | 4.88k | } |
4926 | 90.7k | nbchar = in - ctxt->input->cur; |
4927 | | /* |
4928 | | * save current set of data |
4929 | | */ |
4930 | 90.7k | if (nbchar > 0) { |
4931 | 74.2k | if ((ctxt->sax != NULL) && |
4932 | 74.2k | (ctxt->sax->comment != NULL)) { |
4933 | 74.2k | if (buf == NULL) { |
4934 | 43.0k | if ((*in == '-') && (in[1] == '-')) |
4935 | 13.1k | size = nbchar + 1; |
4936 | 29.8k | else |
4937 | 29.8k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
4938 | 43.0k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4939 | 43.0k | if (buf == NULL) { |
4940 | 0 | xmlErrMemory(ctxt, NULL); |
4941 | 0 | ctxt->instate = state; |
4942 | 0 | return; |
4943 | 0 | } |
4944 | 43.0k | len = 0; |
4945 | 43.0k | } else if (len + nbchar + 1 >= size) { |
4946 | 4.85k | xmlChar *new_buf; |
4947 | 4.85k | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
4948 | 4.85k | new_buf = (xmlChar *) xmlRealloc(buf, |
4949 | 4.85k | size * sizeof(xmlChar)); |
4950 | 4.85k | if (new_buf == NULL) { |
4951 | 0 | xmlFree (buf); |
4952 | 0 | xmlErrMemory(ctxt, NULL); |
4953 | 0 | ctxt->instate = state; |
4954 | 0 | return; |
4955 | 0 | } |
4956 | 4.85k | buf = new_buf; |
4957 | 4.85k | } |
4958 | 74.2k | memcpy(&buf[len], ctxt->input->cur, nbchar); |
4959 | 74.2k | len += nbchar; |
4960 | 74.2k | buf[len] = 0; |
4961 | 74.2k | } |
4962 | 74.2k | } |
4963 | 90.7k | if ((len > XML_MAX_TEXT_LENGTH) && |
4964 | 90.7k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
4965 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4966 | 0 | "Comment too big found", NULL); |
4967 | 0 | xmlFree (buf); |
4968 | 0 | return; |
4969 | 0 | } |
4970 | 90.7k | ctxt->input->cur = in; |
4971 | 90.7k | if (*in == 0xA) { |
4972 | 0 | in++; |
4973 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
4974 | 0 | } |
4975 | 90.7k | if (*in == 0xD) { |
4976 | 2.14k | in++; |
4977 | 2.14k | if (*in == 0xA) { |
4978 | 794 | ctxt->input->cur = in; |
4979 | 794 | in++; |
4980 | 794 | ctxt->input->line++; ctxt->input->col = 1; |
4981 | 794 | goto get_more; |
4982 | 794 | } |
4983 | 1.35k | in--; |
4984 | 1.35k | } |
4985 | 90.0k | SHRINK; |
4986 | 90.0k | GROW; |
4987 | 90.0k | if (ctxt->instate == XML_PARSER_EOF) { |
4988 | 0 | xmlFree(buf); |
4989 | 0 | return; |
4990 | 0 | } |
4991 | 90.0k | in = ctxt->input->cur; |
4992 | 90.0k | if (*in == '-') { |
4993 | 46.9k | if (in[1] == '-') { |
4994 | 34.3k | if (in[2] == '>') { |
4995 | 11.3k | if (ctxt->input->id != inputid) { |
4996 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
4997 | 0 | "comment doesn't start and stop in the" |
4998 | 0 | " same entity\n"); |
4999 | 0 | } |
5000 | 11.3k | SKIP(3); |
5001 | 11.3k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5002 | 11.3k | (!ctxt->disableSAX)) { |
5003 | 3.33k | if (buf != NULL) |
5004 | 1.17k | ctxt->sax->comment(ctxt->userData, buf); |
5005 | 2.16k | else |
5006 | 2.16k | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5007 | 3.33k | } |
5008 | 11.3k | if (buf != NULL) |
5009 | 9.14k | xmlFree(buf); |
5010 | 11.3k | if (ctxt->instate != XML_PARSER_EOF) |
5011 | 11.3k | ctxt->instate = state; |
5012 | 11.3k | return; |
5013 | 11.3k | } |
5014 | 22.9k | if (buf != NULL) { |
5015 | 18.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5016 | 18.0k | "Double hyphen within comment: " |
5017 | 18.0k | "<!--%.50s\n", |
5018 | 18.0k | buf); |
5019 | 18.0k | } else |
5020 | 4.84k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5021 | 4.84k | "Double hyphen within comment\n", NULL); |
5022 | 22.9k | if (ctxt->instate == XML_PARSER_EOF) { |
5023 | 0 | xmlFree(buf); |
5024 | 0 | return; |
5025 | 0 | } |
5026 | 22.9k | in++; |
5027 | 22.9k | ctxt->input->col++; |
5028 | 22.9k | } |
5029 | 35.6k | in++; |
5030 | 35.6k | ctxt->input->col++; |
5031 | 35.6k | goto get_more; |
5032 | 46.9k | } |
5033 | 90.0k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5034 | 42.6k | xmlParseCommentComplex(ctxt, buf, len, size); |
5035 | 42.6k | ctxt->instate = state; |
5036 | 42.6k | return; |
5037 | 54.0k | } |
5038 | | |
5039 | | |
5040 | | /** |
5041 | | * xmlParsePITarget: |
5042 | | * @ctxt: an XML parser context |
5043 | | * |
5044 | | * parse the name of a PI |
5045 | | * |
5046 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5047 | | * |
5048 | | * Returns the PITarget name or NULL |
5049 | | */ |
5050 | | |
5051 | | const xmlChar * |
5052 | 842k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5053 | 842k | const xmlChar *name; |
5054 | | |
5055 | 842k | name = xmlParseName(ctxt); |
5056 | 842k | if ((name != NULL) && |
5057 | 842k | ((name[0] == 'x') || (name[0] == 'X')) && |
5058 | 842k | ((name[1] == 'm') || (name[1] == 'M')) && |
5059 | 842k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5060 | 16.5k | int i; |
5061 | 16.5k | if ((name[0] == 'x') && (name[1] == 'm') && |
5062 | 16.5k | (name[2] == 'l') && (name[3] == 0)) { |
5063 | 6.32k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5064 | 6.32k | "XML declaration allowed only at the start of the document\n"); |
5065 | 6.32k | return(name); |
5066 | 10.2k | } else if (name[3] == 0) { |
5067 | 2.61k | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5068 | 2.61k | return(name); |
5069 | 2.61k | } |
5070 | 20.1k | for (i = 0;;i++) { |
5071 | 20.1k | if (xmlW3CPIs[i] == NULL) break; |
5072 | 15.3k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5073 | 2.79k | return(name); |
5074 | 15.3k | } |
5075 | 4.86k | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5076 | 4.86k | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5077 | 4.86k | NULL, NULL); |
5078 | 4.86k | } |
5079 | 830k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5080 | 2.96k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5081 | 2.96k | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5082 | 2.96k | } |
5083 | 830k | return(name); |
5084 | 842k | } |
5085 | | |
5086 | | #ifdef LIBXML_CATALOG_ENABLED |
5087 | | /** |
5088 | | * xmlParseCatalogPI: |
5089 | | * @ctxt: an XML parser context |
5090 | | * @catalog: the PI value string |
5091 | | * |
5092 | | * parse an XML Catalog Processing Instruction. |
5093 | | * |
5094 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5095 | | * |
5096 | | * Occurs only if allowed by the user and if happening in the Misc |
5097 | | * part of the document before any doctype information |
5098 | | * This will add the given catalog to the parsing context in order |
5099 | | * to be used if there is a resolution need further down in the document |
5100 | | */ |
5101 | | |
5102 | | static void |
5103 | 4.40k | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5104 | 4.40k | xmlChar *URL = NULL; |
5105 | 4.40k | const xmlChar *tmp, *base; |
5106 | 4.40k | xmlChar marker; |
5107 | | |
5108 | 4.40k | tmp = catalog; |
5109 | 4.40k | while (IS_BLANK_CH(*tmp)) tmp++; |
5110 | 4.40k | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5111 | 739 | goto error; |
5112 | 3.66k | tmp += 7; |
5113 | 3.66k | while (IS_BLANK_CH(*tmp)) tmp++; |
5114 | 3.66k | if (*tmp != '=') { |
5115 | 2.25k | return; |
5116 | 2.25k | } |
5117 | 1.40k | tmp++; |
5118 | 1.40k | while (IS_BLANK_CH(*tmp)) tmp++; |
5119 | 1.40k | marker = *tmp; |
5120 | 1.40k | if ((marker != '\'') && (marker != '"')) |
5121 | 418 | goto error; |
5122 | 989 | tmp++; |
5123 | 989 | base = tmp; |
5124 | 2.40k | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5125 | 989 | if (*tmp == 0) |
5126 | 376 | goto error; |
5127 | 613 | URL = xmlStrndup(base, tmp - base); |
5128 | 613 | tmp++; |
5129 | 829 | while (IS_BLANK_CH(*tmp)) tmp++; |
5130 | 613 | if (*tmp != 0) |
5131 | 229 | goto error; |
5132 | | |
5133 | 384 | if (URL != NULL) { |
5134 | 384 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5135 | 384 | xmlFree(URL); |
5136 | 384 | } |
5137 | 384 | return; |
5138 | | |
5139 | 1.76k | error: |
5140 | 1.76k | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5141 | 1.76k | "Catalog PI syntax error: %s\n", |
5142 | 1.76k | catalog, NULL); |
5143 | 1.76k | if (URL != NULL) |
5144 | 229 | xmlFree(URL); |
5145 | 1.76k | } |
5146 | | #endif |
5147 | | |
5148 | | /** |
5149 | | * xmlParsePI: |
5150 | | * @ctxt: an XML parser context |
5151 | | * |
5152 | | * parse an XML Processing Instruction. |
5153 | | * |
5154 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5155 | | * |
5156 | | * The processing is transferred to SAX once parsed. |
5157 | | */ |
5158 | | |
5159 | | void |
5160 | 842k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5161 | 842k | xmlChar *buf = NULL; |
5162 | 842k | size_t len = 0; |
5163 | 842k | size_t size = XML_PARSER_BUFFER_SIZE; |
5164 | 842k | int cur, l; |
5165 | 842k | const xmlChar *target; |
5166 | 842k | xmlParserInputState state; |
5167 | 842k | int count = 0; |
5168 | | |
5169 | 842k | if ((RAW == '<') && (NXT(1) == '?')) { |
5170 | 842k | int inputid = ctxt->input->id; |
5171 | 842k | state = ctxt->instate; |
5172 | 842k | ctxt->instate = XML_PARSER_PI; |
5173 | | /* |
5174 | | * this is a Processing Instruction. |
5175 | | */ |
5176 | 842k | SKIP(2); |
5177 | 842k | SHRINK; |
5178 | | |
5179 | | /* |
5180 | | * Parse the target name and check for special support like |
5181 | | * namespace. |
5182 | | */ |
5183 | 842k | target = xmlParsePITarget(ctxt); |
5184 | 842k | if (target != NULL) { |
5185 | 96.1k | if ((RAW == '?') && (NXT(1) == '>')) { |
5186 | 8.46k | if (inputid != ctxt->input->id) { |
5187 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5188 | 0 | "PI declaration doesn't start and stop in" |
5189 | 0 | " the same entity\n"); |
5190 | 0 | } |
5191 | 8.46k | SKIP(2); |
5192 | | |
5193 | | /* |
5194 | | * SAX: PI detected. |
5195 | | */ |
5196 | 8.46k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5197 | 8.46k | (ctxt->sax->processingInstruction != NULL)) |
5198 | 2.85k | ctxt->sax->processingInstruction(ctxt->userData, |
5199 | 2.85k | target, NULL); |
5200 | 8.46k | if (ctxt->instate != XML_PARSER_EOF) |
5201 | 8.46k | ctxt->instate = state; |
5202 | 8.46k | return; |
5203 | 8.46k | } |
5204 | 87.7k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
5205 | 87.7k | if (buf == NULL) { |
5206 | 0 | xmlErrMemory(ctxt, NULL); |
5207 | 0 | ctxt->instate = state; |
5208 | 0 | return; |
5209 | 0 | } |
5210 | 87.7k | if (SKIP_BLANKS == 0) { |
5211 | 16.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5212 | 16.0k | "ParsePI: PI %s space expected\n", target); |
5213 | 16.0k | } |
5214 | 87.7k | cur = CUR_CHAR(l); |
5215 | 175M | while (IS_CHAR(cur) && /* checked */ |
5216 | 175M | ((cur != '?') || (NXT(1) != '>'))) { |
5217 | 175M | if (len + 5 >= size) { |
5218 | 15.5k | xmlChar *tmp; |
5219 | 15.5k | size_t new_size = size * 2; |
5220 | 15.5k | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5221 | 15.5k | if (tmp == NULL) { |
5222 | 0 | xmlErrMemory(ctxt, NULL); |
5223 | 0 | xmlFree(buf); |
5224 | 0 | ctxt->instate = state; |
5225 | 0 | return; |
5226 | 0 | } |
5227 | 15.5k | buf = tmp; |
5228 | 15.5k | size = new_size; |
5229 | 15.5k | } |
5230 | 175M | count++; |
5231 | 175M | if (count > 50) { |
5232 | 3.41M | SHRINK; |
5233 | 3.41M | GROW; |
5234 | 3.41M | if (ctxt->instate == XML_PARSER_EOF) { |
5235 | 0 | xmlFree(buf); |
5236 | 0 | return; |
5237 | 0 | } |
5238 | 3.41M | count = 0; |
5239 | 3.41M | if ((len > XML_MAX_TEXT_LENGTH) && |
5240 | 3.41M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
5241 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5242 | 0 | "PI %s too big found", target); |
5243 | 0 | xmlFree(buf); |
5244 | 0 | ctxt->instate = state; |
5245 | 0 | return; |
5246 | 0 | } |
5247 | 3.41M | } |
5248 | 175M | COPY_BUF(l,buf,len,cur); |
5249 | 175M | NEXTL(l); |
5250 | 175M | cur = CUR_CHAR(l); |
5251 | 175M | if (cur == 0) { |
5252 | 39.3k | SHRINK; |
5253 | 39.3k | GROW; |
5254 | 39.3k | cur = CUR_CHAR(l); |
5255 | 39.3k | } |
5256 | 175M | } |
5257 | 87.7k | if ((len > XML_MAX_TEXT_LENGTH) && |
5258 | 87.7k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
5259 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5260 | 0 | "PI %s too big found", target); |
5261 | 0 | xmlFree(buf); |
5262 | 0 | ctxt->instate = state; |
5263 | 0 | return; |
5264 | 0 | } |
5265 | 87.7k | buf[len] = 0; |
5266 | 87.7k | if (cur != '?') { |
5267 | 26.9k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5268 | 26.9k | "ParsePI: PI %s never end ...\n", target); |
5269 | 60.7k | } else { |
5270 | 60.7k | if (inputid != ctxt->input->id) { |
5271 | 48.1k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5272 | 48.1k | "PI declaration doesn't start and stop in" |
5273 | 48.1k | " the same entity\n"); |
5274 | 48.1k | } |
5275 | 60.7k | SKIP(2); |
5276 | | |
5277 | 60.7k | #ifdef LIBXML_CATALOG_ENABLED |
5278 | 60.7k | if (((state == XML_PARSER_MISC) || |
5279 | 60.7k | (state == XML_PARSER_START)) && |
5280 | 60.7k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5281 | 4.40k | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5282 | 4.40k | if ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5283 | 4.40k | (allow == XML_CATA_ALLOW_ALL)) |
5284 | 4.40k | xmlParseCatalogPI(ctxt, buf); |
5285 | 4.40k | } |
5286 | 60.7k | #endif |
5287 | | |
5288 | | |
5289 | | /* |
5290 | | * SAX: PI detected. |
5291 | | */ |
5292 | 60.7k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5293 | 60.7k | (ctxt->sax->processingInstruction != NULL)) |
5294 | 2.31k | ctxt->sax->processingInstruction(ctxt->userData, |
5295 | 2.31k | target, buf); |
5296 | 60.7k | } |
5297 | 87.7k | xmlFree(buf); |
5298 | 746k | } else { |
5299 | 746k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5300 | 746k | } |
5301 | 834k | if (ctxt->instate != XML_PARSER_EOF) |
5302 | 834k | ctxt->instate = state; |
5303 | 834k | } |
5304 | 842k | } |
5305 | | |
5306 | | /** |
5307 | | * xmlParseNotationDecl: |
5308 | | * @ctxt: an XML parser context |
5309 | | * |
5310 | | * parse a notation declaration |
5311 | | * |
5312 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5313 | | * |
5314 | | * Hence there is actually 3 choices: |
5315 | | * 'PUBLIC' S PubidLiteral |
5316 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5317 | | * and 'SYSTEM' S SystemLiteral |
5318 | | * |
5319 | | * See the NOTE on xmlParseExternalID(). |
5320 | | */ |
5321 | | |
5322 | | void |
5323 | 193k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5324 | 193k | const xmlChar *name; |
5325 | 193k | xmlChar *Pubid; |
5326 | 193k | xmlChar *Systemid; |
5327 | | |
5328 | 193k | if (CMP10(CUR_PTR, '<', '!', 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5329 | 84.2k | int inputid = ctxt->input->id; |
5330 | 84.2k | SHRINK; |
5331 | 84.2k | SKIP(10); |
5332 | 84.2k | if (SKIP_BLANKS == 0) { |
5333 | 2.09k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5334 | 2.09k | "Space required after '<!NOTATION'\n"); |
5335 | 2.09k | return; |
5336 | 2.09k | } |
5337 | | |
5338 | 82.1k | name = xmlParseName(ctxt); |
5339 | 82.1k | if (name == NULL) { |
5340 | 3.46k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5341 | 3.46k | return; |
5342 | 3.46k | } |
5343 | 78.7k | if (xmlStrchr(name, ':') != NULL) { |
5344 | 3.74k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5345 | 3.74k | "colons are forbidden from notation names '%s'\n", |
5346 | 3.74k | name, NULL, NULL); |
5347 | 3.74k | } |
5348 | 78.7k | if (SKIP_BLANKS == 0) { |
5349 | 13.8k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5350 | 13.8k | "Space required after the NOTATION name'\n"); |
5351 | 13.8k | return; |
5352 | 13.8k | } |
5353 | | |
5354 | | /* |
5355 | | * Parse the IDs. |
5356 | | */ |
5357 | 64.8k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5358 | 64.8k | SKIP_BLANKS; |
5359 | | |
5360 | 64.8k | if (RAW == '>') { |
5361 | 5.03k | if (inputid != ctxt->input->id) { |
5362 | 1.39k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5363 | 1.39k | "Notation declaration doesn't start and stop" |
5364 | 1.39k | " in the same entity\n"); |
5365 | 1.39k | } |
5366 | 5.03k | NEXT; |
5367 | 5.03k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5368 | 5.03k | (ctxt->sax->notationDecl != NULL)) |
5369 | 109 | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5370 | 59.8k | } else { |
5371 | 59.8k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5372 | 59.8k | } |
5373 | 64.8k | if (Systemid != NULL) xmlFree(Systemid); |
5374 | 64.8k | if (Pubid != NULL) xmlFree(Pubid); |
5375 | 64.8k | } |
5376 | 193k | } |
5377 | | |
5378 | | /** |
5379 | | * xmlParseEntityDecl: |
5380 | | * @ctxt: an XML parser context |
5381 | | * |
5382 | | * parse <!ENTITY declarations |
5383 | | * |
5384 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5385 | | * |
5386 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5387 | | * |
5388 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5389 | | * |
5390 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5391 | | * |
5392 | | * [74] PEDef ::= EntityValue | ExternalID |
5393 | | * |
5394 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5395 | | * |
5396 | | * [ VC: Notation Declared ] |
5397 | | * The Name must match the declared name of a notation. |
5398 | | */ |
5399 | | |
5400 | | void |
5401 | 4.17M | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5402 | 4.17M | const xmlChar *name = NULL; |
5403 | 4.17M | xmlChar *value = NULL; |
5404 | 4.17M | xmlChar *URI = NULL, *literal = NULL; |
5405 | 4.17M | const xmlChar *ndata = NULL; |
5406 | 4.17M | int isParameter = 0; |
5407 | 4.17M | xmlChar *orig = NULL; |
5408 | | |
5409 | | /* GROW; done in the caller */ |
5410 | 4.17M | if (CMP8(CUR_PTR, '<', '!', 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5411 | 4.16M | int inputid = ctxt->input->id; |
5412 | 4.16M | SHRINK; |
5413 | 4.16M | SKIP(8); |
5414 | 4.16M | if (SKIP_BLANKS == 0) { |
5415 | 1.73M | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5416 | 1.73M | "Space required after '<!ENTITY'\n"); |
5417 | 1.73M | } |
5418 | | |
5419 | 4.16M | if (RAW == '%') { |
5420 | 10.4k | NEXT; |
5421 | 10.4k | if (SKIP_BLANKS == 0) { |
5422 | 4.46k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5423 | 4.46k | "Space required after '%%'\n"); |
5424 | 4.46k | } |
5425 | 10.4k | isParameter = 1; |
5426 | 10.4k | } |
5427 | | |
5428 | 4.16M | name = xmlParseName(ctxt); |
5429 | 4.16M | if (name == NULL) { |
5430 | 2.39M | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5431 | 2.39M | "xmlParseEntityDecl: no name\n"); |
5432 | 2.39M | return; |
5433 | 2.39M | } |
5434 | 1.76M | if (xmlStrchr(name, ':') != NULL) { |
5435 | 21.8k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5436 | 21.8k | "colons are forbidden from entities names '%s'\n", |
5437 | 21.8k | name, NULL, NULL); |
5438 | 21.8k | } |
5439 | 1.76M | if (SKIP_BLANKS == 0) { |
5440 | 1.72M | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5441 | 1.72M | "Space required after the entity name\n"); |
5442 | 1.72M | } |
5443 | | |
5444 | 1.76M | ctxt->instate = XML_PARSER_ENTITY_DECL; |
5445 | | /* |
5446 | | * handle the various case of definitions... |
5447 | | */ |
5448 | 1.76M | if (isParameter) { |
5449 | 10.4k | if ((RAW == '"') || (RAW == '\'')) { |
5450 | 5.78k | value = xmlParseEntityValue(ctxt, &orig); |
5451 | 5.78k | if (value) { |
5452 | 5.12k | if ((ctxt->sax != NULL) && |
5453 | 5.12k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5454 | 3.97k | ctxt->sax->entityDecl(ctxt->userData, name, |
5455 | 3.97k | XML_INTERNAL_PARAMETER_ENTITY, |
5456 | 3.97k | NULL, NULL, value); |
5457 | 5.12k | } |
5458 | 5.78k | } else { |
5459 | 4.68k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5460 | 4.68k | if ((URI == NULL) && (literal == NULL)) { |
5461 | 92 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5462 | 92 | } |
5463 | 4.68k | if (URI) { |
5464 | 3.39k | xmlURIPtr uri; |
5465 | | |
5466 | 3.39k | uri = xmlParseURI((const char *) URI); |
5467 | 3.39k | if (uri == NULL) { |
5468 | 1.38k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5469 | 1.38k | "Invalid URI: %s\n", URI); |
5470 | | /* |
5471 | | * This really ought to be a well formedness error |
5472 | | * but the XML Core WG decided otherwise c.f. issue |
5473 | | * E26 of the XML erratas. |
5474 | | */ |
5475 | 2.00k | } else { |
5476 | 2.00k | if (uri->fragment != NULL) { |
5477 | | /* |
5478 | | * Okay this is foolish to block those but not |
5479 | | * invalid URIs. |
5480 | | */ |
5481 | 862 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5482 | 1.14k | } else { |
5483 | 1.14k | if ((ctxt->sax != NULL) && |
5484 | 1.14k | (!ctxt->disableSAX) && |
5485 | 1.14k | (ctxt->sax->entityDecl != NULL)) |
5486 | 318 | ctxt->sax->entityDecl(ctxt->userData, name, |
5487 | 318 | XML_EXTERNAL_PARAMETER_ENTITY, |
5488 | 318 | literal, URI, NULL); |
5489 | 1.14k | } |
5490 | 2.00k | xmlFreeURI(uri); |
5491 | 2.00k | } |
5492 | 3.39k | } |
5493 | 4.68k | } |
5494 | 1.75M | } else { |
5495 | 1.75M | if ((RAW == '"') || (RAW == '\'')) { |
5496 | 33.5k | value = xmlParseEntityValue(ctxt, &orig); |
5497 | 33.5k | if ((ctxt->sax != NULL) && |
5498 | 33.5k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5499 | 5.17k | ctxt->sax->entityDecl(ctxt->userData, name, |
5500 | 5.17k | XML_INTERNAL_GENERAL_ENTITY, |
5501 | 5.17k | NULL, NULL, value); |
5502 | | /* |
5503 | | * For expat compatibility in SAX mode. |
5504 | | */ |
5505 | 33.5k | if ((ctxt->myDoc == NULL) || |
5506 | 33.5k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5507 | 1.58k | if (ctxt->myDoc == NULL) { |
5508 | 249 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5509 | 249 | if (ctxt->myDoc == NULL) { |
5510 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5511 | 0 | return; |
5512 | 0 | } |
5513 | 249 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5514 | 249 | } |
5515 | 1.58k | if (ctxt->myDoc->intSubset == NULL) |
5516 | 249 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5517 | 249 | BAD_CAST "fake", NULL, NULL); |
5518 | | |
5519 | 1.58k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5520 | 1.58k | NULL, NULL, value); |
5521 | 1.58k | } |
5522 | 1.72M | } else { |
5523 | 1.72M | URI = xmlParseExternalID(ctxt, &literal, 1); |
5524 | 1.72M | if ((URI == NULL) && (literal == NULL)) { |
5525 | 1.71M | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5526 | 1.71M | } |
5527 | 1.72M | if (URI) { |
5528 | 5.57k | xmlURIPtr uri; |
5529 | | |
5530 | 5.57k | uri = xmlParseURI((const char *)URI); |
5531 | 5.57k | if (uri == NULL) { |
5532 | 2.75k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5533 | 2.75k | "Invalid URI: %s\n", URI); |
5534 | | /* |
5535 | | * This really ought to be a well formedness error |
5536 | | * but the XML Core WG decided otherwise c.f. issue |
5537 | | * E26 of the XML erratas. |
5538 | | */ |
5539 | 2.82k | } else { |
5540 | 2.82k | if (uri->fragment != NULL) { |
5541 | | /* |
5542 | | * Okay this is foolish to block those but not |
5543 | | * invalid URIs. |
5544 | | */ |
5545 | 202 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5546 | 202 | } |
5547 | 2.82k | xmlFreeURI(uri); |
5548 | 2.82k | } |
5549 | 5.57k | } |
5550 | 1.72M | if ((RAW != '>') && (SKIP_BLANKS == 0)) { |
5551 | 218 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5552 | 218 | "Space required before 'NDATA'\n"); |
5553 | 218 | } |
5554 | 1.72M | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5555 | 657 | SKIP(5); |
5556 | 657 | if (SKIP_BLANKS == 0) { |
5557 | 73 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5558 | 73 | "Space required after 'NDATA'\n"); |
5559 | 73 | } |
5560 | 657 | ndata = xmlParseName(ctxt); |
5561 | 657 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5562 | 657 | (ctxt->sax->unparsedEntityDecl != NULL)) |
5563 | 382 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5564 | 382 | literal, URI, ndata); |
5565 | 1.72M | } else { |
5566 | 1.72M | if ((ctxt->sax != NULL) && |
5567 | 1.72M | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5568 | 1.67k | ctxt->sax->entityDecl(ctxt->userData, name, |
5569 | 1.67k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5570 | 1.67k | literal, URI, NULL); |
5571 | | /* |
5572 | | * For expat compatibility in SAX mode. |
5573 | | * assuming the entity replacement was asked for |
5574 | | */ |
5575 | 1.72M | if ((ctxt->replaceEntities != 0) && |
5576 | 1.72M | ((ctxt->myDoc == NULL) || |
5577 | 0 | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5578 | 0 | if (ctxt->myDoc == NULL) { |
5579 | 0 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5580 | 0 | if (ctxt->myDoc == NULL) { |
5581 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5582 | 0 | return; |
5583 | 0 | } |
5584 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5585 | 0 | } |
5586 | | |
5587 | 0 | if (ctxt->myDoc->intSubset == NULL) |
5588 | 0 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5589 | 0 | BAD_CAST "fake", NULL, NULL); |
5590 | 0 | xmlSAX2EntityDecl(ctxt, name, |
5591 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5592 | 0 | literal, URI, NULL); |
5593 | 0 | } |
5594 | 1.72M | } |
5595 | 1.72M | } |
5596 | 1.75M | } |
5597 | 1.76M | if (ctxt->instate == XML_PARSER_EOF) |
5598 | 0 | goto done; |
5599 | 1.76M | SKIP_BLANKS; |
5600 | 1.76M | if (RAW != '>') { |
5601 | 984 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5602 | 984 | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5603 | 984 | xmlHaltParser(ctxt); |
5604 | 1.76M | } else { |
5605 | 1.76M | if (inputid != ctxt->input->id) { |
5606 | 3.14k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5607 | 3.14k | "Entity declaration doesn't start and stop in" |
5608 | 3.14k | " the same entity\n"); |
5609 | 3.14k | } |
5610 | 1.76M | NEXT; |
5611 | 1.76M | } |
5612 | 1.76M | if (orig != NULL) { |
5613 | | /* |
5614 | | * Ugly mechanism to save the raw entity value. |
5615 | | */ |
5616 | 31.9k | xmlEntityPtr cur = NULL; |
5617 | | |
5618 | 31.9k | if (isParameter) { |
5619 | 5.64k | if ((ctxt->sax != NULL) && |
5620 | 5.64k | (ctxt->sax->getParameterEntity != NULL)) |
5621 | 5.64k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5622 | 26.3k | } else { |
5623 | 26.3k | if ((ctxt->sax != NULL) && |
5624 | 26.3k | (ctxt->sax->getEntity != NULL)) |
5625 | 26.3k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5626 | 26.3k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5627 | 18.6k | cur = xmlSAX2GetEntity(ctxt, name); |
5628 | 18.6k | } |
5629 | 26.3k | } |
5630 | 31.9k | if ((cur != NULL) && (cur->orig == NULL)) { |
5631 | 7.64k | cur->orig = orig; |
5632 | 7.64k | orig = NULL; |
5633 | 7.64k | } |
5634 | 31.9k | } |
5635 | | |
5636 | 1.76M | done: |
5637 | 1.76M | if (value != NULL) xmlFree(value); |
5638 | 1.76M | if (URI != NULL) xmlFree(URI); |
5639 | 1.76M | if (literal != NULL) xmlFree(literal); |
5640 | 1.76M | if (orig != NULL) xmlFree(orig); |
5641 | 1.76M | } |
5642 | 4.17M | } |
5643 | | |
5644 | | /** |
5645 | | * xmlParseDefaultDecl: |
5646 | | * @ctxt: an XML parser context |
5647 | | * @value: Receive a possible fixed default value for the attribute |
5648 | | * |
5649 | | * Parse an attribute default declaration |
5650 | | * |
5651 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5652 | | * |
5653 | | * [ VC: Required Attribute ] |
5654 | | * if the default declaration is the keyword #REQUIRED, then the |
5655 | | * attribute must be specified for all elements of the type in the |
5656 | | * attribute-list declaration. |
5657 | | * |
5658 | | * [ VC: Attribute Default Legal ] |
5659 | | * The declared default value must meet the lexical constraints of |
5660 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5661 | | * |
5662 | | * [ VC: Fixed Attribute Default ] |
5663 | | * if an attribute has a default value declared with the #FIXED |
5664 | | * keyword, instances of that attribute must match the default value. |
5665 | | * |
5666 | | * [ WFC: No < in Attribute Values ] |
5667 | | * handled in xmlParseAttValue() |
5668 | | * |
5669 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5670 | | * or XML_ATTRIBUTE_FIXED. |
5671 | | */ |
5672 | | |
5673 | | int |
5674 | 818k | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5675 | 818k | int val; |
5676 | 818k | xmlChar *ret; |
5677 | | |
5678 | 818k | *value = NULL; |
5679 | 818k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5680 | 544 | SKIP(9); |
5681 | 544 | return(XML_ATTRIBUTE_REQUIRED); |
5682 | 544 | } |
5683 | 817k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5684 | 862 | SKIP(8); |
5685 | 862 | return(XML_ATTRIBUTE_IMPLIED); |
5686 | 862 | } |
5687 | 817k | val = XML_ATTRIBUTE_NONE; |
5688 | 817k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5689 | 1.50k | SKIP(6); |
5690 | 1.50k | val = XML_ATTRIBUTE_FIXED; |
5691 | 1.50k | if (SKIP_BLANKS == 0) { |
5692 | 750 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5693 | 750 | "Space required after '#FIXED'\n"); |
5694 | 750 | } |
5695 | 1.50k | } |
5696 | 817k | ret = xmlParseAttValue(ctxt); |
5697 | 817k | ctxt->instate = XML_PARSER_DTD; |
5698 | 817k | if (ret == NULL) { |
5699 | 759k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5700 | 759k | "Attribute default value declaration error\n"); |
5701 | 759k | } else |
5702 | 57.8k | *value = ret; |
5703 | 817k | return(val); |
5704 | 817k | } |
5705 | | |
5706 | | /** |
5707 | | * xmlParseNotationType: |
5708 | | * @ctxt: an XML parser context |
5709 | | * |
5710 | | * parse an Notation attribute type. |
5711 | | * |
5712 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5713 | | * |
5714 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5715 | | * |
5716 | | * [ VC: Notation Attributes ] |
5717 | | * Values of this type must match one of the notation names included |
5718 | | * in the declaration; all notation names in the declaration must be declared. |
5719 | | * |
5720 | | * Returns: the notation attribute tree built while parsing |
5721 | | */ |
5722 | | |
5723 | | xmlEnumerationPtr |
5724 | 13.5k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5725 | 13.5k | const xmlChar *name; |
5726 | 13.5k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5727 | | |
5728 | 13.5k | if (RAW != '(') { |
5729 | 2.13k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5730 | 2.13k | return(NULL); |
5731 | 2.13k | } |
5732 | 11.4k | SHRINK; |
5733 | 12.5k | do { |
5734 | 12.5k | NEXT; |
5735 | 12.5k | SKIP_BLANKS; |
5736 | 12.5k | name = xmlParseName(ctxt); |
5737 | 12.5k | if (name == NULL) { |
5738 | 2.13k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5739 | 2.13k | "Name expected in NOTATION declaration\n"); |
5740 | 2.13k | xmlFreeEnumeration(ret); |
5741 | 2.13k | return(NULL); |
5742 | 2.13k | } |
5743 | 10.4k | tmp = ret; |
5744 | 13.3k | while (tmp != NULL) { |
5745 | 3.53k | if (xmlStrEqual(name, tmp->name)) { |
5746 | 593 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5747 | 593 | "standalone: attribute notation value token %s duplicated\n", |
5748 | 593 | name, NULL); |
5749 | 593 | if (!xmlDictOwns(ctxt->dict, name)) |
5750 | 0 | xmlFree((xmlChar *) name); |
5751 | 593 | break; |
5752 | 593 | } |
5753 | 2.93k | tmp = tmp->next; |
5754 | 2.93k | } |
5755 | 10.4k | if (tmp == NULL) { |
5756 | 9.81k | cur = xmlCreateEnumeration(name); |
5757 | 9.81k | if (cur == NULL) { |
5758 | 0 | xmlFreeEnumeration(ret); |
5759 | 0 | return(NULL); |
5760 | 0 | } |
5761 | 9.81k | if (last == NULL) ret = last = cur; |
5762 | 456 | else { |
5763 | 456 | last->next = cur; |
5764 | 456 | last = cur; |
5765 | 456 | } |
5766 | 9.81k | } |
5767 | 10.4k | SKIP_BLANKS; |
5768 | 10.4k | } while (RAW == '|'); |
5769 | 9.31k | if (RAW != ')') { |
5770 | 6.29k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5771 | 6.29k | xmlFreeEnumeration(ret); |
5772 | 6.29k | return(NULL); |
5773 | 6.29k | } |
5774 | 3.01k | NEXT; |
5775 | 3.01k | return(ret); |
5776 | 9.31k | } |
5777 | | |
5778 | | /** |
5779 | | * xmlParseEnumerationType: |
5780 | | * @ctxt: an XML parser context |
5781 | | * |
5782 | | * parse an Enumeration attribute type. |
5783 | | * |
5784 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5785 | | * |
5786 | | * [ VC: Enumeration ] |
5787 | | * Values of this type must match one of the Nmtoken tokens in |
5788 | | * the declaration |
5789 | | * |
5790 | | * Returns: the enumeration attribute tree built while parsing |
5791 | | */ |
5792 | | |
5793 | | xmlEnumerationPtr |
5794 | 82.5k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
5795 | 82.5k | xmlChar *name; |
5796 | 82.5k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5797 | | |
5798 | 82.5k | if (RAW != '(') { |
5799 | 36.1k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5800 | 36.1k | return(NULL); |
5801 | 36.1k | } |
5802 | 46.4k | SHRINK; |
5803 | 49.7k | do { |
5804 | 49.7k | NEXT; |
5805 | 49.7k | SKIP_BLANKS; |
5806 | 49.7k | name = xmlParseNmtoken(ctxt); |
5807 | 49.7k | if (name == NULL) { |
5808 | 781 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5809 | 781 | return(ret); |
5810 | 781 | } |
5811 | 48.9k | tmp = ret; |
5812 | 54.7k | while (tmp != NULL) { |
5813 | 6.56k | if (xmlStrEqual(name, tmp->name)) { |
5814 | 769 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5815 | 769 | "standalone: attribute enumeration value token %s duplicated\n", |
5816 | 769 | name, NULL); |
5817 | 769 | if (!xmlDictOwns(ctxt->dict, name)) |
5818 | 769 | xmlFree(name); |
5819 | 769 | break; |
5820 | 769 | } |
5821 | 5.79k | tmp = tmp->next; |
5822 | 5.79k | } |
5823 | 48.9k | if (tmp == NULL) { |
5824 | 48.2k | cur = xmlCreateEnumeration(name); |
5825 | 48.2k | if (!xmlDictOwns(ctxt->dict, name)) |
5826 | 48.2k | xmlFree(name); |
5827 | 48.2k | if (cur == NULL) { |
5828 | 0 | xmlFreeEnumeration(ret); |
5829 | 0 | return(NULL); |
5830 | 0 | } |
5831 | 48.2k | if (last == NULL) ret = last = cur; |
5832 | 2.53k | else { |
5833 | 2.53k | last->next = cur; |
5834 | 2.53k | last = cur; |
5835 | 2.53k | } |
5836 | 48.2k | } |
5837 | 48.9k | SKIP_BLANKS; |
5838 | 48.9k | } while (RAW == '|'); |
5839 | 45.6k | if (RAW != ')') { |
5840 | 15.3k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5841 | 15.3k | return(ret); |
5842 | 15.3k | } |
5843 | 30.2k | NEXT; |
5844 | 30.2k | return(ret); |
5845 | 45.6k | } |
5846 | | |
5847 | | /** |
5848 | | * xmlParseEnumeratedType: |
5849 | | * @ctxt: an XML parser context |
5850 | | * @tree: the enumeration tree built while parsing |
5851 | | * |
5852 | | * parse an Enumerated attribute type. |
5853 | | * |
5854 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5855 | | * |
5856 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5857 | | * |
5858 | | * |
5859 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
5860 | | */ |
5861 | | |
5862 | | int |
5863 | 97.2k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5864 | 97.2k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5865 | 14.6k | SKIP(8); |
5866 | 14.6k | if (SKIP_BLANKS == 0) { |
5867 | 1.05k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5868 | 1.05k | "Space required after 'NOTATION'\n"); |
5869 | 1.05k | return(0); |
5870 | 1.05k | } |
5871 | 13.5k | *tree = xmlParseNotationType(ctxt); |
5872 | 13.5k | if (*tree == NULL) return(0); |
5873 | 3.01k | return(XML_ATTRIBUTE_NOTATION); |
5874 | 13.5k | } |
5875 | 82.5k | *tree = xmlParseEnumerationType(ctxt); |
5876 | 82.5k | if (*tree == NULL) return(0); |
5877 | 45.6k | return(XML_ATTRIBUTE_ENUMERATION); |
5878 | 82.5k | } |
5879 | | |
5880 | | /** |
5881 | | * xmlParseAttributeType: |
5882 | | * @ctxt: an XML parser context |
5883 | | * @tree: the enumeration tree built while parsing |
5884 | | * |
5885 | | * parse the Attribute list def for an element |
5886 | | * |
5887 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
5888 | | * |
5889 | | * [55] StringType ::= 'CDATA' |
5890 | | * |
5891 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
5892 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
5893 | | * |
5894 | | * Validity constraints for attribute values syntax are checked in |
5895 | | * xmlValidateAttributeValue() |
5896 | | * |
5897 | | * [ VC: ID ] |
5898 | | * Values of type ID must match the Name production. A name must not |
5899 | | * appear more than once in an XML document as a value of this type; |
5900 | | * i.e., ID values must uniquely identify the elements which bear them. |
5901 | | * |
5902 | | * [ VC: One ID per Element Type ] |
5903 | | * No element type may have more than one ID attribute specified. |
5904 | | * |
5905 | | * [ VC: ID Attribute Default ] |
5906 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
5907 | | * |
5908 | | * [ VC: IDREF ] |
5909 | | * Values of type IDREF must match the Name production, and values |
5910 | | * of type IDREFS must match Names; each IDREF Name must match the value |
5911 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
5912 | | * values must match the value of some ID attribute. |
5913 | | * |
5914 | | * [ VC: Entity Name ] |
5915 | | * Values of type ENTITY must match the Name production, values |
5916 | | * of type ENTITIES must match Names; each Entity Name must match the |
5917 | | * name of an unparsed entity declared in the DTD. |
5918 | | * |
5919 | | * [ VC: Name Token ] |
5920 | | * Values of type NMTOKEN must match the Nmtoken production; values |
5921 | | * of type NMTOKENS must match Nmtokens. |
5922 | | * |
5923 | | * Returns the attribute type |
5924 | | */ |
5925 | | int |
5926 | 892k | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5927 | 892k | SHRINK; |
5928 | 892k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
5929 | 738k | SKIP(5); |
5930 | 738k | return(XML_ATTRIBUTE_CDATA); |
5931 | 738k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
5932 | 5.35k | SKIP(6); |
5933 | 5.35k | return(XML_ATTRIBUTE_IDREFS); |
5934 | 148k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
5935 | 2.05k | SKIP(5); |
5936 | 2.05k | return(XML_ATTRIBUTE_IDREF); |
5937 | 146k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
5938 | 31.8k | SKIP(2); |
5939 | 31.8k | return(XML_ATTRIBUTE_ID); |
5940 | 114k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5941 | 2.94k | SKIP(6); |
5942 | 2.94k | return(XML_ATTRIBUTE_ENTITY); |
5943 | 111k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
5944 | 8.54k | SKIP(8); |
5945 | 8.54k | return(XML_ATTRIBUTE_ENTITIES); |
5946 | 102k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
5947 | 3.10k | SKIP(8); |
5948 | 3.10k | return(XML_ATTRIBUTE_NMTOKENS); |
5949 | 99.6k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
5950 | 2.42k | SKIP(7); |
5951 | 2.42k | return(XML_ATTRIBUTE_NMTOKEN); |
5952 | 2.42k | } |
5953 | 97.2k | return(xmlParseEnumeratedType(ctxt, tree)); |
5954 | 892k | } |
5955 | | |
5956 | | /** |
5957 | | * xmlParseAttributeListDecl: |
5958 | | * @ctxt: an XML parser context |
5959 | | * |
5960 | | * : parse the Attribute list def for an element |
5961 | | * |
5962 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
5963 | | * |
5964 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
5965 | | * |
5966 | | */ |
5967 | | void |
5968 | 888k | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
5969 | 888k | const xmlChar *elemName; |
5970 | 888k | const xmlChar *attrName; |
5971 | 888k | xmlEnumerationPtr tree; |
5972 | | |
5973 | 888k | if (CMP9(CUR_PTR, '<', '!', 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
5974 | 878k | int inputid = ctxt->input->id; |
5975 | | |
5976 | 878k | SKIP(9); |
5977 | 878k | if (SKIP_BLANKS == 0) { |
5978 | 836k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5979 | 836k | "Space required after '<!ATTLIST'\n"); |
5980 | 836k | } |
5981 | 878k | elemName = xmlParseName(ctxt); |
5982 | 878k | if (elemName == NULL) { |
5983 | 2.23k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5984 | 2.23k | "ATTLIST: no name for Element\n"); |
5985 | 2.23k | return; |
5986 | 2.23k | } |
5987 | 876k | SKIP_BLANKS; |
5988 | 876k | GROW; |
5989 | 934k | while ((RAW != '>') && (ctxt->instate != XML_PARSER_EOF)) { |
5990 | 920k | int type; |
5991 | 920k | int def; |
5992 | 920k | xmlChar *defaultValue = NULL; |
5993 | | |
5994 | 920k | GROW; |
5995 | 920k | tree = NULL; |
5996 | 920k | attrName = xmlParseName(ctxt); |
5997 | 920k | if (attrName == NULL) { |
5998 | 18.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5999 | 18.2k | "ATTLIST: no name for Attribute\n"); |
6000 | 18.2k | break; |
6001 | 18.2k | } |
6002 | 901k | GROW; |
6003 | 901k | if (SKIP_BLANKS == 0) { |
6004 | 9.56k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6005 | 9.56k | "Space required after the attribute name\n"); |
6006 | 9.56k | break; |
6007 | 9.56k | } |
6008 | | |
6009 | 892k | type = xmlParseAttributeType(ctxt, &tree); |
6010 | 892k | if (type <= 0) { |
6011 | 48.5k | break; |
6012 | 48.5k | } |
6013 | | |
6014 | 843k | GROW; |
6015 | 843k | if (SKIP_BLANKS == 0) { |
6016 | 25.3k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6017 | 25.3k | "Space required after the attribute type\n"); |
6018 | 25.3k | if (tree != NULL) |
6019 | 20.4k | xmlFreeEnumeration(tree); |
6020 | 25.3k | break; |
6021 | 25.3k | } |
6022 | | |
6023 | 818k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6024 | 818k | if (def <= 0) { |
6025 | 0 | if (defaultValue != NULL) |
6026 | 0 | xmlFree(defaultValue); |
6027 | 0 | if (tree != NULL) |
6028 | 0 | xmlFreeEnumeration(tree); |
6029 | 0 | break; |
6030 | 0 | } |
6031 | 818k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6032 | 55.1k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6033 | | |
6034 | 818k | GROW; |
6035 | 818k | if (RAW != '>') { |
6036 | 807k | if (SKIP_BLANKS == 0) { |
6037 | 760k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6038 | 760k | "Space required after the attribute default value\n"); |
6039 | 760k | if (defaultValue != NULL) |
6040 | 2.61k | xmlFree(defaultValue); |
6041 | 760k | if (tree != NULL) |
6042 | 2.18k | xmlFreeEnumeration(tree); |
6043 | 760k | break; |
6044 | 760k | } |
6045 | 807k | } |
6046 | 58.1k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6047 | 58.1k | (ctxt->sax->attributeDecl != NULL)) |
6048 | 29.7k | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6049 | 29.7k | type, def, defaultValue, tree); |
6050 | 28.3k | else if (tree != NULL) |
6051 | 18.3k | xmlFreeEnumeration(tree); |
6052 | | |
6053 | 58.1k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6054 | 58.1k | (def != XML_ATTRIBUTE_IMPLIED) && |
6055 | 58.1k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6056 | 55.2k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6057 | 55.2k | } |
6058 | 58.1k | if (ctxt->sax2) { |
6059 | 58.1k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6060 | 58.1k | } |
6061 | 58.1k | if (defaultValue != NULL) |
6062 | 55.2k | xmlFree(defaultValue); |
6063 | 58.1k | GROW; |
6064 | 58.1k | } |
6065 | 876k | if (RAW == '>') { |
6066 | 20.5k | if (inputid != ctxt->input->id) { |
6067 | 621 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6068 | 621 | "Attribute list declaration doesn't start and" |
6069 | 621 | " stop in the same entity\n"); |
6070 | 621 | } |
6071 | 20.5k | NEXT; |
6072 | 20.5k | } |
6073 | 876k | } |
6074 | 888k | } |
6075 | | |
6076 | | /** |
6077 | | * xmlParseElementMixedContentDecl: |
6078 | | * @ctxt: an XML parser context |
6079 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6080 | | * |
6081 | | * parse the declaration for a Mixed Element content |
6082 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6083 | | * |
6084 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6085 | | * '(' S? '#PCDATA' S? ')' |
6086 | | * |
6087 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6088 | | * |
6089 | | * [ VC: No Duplicate Types ] |
6090 | | * The same name must not appear more than once in a single |
6091 | | * mixed-content declaration. |
6092 | | * |
6093 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6094 | | */ |
6095 | | xmlElementContentPtr |
6096 | 21.3k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6097 | 21.3k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6098 | 21.3k | const xmlChar *elem = NULL; |
6099 | | |
6100 | 21.3k | GROW; |
6101 | 21.3k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6102 | 21.3k | SKIP(7); |
6103 | 21.3k | SKIP_BLANKS; |
6104 | 21.3k | SHRINK; |
6105 | 21.3k | if (RAW == ')') { |
6106 | 486 | if (ctxt->input->id != inputchk) { |
6107 | 280 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6108 | 280 | "Element content declaration doesn't start and" |
6109 | 280 | " stop in the same entity\n"); |
6110 | 280 | } |
6111 | 486 | NEXT; |
6112 | 486 | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6113 | 486 | if (ret == NULL) |
6114 | 0 | return(NULL); |
6115 | 486 | if (RAW == '*') { |
6116 | 68 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6117 | 68 | NEXT; |
6118 | 68 | } |
6119 | 486 | return(ret); |
6120 | 486 | } |
6121 | 20.8k | if ((RAW == '(') || (RAW == '|')) { |
6122 | 18.4k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6123 | 18.4k | if (ret == NULL) return(NULL); |
6124 | 18.4k | } |
6125 | 47.1k | while ((RAW == '|') && (ctxt->instate != XML_PARSER_EOF)) { |
6126 | 29.8k | NEXT; |
6127 | 29.8k | if (elem == NULL) { |
6128 | 17.6k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6129 | 17.6k | if (ret == NULL) { |
6130 | 0 | xmlFreeDocElementContent(ctxt->myDoc, cur); |
6131 | 0 | return(NULL); |
6132 | 0 | } |
6133 | 17.6k | ret->c1 = cur; |
6134 | 17.6k | if (cur != NULL) |
6135 | 17.6k | cur->parent = ret; |
6136 | 17.6k | cur = ret; |
6137 | 17.6k | } else { |
6138 | 12.1k | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6139 | 12.1k | if (n == NULL) { |
6140 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6141 | 0 | return(NULL); |
6142 | 0 | } |
6143 | 12.1k | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6144 | 12.1k | if (n->c1 != NULL) |
6145 | 12.1k | n->c1->parent = n; |
6146 | 12.1k | cur->c2 = n; |
6147 | 12.1k | if (n != NULL) |
6148 | 12.1k | n->parent = cur; |
6149 | 12.1k | cur = n; |
6150 | 12.1k | } |
6151 | 29.8k | SKIP_BLANKS; |
6152 | 29.8k | elem = xmlParseName(ctxt); |
6153 | 29.8k | if (elem == NULL) { |
6154 | 3.51k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6155 | 3.51k | "xmlParseElementMixedContentDecl : Name expected\n"); |
6156 | 3.51k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6157 | 3.51k | return(NULL); |
6158 | 3.51k | } |
6159 | 26.3k | SKIP_BLANKS; |
6160 | 26.3k | GROW; |
6161 | 26.3k | } |
6162 | 17.3k | if ((RAW == ')') && (NXT(1) == '*')) { |
6163 | 57 | if (elem != NULL) { |
6164 | 57 | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6165 | 57 | XML_ELEMENT_CONTENT_ELEMENT); |
6166 | 57 | if (cur->c2 != NULL) |
6167 | 57 | cur->c2->parent = cur; |
6168 | 57 | } |
6169 | 57 | if (ret != NULL) |
6170 | 57 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6171 | 57 | if (ctxt->input->id != inputchk) { |
6172 | 35 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6173 | 35 | "Element content declaration doesn't start and" |
6174 | 35 | " stop in the same entity\n"); |
6175 | 35 | } |
6176 | 57 | SKIP(2); |
6177 | 17.2k | } else { |
6178 | 17.2k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6179 | 17.2k | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6180 | 17.2k | return(NULL); |
6181 | 17.2k | } |
6182 | | |
6183 | 17.3k | } else { |
6184 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6185 | 0 | } |
6186 | 57 | return(ret); |
6187 | 21.3k | } |
6188 | | |
6189 | | /** |
6190 | | * xmlParseElementChildrenContentDeclPriv: |
6191 | | * @ctxt: an XML parser context |
6192 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6193 | | * @depth: the level of recursion |
6194 | | * |
6195 | | * parse the declaration for a Mixed Element content |
6196 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6197 | | * |
6198 | | * |
6199 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6200 | | * |
6201 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6202 | | * |
6203 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6204 | | * |
6205 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6206 | | * |
6207 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6208 | | * TODO Parameter-entity replacement text must be properly nested |
6209 | | * with parenthesized groups. That is to say, if either of the |
6210 | | * opening or closing parentheses in a choice, seq, or Mixed |
6211 | | * construct is contained in the replacement text for a parameter |
6212 | | * entity, both must be contained in the same replacement text. For |
6213 | | * interoperability, if a parameter-entity reference appears in a |
6214 | | * choice, seq, or Mixed construct, its replacement text should not |
6215 | | * be empty, and neither the first nor last non-blank character of |
6216 | | * the replacement text should be a connector (| or ,). |
6217 | | * |
6218 | | * Returns the tree of xmlElementContentPtr describing the element |
6219 | | * hierarchy. |
6220 | | */ |
6221 | | static xmlElementContentPtr |
6222 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6223 | 8.89M | int depth) { |
6224 | 8.89M | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6225 | 8.89M | const xmlChar *elem; |
6226 | 8.89M | xmlChar type = 0; |
6227 | | |
6228 | 8.89M | if (((depth > 128) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
6229 | 8.89M | (depth > 2048)) { |
6230 | 787 | xmlFatalErrMsgInt(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, |
6231 | 787 | "xmlParseElementChildrenContentDecl : depth %d too deep, use XML_PARSE_HUGE\n", |
6232 | 787 | depth); |
6233 | 787 | return(NULL); |
6234 | 787 | } |
6235 | 8.89M | SKIP_BLANKS; |
6236 | 8.89M | GROW; |
6237 | 8.89M | if (RAW == '(') { |
6238 | 7.94M | int inputid = ctxt->input->id; |
6239 | | |
6240 | | /* Recurse on first child */ |
6241 | 7.94M | NEXT; |
6242 | 7.94M | SKIP_BLANKS; |
6243 | 7.94M | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6244 | 7.94M | depth + 1); |
6245 | 7.94M | if (cur == NULL) |
6246 | 7.90M | return(NULL); |
6247 | 43.0k | SKIP_BLANKS; |
6248 | 43.0k | GROW; |
6249 | 954k | } else { |
6250 | 954k | elem = xmlParseName(ctxt); |
6251 | 954k | if (elem == NULL) { |
6252 | 13.3k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6253 | 13.3k | return(NULL); |
6254 | 13.3k | } |
6255 | 941k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6256 | 941k | if (cur == NULL) { |
6257 | 0 | xmlErrMemory(ctxt, NULL); |
6258 | 0 | return(NULL); |
6259 | 0 | } |
6260 | 941k | GROW; |
6261 | 941k | if (RAW == '?') { |
6262 | 4.97k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6263 | 4.97k | NEXT; |
6264 | 936k | } else if (RAW == '*') { |
6265 | 4.13k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6266 | 4.13k | NEXT; |
6267 | 932k | } else if (RAW == '+') { |
6268 | 5.19k | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6269 | 5.19k | NEXT; |
6270 | 927k | } else { |
6271 | 927k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6272 | 927k | } |
6273 | 941k | GROW; |
6274 | 941k | } |
6275 | 984k | SKIP_BLANKS; |
6276 | 984k | SHRINK; |
6277 | 1.00M | while ((RAW != ')') && (ctxt->instate != XML_PARSER_EOF)) { |
6278 | | /* |
6279 | | * Each loop we parse one separator and one element. |
6280 | | */ |
6281 | 944k | if (RAW == ',') { |
6282 | 12.1k | if (type == 0) type = CUR; |
6283 | | |
6284 | | /* |
6285 | | * Detect "Name | Name , Name" error |
6286 | | */ |
6287 | 3.35k | else if (type != CUR) { |
6288 | 2.09k | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6289 | 2.09k | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6290 | 2.09k | type); |
6291 | 2.09k | if ((last != NULL) && (last != ret)) |
6292 | 2.09k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6293 | 2.09k | if (ret != NULL) |
6294 | 2.09k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6295 | 2.09k | return(NULL); |
6296 | 2.09k | } |
6297 | 10.0k | NEXT; |
6298 | | |
6299 | 10.0k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6300 | 10.0k | if (op == NULL) { |
6301 | 0 | if ((last != NULL) && (last != ret)) |
6302 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6303 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6304 | 0 | return(NULL); |
6305 | 0 | } |
6306 | 10.0k | if (last == NULL) { |
6307 | 8.75k | op->c1 = ret; |
6308 | 8.75k | if (ret != NULL) |
6309 | 8.75k | ret->parent = op; |
6310 | 8.75k | ret = cur = op; |
6311 | 8.75k | } else { |
6312 | 1.25k | cur->c2 = op; |
6313 | 1.25k | if (op != NULL) |
6314 | 1.25k | op->parent = cur; |
6315 | 1.25k | op->c1 = last; |
6316 | 1.25k | if (last != NULL) |
6317 | 1.25k | last->parent = op; |
6318 | 1.25k | cur =op; |
6319 | 1.25k | last = NULL; |
6320 | 1.25k | } |
6321 | 932k | } else if (RAW == '|') { |
6322 | 20.9k | if (type == 0) type = CUR; |
6323 | | |
6324 | | /* |
6325 | | * Detect "Name , Name | Name" error |
6326 | | */ |
6327 | 3.61k | else if (type != CUR) { |
6328 | 786 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6329 | 786 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6330 | 786 | type); |
6331 | 786 | if ((last != NULL) && (last != ret)) |
6332 | 786 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6333 | 786 | if (ret != NULL) |
6334 | 786 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6335 | 786 | return(NULL); |
6336 | 786 | } |
6337 | 20.1k | NEXT; |
6338 | | |
6339 | 20.1k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6340 | 20.1k | if (op == NULL) { |
6341 | 0 | if ((last != NULL) && (last != ret)) |
6342 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6343 | 0 | if (ret != NULL) |
6344 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6345 | 0 | return(NULL); |
6346 | 0 | } |
6347 | 20.1k | if (last == NULL) { |
6348 | 17.3k | op->c1 = ret; |
6349 | 17.3k | if (ret != NULL) |
6350 | 17.3k | ret->parent = op; |
6351 | 17.3k | ret = cur = op; |
6352 | 17.3k | } else { |
6353 | 2.82k | cur->c2 = op; |
6354 | 2.82k | if (op != NULL) |
6355 | 2.82k | op->parent = cur; |
6356 | 2.82k | op->c1 = last; |
6357 | 2.82k | if (last != NULL) |
6358 | 2.82k | last->parent = op; |
6359 | 2.82k | cur =op; |
6360 | 2.82k | last = NULL; |
6361 | 2.82k | } |
6362 | 911k | } else { |
6363 | 911k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6364 | 911k | if ((last != NULL) && (last != ret)) |
6365 | 5.38k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6366 | 911k | if (ret != NULL) |
6367 | 911k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6368 | 911k | return(NULL); |
6369 | 911k | } |
6370 | 30.1k | GROW; |
6371 | 30.1k | SKIP_BLANKS; |
6372 | 30.1k | GROW; |
6373 | 30.1k | if (RAW == '(') { |
6374 | 14.1k | int inputid = ctxt->input->id; |
6375 | | /* Recurse on second child */ |
6376 | 14.1k | NEXT; |
6377 | 14.1k | SKIP_BLANKS; |
6378 | 14.1k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6379 | 14.1k | depth + 1); |
6380 | 14.1k | if (last == NULL) { |
6381 | 11.1k | if (ret != NULL) |
6382 | 11.1k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6383 | 11.1k | return(NULL); |
6384 | 11.1k | } |
6385 | 2.99k | SKIP_BLANKS; |
6386 | 15.9k | } else { |
6387 | 15.9k | elem = xmlParseName(ctxt); |
6388 | 15.9k | if (elem == NULL) { |
6389 | 2.53k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6390 | 2.53k | if (ret != NULL) |
6391 | 2.53k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6392 | 2.53k | return(NULL); |
6393 | 2.53k | } |
6394 | 13.4k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6395 | 13.4k | if (last == NULL) { |
6396 | 0 | if (ret != NULL) |
6397 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6398 | 0 | return(NULL); |
6399 | 0 | } |
6400 | 13.4k | if (RAW == '?') { |
6401 | 1.15k | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6402 | 1.15k | NEXT; |
6403 | 12.2k | } else if (RAW == '*') { |
6404 | 1.96k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6405 | 1.96k | NEXT; |
6406 | 10.3k | } else if (RAW == '+') { |
6407 | 756 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6408 | 756 | NEXT; |
6409 | 9.56k | } else { |
6410 | 9.56k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6411 | 9.56k | } |
6412 | 13.4k | } |
6413 | 16.4k | SKIP_BLANKS; |
6414 | 16.4k | GROW; |
6415 | 16.4k | } |
6416 | 56.7k | if ((cur != NULL) && (last != NULL)) { |
6417 | 4.09k | cur->c2 = last; |
6418 | 4.09k | if (last != NULL) |
6419 | 4.09k | last->parent = cur; |
6420 | 4.09k | } |
6421 | 56.7k | if (ctxt->input->id != inputchk) { |
6422 | 466 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6423 | 466 | "Element content declaration doesn't start and stop in" |
6424 | 466 | " the same entity\n"); |
6425 | 466 | } |
6426 | 56.7k | NEXT; |
6427 | 56.7k | if (RAW == '?') { |
6428 | 5.96k | if (ret != NULL) { |
6429 | 5.96k | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6430 | 5.96k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6431 | 4.86k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6432 | 1.09k | else |
6433 | 1.09k | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6434 | 5.96k | } |
6435 | 5.96k | NEXT; |
6436 | 50.8k | } else if (RAW == '*') { |
6437 | 1.87k | if (ret != NULL) { |
6438 | 1.87k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6439 | 1.87k | cur = ret; |
6440 | | /* |
6441 | | * Some normalization: |
6442 | | * (a | b* | c?)* == (a | b | c)* |
6443 | | */ |
6444 | 5.11k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6445 | 3.24k | if ((cur->c1 != NULL) && |
6446 | 3.24k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6447 | 3.24k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6448 | 1.78k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6449 | 3.24k | if ((cur->c2 != NULL) && |
6450 | 3.24k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6451 | 3.24k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6452 | 800 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6453 | 3.24k | cur = cur->c2; |
6454 | 3.24k | } |
6455 | 1.87k | } |
6456 | 1.87k | NEXT; |
6457 | 48.9k | } else if (RAW == '+') { |
6458 | 2.44k | if (ret != NULL) { |
6459 | 2.44k | int found = 0; |
6460 | | |
6461 | 2.44k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6462 | 2.44k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6463 | 581 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6464 | 1.86k | else |
6465 | 1.86k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6466 | | /* |
6467 | | * Some normalization: |
6468 | | * (a | b*)+ == (a | b)* |
6469 | | * (a | b?)+ == (a | b)* |
6470 | | */ |
6471 | 4.55k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6472 | 2.10k | if ((cur->c1 != NULL) && |
6473 | 2.10k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6474 | 2.10k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6475 | 1.68k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6476 | 1.68k | found = 1; |
6477 | 1.68k | } |
6478 | 2.10k | if ((cur->c2 != NULL) && |
6479 | 2.10k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6480 | 2.10k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6481 | 1.42k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6482 | 1.42k | found = 1; |
6483 | 1.42k | } |
6484 | 2.10k | cur = cur->c2; |
6485 | 2.10k | } |
6486 | 2.44k | if (found) |
6487 | 1.42k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6488 | 2.44k | } |
6489 | 2.44k | NEXT; |
6490 | 2.44k | } |
6491 | 56.7k | return(ret); |
6492 | 984k | } |
6493 | | |
6494 | | /** |
6495 | | * xmlParseElementChildrenContentDecl: |
6496 | | * @ctxt: an XML parser context |
6497 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6498 | | * |
6499 | | * parse the declaration for a Mixed Element content |
6500 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6501 | | * |
6502 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6503 | | * |
6504 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6505 | | * |
6506 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6507 | | * |
6508 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6509 | | * |
6510 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6511 | | * TODO Parameter-entity replacement text must be properly nested |
6512 | | * with parenthesized groups. That is to say, if either of the |
6513 | | * opening or closing parentheses in a choice, seq, or Mixed |
6514 | | * construct is contained in the replacement text for a parameter |
6515 | | * entity, both must be contained in the same replacement text. For |
6516 | | * interoperability, if a parameter-entity reference appears in a |
6517 | | * choice, seq, or Mixed construct, its replacement text should not |
6518 | | * be empty, and neither the first nor last non-blank character of |
6519 | | * the replacement text should be a connector (| or ,). |
6520 | | * |
6521 | | * Returns the tree of xmlElementContentPtr describing the element |
6522 | | * hierarchy. |
6523 | | */ |
6524 | | xmlElementContentPtr |
6525 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6526 | | /* stub left for API/ABI compat */ |
6527 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6528 | 0 | } |
6529 | | |
6530 | | /** |
6531 | | * xmlParseElementContentDecl: |
6532 | | * @ctxt: an XML parser context |
6533 | | * @name: the name of the element being defined. |
6534 | | * @result: the Element Content pointer will be stored here if any |
6535 | | * |
6536 | | * parse the declaration for an Element content either Mixed or Children, |
6537 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6538 | | * |
6539 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6540 | | * |
6541 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6542 | | */ |
6543 | | |
6544 | | int |
6545 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6546 | 962k | xmlElementContentPtr *result) { |
6547 | | |
6548 | 962k | xmlElementContentPtr tree = NULL; |
6549 | 962k | int inputid = ctxt->input->id; |
6550 | 962k | int res; |
6551 | | |
6552 | 962k | *result = NULL; |
6553 | | |
6554 | 962k | if (RAW != '(') { |
6555 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6556 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6557 | 0 | return(-1); |
6558 | 0 | } |
6559 | 962k | NEXT; |
6560 | 962k | GROW; |
6561 | 962k | if (ctxt->instate == XML_PARSER_EOF) |
6562 | 0 | return(-1); |
6563 | 962k | SKIP_BLANKS; |
6564 | 962k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6565 | 21.3k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6566 | 21.3k | res = XML_ELEMENT_TYPE_MIXED; |
6567 | 941k | } else { |
6568 | 941k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6569 | 941k | res = XML_ELEMENT_TYPE_ELEMENT; |
6570 | 941k | } |
6571 | 962k | SKIP_BLANKS; |
6572 | 962k | *result = tree; |
6573 | 962k | return(res); |
6574 | 962k | } |
6575 | | |
6576 | | /** |
6577 | | * xmlParseElementDecl: |
6578 | | * @ctxt: an XML parser context |
6579 | | * |
6580 | | * parse an Element declaration. |
6581 | | * |
6582 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6583 | | * |
6584 | | * [ VC: Unique Element Type Declaration ] |
6585 | | * No element type may be declared more than once |
6586 | | * |
6587 | | * Returns the type of the element, or -1 in case of error |
6588 | | */ |
6589 | | int |
6590 | 1.13M | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6591 | 1.13M | const xmlChar *name; |
6592 | 1.13M | int ret = -1; |
6593 | 1.13M | xmlElementContentPtr content = NULL; |
6594 | | |
6595 | | /* GROW; done in the caller */ |
6596 | 1.13M | if (CMP9(CUR_PTR, '<', '!', 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6597 | 1.12M | int inputid = ctxt->input->id; |
6598 | | |
6599 | 1.12M | SKIP(9); |
6600 | 1.12M | if (SKIP_BLANKS == 0) { |
6601 | 3.14k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6602 | 3.14k | "Space required after 'ELEMENT'\n"); |
6603 | 3.14k | return(-1); |
6604 | 3.14k | } |
6605 | 1.11M | name = xmlParseName(ctxt); |
6606 | 1.11M | if (name == NULL) { |
6607 | 42.7k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6608 | 42.7k | "xmlParseElementDecl: no name for Element\n"); |
6609 | 42.7k | return(-1); |
6610 | 42.7k | } |
6611 | 1.07M | if (SKIP_BLANKS == 0) { |
6612 | 1.03M | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6613 | 1.03M | "Space required after the element name\n"); |
6614 | 1.03M | } |
6615 | 1.07M | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6616 | 1.62k | SKIP(5); |
6617 | | /* |
6618 | | * Element must always be empty. |
6619 | | */ |
6620 | 1.62k | ret = XML_ELEMENT_TYPE_EMPTY; |
6621 | 1.07M | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6622 | 1.07M | (NXT(2) == 'Y')) { |
6623 | 1.01k | SKIP(3); |
6624 | | /* |
6625 | | * Element is a generic container. |
6626 | | */ |
6627 | 1.01k | ret = XML_ELEMENT_TYPE_ANY; |
6628 | 1.07M | } else if (RAW == '(') { |
6629 | 962k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6630 | 962k | } else { |
6631 | | /* |
6632 | | * [ WFC: PEs in Internal Subset ] error handling. |
6633 | | */ |
6634 | 108k | if ((RAW == '%') && (ctxt->external == 0) && |
6635 | 108k | (ctxt->inputNr == 1)) { |
6636 | 93 | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_IN_INT_SUBSET, |
6637 | 93 | "PEReference: forbidden within markup decl in internal subset\n"); |
6638 | 108k | } else { |
6639 | 108k | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6640 | 108k | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6641 | 108k | } |
6642 | 108k | return(-1); |
6643 | 108k | } |
6644 | | |
6645 | 965k | SKIP_BLANKS; |
6646 | | |
6647 | 965k | if (RAW != '>') { |
6648 | 945k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6649 | 945k | if (content != NULL) { |
6650 | 5.07k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6651 | 5.07k | } |
6652 | 945k | } else { |
6653 | 20.1k | if (inputid != ctxt->input->id) { |
6654 | 1.34k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6655 | 1.34k | "Element declaration doesn't start and stop in" |
6656 | 1.34k | " the same entity\n"); |
6657 | 1.34k | } |
6658 | | |
6659 | 20.1k | NEXT; |
6660 | 20.1k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6661 | 20.1k | (ctxt->sax->elementDecl != NULL)) { |
6662 | 2.25k | if (content != NULL) |
6663 | 1.31k | content->parent = NULL; |
6664 | 2.25k | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6665 | 2.25k | content); |
6666 | 2.25k | if ((content != NULL) && (content->parent == NULL)) { |
6667 | | /* |
6668 | | * this is a trick: if xmlAddElementDecl is called, |
6669 | | * instead of copying the full tree it is plugged directly |
6670 | | * if called from the parser. Avoid duplicating the |
6671 | | * interfaces or change the API/ABI |
6672 | | */ |
6673 | 1.02k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6674 | 1.02k | } |
6675 | 17.9k | } else if (content != NULL) { |
6676 | 4.90k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6677 | 4.90k | } |
6678 | 20.1k | } |
6679 | 965k | } |
6680 | 975k | return(ret); |
6681 | 1.13M | } |
6682 | | |
6683 | | /** |
6684 | | * xmlParseConditionalSections |
6685 | | * @ctxt: an XML parser context |
6686 | | * |
6687 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6688 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6689 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6690 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6691 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6692 | | */ |
6693 | | |
6694 | | static void |
6695 | 0 | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6696 | 0 | int *inputIds = NULL; |
6697 | 0 | size_t inputIdsSize = 0; |
6698 | 0 | size_t depth = 0; |
6699 | |
|
6700 | 0 | while (ctxt->instate != XML_PARSER_EOF) { |
6701 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6702 | 0 | int id = ctxt->input->id; |
6703 | |
|
6704 | 0 | SKIP(3); |
6705 | 0 | SKIP_BLANKS; |
6706 | |
|
6707 | 0 | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6708 | 0 | SKIP(7); |
6709 | 0 | SKIP_BLANKS; |
6710 | 0 | if (RAW != '[') { |
6711 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6712 | 0 | xmlHaltParser(ctxt); |
6713 | 0 | goto error; |
6714 | 0 | } |
6715 | 0 | if (ctxt->input->id != id) { |
6716 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6717 | 0 | "All markup of the conditional section is" |
6718 | 0 | " not in the same entity\n"); |
6719 | 0 | } |
6720 | 0 | NEXT; |
6721 | |
|
6722 | 0 | if (inputIdsSize <= depth) { |
6723 | 0 | int *tmp; |
6724 | |
|
6725 | 0 | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
6726 | 0 | tmp = (int *) xmlRealloc(inputIds, |
6727 | 0 | inputIdsSize * sizeof(int)); |
6728 | 0 | if (tmp == NULL) { |
6729 | 0 | xmlErrMemory(ctxt, NULL); |
6730 | 0 | goto error; |
6731 | 0 | } |
6732 | 0 | inputIds = tmp; |
6733 | 0 | } |
6734 | 0 | inputIds[depth] = id; |
6735 | 0 | depth++; |
6736 | 0 | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6737 | 0 | int state; |
6738 | 0 | xmlParserInputState instate; |
6739 | 0 | size_t ignoreDepth = 0; |
6740 | |
|
6741 | 0 | SKIP(6); |
6742 | 0 | SKIP_BLANKS; |
6743 | 0 | if (RAW != '[') { |
6744 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6745 | 0 | xmlHaltParser(ctxt); |
6746 | 0 | goto error; |
6747 | 0 | } |
6748 | 0 | if (ctxt->input->id != id) { |
6749 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6750 | 0 | "All markup of the conditional section is" |
6751 | 0 | " not in the same entity\n"); |
6752 | 0 | } |
6753 | 0 | NEXT; |
6754 | | |
6755 | | /* |
6756 | | * Parse up to the end of the conditional section but disable |
6757 | | * SAX event generating DTD building in the meantime |
6758 | | */ |
6759 | 0 | state = ctxt->disableSAX; |
6760 | 0 | instate = ctxt->instate; |
6761 | 0 | if (ctxt->recovery == 0) ctxt->disableSAX = 1; |
6762 | 0 | ctxt->instate = XML_PARSER_IGNORE; |
6763 | |
|
6764 | 0 | while (RAW != 0) { |
6765 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6766 | 0 | SKIP(3); |
6767 | 0 | ignoreDepth++; |
6768 | | /* Check for integer overflow */ |
6769 | 0 | if (ignoreDepth == 0) { |
6770 | 0 | xmlErrMemory(ctxt, NULL); |
6771 | 0 | goto error; |
6772 | 0 | } |
6773 | 0 | } else if ((RAW == ']') && (NXT(1) == ']') && |
6774 | 0 | (NXT(2) == '>')) { |
6775 | 0 | if (ignoreDepth == 0) |
6776 | 0 | break; |
6777 | 0 | SKIP(3); |
6778 | 0 | ignoreDepth--; |
6779 | 0 | } else { |
6780 | 0 | NEXT; |
6781 | 0 | } |
6782 | 0 | } |
6783 | | |
6784 | 0 | ctxt->disableSAX = state; |
6785 | 0 | ctxt->instate = instate; |
6786 | |
|
6787 | 0 | if (RAW == 0) { |
6788 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6789 | 0 | goto error; |
6790 | 0 | } |
6791 | 0 | if (ctxt->input->id != id) { |
6792 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6793 | 0 | "All markup of the conditional section is" |
6794 | 0 | " not in the same entity\n"); |
6795 | 0 | } |
6796 | 0 | SKIP(3); |
6797 | 0 | } else { |
6798 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6799 | 0 | xmlHaltParser(ctxt); |
6800 | 0 | goto error; |
6801 | 0 | } |
6802 | 0 | } else if ((depth > 0) && |
6803 | 0 | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6804 | 0 | depth--; |
6805 | 0 | if (ctxt->input->id != inputIds[depth]) { |
6806 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6807 | 0 | "All markup of the conditional section is not" |
6808 | 0 | " in the same entity\n"); |
6809 | 0 | } |
6810 | 0 | SKIP(3); |
6811 | 0 | } else { |
6812 | 0 | int id = ctxt->input->id; |
6813 | 0 | unsigned long cons = CUR_CONSUMED; |
6814 | |
|
6815 | 0 | xmlParseMarkupDecl(ctxt); |
6816 | |
|
6817 | 0 | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
6818 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6819 | 0 | xmlHaltParser(ctxt); |
6820 | 0 | goto error; |
6821 | 0 | } |
6822 | 0 | } |
6823 | | |
6824 | 0 | if (depth == 0) |
6825 | 0 | break; |
6826 | | |
6827 | 0 | SKIP_BLANKS; |
6828 | 0 | GROW; |
6829 | 0 | } |
6830 | | |
6831 | 0 | error: |
6832 | 0 | xmlFree(inputIds); |
6833 | 0 | } |
6834 | | |
6835 | | /** |
6836 | | * xmlParseMarkupDecl: |
6837 | | * @ctxt: an XML parser context |
6838 | | * |
6839 | | * parse Markup declarations |
6840 | | * |
6841 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6842 | | * NotationDecl | PI | Comment |
6843 | | * |
6844 | | * [ VC: Proper Declaration/PE Nesting ] |
6845 | | * Parameter-entity replacement text must be properly nested with |
6846 | | * markup declarations. That is to say, if either the first character |
6847 | | * or the last character of a markup declaration (markupdecl above) is |
6848 | | * contained in the replacement text for a parameter-entity reference, |
6849 | | * both must be contained in the same replacement text. |
6850 | | * |
6851 | | * [ WFC: PEs in Internal Subset ] |
6852 | | * In the internal DTD subset, parameter-entity references can occur |
6853 | | * only where markup declarations can occur, not within markup declarations. |
6854 | | * (This does not apply to references that occur in external parameter |
6855 | | * entities or to the external subset.) |
6856 | | */ |
6857 | | void |
6858 | 8.46M | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
6859 | 8.46M | GROW; |
6860 | 8.46M | if (CUR == '<') { |
6861 | 7.34M | if (NXT(1) == '!') { |
6862 | 6.48M | switch (NXT(2)) { |
6863 | 5.35M | case 'E': |
6864 | 5.35M | if (NXT(3) == 'L') |
6865 | 1.13M | xmlParseElementDecl(ctxt); |
6866 | 4.22M | else if (NXT(3) == 'N') |
6867 | 4.17M | xmlParseEntityDecl(ctxt); |
6868 | 5.35M | break; |
6869 | 888k | case 'A': |
6870 | 888k | xmlParseAttributeListDecl(ctxt); |
6871 | 888k | break; |
6872 | 193k | case 'N': |
6873 | 193k | xmlParseNotationDecl(ctxt); |
6874 | 193k | break; |
6875 | 46.9k | case '-': |
6876 | 46.9k | xmlParseComment(ctxt); |
6877 | 46.9k | break; |
6878 | 4.04k | default: |
6879 | | /* there is an error but it will be detected later */ |
6880 | 4.04k | break; |
6881 | 6.48M | } |
6882 | 6.48M | } else if (NXT(1) == '?') { |
6883 | 829k | xmlParsePI(ctxt); |
6884 | 829k | } |
6885 | 7.34M | } |
6886 | | |
6887 | | /* |
6888 | | * detect requirement to exit there and act accordingly |
6889 | | * and avoid having instate overridden later on |
6890 | | */ |
6891 | 8.46M | if (ctxt->instate == XML_PARSER_EOF) |
6892 | 984 | return; |
6893 | | |
6894 | 8.46M | ctxt->instate = XML_PARSER_DTD; |
6895 | 8.46M | } |
6896 | | |
6897 | | /** |
6898 | | * xmlParseTextDecl: |
6899 | | * @ctxt: an XML parser context |
6900 | | * |
6901 | | * parse an XML declaration header for external entities |
6902 | | * |
6903 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
6904 | | */ |
6905 | | |
6906 | | void |
6907 | 0 | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
6908 | 0 | xmlChar *version; |
6909 | 0 | const xmlChar *encoding; |
6910 | 0 | int oldstate; |
6911 | | |
6912 | | /* |
6913 | | * We know that '<?xml' is here. |
6914 | | */ |
6915 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
6916 | 0 | SKIP(5); |
6917 | 0 | } else { |
6918 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
6919 | 0 | return; |
6920 | 0 | } |
6921 | | |
6922 | | /* Avoid expansion of parameter entities when skipping blanks. */ |
6923 | 0 | oldstate = ctxt->instate; |
6924 | 0 | ctxt->instate = XML_PARSER_START; |
6925 | |
|
6926 | 0 | if (SKIP_BLANKS == 0) { |
6927 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6928 | 0 | "Space needed after '<?xml'\n"); |
6929 | 0 | } |
6930 | | |
6931 | | /* |
6932 | | * We may have the VersionInfo here. |
6933 | | */ |
6934 | 0 | version = xmlParseVersionInfo(ctxt); |
6935 | 0 | if (version == NULL) |
6936 | 0 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
6937 | 0 | else { |
6938 | 0 | if (SKIP_BLANKS == 0) { |
6939 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6940 | 0 | "Space needed here\n"); |
6941 | 0 | } |
6942 | 0 | } |
6943 | 0 | ctxt->input->version = version; |
6944 | | |
6945 | | /* |
6946 | | * We must have the encoding declaration |
6947 | | */ |
6948 | 0 | encoding = xmlParseEncodingDecl(ctxt); |
6949 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
6950 | | /* |
6951 | | * The XML REC instructs us to stop parsing right here |
6952 | | */ |
6953 | 0 | ctxt->instate = oldstate; |
6954 | 0 | return; |
6955 | 0 | } |
6956 | 0 | if ((encoding == NULL) && (ctxt->errNo == XML_ERR_OK)) { |
6957 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_MISSING_ENCODING, |
6958 | 0 | "Missing encoding in text declaration\n"); |
6959 | 0 | } |
6960 | |
|
6961 | 0 | SKIP_BLANKS; |
6962 | 0 | if ((RAW == '?') && (NXT(1) == '>')) { |
6963 | 0 | SKIP(2); |
6964 | 0 | } else if (RAW == '>') { |
6965 | | /* Deprecated old WD ... */ |
6966 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
6967 | 0 | NEXT; |
6968 | 0 | } else { |
6969 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
6970 | 0 | MOVETO_ENDTAG(CUR_PTR); |
6971 | 0 | NEXT; |
6972 | 0 | } |
6973 | |
|
6974 | 0 | ctxt->instate = oldstate; |
6975 | 0 | } |
6976 | | |
6977 | | /** |
6978 | | * xmlParseExternalSubset: |
6979 | | * @ctxt: an XML parser context |
6980 | | * @ExternalID: the external identifier |
6981 | | * @SystemID: the system identifier (or URL) |
6982 | | * |
6983 | | * parse Markup declarations from an external subset |
6984 | | * |
6985 | | * [30] extSubset ::= textDecl? extSubsetDecl |
6986 | | * |
6987 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
6988 | | */ |
6989 | | void |
6990 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
6991 | 0 | const xmlChar *SystemID) { |
6992 | 0 | xmlDetectSAX2(ctxt); |
6993 | 0 | GROW; |
6994 | |
|
6995 | 0 | if ((ctxt->encoding == NULL) && |
6996 | 0 | (ctxt->input->end - ctxt->input->cur >= 4)) { |
6997 | 0 | xmlChar start[4]; |
6998 | 0 | xmlCharEncoding enc; |
6999 | |
|
7000 | 0 | start[0] = RAW; |
7001 | 0 | start[1] = NXT(1); |
7002 | 0 | start[2] = NXT(2); |
7003 | 0 | start[3] = NXT(3); |
7004 | 0 | enc = xmlDetectCharEncoding(start, 4); |
7005 | 0 | if (enc != XML_CHAR_ENCODING_NONE) |
7006 | 0 | xmlSwitchEncoding(ctxt, enc); |
7007 | 0 | } |
7008 | |
|
7009 | 0 | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7010 | 0 | xmlParseTextDecl(ctxt); |
7011 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7012 | | /* |
7013 | | * The XML REC instructs us to stop parsing right here |
7014 | | */ |
7015 | 0 | xmlHaltParser(ctxt); |
7016 | 0 | return; |
7017 | 0 | } |
7018 | 0 | } |
7019 | 0 | if (ctxt->myDoc == NULL) { |
7020 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7021 | 0 | if (ctxt->myDoc == NULL) { |
7022 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
7023 | 0 | return; |
7024 | 0 | } |
7025 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7026 | 0 | } |
7027 | 0 | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL)) |
7028 | 0 | xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID); |
7029 | |
|
7030 | 0 | ctxt->instate = XML_PARSER_DTD; |
7031 | 0 | ctxt->external = 1; |
7032 | 0 | SKIP_BLANKS; |
7033 | 0 | while (((RAW == '<') && (NXT(1) == '?')) || |
7034 | 0 | ((RAW == '<') && (NXT(1) == '!')) || |
7035 | 0 | (RAW == '%')) { |
7036 | 0 | int id = ctxt->input->id; |
7037 | 0 | unsigned long cons = CUR_CONSUMED; |
7038 | |
|
7039 | 0 | GROW; |
7040 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7041 | 0 | xmlParseConditionalSections(ctxt); |
7042 | 0 | } else |
7043 | 0 | xmlParseMarkupDecl(ctxt); |
7044 | 0 | SKIP_BLANKS; |
7045 | |
|
7046 | 0 | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
7047 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7048 | 0 | break; |
7049 | 0 | } |
7050 | 0 | } |
7051 | |
|
7052 | 0 | if (RAW != 0) { |
7053 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7054 | 0 | } |
7055 | |
|
7056 | 0 | } |
7057 | | |
7058 | | /** |
7059 | | * xmlParseReference: |
7060 | | * @ctxt: an XML parser context |
7061 | | * |
7062 | | * parse and handle entity references in content, depending on the SAX |
7063 | | * interface, this may end-up in a call to character() if this is a |
7064 | | * CharRef, a predefined entity, if there is no reference() callback. |
7065 | | * or if the parser was asked to switch to that mode. |
7066 | | * |
7067 | | * [67] Reference ::= EntityRef | CharRef |
7068 | | */ |
7069 | | void |
7070 | 534k | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7071 | 534k | xmlEntityPtr ent; |
7072 | 534k | xmlChar *val; |
7073 | 534k | int was_checked; |
7074 | 534k | xmlNodePtr list = NULL; |
7075 | 534k | xmlParserErrors ret = XML_ERR_OK; |
7076 | | |
7077 | | |
7078 | 534k | if (RAW != '&') |
7079 | 0 | return; |
7080 | | |
7081 | | /* |
7082 | | * Simple case of a CharRef |
7083 | | */ |
7084 | 534k | if (NXT(1) == '#') { |
7085 | 28.9k | int i = 0; |
7086 | 28.9k | xmlChar out[16]; |
7087 | 28.9k | int hex = NXT(2); |
7088 | 28.9k | int value = xmlParseCharRef(ctxt); |
7089 | | |
7090 | 28.9k | if (value == 0) |
7091 | 12.9k | return; |
7092 | 15.9k | if (ctxt->charset != XML_CHAR_ENCODING_UTF8) { |
7093 | | /* |
7094 | | * So we are using non-UTF-8 buffers |
7095 | | * Check that the char fit on 8bits, if not |
7096 | | * generate a CharRef. |
7097 | | */ |
7098 | 8.61k | if (value <= 0xFF) { |
7099 | 8.04k | out[0] = value; |
7100 | 8.04k | out[1] = 0; |
7101 | 8.04k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7102 | 8.04k | (!ctxt->disableSAX)) |
7103 | 0 | ctxt->sax->characters(ctxt->userData, out, 1); |
7104 | 8.04k | } else { |
7105 | 567 | if ((hex == 'x') || (hex == 'X')) |
7106 | 224 | snprintf((char *)out, sizeof(out), "#x%X", value); |
7107 | 343 | else |
7108 | 343 | snprintf((char *)out, sizeof(out), "#%d", value); |
7109 | 567 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7110 | 567 | (!ctxt->disableSAX)) |
7111 | 0 | ctxt->sax->reference(ctxt->userData, out); |
7112 | 567 | } |
7113 | 8.61k | } else { |
7114 | | /* |
7115 | | * Just encode the value in UTF-8 |
7116 | | */ |
7117 | 7.36k | COPY_BUF(0 ,out, i, value); |
7118 | 7.36k | out[i] = 0; |
7119 | 7.36k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7120 | 7.36k | (!ctxt->disableSAX)) |
7121 | 2.97k | ctxt->sax->characters(ctxt->userData, out, i); |
7122 | 7.36k | } |
7123 | 15.9k | return; |
7124 | 28.9k | } |
7125 | | |
7126 | | /* |
7127 | | * We are seeing an entity reference |
7128 | | */ |
7129 | 505k | ent = xmlParseEntityRef(ctxt); |
7130 | 505k | if (ent == NULL) return; |
7131 | 481k | if (!ctxt->wellFormed) |
7132 | 29.5k | return; |
7133 | 451k | was_checked = ent->checked; |
7134 | | |
7135 | | /* special case of predefined entities */ |
7136 | 451k | if ((ent->name == NULL) || |
7137 | 451k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7138 | 4.94k | val = ent->content; |
7139 | 4.94k | if (val == NULL) return; |
7140 | | /* |
7141 | | * inline the entity. |
7142 | | */ |
7143 | 4.94k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7144 | 4.94k | (!ctxt->disableSAX)) |
7145 | 4.94k | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7146 | 4.94k | return; |
7147 | 4.94k | } |
7148 | | |
7149 | | /* |
7150 | | * The first reference to the entity trigger a parsing phase |
7151 | | * where the ent->children is filled with the result from |
7152 | | * the parsing. |
7153 | | * Note: external parsed entities will not be loaded, it is not |
7154 | | * required for a non-validating parser, unless the parsing option |
7155 | | * of validating, or substituting entities were given. Doing so is |
7156 | | * far more secure as the parser will only process data coming from |
7157 | | * the document entity by default. |
7158 | | */ |
7159 | 446k | if (((ent->checked == 0) || |
7160 | 446k | ((ent->children == NULL) && (ctxt->options & XML_PARSE_NOENT))) && |
7161 | 446k | ((ent->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY) || |
7162 | 299k | (ctxt->options & (XML_PARSE_NOENT | XML_PARSE_DTDVALID)))) { |
7163 | 299k | unsigned long oldnbent = ctxt->nbentities, diff; |
7164 | | |
7165 | | /* |
7166 | | * This is a bit hackish but this seems the best |
7167 | | * way to make sure both SAX and DOM entity support |
7168 | | * behaves okay. |
7169 | | */ |
7170 | 299k | void *user_data; |
7171 | 299k | if (ctxt->userData == ctxt) |
7172 | 299k | user_data = NULL; |
7173 | 0 | else |
7174 | 0 | user_data = ctxt->userData; |
7175 | | |
7176 | | /* |
7177 | | * Check that this entity is well formed |
7178 | | * 4.3.2: An internal general parsed entity is well-formed |
7179 | | * if its replacement text matches the production labeled |
7180 | | * content. |
7181 | | */ |
7182 | 299k | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7183 | 299k | ctxt->depth++; |
7184 | 299k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, ent->content, |
7185 | 299k | user_data, &list); |
7186 | 299k | ctxt->depth--; |
7187 | | |
7188 | 299k | } else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7189 | 0 | ctxt->depth++; |
7190 | 0 | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, ctxt->sax, |
7191 | 0 | user_data, ctxt->depth, ent->URI, |
7192 | 0 | ent->ExternalID, &list); |
7193 | 0 | ctxt->depth--; |
7194 | 0 | } else { |
7195 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7196 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7197 | 0 | "invalid entity type found\n", NULL); |
7198 | 0 | } |
7199 | | |
7200 | | /* |
7201 | | * Store the number of entities needing parsing for this entity |
7202 | | * content and do checkings |
7203 | | */ |
7204 | 299k | diff = ctxt->nbentities - oldnbent + 1; |
7205 | 299k | if (diff > INT_MAX / 2) |
7206 | 297 | diff = INT_MAX / 2; |
7207 | 299k | ent->checked = diff * 2; |
7208 | 299k | if ((ent->content != NULL) && (xmlStrchr(ent->content, '<'))) |
7209 | 105k | ent->checked |= 1; |
7210 | 299k | if (ret == XML_ERR_ENTITY_LOOP) { |
7211 | 232k | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7212 | 232k | xmlHaltParser(ctxt); |
7213 | 232k | xmlFreeNodeList(list); |
7214 | 232k | return; |
7215 | 232k | } |
7216 | 66.6k | if (xmlParserEntityCheck(ctxt, 0, ent, 0)) { |
7217 | 0 | xmlFreeNodeList(list); |
7218 | 0 | return; |
7219 | 0 | } |
7220 | | |
7221 | 66.6k | if ((ret == XML_ERR_OK) && (list != NULL)) { |
7222 | 255 | if (((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7223 | 255 | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY))&& |
7224 | 255 | (ent->children == NULL)) { |
7225 | 109 | ent->children = list; |
7226 | | /* |
7227 | | * Prune it directly in the generated document |
7228 | | * except for single text nodes. |
7229 | | */ |
7230 | 109 | if ((ctxt->replaceEntities == 0) || |
7231 | 109 | (ctxt->parseMode == XML_PARSE_READER) || |
7232 | 109 | ((list->type == XML_TEXT_NODE) && |
7233 | 109 | (list->next == NULL))) { |
7234 | 109 | ent->owner = 1; |
7235 | 578 | while (list != NULL) { |
7236 | 469 | list->parent = (xmlNodePtr) ent; |
7237 | 469 | xmlSetTreeDoc(list, ent->doc); |
7238 | 469 | if (list->next == NULL) |
7239 | 109 | ent->last = list; |
7240 | 469 | list = list->next; |
7241 | 469 | } |
7242 | 109 | list = NULL; |
7243 | 109 | } else { |
7244 | 0 | ent->owner = 0; |
7245 | 0 | while (list != NULL) { |
7246 | 0 | list->parent = (xmlNodePtr) ctxt->node; |
7247 | 0 | list->doc = ctxt->myDoc; |
7248 | 0 | if (list->next == NULL) |
7249 | 0 | ent->last = list; |
7250 | 0 | list = list->next; |
7251 | 0 | } |
7252 | 0 | list = ent->children; |
7253 | | #ifdef LIBXML_LEGACY_ENABLED |
7254 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7255 | | xmlAddEntityReference(ent, list, NULL); |
7256 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7257 | 0 | } |
7258 | 146 | } else { |
7259 | 146 | xmlFreeNodeList(list); |
7260 | 146 | list = NULL; |
7261 | 146 | } |
7262 | 66.3k | } else if ((ret != XML_ERR_OK) && |
7263 | 66.3k | (ret != XML_WAR_UNDECLARED_ENTITY)) { |
7264 | 66.1k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7265 | 66.1k | "Entity '%s' failed to parse\n", ent->name); |
7266 | 66.1k | if (ent->content != NULL) |
7267 | 66.1k | ent->content[0] = 0; |
7268 | 66.1k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7269 | 66.1k | } else if (list != NULL) { |
7270 | 0 | xmlFreeNodeList(list); |
7271 | 0 | list = NULL; |
7272 | 0 | } |
7273 | 66.6k | if (ent->checked == 0) |
7274 | 0 | ent->checked = 2; |
7275 | | |
7276 | | /* Prevent entity from being parsed and expanded twice (Bug 760367). */ |
7277 | 66.6k | was_checked = 0; |
7278 | 147k | } else if (ent->checked != 1) { |
7279 | 147k | ctxt->nbentities += ent->checked / 2; |
7280 | 147k | } |
7281 | | |
7282 | | /* |
7283 | | * Now that the entity content has been gathered |
7284 | | * provide it to the application, this can take different forms based |
7285 | | * on the parsing modes. |
7286 | | */ |
7287 | 213k | if (ent->children == NULL) { |
7288 | | /* |
7289 | | * Probably running in SAX mode and the callbacks don't |
7290 | | * build the entity content. So unless we already went |
7291 | | * though parsing for first checking go though the entity |
7292 | | * content to generate callbacks associated to the entity |
7293 | | */ |
7294 | 200k | if (was_checked != 0) { |
7295 | 133k | void *user_data; |
7296 | | /* |
7297 | | * This is a bit hackish but this seems the best |
7298 | | * way to make sure both SAX and DOM entity support |
7299 | | * behaves okay. |
7300 | | */ |
7301 | 133k | if (ctxt->userData == ctxt) |
7302 | 133k | user_data = NULL; |
7303 | 0 | else |
7304 | 0 | user_data = ctxt->userData; |
7305 | | |
7306 | 133k | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7307 | 133k | ctxt->depth++; |
7308 | 133k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, |
7309 | 133k | ent->content, user_data, NULL); |
7310 | 133k | ctxt->depth--; |
7311 | 133k | } else if (ent->etype == |
7312 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7313 | 0 | ctxt->depth++; |
7314 | 0 | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, |
7315 | 0 | ctxt->sax, user_data, ctxt->depth, |
7316 | 0 | ent->URI, ent->ExternalID, NULL); |
7317 | 0 | ctxt->depth--; |
7318 | 0 | } else { |
7319 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7320 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7321 | 0 | "invalid entity type found\n", NULL); |
7322 | 0 | } |
7323 | 133k | if (ret == XML_ERR_ENTITY_LOOP) { |
7324 | 131 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7325 | 131 | return; |
7326 | 131 | } |
7327 | 133k | } |
7328 | 199k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7329 | 199k | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7330 | | /* |
7331 | | * Entity reference callback comes second, it's somewhat |
7332 | | * superfluous but a compatibility to historical behaviour |
7333 | | */ |
7334 | 133k | ctxt->sax->reference(ctxt->userData, ent->name); |
7335 | 133k | } |
7336 | 199k | return; |
7337 | 200k | } |
7338 | | |
7339 | | /* |
7340 | | * If we didn't get any children for the entity being built |
7341 | | */ |
7342 | 13.6k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7343 | 13.6k | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7344 | | /* |
7345 | | * Create a node. |
7346 | | */ |
7347 | 13.6k | ctxt->sax->reference(ctxt->userData, ent->name); |
7348 | 13.6k | return; |
7349 | 13.6k | } |
7350 | | |
7351 | 4 | if ((ctxt->replaceEntities) || (ent->children == NULL)) { |
7352 | | /* |
7353 | | * There is a problem on the handling of _private for entities |
7354 | | * (bug 155816): Should we copy the content of the field from |
7355 | | * the entity (possibly overwriting some value set by the user |
7356 | | * when a copy is created), should we leave it alone, or should |
7357 | | * we try to take care of different situations? The problem |
7358 | | * is exacerbated by the usage of this field by the xmlReader. |
7359 | | * To fix this bug, we look at _private on the created node |
7360 | | * and, if it's NULL, we copy in whatever was in the entity. |
7361 | | * If it's not NULL we leave it alone. This is somewhat of a |
7362 | | * hack - maybe we should have further tests to determine |
7363 | | * what to do. |
7364 | | */ |
7365 | 0 | if ((ctxt->node != NULL) && (ent->children != NULL)) { |
7366 | | /* |
7367 | | * Seems we are generating the DOM content, do |
7368 | | * a simple tree copy for all references except the first |
7369 | | * In the first occurrence list contains the replacement. |
7370 | | */ |
7371 | 0 | if (((list == NULL) && (ent->owner == 0)) || |
7372 | 0 | (ctxt->parseMode == XML_PARSE_READER)) { |
7373 | 0 | xmlNodePtr nw = NULL, cur, firstChild = NULL; |
7374 | | |
7375 | | /* |
7376 | | * We are copying here, make sure there is no abuse |
7377 | | */ |
7378 | 0 | ctxt->sizeentcopy += ent->length + 5; |
7379 | 0 | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7380 | 0 | return; |
7381 | | |
7382 | | /* |
7383 | | * when operating on a reader, the entities definitions |
7384 | | * are always owning the entities subtree. |
7385 | | if (ctxt->parseMode == XML_PARSE_READER) |
7386 | | ent->owner = 1; |
7387 | | */ |
7388 | | |
7389 | 0 | cur = ent->children; |
7390 | 0 | while (cur != NULL) { |
7391 | 0 | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7392 | 0 | if (nw != NULL) { |
7393 | 0 | if (nw->_private == NULL) |
7394 | 0 | nw->_private = cur->_private; |
7395 | 0 | if (firstChild == NULL){ |
7396 | 0 | firstChild = nw; |
7397 | 0 | } |
7398 | 0 | nw = xmlAddChild(ctxt->node, nw); |
7399 | 0 | } |
7400 | 0 | if (cur == ent->last) { |
7401 | | /* |
7402 | | * needed to detect some strange empty |
7403 | | * node cases in the reader tests |
7404 | | */ |
7405 | 0 | if ((ctxt->parseMode == XML_PARSE_READER) && |
7406 | 0 | (nw != NULL) && |
7407 | 0 | (nw->type == XML_ELEMENT_NODE) && |
7408 | 0 | (nw->children == NULL)) |
7409 | 0 | nw->extra = 1; |
7410 | |
|
7411 | 0 | break; |
7412 | 0 | } |
7413 | 0 | cur = cur->next; |
7414 | 0 | } |
7415 | | #ifdef LIBXML_LEGACY_ENABLED |
7416 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7417 | | xmlAddEntityReference(ent, firstChild, nw); |
7418 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7419 | 0 | } else if ((list == NULL) || (ctxt->inputNr > 0)) { |
7420 | 0 | xmlNodePtr nw = NULL, cur, next, last, |
7421 | 0 | firstChild = NULL; |
7422 | | |
7423 | | /* |
7424 | | * We are copying here, make sure there is no abuse |
7425 | | */ |
7426 | 0 | ctxt->sizeentcopy += ent->length + 5; |
7427 | 0 | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7428 | 0 | return; |
7429 | | |
7430 | | /* |
7431 | | * Copy the entity child list and make it the new |
7432 | | * entity child list. The goal is to make sure any |
7433 | | * ID or REF referenced will be the one from the |
7434 | | * document content and not the entity copy. |
7435 | | */ |
7436 | 0 | cur = ent->children; |
7437 | 0 | ent->children = NULL; |
7438 | 0 | last = ent->last; |
7439 | 0 | ent->last = NULL; |
7440 | 0 | while (cur != NULL) { |
7441 | 0 | next = cur->next; |
7442 | 0 | cur->next = NULL; |
7443 | 0 | cur->parent = NULL; |
7444 | 0 | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7445 | 0 | if (nw != NULL) { |
7446 | 0 | if (nw->_private == NULL) |
7447 | 0 | nw->_private = cur->_private; |
7448 | 0 | if (firstChild == NULL){ |
7449 | 0 | firstChild = cur; |
7450 | 0 | } |
7451 | 0 | xmlAddChild((xmlNodePtr) ent, nw); |
7452 | 0 | xmlAddChild(ctxt->node, cur); |
7453 | 0 | } |
7454 | 0 | if (cur == last) |
7455 | 0 | break; |
7456 | 0 | cur = next; |
7457 | 0 | } |
7458 | 0 | if (ent->owner == 0) |
7459 | 0 | ent->owner = 1; |
7460 | | #ifdef LIBXML_LEGACY_ENABLED |
7461 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7462 | | xmlAddEntityReference(ent, firstChild, nw); |
7463 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7464 | 0 | } else { |
7465 | 0 | const xmlChar *nbktext; |
7466 | | |
7467 | | /* |
7468 | | * the name change is to avoid coalescing of the |
7469 | | * node with a possible previous text one which |
7470 | | * would make ent->children a dangling pointer |
7471 | | */ |
7472 | 0 | nbktext = xmlDictLookup(ctxt->dict, BAD_CAST "nbktext", |
7473 | 0 | -1); |
7474 | 0 | if (ent->children->type == XML_TEXT_NODE) |
7475 | 0 | ent->children->name = nbktext; |
7476 | 0 | if ((ent->last != ent->children) && |
7477 | 0 | (ent->last->type == XML_TEXT_NODE)) |
7478 | 0 | ent->last->name = nbktext; |
7479 | 0 | xmlAddChildList(ctxt->node, ent->children); |
7480 | 0 | } |
7481 | | |
7482 | | /* |
7483 | | * This is to avoid a nasty side effect, see |
7484 | | * characters() in SAX.c |
7485 | | */ |
7486 | 0 | ctxt->nodemem = 0; |
7487 | 0 | ctxt->nodelen = 0; |
7488 | 0 | return; |
7489 | 0 | } |
7490 | 0 | } |
7491 | 4 | } |
7492 | | |
7493 | | /** |
7494 | | * xmlParseEntityRef: |
7495 | | * @ctxt: an XML parser context |
7496 | | * |
7497 | | * parse ENTITY references declarations |
7498 | | * |
7499 | | * [68] EntityRef ::= '&' Name ';' |
7500 | | * |
7501 | | * [ WFC: Entity Declared ] |
7502 | | * In a document without any DTD, a document with only an internal DTD |
7503 | | * subset which contains no parameter entity references, or a document |
7504 | | * with "standalone='yes'", the Name given in the entity reference |
7505 | | * must match that in an entity declaration, except that well-formed |
7506 | | * documents need not declare any of the following entities: amp, lt, |
7507 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7508 | | * any reference to it. Similarly, the declaration of a general entity |
7509 | | * must precede any reference to it which appears in a default value in an |
7510 | | * attribute-list declaration. Note that if entities are declared in the |
7511 | | * external subset or in external parameter entities, a non-validating |
7512 | | * processor is not obligated to read and process their declarations; |
7513 | | * for such documents, the rule that an entity must be declared is a |
7514 | | * well-formedness constraint only if standalone='yes'. |
7515 | | * |
7516 | | * [ WFC: Parsed Entity ] |
7517 | | * An entity reference must not contain the name of an unparsed entity |
7518 | | * |
7519 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7520 | | */ |
7521 | | xmlEntityPtr |
7522 | 974k | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7523 | 974k | const xmlChar *name; |
7524 | 974k | xmlEntityPtr ent = NULL; |
7525 | | |
7526 | 974k | GROW; |
7527 | 974k | if (ctxt->instate == XML_PARSER_EOF) |
7528 | 0 | return(NULL); |
7529 | | |
7530 | 974k | if (RAW != '&') |
7531 | 0 | return(NULL); |
7532 | 974k | NEXT; |
7533 | 974k | name = xmlParseName(ctxt); |
7534 | 974k | if (name == NULL) { |
7535 | 78.5k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7536 | 78.5k | "xmlParseEntityRef: no name\n"); |
7537 | 78.5k | return(NULL); |
7538 | 78.5k | } |
7539 | 896k | if (RAW != ';') { |
7540 | 26.2k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7541 | 26.2k | return(NULL); |
7542 | 26.2k | } |
7543 | 870k | NEXT; |
7544 | | |
7545 | | /* |
7546 | | * Predefined entities override any extra definition |
7547 | | */ |
7548 | 870k | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7549 | 870k | ent = xmlGetPredefinedEntity(name); |
7550 | 870k | if (ent != NULL) |
7551 | 18.0k | return(ent); |
7552 | 870k | } |
7553 | | |
7554 | | /* |
7555 | | * Increase the number of entity references parsed |
7556 | | */ |
7557 | 852k | ctxt->nbentities++; |
7558 | | |
7559 | | /* |
7560 | | * Ask first SAX for entity resolution, otherwise try the |
7561 | | * entities which may have stored in the parser context. |
7562 | | */ |
7563 | 852k | if (ctxt->sax != NULL) { |
7564 | 852k | if (ctxt->sax->getEntity != NULL) |
7565 | 852k | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7566 | 852k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7567 | 852k | (ctxt->options & XML_PARSE_OLDSAX)) |
7568 | 0 | ent = xmlGetPredefinedEntity(name); |
7569 | 852k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7570 | 852k | (ctxt->userData==ctxt)) { |
7571 | 11.9k | ent = xmlSAX2GetEntity(ctxt, name); |
7572 | 11.9k | } |
7573 | 852k | } |
7574 | 852k | if (ctxt->instate == XML_PARSER_EOF) |
7575 | 0 | return(NULL); |
7576 | | /* |
7577 | | * [ WFC: Entity Declared ] |
7578 | | * In a document without any DTD, a document with only an |
7579 | | * internal DTD subset which contains no parameter entity |
7580 | | * references, or a document with "standalone='yes'", the |
7581 | | * Name given in the entity reference must match that in an |
7582 | | * entity declaration, except that well-formed documents |
7583 | | * need not declare any of the following entities: amp, lt, |
7584 | | * gt, apos, quot. |
7585 | | * The declaration of a parameter entity must precede any |
7586 | | * reference to it. |
7587 | | * Similarly, the declaration of a general entity must |
7588 | | * precede any reference to it which appears in a default |
7589 | | * value in an attribute-list declaration. Note that if |
7590 | | * entities are declared in the external subset or in |
7591 | | * external parameter entities, a non-validating processor |
7592 | | * is not obligated to read and process their declarations; |
7593 | | * for such documents, the rule that an entity must be |
7594 | | * declared is a well-formedness constraint only if |
7595 | | * standalone='yes'. |
7596 | | */ |
7597 | 852k | if (ent == NULL) { |
7598 | 53.9k | if ((ctxt->standalone == 1) || |
7599 | 53.9k | ((ctxt->hasExternalSubset == 0) && |
7600 | 53.5k | (ctxt->hasPErefs == 0))) { |
7601 | 40.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7602 | 40.0k | "Entity '%s' not defined\n", name); |
7603 | 40.0k | } else { |
7604 | 13.8k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7605 | 13.8k | "Entity '%s' not defined\n", name); |
7606 | 13.8k | if ((ctxt->inSubset == 0) && |
7607 | 13.8k | (ctxt->sax != NULL) && |
7608 | 13.8k | (ctxt->sax->reference != NULL)) { |
7609 | 12.7k | ctxt->sax->reference(ctxt->userData, name); |
7610 | 12.7k | } |
7611 | 13.8k | } |
7612 | 53.9k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7613 | 53.9k | ctxt->valid = 0; |
7614 | 53.9k | } |
7615 | | |
7616 | | /* |
7617 | | * [ WFC: Parsed Entity ] |
7618 | | * An entity reference must not contain the name of an |
7619 | | * unparsed entity |
7620 | | */ |
7621 | 798k | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7622 | 194 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7623 | 194 | "Entity reference to unparsed entity %s\n", name); |
7624 | 194 | } |
7625 | | |
7626 | | /* |
7627 | | * [ WFC: No External Entity References ] |
7628 | | * Attribute values cannot contain direct or indirect |
7629 | | * entity references to external entities. |
7630 | | */ |
7631 | 797k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7632 | 797k | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7633 | 312k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7634 | 312k | "Attribute references external entity '%s'\n", name); |
7635 | 312k | } |
7636 | | /* |
7637 | | * [ WFC: No < in Attribute Values ] |
7638 | | * The replacement text of any entity referred to directly or |
7639 | | * indirectly in an attribute value (other than "<") must |
7640 | | * not contain a <. |
7641 | | */ |
7642 | 485k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7643 | 485k | (ent != NULL) && |
7644 | 485k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
7645 | 19.4k | if (((ent->checked & 1) || (ent->checked == 0)) && |
7646 | 19.4k | (ent->content != NULL) && (xmlStrchr(ent->content, '<'))) { |
7647 | 395 | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7648 | 395 | "'<' in entity '%s' is not allowed in attributes values\n", name); |
7649 | 395 | } |
7650 | 19.4k | } |
7651 | | |
7652 | | /* |
7653 | | * Internal check, no parameter entities here ... |
7654 | | */ |
7655 | 466k | else { |
7656 | 466k | switch (ent->etype) { |
7657 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7658 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7659 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7660 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7661 | 0 | name); |
7662 | 0 | break; |
7663 | 466k | default: |
7664 | 466k | break; |
7665 | 466k | } |
7666 | 466k | } |
7667 | | |
7668 | | /* |
7669 | | * [ WFC: No Recursion ] |
7670 | | * A parsed entity must not contain a recursive reference |
7671 | | * to itself, either directly or indirectly. |
7672 | | * Done somewhere else |
7673 | | */ |
7674 | 852k | return(ent); |
7675 | 852k | } |
7676 | | |
7677 | | /** |
7678 | | * xmlParseStringEntityRef: |
7679 | | * @ctxt: an XML parser context |
7680 | | * @str: a pointer to an index in the string |
7681 | | * |
7682 | | * parse ENTITY references declarations, but this version parses it from |
7683 | | * a string value. |
7684 | | * |
7685 | | * [68] EntityRef ::= '&' Name ';' |
7686 | | * |
7687 | | * [ WFC: Entity Declared ] |
7688 | | * In a document without any DTD, a document with only an internal DTD |
7689 | | * subset which contains no parameter entity references, or a document |
7690 | | * with "standalone='yes'", the Name given in the entity reference |
7691 | | * must match that in an entity declaration, except that well-formed |
7692 | | * documents need not declare any of the following entities: amp, lt, |
7693 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7694 | | * any reference to it. Similarly, the declaration of a general entity |
7695 | | * must precede any reference to it which appears in a default value in an |
7696 | | * attribute-list declaration. Note that if entities are declared in the |
7697 | | * external subset or in external parameter entities, a non-validating |
7698 | | * processor is not obligated to read and process their declarations; |
7699 | | * for such documents, the rule that an entity must be declared is a |
7700 | | * well-formedness constraint only if standalone='yes'. |
7701 | | * |
7702 | | * [ WFC: Parsed Entity ] |
7703 | | * An entity reference must not contain the name of an unparsed entity |
7704 | | * |
7705 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7706 | | * is updated to the current location in the string. |
7707 | | */ |
7708 | | static xmlEntityPtr |
7709 | 2.01M | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7710 | 2.01M | xmlChar *name; |
7711 | 2.01M | const xmlChar *ptr; |
7712 | 2.01M | xmlChar cur; |
7713 | 2.01M | xmlEntityPtr ent = NULL; |
7714 | | |
7715 | 2.01M | if ((str == NULL) || (*str == NULL)) |
7716 | 0 | return(NULL); |
7717 | 2.01M | ptr = *str; |
7718 | 2.01M | cur = *ptr; |
7719 | 2.01M | if (cur != '&') |
7720 | 0 | return(NULL); |
7721 | | |
7722 | 2.01M | ptr++; |
7723 | 2.01M | name = xmlParseStringName(ctxt, &ptr); |
7724 | 2.01M | if (name == NULL) { |
7725 | 11.3k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7726 | 11.3k | "xmlParseStringEntityRef: no name\n"); |
7727 | 11.3k | *str = ptr; |
7728 | 11.3k | return(NULL); |
7729 | 11.3k | } |
7730 | 2.00M | if (*ptr != ';') { |
7731 | 5.74k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7732 | 5.74k | xmlFree(name); |
7733 | 5.74k | *str = ptr; |
7734 | 5.74k | return(NULL); |
7735 | 5.74k | } |
7736 | 1.99M | ptr++; |
7737 | | |
7738 | | |
7739 | | /* |
7740 | | * Predefined entities override any extra definition |
7741 | | */ |
7742 | 1.99M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7743 | 1.99M | ent = xmlGetPredefinedEntity(name); |
7744 | 1.99M | if (ent != NULL) { |
7745 | 15.2k | xmlFree(name); |
7746 | 15.2k | *str = ptr; |
7747 | 15.2k | return(ent); |
7748 | 15.2k | } |
7749 | 1.99M | } |
7750 | | |
7751 | | /* |
7752 | | * Increase the number of entity references parsed |
7753 | | */ |
7754 | 1.98M | ctxt->nbentities++; |
7755 | | |
7756 | | /* |
7757 | | * Ask first SAX for entity resolution, otherwise try the |
7758 | | * entities which may have stored in the parser context. |
7759 | | */ |
7760 | 1.98M | if (ctxt->sax != NULL) { |
7761 | 1.98M | if (ctxt->sax->getEntity != NULL) |
7762 | 1.98M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7763 | 1.98M | if ((ent == NULL) && (ctxt->options & XML_PARSE_OLDSAX)) |
7764 | 0 | ent = xmlGetPredefinedEntity(name); |
7765 | 1.98M | if ((ent == NULL) && (ctxt->userData==ctxt)) { |
7766 | 219k | ent = xmlSAX2GetEntity(ctxt, name); |
7767 | 219k | } |
7768 | 1.98M | } |
7769 | 1.98M | if (ctxt->instate == XML_PARSER_EOF) { |
7770 | 0 | xmlFree(name); |
7771 | 0 | return(NULL); |
7772 | 0 | } |
7773 | | |
7774 | | /* |
7775 | | * [ WFC: Entity Declared ] |
7776 | | * In a document without any DTD, a document with only an |
7777 | | * internal DTD subset which contains no parameter entity |
7778 | | * references, or a document with "standalone='yes'", the |
7779 | | * Name given in the entity reference must match that in an |
7780 | | * entity declaration, except that well-formed documents |
7781 | | * need not declare any of the following entities: amp, lt, |
7782 | | * gt, apos, quot. |
7783 | | * The declaration of a parameter entity must precede any |
7784 | | * reference to it. |
7785 | | * Similarly, the declaration of a general entity must |
7786 | | * precede any reference to it which appears in a default |
7787 | | * value in an attribute-list declaration. Note that if |
7788 | | * entities are declared in the external subset or in |
7789 | | * external parameter entities, a non-validating processor |
7790 | | * is not obligated to read and process their declarations; |
7791 | | * for such documents, the rule that an entity must be |
7792 | | * declared is a well-formedness constraint only if |
7793 | | * standalone='yes'. |
7794 | | */ |
7795 | 1.98M | if (ent == NULL) { |
7796 | 219k | if ((ctxt->standalone == 1) || |
7797 | 219k | ((ctxt->hasExternalSubset == 0) && |
7798 | 219k | (ctxt->hasPErefs == 0))) { |
7799 | 214k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7800 | 214k | "Entity '%s' not defined\n", name); |
7801 | 214k | } else { |
7802 | 5.15k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7803 | 5.15k | "Entity '%s' not defined\n", |
7804 | 5.15k | name); |
7805 | 5.15k | } |
7806 | 219k | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7807 | | /* TODO ? check regressions ctxt->valid = 0; */ |
7808 | 219k | } |
7809 | | |
7810 | | /* |
7811 | | * [ WFC: Parsed Entity ] |
7812 | | * An entity reference must not contain the name of an |
7813 | | * unparsed entity |
7814 | | */ |
7815 | 1.76M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7816 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7817 | 0 | "Entity reference to unparsed entity %s\n", name); |
7818 | 0 | } |
7819 | | |
7820 | | /* |
7821 | | * [ WFC: No External Entity References ] |
7822 | | * Attribute values cannot contain direct or indirect |
7823 | | * entity references to external entities. |
7824 | | */ |
7825 | 1.76M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7826 | 1.76M | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7827 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7828 | 0 | "Attribute references external entity '%s'\n", name); |
7829 | 0 | } |
7830 | | /* |
7831 | | * [ WFC: No < in Attribute Values ] |
7832 | | * The replacement text of any entity referred to directly or |
7833 | | * indirectly in an attribute value (other than "<") must |
7834 | | * not contain a <. |
7835 | | */ |
7836 | 1.76M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7837 | 1.76M | (ent != NULL) && (ent->content != NULL) && |
7838 | 1.76M | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
7839 | 1.76M | (xmlStrchr(ent->content, '<'))) { |
7840 | 173k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7841 | 173k | "'<' in entity '%s' is not allowed in attributes values\n", |
7842 | 173k | name); |
7843 | 173k | } |
7844 | | |
7845 | | /* |
7846 | | * Internal check, no parameter entities here ... |
7847 | | */ |
7848 | 1.58M | else { |
7849 | 1.58M | switch (ent->etype) { |
7850 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7851 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7852 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7853 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7854 | 0 | name); |
7855 | 0 | break; |
7856 | 1.58M | default: |
7857 | 1.58M | break; |
7858 | 1.58M | } |
7859 | 1.58M | } |
7860 | | |
7861 | | /* |
7862 | | * [ WFC: No Recursion ] |
7863 | | * A parsed entity must not contain a recursive reference |
7864 | | * to itself, either directly or indirectly. |
7865 | | * Done somewhere else |
7866 | | */ |
7867 | | |
7868 | 1.98M | xmlFree(name); |
7869 | 1.98M | *str = ptr; |
7870 | 1.98M | return(ent); |
7871 | 1.98M | } |
7872 | | |
7873 | | /** |
7874 | | * xmlParsePEReference: |
7875 | | * @ctxt: an XML parser context |
7876 | | * |
7877 | | * parse PEReference declarations |
7878 | | * The entity content is handled directly by pushing it's content as |
7879 | | * a new input stream. |
7880 | | * |
7881 | | * [69] PEReference ::= '%' Name ';' |
7882 | | * |
7883 | | * [ WFC: No Recursion ] |
7884 | | * A parsed entity must not contain a recursive |
7885 | | * reference to itself, either directly or indirectly. |
7886 | | * |
7887 | | * [ WFC: Entity Declared ] |
7888 | | * In a document without any DTD, a document with only an internal DTD |
7889 | | * subset which contains no parameter entity references, or a document |
7890 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7891 | | * entity must precede any reference to it... |
7892 | | * |
7893 | | * [ VC: Entity Declared ] |
7894 | | * In a document with an external subset or external parameter entities |
7895 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7896 | | * must precede any reference to it... |
7897 | | * |
7898 | | * [ WFC: In DTD ] |
7899 | | * Parameter-entity references may only appear in the DTD. |
7900 | | * NOTE: misleading but this is handled. |
7901 | | */ |
7902 | | void |
7903 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7904 | 14.0M | { |
7905 | 14.0M | const xmlChar *name; |
7906 | 14.0M | xmlEntityPtr entity = NULL; |
7907 | 14.0M | xmlParserInputPtr input; |
7908 | | |
7909 | 14.0M | if (RAW != '%') |
7910 | 8.43M | return; |
7911 | 5.66M | NEXT; |
7912 | 5.66M | name = xmlParseName(ctxt); |
7913 | 5.66M | if (name == NULL) { |
7914 | 9.00k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7915 | 9.00k | return; |
7916 | 9.00k | } |
7917 | 5.65M | if (xmlParserDebugEntities) |
7918 | 0 | xmlGenericError(xmlGenericErrorContext, |
7919 | 0 | "PEReference: %s\n", name); |
7920 | 5.65M | if (RAW != ';') { |
7921 | 4.56k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7922 | 4.56k | return; |
7923 | 4.56k | } |
7924 | | |
7925 | 5.64M | NEXT; |
7926 | | |
7927 | | /* |
7928 | | * Increase the number of entity references parsed |
7929 | | */ |
7930 | 5.64M | ctxt->nbentities++; |
7931 | | |
7932 | | /* |
7933 | | * Request the entity from SAX |
7934 | | */ |
7935 | 5.64M | if ((ctxt->sax != NULL) && |
7936 | 5.64M | (ctxt->sax->getParameterEntity != NULL)) |
7937 | 5.64M | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7938 | 5.64M | if (ctxt->instate == XML_PARSER_EOF) |
7939 | 0 | return; |
7940 | 5.64M | if (entity == NULL) { |
7941 | | /* |
7942 | | * [ WFC: Entity Declared ] |
7943 | | * In a document without any DTD, a document with only an |
7944 | | * internal DTD subset which contains no parameter entity |
7945 | | * references, or a document with "standalone='yes'", ... |
7946 | | * ... The declaration of a parameter entity must precede |
7947 | | * any reference to it... |
7948 | | */ |
7949 | 6.73k | if ((ctxt->standalone == 1) || |
7950 | 6.73k | ((ctxt->hasExternalSubset == 0) && |
7951 | 6.53k | (ctxt->hasPErefs == 0))) { |
7952 | 426 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7953 | 426 | "PEReference: %%%s; not found\n", |
7954 | 426 | name); |
7955 | 6.30k | } else { |
7956 | | /* |
7957 | | * [ VC: Entity Declared ] |
7958 | | * In a document with an external subset or external |
7959 | | * parameter entities with "standalone='no'", ... |
7960 | | * ... The declaration of a parameter entity must |
7961 | | * precede any reference to it... |
7962 | | */ |
7963 | 6.30k | if ((ctxt->validate) && (ctxt->vctxt.error != NULL)) { |
7964 | 0 | xmlValidityError(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7965 | 0 | "PEReference: %%%s; not found\n", |
7966 | 0 | name, NULL); |
7967 | 0 | } else |
7968 | 6.30k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7969 | 6.30k | "PEReference: %%%s; not found\n", |
7970 | 6.30k | name, NULL); |
7971 | 6.30k | ctxt->valid = 0; |
7972 | 6.30k | } |
7973 | 6.73k | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
7974 | 5.64M | } else { |
7975 | | /* |
7976 | | * Internal checking in case the entity quest barfed |
7977 | | */ |
7978 | 5.64M | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7979 | 5.64M | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7980 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7981 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7982 | 0 | name, NULL); |
7983 | 5.64M | } else { |
7984 | 5.64M | xmlChar start[4]; |
7985 | 5.64M | xmlCharEncoding enc; |
7986 | | |
7987 | 5.64M | if (xmlParserEntityCheck(ctxt, 0, entity, 0)) |
7988 | 0 | return; |
7989 | | |
7990 | 5.64M | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
7991 | 5.64M | ((ctxt->options & XML_PARSE_NOENT) == 0) && |
7992 | 5.64M | ((ctxt->options & XML_PARSE_DTDVALID) == 0) && |
7993 | 5.64M | ((ctxt->options & XML_PARSE_DTDLOAD) == 0) && |
7994 | 5.64M | ((ctxt->options & XML_PARSE_DTDATTR) == 0) && |
7995 | 5.64M | (ctxt->replaceEntities == 0) && |
7996 | 5.64M | (ctxt->validate == 0)) |
7997 | 67 | return; |
7998 | | |
7999 | 5.64M | input = xmlNewEntityInputStream(ctxt, entity); |
8000 | 5.64M | if (xmlPushInput(ctxt, input) < 0) { |
8001 | 4.07k | xmlFreeInputStream(input); |
8002 | 4.07k | return; |
8003 | 4.07k | } |
8004 | | |
8005 | 5.63M | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
8006 | | /* |
8007 | | * Get the 4 first bytes and decode the charset |
8008 | | * if enc != XML_CHAR_ENCODING_NONE |
8009 | | * plug some encoding conversion routines. |
8010 | | * Note that, since we may have some non-UTF8 |
8011 | | * encoding (like UTF16, bug 135229), the 'length' |
8012 | | * is not known, but we can calculate based upon |
8013 | | * the amount of data in the buffer. |
8014 | | */ |
8015 | 0 | GROW |
8016 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
8017 | 0 | return; |
8018 | 0 | if ((ctxt->input->end - ctxt->input->cur)>=4) { |
8019 | 0 | start[0] = RAW; |
8020 | 0 | start[1] = NXT(1); |
8021 | 0 | start[2] = NXT(2); |
8022 | 0 | start[3] = NXT(3); |
8023 | 0 | enc = xmlDetectCharEncoding(start, 4); |
8024 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
8025 | 0 | xmlSwitchEncoding(ctxt, enc); |
8026 | 0 | } |
8027 | 0 | } |
8028 | |
|
8029 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
8030 | 0 | (IS_BLANK_CH(NXT(5)))) { |
8031 | 0 | xmlParseTextDecl(ctxt); |
8032 | 0 | } |
8033 | 0 | } |
8034 | 5.63M | } |
8035 | 5.64M | } |
8036 | 5.64M | ctxt->hasPErefs = 1; |
8037 | 5.64M | } |
8038 | | |
8039 | | /** |
8040 | | * xmlLoadEntityContent: |
8041 | | * @ctxt: an XML parser context |
8042 | | * @entity: an unloaded system entity |
8043 | | * |
8044 | | * Load the original content of the given system entity from the |
8045 | | * ExternalID/SystemID given. This is to be used for Included in Literal |
8046 | | * http://www.w3.org/TR/REC-xml/#inliteral processing of entities references |
8047 | | * |
8048 | | * Returns 0 in case of success and -1 in case of failure |
8049 | | */ |
8050 | | static int |
8051 | 0 | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
8052 | 0 | xmlParserInputPtr input; |
8053 | 0 | xmlBufferPtr buf; |
8054 | 0 | int l, c; |
8055 | 0 | int count = 0; |
8056 | |
|
8057 | 0 | if ((ctxt == NULL) || (entity == NULL) || |
8058 | 0 | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
8059 | 0 | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
8060 | 0 | (entity->content != NULL)) { |
8061 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8062 | 0 | "xmlLoadEntityContent parameter error"); |
8063 | 0 | return(-1); |
8064 | 0 | } |
8065 | | |
8066 | 0 | if (xmlParserDebugEntities) |
8067 | 0 | xmlGenericError(xmlGenericErrorContext, |
8068 | 0 | "Reading %s entity content input\n", entity->name); |
8069 | |
|
8070 | 0 | buf = xmlBufferCreate(); |
8071 | 0 | if (buf == NULL) { |
8072 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8073 | 0 | "xmlLoadEntityContent parameter error"); |
8074 | 0 | return(-1); |
8075 | 0 | } |
8076 | 0 | xmlBufferSetAllocationScheme(buf, XML_BUFFER_ALLOC_DOUBLEIT); |
8077 | |
|
8078 | 0 | input = xmlNewEntityInputStream(ctxt, entity); |
8079 | 0 | if (input == NULL) { |
8080 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8081 | 0 | "xmlLoadEntityContent input error"); |
8082 | 0 | xmlBufferFree(buf); |
8083 | 0 | return(-1); |
8084 | 0 | } |
8085 | | |
8086 | | /* |
8087 | | * Push the entity as the current input, read char by char |
8088 | | * saving to the buffer until the end of the entity or an error |
8089 | | */ |
8090 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
8091 | 0 | xmlBufferFree(buf); |
8092 | 0 | return(-1); |
8093 | 0 | } |
8094 | | |
8095 | 0 | GROW; |
8096 | 0 | c = CUR_CHAR(l); |
8097 | 0 | while ((ctxt->input == input) && (ctxt->input->cur < ctxt->input->end) && |
8098 | 0 | (IS_CHAR(c))) { |
8099 | 0 | xmlBufferAdd(buf, ctxt->input->cur, l); |
8100 | 0 | if (count++ > XML_PARSER_CHUNK_SIZE) { |
8101 | 0 | count = 0; |
8102 | 0 | GROW; |
8103 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
8104 | 0 | xmlBufferFree(buf); |
8105 | 0 | return(-1); |
8106 | 0 | } |
8107 | 0 | } |
8108 | 0 | NEXTL(l); |
8109 | 0 | c = CUR_CHAR(l); |
8110 | 0 | if (c == 0) { |
8111 | 0 | count = 0; |
8112 | 0 | GROW; |
8113 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
8114 | 0 | xmlBufferFree(buf); |
8115 | 0 | return(-1); |
8116 | 0 | } |
8117 | 0 | c = CUR_CHAR(l); |
8118 | 0 | } |
8119 | 0 | } |
8120 | | |
8121 | 0 | if ((ctxt->input == input) && (ctxt->input->cur >= ctxt->input->end)) { |
8122 | 0 | xmlPopInput(ctxt); |
8123 | 0 | } else if (!IS_CHAR(c)) { |
8124 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8125 | 0 | "xmlLoadEntityContent: invalid char value %d\n", |
8126 | 0 | c); |
8127 | 0 | xmlBufferFree(buf); |
8128 | 0 | return(-1); |
8129 | 0 | } |
8130 | 0 | entity->content = buf->content; |
8131 | 0 | buf->content = NULL; |
8132 | 0 | xmlBufferFree(buf); |
8133 | |
|
8134 | 0 | return(0); |
8135 | 0 | } |
8136 | | |
8137 | | /** |
8138 | | * xmlParseStringPEReference: |
8139 | | * @ctxt: an XML parser context |
8140 | | * @str: a pointer to an index in the string |
8141 | | * |
8142 | | * parse PEReference declarations |
8143 | | * |
8144 | | * [69] PEReference ::= '%' Name ';' |
8145 | | * |
8146 | | * [ WFC: No Recursion ] |
8147 | | * A parsed entity must not contain a recursive |
8148 | | * reference to itself, either directly or indirectly. |
8149 | | * |
8150 | | * [ WFC: Entity Declared ] |
8151 | | * In a document without any DTD, a document with only an internal DTD |
8152 | | * subset which contains no parameter entity references, or a document |
8153 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8154 | | * entity must precede any reference to it... |
8155 | | * |
8156 | | * [ VC: Entity Declared ] |
8157 | | * In a document with an external subset or external parameter entities |
8158 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8159 | | * must precede any reference to it... |
8160 | | * |
8161 | | * [ WFC: In DTD ] |
8162 | | * Parameter-entity references may only appear in the DTD. |
8163 | | * NOTE: misleading but this is handled. |
8164 | | * |
8165 | | * Returns the string of the entity content. |
8166 | | * str is updated to the current value of the index |
8167 | | */ |
8168 | | static xmlEntityPtr |
8169 | 37.8k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8170 | 37.8k | const xmlChar *ptr; |
8171 | 37.8k | xmlChar cur; |
8172 | 37.8k | xmlChar *name; |
8173 | 37.8k | xmlEntityPtr entity = NULL; |
8174 | | |
8175 | 37.8k | if ((str == NULL) || (*str == NULL)) return(NULL); |
8176 | 37.8k | ptr = *str; |
8177 | 37.8k | cur = *ptr; |
8178 | 37.8k | if (cur != '%') |
8179 | 0 | return(NULL); |
8180 | 37.8k | ptr++; |
8181 | 37.8k | name = xmlParseStringName(ctxt, &ptr); |
8182 | 37.8k | if (name == NULL) { |
8183 | 14.5k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8184 | 14.5k | "xmlParseStringPEReference: no name\n"); |
8185 | 14.5k | *str = ptr; |
8186 | 14.5k | return(NULL); |
8187 | 14.5k | } |
8188 | 23.2k | cur = *ptr; |
8189 | 23.2k | if (cur != ';') { |
8190 | 6.26k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8191 | 6.26k | xmlFree(name); |
8192 | 6.26k | *str = ptr; |
8193 | 6.26k | return(NULL); |
8194 | 6.26k | } |
8195 | 16.9k | ptr++; |
8196 | | |
8197 | | /* |
8198 | | * Increase the number of entity references parsed |
8199 | | */ |
8200 | 16.9k | ctxt->nbentities++; |
8201 | | |
8202 | | /* |
8203 | | * Request the entity from SAX |
8204 | | */ |
8205 | 16.9k | if ((ctxt->sax != NULL) && |
8206 | 16.9k | (ctxt->sax->getParameterEntity != NULL)) |
8207 | 16.9k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8208 | 16.9k | if (ctxt->instate == XML_PARSER_EOF) { |
8209 | 0 | xmlFree(name); |
8210 | 0 | *str = ptr; |
8211 | 0 | return(NULL); |
8212 | 0 | } |
8213 | 16.9k | if (entity == NULL) { |
8214 | | /* |
8215 | | * [ WFC: Entity Declared ] |
8216 | | * In a document without any DTD, a document with only an |
8217 | | * internal DTD subset which contains no parameter entity |
8218 | | * references, or a document with "standalone='yes'", ... |
8219 | | * ... The declaration of a parameter entity must precede |
8220 | | * any reference to it... |
8221 | | */ |
8222 | 2.16k | if ((ctxt->standalone == 1) || |
8223 | 2.16k | ((ctxt->hasExternalSubset == 0) && (ctxt->hasPErefs == 0))) { |
8224 | 10 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8225 | 10 | "PEReference: %%%s; not found\n", name); |
8226 | 2.15k | } else { |
8227 | | /* |
8228 | | * [ VC: Entity Declared ] |
8229 | | * In a document with an external subset or external |
8230 | | * parameter entities with "standalone='no'", ... |
8231 | | * ... The declaration of a parameter entity must |
8232 | | * precede any reference to it... |
8233 | | */ |
8234 | 2.15k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8235 | 2.15k | "PEReference: %%%s; not found\n", |
8236 | 2.15k | name, NULL); |
8237 | 2.15k | ctxt->valid = 0; |
8238 | 2.15k | } |
8239 | 2.16k | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
8240 | 14.8k | } else { |
8241 | | /* |
8242 | | * Internal checking in case the entity quest barfed |
8243 | | */ |
8244 | 14.8k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8245 | 14.8k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8246 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8247 | 0 | "%%%s; is not a parameter entity\n", |
8248 | 0 | name, NULL); |
8249 | 0 | } |
8250 | 14.8k | } |
8251 | 16.9k | ctxt->hasPErefs = 1; |
8252 | 16.9k | xmlFree(name); |
8253 | 16.9k | *str = ptr; |
8254 | 16.9k | return(entity); |
8255 | 16.9k | } |
8256 | | |
8257 | | /** |
8258 | | * xmlParseDocTypeDecl: |
8259 | | * @ctxt: an XML parser context |
8260 | | * |
8261 | | * parse a DOCTYPE declaration |
8262 | | * |
8263 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8264 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8265 | | * |
8266 | | * [ VC: Root Element Type ] |
8267 | | * The Name in the document type declaration must match the element |
8268 | | * type of the root element. |
8269 | | */ |
8270 | | |
8271 | | void |
8272 | 9.57k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8273 | 9.57k | const xmlChar *name = NULL; |
8274 | 9.57k | xmlChar *ExternalID = NULL; |
8275 | 9.57k | xmlChar *URI = NULL; |
8276 | | |
8277 | | /* |
8278 | | * We know that '<!DOCTYPE' has been detected. |
8279 | | */ |
8280 | 9.57k | SKIP(9); |
8281 | | |
8282 | 9.57k | SKIP_BLANKS; |
8283 | | |
8284 | | /* |
8285 | | * Parse the DOCTYPE name. |
8286 | | */ |
8287 | 9.57k | name = xmlParseName(ctxt); |
8288 | 9.57k | if (name == NULL) { |
8289 | 1.59k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8290 | 1.59k | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8291 | 1.59k | } |
8292 | 9.57k | ctxt->intSubName = name; |
8293 | | |
8294 | 9.57k | SKIP_BLANKS; |
8295 | | |
8296 | | /* |
8297 | | * Check for SystemID and ExternalID |
8298 | | */ |
8299 | 9.57k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8300 | | |
8301 | 9.57k | if ((URI != NULL) || (ExternalID != NULL)) { |
8302 | 334 | ctxt->hasExternalSubset = 1; |
8303 | 334 | } |
8304 | 9.57k | ctxt->extSubURI = URI; |
8305 | 9.57k | ctxt->extSubSystem = ExternalID; |
8306 | | |
8307 | 9.57k | SKIP_BLANKS; |
8308 | | |
8309 | | /* |
8310 | | * Create and update the internal subset. |
8311 | | */ |
8312 | 9.57k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8313 | 9.57k | (!ctxt->disableSAX)) |
8314 | 7.50k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8315 | 9.57k | if (ctxt->instate == XML_PARSER_EOF) |
8316 | 0 | return; |
8317 | | |
8318 | | /* |
8319 | | * Is there any internal subset declarations ? |
8320 | | * they are handled separately in xmlParseInternalSubset() |
8321 | | */ |
8322 | 9.57k | if (RAW == '[') |
8323 | 9.06k | return; |
8324 | | |
8325 | | /* |
8326 | | * We should be at the end of the DOCTYPE declaration. |
8327 | | */ |
8328 | 512 | if (RAW != '>') { |
8329 | 316 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8330 | 316 | } |
8331 | 512 | NEXT; |
8332 | 512 | } |
8333 | | |
8334 | | /** |
8335 | | * xmlParseInternalSubset: |
8336 | | * @ctxt: an XML parser context |
8337 | | * |
8338 | | * parse the internal subset declaration |
8339 | | * |
8340 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8341 | | */ |
8342 | | |
8343 | | static void |
8344 | 9.13k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8345 | | /* |
8346 | | * Is there any DTD definition ? |
8347 | | */ |
8348 | 9.13k | if (RAW == '[') { |
8349 | 9.13k | int baseInputNr = ctxt->inputNr; |
8350 | 9.13k | ctxt->instate = XML_PARSER_DTD; |
8351 | 9.13k | NEXT; |
8352 | | /* |
8353 | | * Parse the succession of Markup declarations and |
8354 | | * PEReferences. |
8355 | | * Subsequence (markupdecl | PEReference | S)* |
8356 | | */ |
8357 | 8.46M | while (((RAW != ']') || (ctxt->inputNr > baseInputNr)) && |
8358 | 8.46M | (ctxt->instate != XML_PARSER_EOF)) { |
8359 | 8.46M | int id = ctxt->input->id; |
8360 | 8.46M | unsigned long cons = CUR_CONSUMED; |
8361 | | |
8362 | 8.46M | SKIP_BLANKS; |
8363 | 8.46M | xmlParseMarkupDecl(ctxt); |
8364 | 8.46M | xmlParsePEReference(ctxt); |
8365 | | |
8366 | | /* |
8367 | | * Conditional sections are allowed from external entities included |
8368 | | * by PE References in the internal subset. |
8369 | | */ |
8370 | 8.46M | if ((ctxt->inputNr > 1) && (ctxt->input->filename != NULL) && |
8371 | 8.46M | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8372 | 0 | xmlParseConditionalSections(ctxt); |
8373 | 0 | } |
8374 | | |
8375 | 8.46M | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
8376 | 1.32M | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8377 | 1.32M | "xmlParseInternalSubset: error detected in Markup declaration\n"); |
8378 | 1.32M | if (ctxt->inputNr > baseInputNr) |
8379 | 1.31M | xmlPopInput(ctxt); |
8380 | 6.23k | else |
8381 | 6.23k | break; |
8382 | 1.32M | } |
8383 | 8.46M | } |
8384 | 9.13k | if (RAW == ']') { |
8385 | 1.91k | NEXT; |
8386 | 1.91k | SKIP_BLANKS; |
8387 | 1.91k | } |
8388 | 9.13k | } |
8389 | | |
8390 | | /* |
8391 | | * We should be at the end of the DOCTYPE declaration. |
8392 | | */ |
8393 | 9.13k | if (RAW != '>') { |
8394 | 7.04k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8395 | 7.04k | return; |
8396 | 7.04k | } |
8397 | 2.09k | NEXT; |
8398 | 2.09k | } |
8399 | | |
8400 | | #ifdef LIBXML_SAX1_ENABLED |
8401 | | /** |
8402 | | * xmlParseAttribute: |
8403 | | * @ctxt: an XML parser context |
8404 | | * @value: a xmlChar ** used to store the value of the attribute |
8405 | | * |
8406 | | * parse an attribute |
8407 | | * |
8408 | | * [41] Attribute ::= Name Eq AttValue |
8409 | | * |
8410 | | * [ WFC: No External Entity References ] |
8411 | | * Attribute values cannot contain direct or indirect entity references |
8412 | | * to external entities. |
8413 | | * |
8414 | | * [ WFC: No < in Attribute Values ] |
8415 | | * The replacement text of any entity referred to directly or indirectly in |
8416 | | * an attribute value (other than "<") must not contain a <. |
8417 | | * |
8418 | | * [ VC: Attribute Value Type ] |
8419 | | * The attribute must have been declared; the value must be of the type |
8420 | | * declared for it. |
8421 | | * |
8422 | | * [25] Eq ::= S? '=' S? |
8423 | | * |
8424 | | * With namespace: |
8425 | | * |
8426 | | * [NS 11] Attribute ::= QName Eq AttValue |
8427 | | * |
8428 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8429 | | * definition. |
8430 | | * |
8431 | | * Returns the attribute name, and the value in *value. |
8432 | | */ |
8433 | | |
8434 | | const xmlChar * |
8435 | 0 | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8436 | 0 | const xmlChar *name; |
8437 | 0 | xmlChar *val; |
8438 | |
|
8439 | 0 | *value = NULL; |
8440 | 0 | GROW; |
8441 | 0 | name = xmlParseName(ctxt); |
8442 | 0 | if (name == NULL) { |
8443 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8444 | 0 | "error parsing attribute name\n"); |
8445 | 0 | return(NULL); |
8446 | 0 | } |
8447 | | |
8448 | | /* |
8449 | | * read the value |
8450 | | */ |
8451 | 0 | SKIP_BLANKS; |
8452 | 0 | if (RAW == '=') { |
8453 | 0 | NEXT; |
8454 | 0 | SKIP_BLANKS; |
8455 | 0 | val = xmlParseAttValue(ctxt); |
8456 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
8457 | 0 | } else { |
8458 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8459 | 0 | "Specification mandates value for attribute %s\n", name); |
8460 | 0 | return(NULL); |
8461 | 0 | } |
8462 | | |
8463 | | /* |
8464 | | * Check that xml:lang conforms to the specification |
8465 | | * No more registered as an error, just generate a warning now |
8466 | | * since this was deprecated in XML second edition |
8467 | | */ |
8468 | 0 | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8469 | 0 | if (!xmlCheckLanguageID(val)) { |
8470 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8471 | 0 | "Malformed value for xml:lang : %s\n", |
8472 | 0 | val, NULL); |
8473 | 0 | } |
8474 | 0 | } |
8475 | | |
8476 | | /* |
8477 | | * Check that xml:space conforms to the specification |
8478 | | */ |
8479 | 0 | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8480 | 0 | if (xmlStrEqual(val, BAD_CAST "default")) |
8481 | 0 | *(ctxt->space) = 0; |
8482 | 0 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8483 | 0 | *(ctxt->space) = 1; |
8484 | 0 | else { |
8485 | 0 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8486 | 0 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8487 | 0 | val, NULL); |
8488 | 0 | } |
8489 | 0 | } |
8490 | |
|
8491 | 0 | *value = val; |
8492 | 0 | return(name); |
8493 | 0 | } |
8494 | | |
8495 | | /** |
8496 | | * xmlParseStartTag: |
8497 | | * @ctxt: an XML parser context |
8498 | | * |
8499 | | * parse a start of tag either for rule element or |
8500 | | * EmptyElement. In both case we don't parse the tag closing chars. |
8501 | | * |
8502 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8503 | | * |
8504 | | * [ WFC: Unique Att Spec ] |
8505 | | * No attribute name may appear more than once in the same start-tag or |
8506 | | * empty-element tag. |
8507 | | * |
8508 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8509 | | * |
8510 | | * [ WFC: Unique Att Spec ] |
8511 | | * No attribute name may appear more than once in the same start-tag or |
8512 | | * empty-element tag. |
8513 | | * |
8514 | | * With namespace: |
8515 | | * |
8516 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8517 | | * |
8518 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8519 | | * |
8520 | | * Returns the element name parsed |
8521 | | */ |
8522 | | |
8523 | | const xmlChar * |
8524 | 0 | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8525 | 0 | const xmlChar *name; |
8526 | 0 | const xmlChar *attname; |
8527 | 0 | xmlChar *attvalue; |
8528 | 0 | const xmlChar **atts = ctxt->atts; |
8529 | 0 | int nbatts = 0; |
8530 | 0 | int maxatts = ctxt->maxatts; |
8531 | 0 | int i; |
8532 | |
|
8533 | 0 | if (RAW != '<') return(NULL); |
8534 | 0 | NEXT1; |
8535 | |
|
8536 | 0 | name = xmlParseName(ctxt); |
8537 | 0 | if (name == NULL) { |
8538 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8539 | 0 | "xmlParseStartTag: invalid element name\n"); |
8540 | 0 | return(NULL); |
8541 | 0 | } |
8542 | | |
8543 | | /* |
8544 | | * Now parse the attributes, it ends up with the ending |
8545 | | * |
8546 | | * (S Attribute)* S? |
8547 | | */ |
8548 | 0 | SKIP_BLANKS; |
8549 | 0 | GROW; |
8550 | |
|
8551 | 0 | while (((RAW != '>') && |
8552 | 0 | ((RAW != '/') || (NXT(1) != '>')) && |
8553 | 0 | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
8554 | 0 | int id = ctxt->input->id; |
8555 | 0 | unsigned long cons = CUR_CONSUMED; |
8556 | |
|
8557 | 0 | attname = xmlParseAttribute(ctxt, &attvalue); |
8558 | 0 | if ((attname != NULL) && (attvalue != NULL)) { |
8559 | | /* |
8560 | | * [ WFC: Unique Att Spec ] |
8561 | | * No attribute name may appear more than once in the same |
8562 | | * start-tag or empty-element tag. |
8563 | | */ |
8564 | 0 | for (i = 0; i < nbatts;i += 2) { |
8565 | 0 | if (xmlStrEqual(atts[i], attname)) { |
8566 | 0 | xmlErrAttributeDup(ctxt, NULL, attname); |
8567 | 0 | xmlFree(attvalue); |
8568 | 0 | goto failed; |
8569 | 0 | } |
8570 | 0 | } |
8571 | | /* |
8572 | | * Add the pair to atts |
8573 | | */ |
8574 | 0 | if (atts == NULL) { |
8575 | 0 | maxatts = 22; /* allow for 10 attrs by default */ |
8576 | 0 | atts = (const xmlChar **) |
8577 | 0 | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8578 | 0 | if (atts == NULL) { |
8579 | 0 | xmlErrMemory(ctxt, NULL); |
8580 | 0 | if (attvalue != NULL) |
8581 | 0 | xmlFree(attvalue); |
8582 | 0 | goto failed; |
8583 | 0 | } |
8584 | 0 | ctxt->atts = atts; |
8585 | 0 | ctxt->maxatts = maxatts; |
8586 | 0 | } else if (nbatts + 4 > maxatts) { |
8587 | 0 | const xmlChar **n; |
8588 | |
|
8589 | 0 | maxatts *= 2; |
8590 | 0 | n = (const xmlChar **) xmlRealloc((void *) atts, |
8591 | 0 | maxatts * sizeof(const xmlChar *)); |
8592 | 0 | if (n == NULL) { |
8593 | 0 | xmlErrMemory(ctxt, NULL); |
8594 | 0 | if (attvalue != NULL) |
8595 | 0 | xmlFree(attvalue); |
8596 | 0 | goto failed; |
8597 | 0 | } |
8598 | 0 | atts = n; |
8599 | 0 | ctxt->atts = atts; |
8600 | 0 | ctxt->maxatts = maxatts; |
8601 | 0 | } |
8602 | 0 | atts[nbatts++] = attname; |
8603 | 0 | atts[nbatts++] = attvalue; |
8604 | 0 | atts[nbatts] = NULL; |
8605 | 0 | atts[nbatts + 1] = NULL; |
8606 | 0 | } else { |
8607 | 0 | if (attvalue != NULL) |
8608 | 0 | xmlFree(attvalue); |
8609 | 0 | } |
8610 | | |
8611 | 0 | failed: |
8612 | |
|
8613 | 0 | GROW |
8614 | 0 | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8615 | 0 | break; |
8616 | 0 | if (SKIP_BLANKS == 0) { |
8617 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8618 | 0 | "attributes construct error\n"); |
8619 | 0 | } |
8620 | 0 | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
8621 | 0 | (attname == NULL) && (attvalue == NULL)) { |
8622 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
8623 | 0 | "xmlParseStartTag: problem parsing attributes\n"); |
8624 | 0 | break; |
8625 | 0 | } |
8626 | 0 | SHRINK; |
8627 | 0 | GROW; |
8628 | 0 | } |
8629 | | |
8630 | | /* |
8631 | | * SAX: Start of Element ! |
8632 | | */ |
8633 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8634 | 0 | (!ctxt->disableSAX)) { |
8635 | 0 | if (nbatts > 0) |
8636 | 0 | ctxt->sax->startElement(ctxt->userData, name, atts); |
8637 | 0 | else |
8638 | 0 | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8639 | 0 | } |
8640 | |
|
8641 | 0 | if (atts != NULL) { |
8642 | | /* Free only the content strings */ |
8643 | 0 | for (i = 1;i < nbatts;i+=2) |
8644 | 0 | if (atts[i] != NULL) |
8645 | 0 | xmlFree((xmlChar *) atts[i]); |
8646 | 0 | } |
8647 | 0 | return(name); |
8648 | 0 | } |
8649 | | |
8650 | | /** |
8651 | | * xmlParseEndTag1: |
8652 | | * @ctxt: an XML parser context |
8653 | | * @line: line of the start tag |
8654 | | * @nsNr: number of namespaces on the start tag |
8655 | | * |
8656 | | * parse an end of tag |
8657 | | * |
8658 | | * [42] ETag ::= '</' Name S? '>' |
8659 | | * |
8660 | | * With namespace |
8661 | | * |
8662 | | * [NS 9] ETag ::= '</' QName S? '>' |
8663 | | */ |
8664 | | |
8665 | | static void |
8666 | 0 | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8667 | 0 | const xmlChar *name; |
8668 | |
|
8669 | 0 | GROW; |
8670 | 0 | if ((RAW != '<') || (NXT(1) != '/')) { |
8671 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8672 | 0 | "xmlParseEndTag: '</' not found\n"); |
8673 | 0 | return; |
8674 | 0 | } |
8675 | 0 | SKIP(2); |
8676 | |
|
8677 | 0 | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8678 | | |
8679 | | /* |
8680 | | * We should definitely be at the ending "S? '>'" part |
8681 | | */ |
8682 | 0 | GROW; |
8683 | 0 | SKIP_BLANKS; |
8684 | 0 | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8685 | 0 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8686 | 0 | } else |
8687 | 0 | NEXT1; |
8688 | | |
8689 | | /* |
8690 | | * [ WFC: Element Type Match ] |
8691 | | * The Name in an element's end-tag must match the element type in the |
8692 | | * start-tag. |
8693 | | * |
8694 | | */ |
8695 | 0 | if (name != (xmlChar*)1) { |
8696 | 0 | if (name == NULL) name = BAD_CAST "unparsable"; |
8697 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8698 | 0 | "Opening and ending tag mismatch: %s line %d and %s\n", |
8699 | 0 | ctxt->name, line, name); |
8700 | 0 | } |
8701 | | |
8702 | | /* |
8703 | | * SAX: End of Tag |
8704 | | */ |
8705 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8706 | 0 | (!ctxt->disableSAX)) |
8707 | 0 | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8708 | |
|
8709 | 0 | namePop(ctxt); |
8710 | 0 | spacePop(ctxt); |
8711 | 0 | return; |
8712 | 0 | } |
8713 | | |
8714 | | /** |
8715 | | * xmlParseEndTag: |
8716 | | * @ctxt: an XML parser context |
8717 | | * |
8718 | | * parse an end of tag |
8719 | | * |
8720 | | * [42] ETag ::= '</' Name S? '>' |
8721 | | * |
8722 | | * With namespace |
8723 | | * |
8724 | | * [NS 9] ETag ::= '</' QName S? '>' |
8725 | | */ |
8726 | | |
8727 | | void |
8728 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8729 | 0 | xmlParseEndTag1(ctxt, 0); |
8730 | 0 | } |
8731 | | #endif /* LIBXML_SAX1_ENABLED */ |
8732 | | |
8733 | | /************************************************************************ |
8734 | | * * |
8735 | | * SAX 2 specific operations * |
8736 | | * * |
8737 | | ************************************************************************/ |
8738 | | |
8739 | | /* |
8740 | | * xmlGetNamespace: |
8741 | | * @ctxt: an XML parser context |
8742 | | * @prefix: the prefix to lookup |
8743 | | * |
8744 | | * Lookup the namespace name for the @prefix (which ca be NULL) |
8745 | | * The prefix must come from the @ctxt->dict dictionary |
8746 | | * |
8747 | | * Returns the namespace name or NULL if not bound |
8748 | | */ |
8749 | | static const xmlChar * |
8750 | 29.9M | xmlGetNamespace(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
8751 | 29.9M | int i; |
8752 | | |
8753 | 29.9M | if (prefix == ctxt->str_xml) return(ctxt->str_xml_ns); |
8754 | 307M | for (i = ctxt->nsNr - 2;i >= 0;i-=2) |
8755 | 306M | if (ctxt->nsTab[i] == prefix) { |
8756 | 28.9M | if ((prefix == NULL) && (*ctxt->nsTab[i + 1] == 0)) |
8757 | 8.26M | return(NULL); |
8758 | 20.6M | return(ctxt->nsTab[i + 1]); |
8759 | 28.9M | } |
8760 | 992k | return(NULL); |
8761 | 29.8M | } |
8762 | | |
8763 | | /** |
8764 | | * xmlParseQName: |
8765 | | * @ctxt: an XML parser context |
8766 | | * @prefix: pointer to store the prefix part |
8767 | | * |
8768 | | * parse an XML Namespace QName |
8769 | | * |
8770 | | * [6] QName ::= (Prefix ':')? LocalPart |
8771 | | * [7] Prefix ::= NCName |
8772 | | * [8] LocalPart ::= NCName |
8773 | | * |
8774 | | * Returns the Name parsed or NULL |
8775 | | */ |
8776 | | |
8777 | | static const xmlChar * |
8778 | 6.22M | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8779 | 6.22M | const xmlChar *l, *p; |
8780 | | |
8781 | 6.22M | GROW; |
8782 | | |
8783 | 6.22M | l = xmlParseNCName(ctxt); |
8784 | 6.22M | if (l == NULL) { |
8785 | 294k | if (CUR == ':') { |
8786 | 9.58k | l = xmlParseName(ctxt); |
8787 | 9.58k | if (l != NULL) { |
8788 | 9.58k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8789 | 9.58k | "Failed to parse QName '%s'\n", l, NULL, NULL); |
8790 | 9.58k | *prefix = NULL; |
8791 | 9.58k | return(l); |
8792 | 9.58k | } |
8793 | 9.58k | } |
8794 | 284k | return(NULL); |
8795 | 294k | } |
8796 | 5.93M | if (CUR == ':') { |
8797 | 177k | NEXT; |
8798 | 177k | p = l; |
8799 | 177k | l = xmlParseNCName(ctxt); |
8800 | 177k | if (l == NULL) { |
8801 | 17.7k | xmlChar *tmp; |
8802 | | |
8803 | 17.7k | if (ctxt->instate == XML_PARSER_EOF) |
8804 | 0 | return(NULL); |
8805 | 17.7k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8806 | 17.7k | "Failed to parse QName '%s:'\n", p, NULL, NULL); |
8807 | 17.7k | l = xmlParseNmtoken(ctxt); |
8808 | 17.7k | if (l == NULL) { |
8809 | 14.3k | if (ctxt->instate == XML_PARSER_EOF) |
8810 | 0 | return(NULL); |
8811 | 14.3k | tmp = xmlBuildQName(BAD_CAST "", p, NULL, 0); |
8812 | 14.3k | } else { |
8813 | 3.40k | tmp = xmlBuildQName(l, p, NULL, 0); |
8814 | 3.40k | xmlFree((char *)l); |
8815 | 3.40k | } |
8816 | 17.7k | p = xmlDictLookup(ctxt->dict, tmp, -1); |
8817 | 17.7k | if (tmp != NULL) xmlFree(tmp); |
8818 | 17.7k | *prefix = NULL; |
8819 | 17.7k | return(p); |
8820 | 17.7k | } |
8821 | 159k | if (CUR == ':') { |
8822 | 16.1k | xmlChar *tmp; |
8823 | | |
8824 | 16.1k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8825 | 16.1k | "Failed to parse QName '%s:%s:'\n", p, l, NULL); |
8826 | 16.1k | NEXT; |
8827 | 16.1k | tmp = (xmlChar *) xmlParseName(ctxt); |
8828 | 16.1k | if (tmp != NULL) { |
8829 | 13.8k | tmp = xmlBuildQName(tmp, l, NULL, 0); |
8830 | 13.8k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8831 | 13.8k | if (tmp != NULL) xmlFree(tmp); |
8832 | 13.8k | *prefix = p; |
8833 | 13.8k | return(l); |
8834 | 13.8k | } |
8835 | 2.24k | if (ctxt->instate == XML_PARSER_EOF) |
8836 | 0 | return(NULL); |
8837 | 2.24k | tmp = xmlBuildQName(BAD_CAST "", l, NULL, 0); |
8838 | 2.24k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8839 | 2.24k | if (tmp != NULL) xmlFree(tmp); |
8840 | 2.24k | *prefix = p; |
8841 | 2.24k | return(l); |
8842 | 2.24k | } |
8843 | 143k | *prefix = p; |
8844 | 143k | } else |
8845 | 5.75M | *prefix = NULL; |
8846 | 5.90M | return(l); |
8847 | 5.93M | } |
8848 | | |
8849 | | /** |
8850 | | * xmlParseQNameAndCompare: |
8851 | | * @ctxt: an XML parser context |
8852 | | * @name: the localname |
8853 | | * @prefix: the prefix, if any. |
8854 | | * |
8855 | | * parse an XML name and compares for match |
8856 | | * (specialized for endtag parsing) |
8857 | | * |
8858 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8859 | | * and the name for mismatch |
8860 | | */ |
8861 | | |
8862 | | static const xmlChar * |
8863 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8864 | 9.87k | xmlChar const *prefix) { |
8865 | 9.87k | const xmlChar *cmp; |
8866 | 9.87k | const xmlChar *in; |
8867 | 9.87k | const xmlChar *ret; |
8868 | 9.87k | const xmlChar *prefix2; |
8869 | | |
8870 | 9.87k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8871 | | |
8872 | 9.87k | GROW; |
8873 | 9.87k | in = ctxt->input->cur; |
8874 | | |
8875 | 9.87k | cmp = prefix; |
8876 | 26.3k | while (*in != 0 && *in == *cmp) { |
8877 | 16.5k | ++in; |
8878 | 16.5k | ++cmp; |
8879 | 16.5k | } |
8880 | 9.87k | if ((*cmp == 0) && (*in == ':')) { |
8881 | 6.44k | in++; |
8882 | 6.44k | cmp = name; |
8883 | 16.9k | while (*in != 0 && *in == *cmp) { |
8884 | 10.4k | ++in; |
8885 | 10.4k | ++cmp; |
8886 | 10.4k | } |
8887 | 6.44k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8888 | | /* success */ |
8889 | 4.71k | ctxt->input->col += in - ctxt->input->cur; |
8890 | 4.71k | ctxt->input->cur = in; |
8891 | 4.71k | return((const xmlChar*) 1); |
8892 | 4.71k | } |
8893 | 6.44k | } |
8894 | | /* |
8895 | | * all strings coms from the dictionary, equality can be done directly |
8896 | | */ |
8897 | 5.16k | ret = xmlParseQName (ctxt, &prefix2); |
8898 | 5.16k | if ((ret == name) && (prefix == prefix2)) |
8899 | 952 | return((const xmlChar*) 1); |
8900 | 4.21k | return ret; |
8901 | 5.16k | } |
8902 | | |
8903 | | /** |
8904 | | * xmlParseAttValueInternal: |
8905 | | * @ctxt: an XML parser context |
8906 | | * @len: attribute len result |
8907 | | * @alloc: whether the attribute was reallocated as a new string |
8908 | | * @normalize: if 1 then further non-CDATA normalization must be done |
8909 | | * |
8910 | | * parse a value for an attribute. |
8911 | | * NOTE: if no normalization is needed, the routine will return pointers |
8912 | | * directly from the data buffer. |
8913 | | * |
8914 | | * 3.3.3 Attribute-Value Normalization: |
8915 | | * Before the value of an attribute is passed to the application or |
8916 | | * checked for validity, the XML processor must normalize it as follows: |
8917 | | * - a character reference is processed by appending the referenced |
8918 | | * character to the attribute value |
8919 | | * - an entity reference is processed by recursively processing the |
8920 | | * replacement text of the entity |
8921 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
8922 | | * appending #x20 to the normalized value, except that only a single |
8923 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
8924 | | * parsed entity or the literal entity value of an internal parsed entity |
8925 | | * - other characters are processed by appending them to the normalized value |
8926 | | * If the declared value is not CDATA, then the XML processor must further |
8927 | | * process the normalized attribute value by discarding any leading and |
8928 | | * trailing space (#x20) characters, and by replacing sequences of space |
8929 | | * (#x20) characters by a single space (#x20) character. |
8930 | | * All attributes for which no declaration has been read should be treated |
8931 | | * by a non-validating parser as if declared CDATA. |
8932 | | * |
8933 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
8934 | | * caller if it was copied, this can be detected by val[*len] == 0. |
8935 | | */ |
8936 | | |
8937 | | #define GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) \ |
8938 | 4.04k | const xmlChar *oldbase = ctxt->input->base;\ |
8939 | 4.04k | GROW;\ |
8940 | 4.04k | if (ctxt->instate == XML_PARSER_EOF)\ |
8941 | 4.04k | return(NULL);\ |
8942 | 4.04k | if (oldbase != ctxt->input->base) {\ |
8943 | 1 | ptrdiff_t delta = ctxt->input->base - oldbase;\ |
8944 | 1 | start = start + delta;\ |
8945 | 1 | in = in + delta;\ |
8946 | 1 | }\ |
8947 | 4.04k | end = ctxt->input->end; |
8948 | | |
8949 | | static xmlChar * |
8950 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *len, int *alloc, |
8951 | | int normalize) |
8952 | 1.07M | { |
8953 | 1.07M | xmlChar limit = 0; |
8954 | 1.07M | const xmlChar *in = NULL, *start, *end, *last; |
8955 | 1.07M | xmlChar *ret = NULL; |
8956 | 1.07M | int line, col; |
8957 | | |
8958 | 1.07M | GROW; |
8959 | 1.07M | in = (xmlChar *) CUR_PTR; |
8960 | 1.07M | line = ctxt->input->line; |
8961 | 1.07M | col = ctxt->input->col; |
8962 | 1.07M | if (*in != '"' && *in != '\'') { |
8963 | 764k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
8964 | 764k | return (NULL); |
8965 | 764k | } |
8966 | 315k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
8967 | | |
8968 | | /* |
8969 | | * try to handle in this routine the most common case where no |
8970 | | * allocation of a new string is required and where content is |
8971 | | * pure ASCII. |
8972 | | */ |
8973 | 315k | limit = *in++; |
8974 | 315k | col++; |
8975 | 315k | end = ctxt->input->end; |
8976 | 315k | start = in; |
8977 | 315k | if (in >= end) { |
8978 | 183 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
8979 | 183 | } |
8980 | 315k | if (normalize) { |
8981 | | /* |
8982 | | * Skip any leading spaces |
8983 | | */ |
8984 | 7.67k | while ((in < end) && (*in != limit) && |
8985 | 7.67k | ((*in == 0x20) || (*in == 0x9) || |
8986 | 7.02k | (*in == 0xA) || (*in == 0xD))) { |
8987 | 1.90k | if (*in == 0xA) { |
8988 | 615 | line++; col = 1; |
8989 | 1.29k | } else { |
8990 | 1.29k | col++; |
8991 | 1.29k | } |
8992 | 1.90k | in++; |
8993 | 1.90k | start = in; |
8994 | 1.90k | if (in >= end) { |
8995 | 17 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
8996 | 17 | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
8997 | 17 | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
8998 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
8999 | 0 | "AttValue length too long\n"); |
9000 | 0 | return(NULL); |
9001 | 0 | } |
9002 | 17 | } |
9003 | 1.90k | } |
9004 | 14.2k | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9005 | 14.2k | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9006 | 8.78k | col++; |
9007 | 8.78k | if ((*in++ == 0x20) && (*in == 0x20)) break; |
9008 | 8.50k | if (in >= end) { |
9009 | 30 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9010 | 30 | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
9011 | 30 | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9012 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9013 | 0 | "AttValue length too long\n"); |
9014 | 0 | return(NULL); |
9015 | 0 | } |
9016 | 30 | } |
9017 | 8.50k | } |
9018 | 5.76k | last = in; |
9019 | | /* |
9020 | | * skip the trailing blanks |
9021 | | */ |
9022 | 6.08k | while ((last[-1] == 0x20) && (last > start)) last--; |
9023 | 8.15k | while ((in < end) && (*in != limit) && |
9024 | 8.15k | ((*in == 0x20) || (*in == 0x9) || |
9025 | 6.77k | (*in == 0xA) || (*in == 0xD))) { |
9026 | 2.38k | if (*in == 0xA) { |
9027 | 467 | line++, col = 1; |
9028 | 1.91k | } else { |
9029 | 1.91k | col++; |
9030 | 1.91k | } |
9031 | 2.38k | in++; |
9032 | 2.38k | if (in >= end) { |
9033 | 31 | const xmlChar *oldbase = ctxt->input->base; |
9034 | 31 | GROW; |
9035 | 31 | if (ctxt->instate == XML_PARSER_EOF) |
9036 | 0 | return(NULL); |
9037 | 31 | if (oldbase != ctxt->input->base) { |
9038 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase; |
9039 | 0 | start = start + delta; |
9040 | 0 | in = in + delta; |
9041 | 0 | last = last + delta; |
9042 | 0 | } |
9043 | 31 | end = ctxt->input->end; |
9044 | 31 | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
9045 | 31 | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9046 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9047 | 0 | "AttValue length too long\n"); |
9048 | 0 | return(NULL); |
9049 | 0 | } |
9050 | 31 | } |
9051 | 2.38k | } |
9052 | 5.76k | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
9053 | 5.76k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9054 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9055 | 0 | "AttValue length too long\n"); |
9056 | 0 | return(NULL); |
9057 | 0 | } |
9058 | 5.76k | if (*in != limit) goto need_complex; |
9059 | 309k | } else { |
9060 | 1.93M | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9061 | 1.93M | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9062 | 1.63M | in++; |
9063 | 1.63M | col++; |
9064 | 1.63M | if (in >= end) { |
9065 | 3.81k | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9066 | 3.81k | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
9067 | 3.81k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9068 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9069 | 0 | "AttValue length too long\n"); |
9070 | 0 | return(NULL); |
9071 | 0 | } |
9072 | 3.81k | } |
9073 | 1.63M | } |
9074 | 309k | last = in; |
9075 | 309k | if (((in - start) > XML_MAX_TEXT_LENGTH) && |
9076 | 309k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9077 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9078 | 0 | "AttValue length too long\n"); |
9079 | 0 | return(NULL); |
9080 | 0 | } |
9081 | 309k | if (*in != limit) goto need_complex; |
9082 | 309k | } |
9083 | 216k | in++; |
9084 | 216k | col++; |
9085 | 216k | if (len != NULL) { |
9086 | 171k | *len = last - start; |
9087 | 171k | ret = (xmlChar *) start; |
9088 | 171k | } else { |
9089 | 44.6k | if (alloc) *alloc = 1; |
9090 | 44.6k | ret = xmlStrndup(start, last - start); |
9091 | 44.6k | } |
9092 | 216k | CUR_PTR = in; |
9093 | 216k | ctxt->input->line = line; |
9094 | 216k | ctxt->input->col = col; |
9095 | 216k | if (alloc) *alloc = 0; |
9096 | 216k | return ret; |
9097 | 98.8k | need_complex: |
9098 | 98.8k | if (alloc) *alloc = 1; |
9099 | 98.8k | return xmlParseAttValueComplex(ctxt, len, normalize); |
9100 | 315k | } |
9101 | | |
9102 | | /** |
9103 | | * xmlParseAttribute2: |
9104 | | * @ctxt: an XML parser context |
9105 | | * @pref: the element prefix |
9106 | | * @elem: the element name |
9107 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
9108 | | * @value: a xmlChar ** used to store the value of the attribute |
9109 | | * @len: an int * to save the length of the attribute |
9110 | | * @alloc: an int * to indicate if the attribute was allocated |
9111 | | * |
9112 | | * parse an attribute in the new SAX2 framework. |
9113 | | * |
9114 | | * Returns the attribute name, and the value in *value, . |
9115 | | */ |
9116 | | |
9117 | | static const xmlChar * |
9118 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
9119 | | const xmlChar * pref, const xmlChar * elem, |
9120 | | const xmlChar ** prefix, xmlChar ** value, |
9121 | | int *len, int *alloc) |
9122 | 431k | { |
9123 | 431k | const xmlChar *name; |
9124 | 431k | xmlChar *val, *internal_val = NULL; |
9125 | 431k | int normalize = 0; |
9126 | | |
9127 | 431k | *value = NULL; |
9128 | 431k | GROW; |
9129 | 431k | name = xmlParseQName(ctxt, prefix); |
9130 | 431k | if (name == NULL) { |
9131 | 154k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9132 | 154k | "error parsing attribute name\n"); |
9133 | 154k | return (NULL); |
9134 | 154k | } |
9135 | | |
9136 | | /* |
9137 | | * get the type if needed |
9138 | | */ |
9139 | 276k | if (ctxt->attsSpecial != NULL) { |
9140 | 31.3k | int type; |
9141 | | |
9142 | 31.3k | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
9143 | 31.3k | pref, elem, *prefix, name); |
9144 | 31.3k | if (type != 0) |
9145 | 6.20k | normalize = 1; |
9146 | 31.3k | } |
9147 | | |
9148 | | /* |
9149 | | * read the value |
9150 | | */ |
9151 | 276k | SKIP_BLANKS; |
9152 | 276k | if (RAW == '=') { |
9153 | 262k | NEXT; |
9154 | 262k | SKIP_BLANKS; |
9155 | 262k | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize); |
9156 | 262k | if (normalize) { |
9157 | | /* |
9158 | | * Sometimes a second normalisation pass for spaces is needed |
9159 | | * but that only happens if charrefs or entities references |
9160 | | * have been used in the attribute value, i.e. the attribute |
9161 | | * value have been extracted in an allocated string already. |
9162 | | */ |
9163 | 5.90k | if (*alloc) { |
9164 | 4.47k | const xmlChar *val2; |
9165 | | |
9166 | 4.47k | val2 = xmlAttrNormalizeSpace2(ctxt, val, len); |
9167 | 4.47k | if ((val2 != NULL) && (val2 != val)) { |
9168 | 441 | xmlFree(val); |
9169 | 441 | val = (xmlChar *) val2; |
9170 | 441 | } |
9171 | 4.47k | } |
9172 | 5.90k | } |
9173 | 262k | ctxt->instate = XML_PARSER_CONTENT; |
9174 | 262k | } else { |
9175 | 13.9k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
9176 | 13.9k | "Specification mandates value for attribute %s\n", |
9177 | 13.9k | name); |
9178 | 13.9k | return (NULL); |
9179 | 13.9k | } |
9180 | | |
9181 | 262k | if (*prefix == ctxt->str_xml) { |
9182 | | /* |
9183 | | * Check that xml:lang conforms to the specification |
9184 | | * No more registered as an error, just generate a warning now |
9185 | | * since this was deprecated in XML second edition |
9186 | | */ |
9187 | 50.4k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
9188 | 0 | internal_val = xmlStrndup(val, *len); |
9189 | 0 | if (!xmlCheckLanguageID(internal_val)) { |
9190 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
9191 | 0 | "Malformed value for xml:lang : %s\n", |
9192 | 0 | internal_val, NULL); |
9193 | 0 | } |
9194 | 0 | } |
9195 | | |
9196 | | /* |
9197 | | * Check that xml:space conforms to the specification |
9198 | | */ |
9199 | 50.4k | if (xmlStrEqual(name, BAD_CAST "space")) { |
9200 | 3.70k | internal_val = xmlStrndup(val, *len); |
9201 | 3.70k | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
9202 | 195 | *(ctxt->space) = 0; |
9203 | 3.50k | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
9204 | 577 | *(ctxt->space) = 1; |
9205 | 2.92k | else { |
9206 | 2.92k | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
9207 | 2.92k | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
9208 | 2.92k | internal_val, NULL); |
9209 | 2.92k | } |
9210 | 3.70k | } |
9211 | 50.4k | if (internal_val) { |
9212 | 3.11k | xmlFree(internal_val); |
9213 | 3.11k | } |
9214 | 50.4k | } |
9215 | | |
9216 | 262k | *value = val; |
9217 | 262k | return (name); |
9218 | 276k | } |
9219 | | /** |
9220 | | * xmlParseStartTag2: |
9221 | | * @ctxt: an XML parser context |
9222 | | * |
9223 | | * parse a start of tag either for rule element or |
9224 | | * EmptyElement. In both case we don't parse the tag closing chars. |
9225 | | * This routine is called when running SAX2 parsing |
9226 | | * |
9227 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
9228 | | * |
9229 | | * [ WFC: Unique Att Spec ] |
9230 | | * No attribute name may appear more than once in the same start-tag or |
9231 | | * empty-element tag. |
9232 | | * |
9233 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
9234 | | * |
9235 | | * [ WFC: Unique Att Spec ] |
9236 | | * No attribute name may appear more than once in the same start-tag or |
9237 | | * empty-element tag. |
9238 | | * |
9239 | | * With namespace: |
9240 | | * |
9241 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
9242 | | * |
9243 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
9244 | | * |
9245 | | * Returns the element name parsed |
9246 | | */ |
9247 | | |
9248 | | static const xmlChar * |
9249 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
9250 | 5.79M | const xmlChar **URI, int *tlen) { |
9251 | 5.79M | const xmlChar *localname; |
9252 | 5.79M | const xmlChar *prefix; |
9253 | 5.79M | const xmlChar *attname; |
9254 | 5.79M | const xmlChar *aprefix; |
9255 | 5.79M | const xmlChar *nsname; |
9256 | 5.79M | xmlChar *attvalue; |
9257 | 5.79M | const xmlChar **atts = ctxt->atts; |
9258 | 5.79M | int maxatts = ctxt->maxatts; |
9259 | 5.79M | int nratts, nbatts, nbdef, inputid; |
9260 | 5.79M | int i, j, nbNs, attval; |
9261 | 5.79M | unsigned long cur; |
9262 | 5.79M | int nsNr = ctxt->nsNr; |
9263 | | |
9264 | 5.79M | if (RAW != '<') return(NULL); |
9265 | 5.79M | NEXT1; |
9266 | | |
9267 | | /* |
9268 | | * NOTE: it is crucial with the SAX2 API to never call SHRINK beyond that |
9269 | | * point since the attribute values may be stored as pointers to |
9270 | | * the buffer and calling SHRINK would destroy them ! |
9271 | | * The Shrinking is only possible once the full set of attribute |
9272 | | * callbacks have been done. |
9273 | | */ |
9274 | 5.79M | SHRINK; |
9275 | 5.79M | cur = ctxt->input->cur - ctxt->input->base; |
9276 | 5.79M | inputid = ctxt->input->id; |
9277 | 5.79M | nbatts = 0; |
9278 | 5.79M | nratts = 0; |
9279 | 5.79M | nbdef = 0; |
9280 | 5.79M | nbNs = 0; |
9281 | 5.79M | attval = 0; |
9282 | | /* Forget any namespaces added during an earlier parse of this element. */ |
9283 | 5.79M | ctxt->nsNr = nsNr; |
9284 | | |
9285 | 5.79M | localname = xmlParseQName(ctxt, &prefix); |
9286 | 5.79M | if (localname == NULL) { |
9287 | 129k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9288 | 129k | "StartTag: invalid element name\n"); |
9289 | 129k | return(NULL); |
9290 | 129k | } |
9291 | 5.66M | *tlen = ctxt->input->cur - ctxt->input->base - cur; |
9292 | | |
9293 | | /* |
9294 | | * Now parse the attributes, it ends up with the ending |
9295 | | * |
9296 | | * (S Attribute)* S? |
9297 | | */ |
9298 | 5.66M | SKIP_BLANKS; |
9299 | 5.66M | GROW; |
9300 | | |
9301 | 5.71M | while (((RAW != '>') && |
9302 | 5.71M | ((RAW != '/') || (NXT(1) != '>')) && |
9303 | 5.71M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
9304 | 431k | int id = ctxt->input->id; |
9305 | 431k | unsigned long cons = CUR_CONSUMED; |
9306 | 431k | int len = -1, alloc = 0; |
9307 | | |
9308 | 431k | attname = xmlParseAttribute2(ctxt, prefix, localname, |
9309 | 431k | &aprefix, &attvalue, &len, &alloc); |
9310 | 431k | if ((attname == NULL) || (attvalue == NULL)) |
9311 | 173k | goto next_attr; |
9312 | 257k | if (len < 0) len = xmlStrlen(attvalue); |
9313 | | |
9314 | 257k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9315 | 93.2k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9316 | 93.2k | xmlURIPtr uri; |
9317 | | |
9318 | 93.2k | if (URL == NULL) { |
9319 | 0 | xmlErrMemory(ctxt, "dictionary allocation failure"); |
9320 | 0 | if ((attvalue != NULL) && (alloc != 0)) |
9321 | 0 | xmlFree(attvalue); |
9322 | 0 | localname = NULL; |
9323 | 0 | goto done; |
9324 | 0 | } |
9325 | 93.2k | if (*URL != 0) { |
9326 | 91.8k | uri = xmlParseURI((const char *) URL); |
9327 | 91.8k | if (uri == NULL) { |
9328 | 56.6k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9329 | 56.6k | "xmlns: '%s' is not a valid URI\n", |
9330 | 56.6k | URL, NULL, NULL); |
9331 | 56.6k | } else { |
9332 | 35.2k | if (uri->scheme == NULL) { |
9333 | 20.3k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9334 | 20.3k | "xmlns: URI %s is not absolute\n", |
9335 | 20.3k | URL, NULL, NULL); |
9336 | 20.3k | } |
9337 | 35.2k | xmlFreeURI(uri); |
9338 | 35.2k | } |
9339 | 91.8k | if (URL == ctxt->str_xml_ns) { |
9340 | 577 | if (attname != ctxt->str_xml) { |
9341 | 577 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9342 | 577 | "xml namespace URI cannot be the default namespace\n", |
9343 | 577 | NULL, NULL, NULL); |
9344 | 577 | } |
9345 | 577 | goto next_attr; |
9346 | 577 | } |
9347 | 91.2k | if ((len == 29) && |
9348 | 91.2k | (xmlStrEqual(URL, |
9349 | 2.83k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9350 | 896 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9351 | 896 | "reuse of the xmlns namespace name is forbidden\n", |
9352 | 896 | NULL, NULL, NULL); |
9353 | 896 | goto next_attr; |
9354 | 896 | } |
9355 | 91.2k | } |
9356 | | /* |
9357 | | * check that it's not a defined namespace |
9358 | | */ |
9359 | 96.0k | for (j = 1;j <= nbNs;j++) |
9360 | 6.88k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9361 | 2.64k | break; |
9362 | 91.7k | if (j <= nbNs) |
9363 | 2.64k | xmlErrAttributeDup(ctxt, NULL, attname); |
9364 | 89.1k | else |
9365 | 89.1k | if (nsPush(ctxt, NULL, URL) > 0) nbNs++; |
9366 | | |
9367 | 164k | } else if (aprefix == ctxt->str_xmlns) { |
9368 | 44.3k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9369 | 44.3k | xmlURIPtr uri; |
9370 | | |
9371 | 44.3k | if (attname == ctxt->str_xml) { |
9372 | 1.37k | if (URL != ctxt->str_xml_ns) { |
9373 | 1.18k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9374 | 1.18k | "xml namespace prefix mapped to wrong URI\n", |
9375 | 1.18k | NULL, NULL, NULL); |
9376 | 1.18k | } |
9377 | | /* |
9378 | | * Do not keep a namespace definition node |
9379 | | */ |
9380 | 1.37k | goto next_attr; |
9381 | 1.37k | } |
9382 | 43.0k | if (URL == ctxt->str_xml_ns) { |
9383 | 676 | if (attname != ctxt->str_xml) { |
9384 | 676 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9385 | 676 | "xml namespace URI mapped to wrong prefix\n", |
9386 | 676 | NULL, NULL, NULL); |
9387 | 676 | } |
9388 | 676 | goto next_attr; |
9389 | 676 | } |
9390 | 42.3k | if (attname == ctxt->str_xmlns) { |
9391 | 1.14k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9392 | 1.14k | "redefinition of the xmlns prefix is forbidden\n", |
9393 | 1.14k | NULL, NULL, NULL); |
9394 | 1.14k | goto next_attr; |
9395 | 1.14k | } |
9396 | 41.1k | if ((len == 29) && |
9397 | 41.1k | (xmlStrEqual(URL, |
9398 | 1.32k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9399 | 579 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9400 | 579 | "reuse of the xmlns namespace name is forbidden\n", |
9401 | 579 | NULL, NULL, NULL); |
9402 | 579 | goto next_attr; |
9403 | 579 | } |
9404 | 40.6k | if ((URL == NULL) || (URL[0] == 0)) { |
9405 | 1.21k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9406 | 1.21k | "xmlns:%s: Empty XML namespace is not allowed\n", |
9407 | 1.21k | attname, NULL, NULL); |
9408 | 1.21k | goto next_attr; |
9409 | 39.3k | } else { |
9410 | 39.3k | uri = xmlParseURI((const char *) URL); |
9411 | 39.3k | if (uri == NULL) { |
9412 | 16.7k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9413 | 16.7k | "xmlns:%s: '%s' is not a valid URI\n", |
9414 | 16.7k | attname, URL, NULL); |
9415 | 22.6k | } else { |
9416 | 22.6k | if ((ctxt->pedantic) && (uri->scheme == NULL)) { |
9417 | 0 | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9418 | 0 | "xmlns:%s: URI %s is not absolute\n", |
9419 | 0 | attname, URL, NULL); |
9420 | 0 | } |
9421 | 22.6k | xmlFreeURI(uri); |
9422 | 22.6k | } |
9423 | 39.3k | } |
9424 | | |
9425 | | /* |
9426 | | * check that it's not a defined namespace |
9427 | | */ |
9428 | 46.9k | for (j = 1;j <= nbNs;j++) |
9429 | 8.78k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9430 | 1.25k | break; |
9431 | 39.3k | if (j <= nbNs) |
9432 | 1.25k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9433 | 38.1k | else |
9434 | 38.1k | if (nsPush(ctxt, attname, URL) > 0) nbNs++; |
9435 | | |
9436 | 119k | } else { |
9437 | | /* |
9438 | | * Add the pair to atts |
9439 | | */ |
9440 | 119k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9441 | 55.1k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9442 | 0 | goto next_attr; |
9443 | 0 | } |
9444 | 55.1k | maxatts = ctxt->maxatts; |
9445 | 55.1k | atts = ctxt->atts; |
9446 | 55.1k | } |
9447 | 119k | ctxt->attallocs[nratts++] = alloc; |
9448 | 119k | atts[nbatts++] = attname; |
9449 | 119k | atts[nbatts++] = aprefix; |
9450 | | /* |
9451 | | * The namespace URI field is used temporarily to point at the |
9452 | | * base of the current input buffer for non-alloced attributes. |
9453 | | * When the input buffer is reallocated, all the pointers become |
9454 | | * invalid, but they can be reconstructed later. |
9455 | | */ |
9456 | 119k | if (alloc) |
9457 | 51.8k | atts[nbatts++] = NULL; |
9458 | 67.9k | else |
9459 | 67.9k | atts[nbatts++] = ctxt->input->base; |
9460 | 119k | atts[nbatts++] = attvalue; |
9461 | 119k | attvalue += len; |
9462 | 119k | atts[nbatts++] = attvalue; |
9463 | | /* |
9464 | | * tag if some deallocation is needed |
9465 | | */ |
9466 | 119k | if (alloc != 0) attval = 1; |
9467 | 119k | attvalue = NULL; /* moved into atts */ |
9468 | 119k | } |
9469 | | |
9470 | 431k | next_attr: |
9471 | 431k | if ((attvalue != NULL) && (alloc != 0)) { |
9472 | 33.7k | xmlFree(attvalue); |
9473 | 33.7k | attvalue = NULL; |
9474 | 33.7k | } |
9475 | | |
9476 | 431k | GROW |
9477 | 431k | if (ctxt->instate == XML_PARSER_EOF) |
9478 | 0 | break; |
9479 | 431k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9480 | 177k | break; |
9481 | 253k | if (SKIP_BLANKS == 0) { |
9482 | 202k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9483 | 202k | "attributes construct error\n"); |
9484 | 202k | break; |
9485 | 202k | } |
9486 | 50.6k | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
9487 | 50.6k | (attname == NULL) && (attvalue == NULL)) { |
9488 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9489 | 0 | "xmlParseStartTag: problem parsing attributes\n"); |
9490 | 0 | break; |
9491 | 0 | } |
9492 | 50.6k | GROW; |
9493 | 50.6k | } |
9494 | | |
9495 | 5.66M | if (ctxt->input->id != inputid) { |
9496 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9497 | 0 | "Unexpected change of input\n"); |
9498 | 0 | localname = NULL; |
9499 | 0 | goto done; |
9500 | 0 | } |
9501 | | |
9502 | | /* Reconstruct attribute value pointers. */ |
9503 | 5.78M | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9504 | 119k | if (atts[i+2] != NULL) { |
9505 | | /* |
9506 | | * Arithmetic on dangling pointers is technically undefined |
9507 | | * behavior, but well... |
9508 | | */ |
9509 | 67.9k | ptrdiff_t offset = ctxt->input->base - atts[i+2]; |
9510 | 67.9k | atts[i+2] = NULL; /* Reset repurposed namespace URI */ |
9511 | 67.9k | atts[i+3] += offset; /* value */ |
9512 | 67.9k | atts[i+4] += offset; /* valuend */ |
9513 | 67.9k | } |
9514 | 119k | } |
9515 | | |
9516 | | /* |
9517 | | * The attributes defaulting |
9518 | | */ |
9519 | 5.66M | if (ctxt->attsDefault != NULL) { |
9520 | 4.20M | xmlDefAttrsPtr defaults; |
9521 | | |
9522 | 4.20M | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9523 | 4.20M | if (defaults != NULL) { |
9524 | 32.3M | for (i = 0;i < defaults->nbAttrs;i++) { |
9525 | 28.3M | attname = defaults->values[5 * i]; |
9526 | 28.3M | aprefix = defaults->values[5 * i + 1]; |
9527 | | |
9528 | | /* |
9529 | | * special work for namespaces defaulted defs |
9530 | | */ |
9531 | 28.3M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9532 | | /* |
9533 | | * check that it's not a defined namespace |
9534 | | */ |
9535 | 24.2M | for (j = 1;j <= nbNs;j++) |
9536 | 20.1M | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9537 | 2.37k | break; |
9538 | 4.08M | if (j <= nbNs) continue; |
9539 | | |
9540 | 4.08M | nsname = xmlGetNamespace(ctxt, NULL); |
9541 | 4.08M | if (nsname != defaults->values[5 * i + 2]) { |
9542 | 4.08M | if (nsPush(ctxt, NULL, |
9543 | 4.08M | defaults->values[5 * i + 2]) > 0) |
9544 | 4.08M | nbNs++; |
9545 | 4.08M | } |
9546 | 24.2M | } else if (aprefix == ctxt->str_xmlns) { |
9547 | | /* |
9548 | | * check that it's not a defined namespace |
9549 | | */ |
9550 | 60.3M | for (j = 1;j <= nbNs;j++) |
9551 | 40.2M | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9552 | 517 | break; |
9553 | 20.1M | if (j <= nbNs) continue; |
9554 | | |
9555 | 20.1M | nsname = xmlGetNamespace(ctxt, attname); |
9556 | 20.1M | if (nsname != defaults->values[2]) { |
9557 | 20.1M | if (nsPush(ctxt, attname, |
9558 | 20.1M | defaults->values[5 * i + 2]) > 0) |
9559 | 20.1M | nbNs++; |
9560 | 20.1M | } |
9561 | 20.1M | } else { |
9562 | | /* |
9563 | | * check that it's not a defined attribute |
9564 | | */ |
9565 | 4.12M | for (j = 0;j < nbatts;j+=5) { |
9566 | 38.2k | if ((attname == atts[j]) && (aprefix == atts[j+1])) |
9567 | 1.17k | break; |
9568 | 38.2k | } |
9569 | 4.08M | if (j < nbatts) continue; |
9570 | | |
9571 | 4.08M | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9572 | 1.12k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9573 | 0 | localname = NULL; |
9574 | 0 | goto done; |
9575 | 0 | } |
9576 | 1.12k | maxatts = ctxt->maxatts; |
9577 | 1.12k | atts = ctxt->atts; |
9578 | 1.12k | } |
9579 | 4.08M | atts[nbatts++] = attname; |
9580 | 4.08M | atts[nbatts++] = aprefix; |
9581 | 4.08M | if (aprefix == NULL) |
9582 | 4.08M | atts[nbatts++] = NULL; |
9583 | 3.69k | else |
9584 | 3.69k | atts[nbatts++] = xmlGetNamespace(ctxt, aprefix); |
9585 | 4.08M | atts[nbatts++] = defaults->values[5 * i + 2]; |
9586 | 4.08M | atts[nbatts++] = defaults->values[5 * i + 3]; |
9587 | 4.08M | if ((ctxt->standalone == 1) && |
9588 | 4.08M | (defaults->values[5 * i + 4] != NULL)) { |
9589 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9590 | 0 | "standalone: attribute %s on %s defaulted from external subset\n", |
9591 | 0 | attname, localname); |
9592 | 0 | } |
9593 | 4.08M | nbdef++; |
9594 | 4.08M | } |
9595 | 28.3M | } |
9596 | 4.09M | } |
9597 | 4.20M | } |
9598 | | |
9599 | | /* |
9600 | | * The attributes checkings |
9601 | | */ |
9602 | 9.86M | for (i = 0; i < nbatts;i += 5) { |
9603 | | /* |
9604 | | * The default namespace does not apply to attribute names. |
9605 | | */ |
9606 | 4.20M | if (atts[i + 1] != NULL) { |
9607 | 70.5k | nsname = xmlGetNamespace(ctxt, atts[i + 1]); |
9608 | 70.5k | if (nsname == NULL) { |
9609 | 16.5k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9610 | 16.5k | "Namespace prefix %s for %s on %s is not defined\n", |
9611 | 16.5k | atts[i + 1], atts[i], localname); |
9612 | 16.5k | } |
9613 | 70.5k | atts[i + 2] = nsname; |
9614 | 70.5k | } else |
9615 | 4.13M | nsname = NULL; |
9616 | | /* |
9617 | | * [ WFC: Unique Att Spec ] |
9618 | | * No attribute name may appear more than once in the same |
9619 | | * start-tag or empty-element tag. |
9620 | | * As extended by the Namespace in XML REC. |
9621 | | */ |
9622 | 4.27M | for (j = 0; j < i;j += 5) { |
9623 | 77.2k | if (atts[i] == atts[j]) { |
9624 | 9.72k | if (atts[i+1] == atts[j+1]) { |
9625 | 6.59k | xmlErrAttributeDup(ctxt, atts[i+1], atts[i]); |
9626 | 6.59k | break; |
9627 | 6.59k | } |
9628 | 3.12k | if ((nsname != NULL) && (atts[j + 2] == nsname)) { |
9629 | 315 | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9630 | 315 | "Namespaced Attribute %s in '%s' redefined\n", |
9631 | 315 | atts[i], nsname, NULL); |
9632 | 315 | break; |
9633 | 315 | } |
9634 | 3.12k | } |
9635 | 77.2k | } |
9636 | 4.20M | } |
9637 | | |
9638 | 5.66M | nsname = xmlGetNamespace(ctxt, prefix); |
9639 | 5.66M | if ((prefix != NULL) && (nsname == NULL)) { |
9640 | 27.1k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9641 | 27.1k | "Namespace prefix %s on %s is not defined\n", |
9642 | 27.1k | prefix, localname, NULL); |
9643 | 27.1k | } |
9644 | 5.66M | *pref = prefix; |
9645 | 5.66M | *URI = nsname; |
9646 | | |
9647 | | /* |
9648 | | * SAX: Start of Element ! |
9649 | | */ |
9650 | 5.66M | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9651 | 5.66M | (!ctxt->disableSAX)) { |
9652 | 533k | if (nbNs > 0) |
9653 | 124k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9654 | 124k | nsname, nbNs, &ctxt->nsTab[ctxt->nsNr - 2 * nbNs], |
9655 | 124k | nbatts / 5, nbdef, atts); |
9656 | 409k | else |
9657 | 409k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9658 | 409k | nsname, 0, NULL, nbatts / 5, nbdef, atts); |
9659 | 533k | } |
9660 | | |
9661 | 5.66M | done: |
9662 | | /* |
9663 | | * Free up attribute allocated strings if needed |
9664 | | */ |
9665 | 5.66M | if (attval != 0) { |
9666 | 105k | for (i = 3,j = 0; j < nratts;i += 5,j++) |
9667 | 54.7k | if ((ctxt->attallocs[j] != 0) && (atts[i] != NULL)) |
9668 | 51.8k | xmlFree((xmlChar *) atts[i]); |
9669 | 50.8k | } |
9670 | | |
9671 | 5.66M | return(localname); |
9672 | 5.66M | } |
9673 | | |
9674 | | /** |
9675 | | * xmlParseEndTag2: |
9676 | | * @ctxt: an XML parser context |
9677 | | * @line: line of the start tag |
9678 | | * @nsNr: number of namespaces on the start tag |
9679 | | * |
9680 | | * parse an end of tag |
9681 | | * |
9682 | | * [42] ETag ::= '</' Name S? '>' |
9683 | | * |
9684 | | * With namespace |
9685 | | * |
9686 | | * [NS 9] ETag ::= '</' QName S? '>' |
9687 | | */ |
9688 | | |
9689 | | static void |
9690 | 201k | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9691 | 201k | const xmlChar *name; |
9692 | | |
9693 | 201k | GROW; |
9694 | 201k | if ((RAW != '<') || (NXT(1) != '/')) { |
9695 | 0 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9696 | 0 | return; |
9697 | 0 | } |
9698 | 201k | SKIP(2); |
9699 | | |
9700 | 201k | if (tag->prefix == NULL) |
9701 | 191k | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9702 | 9.87k | else |
9703 | 9.87k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9704 | | |
9705 | | /* |
9706 | | * We should definitely be at the ending "S? '>'" part |
9707 | | */ |
9708 | 201k | GROW; |
9709 | 201k | if (ctxt->instate == XML_PARSER_EOF) |
9710 | 0 | return; |
9711 | 201k | SKIP_BLANKS; |
9712 | 201k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9713 | 15.6k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9714 | 15.6k | } else |
9715 | 185k | NEXT1; |
9716 | | |
9717 | | /* |
9718 | | * [ WFC: Element Type Match ] |
9719 | | * The Name in an element's end-tag must match the element type in the |
9720 | | * start-tag. |
9721 | | * |
9722 | | */ |
9723 | 201k | if (name != (xmlChar*)1) { |
9724 | 19.6k | if (name == NULL) name = BAD_CAST "unparsable"; |
9725 | 19.6k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9726 | 19.6k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9727 | 19.6k | ctxt->name, tag->line, name); |
9728 | 19.6k | } |
9729 | | |
9730 | | /* |
9731 | | * SAX: End of Tag |
9732 | | */ |
9733 | 201k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9734 | 201k | (!ctxt->disableSAX)) |
9735 | 150k | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9736 | 150k | tag->URI); |
9737 | | |
9738 | 201k | spacePop(ctxt); |
9739 | 201k | if (tag->nsNr != 0) |
9740 | 9.97k | nsPop(ctxt, tag->nsNr); |
9741 | 201k | } |
9742 | | |
9743 | | /** |
9744 | | * xmlParseCDSect: |
9745 | | * @ctxt: an XML parser context |
9746 | | * |
9747 | | * Parse escaped pure raw content. |
9748 | | * |
9749 | | * [18] CDSect ::= CDStart CData CDEnd |
9750 | | * |
9751 | | * [19] CDStart ::= '<![CDATA[' |
9752 | | * |
9753 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9754 | | * |
9755 | | * [21] CDEnd ::= ']]>' |
9756 | | */ |
9757 | | void |
9758 | 13.7k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9759 | 13.7k | xmlChar *buf = NULL; |
9760 | 13.7k | int len = 0; |
9761 | 13.7k | int size = XML_PARSER_BUFFER_SIZE; |
9762 | 13.7k | int r, rl; |
9763 | 13.7k | int s, sl; |
9764 | 13.7k | int cur, l; |
9765 | 13.7k | int count = 0; |
9766 | | |
9767 | | /* Check 2.6.0 was NXT(0) not RAW */ |
9768 | 13.7k | if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9769 | 13.7k | SKIP(9); |
9770 | 13.7k | } else |
9771 | 0 | return; |
9772 | | |
9773 | 13.7k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
9774 | 13.7k | r = CUR_CHAR(rl); |
9775 | 13.7k | if (!IS_CHAR(r)) { |
9776 | 1.00k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9777 | 1.00k | ctxt->instate = XML_PARSER_CONTENT; |
9778 | 1.00k | return; |
9779 | 1.00k | } |
9780 | 12.6k | NEXTL(rl); |
9781 | 12.6k | s = CUR_CHAR(sl); |
9782 | 12.6k | if (!IS_CHAR(s)) { |
9783 | 2.44k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9784 | 2.44k | ctxt->instate = XML_PARSER_CONTENT; |
9785 | 2.44k | return; |
9786 | 2.44k | } |
9787 | 10.2k | NEXTL(sl); |
9788 | 10.2k | cur = CUR_CHAR(l); |
9789 | 10.2k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
9790 | 10.2k | if (buf == NULL) { |
9791 | 0 | xmlErrMemory(ctxt, NULL); |
9792 | 0 | return; |
9793 | 0 | } |
9794 | 67.6M | while (IS_CHAR(cur) && |
9795 | 67.6M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9796 | 67.6M | if (len + 5 >= size) { |
9797 | 7.95k | xmlChar *tmp; |
9798 | | |
9799 | 7.95k | if ((size > XML_MAX_TEXT_LENGTH) && |
9800 | 7.95k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
9801 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9802 | 0 | "CData section too big found", NULL); |
9803 | 0 | xmlFree (buf); |
9804 | 0 | return; |
9805 | 0 | } |
9806 | 7.95k | tmp = (xmlChar *) xmlRealloc(buf, size * 2 * sizeof(xmlChar)); |
9807 | 7.95k | if (tmp == NULL) { |
9808 | 0 | xmlFree(buf); |
9809 | 0 | xmlErrMemory(ctxt, NULL); |
9810 | 0 | return; |
9811 | 0 | } |
9812 | 7.95k | buf = tmp; |
9813 | 7.95k | size *= 2; |
9814 | 7.95k | } |
9815 | 67.6M | COPY_BUF(rl,buf,len,r); |
9816 | 67.6M | r = s; |
9817 | 67.6M | rl = sl; |
9818 | 67.6M | s = cur; |
9819 | 67.6M | sl = l; |
9820 | 67.6M | count++; |
9821 | 67.6M | if (count > 50) { |
9822 | 1.32M | SHRINK; |
9823 | 1.32M | GROW; |
9824 | 1.32M | if (ctxt->instate == XML_PARSER_EOF) { |
9825 | 0 | xmlFree(buf); |
9826 | 0 | return; |
9827 | 0 | } |
9828 | 1.32M | count = 0; |
9829 | 1.32M | } |
9830 | 67.6M | NEXTL(l); |
9831 | 67.6M | cur = CUR_CHAR(l); |
9832 | 67.6M | } |
9833 | 10.2k | buf[len] = 0; |
9834 | 10.2k | ctxt->instate = XML_PARSER_CONTENT; |
9835 | 10.2k | if (cur != '>') { |
9836 | 6.00k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9837 | 6.00k | "CData section not finished\n%.50s\n", buf); |
9838 | 6.00k | xmlFree(buf); |
9839 | 6.00k | return; |
9840 | 6.00k | } |
9841 | 4.24k | NEXTL(l); |
9842 | | |
9843 | | /* |
9844 | | * OK the buffer is to be consumed as cdata. |
9845 | | */ |
9846 | 4.24k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9847 | 942 | if (ctxt->sax->cdataBlock != NULL) |
9848 | 942 | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9849 | 0 | else if (ctxt->sax->characters != NULL) |
9850 | 0 | ctxt->sax->characters(ctxt->userData, buf, len); |
9851 | 942 | } |
9852 | 4.24k | xmlFree(buf); |
9853 | 4.24k | } |
9854 | | |
9855 | | /** |
9856 | | * xmlParseContentInternal: |
9857 | | * @ctxt: an XML parser context |
9858 | | * |
9859 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9860 | | * unexpected EOF to the caller. |
9861 | | */ |
9862 | | |
9863 | | static void |
9864 | 307k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9865 | 307k | int nameNr = ctxt->nameNr; |
9866 | | |
9867 | 307k | GROW; |
9868 | 7.62M | while ((RAW != 0) && |
9869 | 7.62M | (ctxt->instate != XML_PARSER_EOF)) { |
9870 | 7.37M | int id = ctxt->input->id; |
9871 | 7.37M | unsigned long cons = CUR_CONSUMED; |
9872 | 7.37M | const xmlChar *cur = ctxt->input->cur; |
9873 | | |
9874 | | /* |
9875 | | * First case : a Processing Instruction. |
9876 | | */ |
9877 | 7.37M | if ((*cur == '<') && (cur[1] == '?')) { |
9878 | 6.16k | xmlParsePI(ctxt); |
9879 | 6.16k | } |
9880 | | |
9881 | | /* |
9882 | | * Second case : a CDSection |
9883 | | */ |
9884 | | /* 2.6.0 test was *cur not RAW */ |
9885 | 7.37M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9886 | 13.7k | xmlParseCDSect(ctxt); |
9887 | 13.7k | } |
9888 | | |
9889 | | /* |
9890 | | * Third case : a comment |
9891 | | */ |
9892 | 7.35M | else if ((*cur == '<') && (NXT(1) == '!') && |
9893 | 7.35M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9894 | 7.66k | xmlParseComment(ctxt); |
9895 | 7.66k | ctxt->instate = XML_PARSER_CONTENT; |
9896 | 7.66k | } |
9897 | | |
9898 | | /* |
9899 | | * Fourth case : a sub-element. |
9900 | | */ |
9901 | 7.34M | else if (*cur == '<') { |
9902 | 5.98M | if (NXT(1) == '/') { |
9903 | 202k | if (ctxt->nameNr <= nameNr) |
9904 | 1.71k | break; |
9905 | 200k | xmlParseElementEnd(ctxt); |
9906 | 5.78M | } else { |
9907 | 5.78M | xmlParseElementStart(ctxt); |
9908 | 5.78M | } |
9909 | 5.98M | } |
9910 | | |
9911 | | /* |
9912 | | * Fifth case : a reference. If if has not been resolved, |
9913 | | * parsing returns it's Name, create the node |
9914 | | */ |
9915 | | |
9916 | 1.36M | else if (*cur == '&') { |
9917 | 534k | xmlParseReference(ctxt); |
9918 | 534k | } |
9919 | | |
9920 | | /* |
9921 | | * Last case, text. Note that References are handled directly. |
9922 | | */ |
9923 | 830k | else { |
9924 | 830k | xmlParseCharData(ctxt, 0); |
9925 | 830k | } |
9926 | | |
9927 | 7.37M | GROW; |
9928 | 7.37M | SHRINK; |
9929 | | |
9930 | 7.37M | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
9931 | 53.4k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9932 | 53.4k | "detected an error in element content\n"); |
9933 | 53.4k | xmlHaltParser(ctxt); |
9934 | 53.4k | break; |
9935 | 53.4k | } |
9936 | 7.37M | } |
9937 | 307k | } |
9938 | | |
9939 | | /** |
9940 | | * xmlParseContent: |
9941 | | * @ctxt: an XML parser context |
9942 | | * |
9943 | | * Parse a content sequence. Stops at EOF or '</'. |
9944 | | * |
9945 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
9946 | | */ |
9947 | | |
9948 | | void |
9949 | 300k | xmlParseContent(xmlParserCtxtPtr ctxt) { |
9950 | 300k | int nameNr = ctxt->nameNr; |
9951 | | |
9952 | 300k | xmlParseContentInternal(ctxt); |
9953 | | |
9954 | 300k | if ((ctxt->instate != XML_PARSER_EOF) && (ctxt->nameNr > nameNr)) { |
9955 | 38.1k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9956 | 38.1k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9957 | 38.1k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9958 | 38.1k | "Premature end of data in tag %s line %d\n", |
9959 | 38.1k | name, line, NULL); |
9960 | 38.1k | } |
9961 | 300k | } |
9962 | | |
9963 | | /** |
9964 | | * xmlParseElement: |
9965 | | * @ctxt: an XML parser context |
9966 | | * |
9967 | | * parse an XML element |
9968 | | * |
9969 | | * [39] element ::= EmptyElemTag | STag content ETag |
9970 | | * |
9971 | | * [ WFC: Element Type Match ] |
9972 | | * The Name in an element's end-tag must match the element type in the |
9973 | | * start-tag. |
9974 | | * |
9975 | | */ |
9976 | | |
9977 | | void |
9978 | 10.2k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
9979 | 10.2k | if (xmlParseElementStart(ctxt) != 0) |
9980 | 3.73k | return; |
9981 | | |
9982 | 6.54k | xmlParseContentInternal(ctxt); |
9983 | 6.54k | if (ctxt->instate == XML_PARSER_EOF) |
9984 | 653 | return; |
9985 | | |
9986 | 5.88k | if (CUR == 0) { |
9987 | 5.73k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9988 | 5.73k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9989 | 5.73k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9990 | 5.73k | "Premature end of data in tag %s line %d\n", |
9991 | 5.73k | name, line, NULL); |
9992 | 5.73k | return; |
9993 | 5.73k | } |
9994 | | |
9995 | 154 | xmlParseElementEnd(ctxt); |
9996 | 154 | } |
9997 | | |
9998 | | /** |
9999 | | * xmlParseElementStart: |
10000 | | * @ctxt: an XML parser context |
10001 | | * |
10002 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
10003 | | * opening tag was parsed, 1 if an empty element was parsed. |
10004 | | */ |
10005 | | static int |
10006 | 5.79M | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
10007 | 5.79M | const xmlChar *name; |
10008 | 5.79M | const xmlChar *prefix = NULL; |
10009 | 5.79M | const xmlChar *URI = NULL; |
10010 | 5.79M | xmlParserNodeInfo node_info; |
10011 | 5.79M | int line, tlen = 0; |
10012 | 5.79M | xmlNodePtr ret; |
10013 | 5.79M | int nsNr = ctxt->nsNr; |
10014 | | |
10015 | 5.79M | if (((unsigned int) ctxt->nameNr > xmlParserMaxDepth) && |
10016 | 5.79M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
10017 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
10018 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
10019 | 0 | xmlParserMaxDepth); |
10020 | 0 | xmlHaltParser(ctxt); |
10021 | 0 | return(-1); |
10022 | 0 | } |
10023 | | |
10024 | | /* Capture start position */ |
10025 | 5.79M | if (ctxt->record_info) { |
10026 | 0 | node_info.begin_pos = ctxt->input->consumed + |
10027 | 0 | (CUR_PTR - ctxt->input->base); |
10028 | 0 | node_info.begin_line = ctxt->input->line; |
10029 | 0 | } |
10030 | | |
10031 | 5.79M | if (ctxt->spaceNr == 0) |
10032 | 0 | spacePush(ctxt, -1); |
10033 | 5.79M | else if (*ctxt->space == -2) |
10034 | 0 | spacePush(ctxt, -1); |
10035 | 5.79M | else |
10036 | 5.79M | spacePush(ctxt, *ctxt->space); |
10037 | | |
10038 | 5.79M | line = ctxt->input->line; |
10039 | 5.79M | #ifdef LIBXML_SAX1_ENABLED |
10040 | 5.79M | if (ctxt->sax2) |
10041 | 5.79M | #endif /* LIBXML_SAX1_ENABLED */ |
10042 | 5.79M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
10043 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10044 | 0 | else |
10045 | 0 | name = xmlParseStartTag(ctxt); |
10046 | 5.79M | #endif /* LIBXML_SAX1_ENABLED */ |
10047 | 5.79M | if (ctxt->instate == XML_PARSER_EOF) |
10048 | 0 | return(-1); |
10049 | 5.79M | if (name == NULL) { |
10050 | 129k | spacePop(ctxt); |
10051 | 129k | return(-1); |
10052 | 129k | } |
10053 | 5.66M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
10054 | 5.66M | ret = ctxt->node; |
10055 | | |
10056 | 5.66M | #ifdef LIBXML_VALID_ENABLED |
10057 | | /* |
10058 | | * [ VC: Root Element Type ] |
10059 | | * The Name in the document type declaration must match the element |
10060 | | * type of the root element. |
10061 | | */ |
10062 | 5.66M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10063 | 5.66M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10064 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10065 | 5.66M | #endif /* LIBXML_VALID_ENABLED */ |
10066 | | |
10067 | | /* |
10068 | | * Check for an Empty Element. |
10069 | | */ |
10070 | 5.66M | if ((RAW == '/') && (NXT(1) == '>')) { |
10071 | 60.3k | SKIP(2); |
10072 | 60.3k | if (ctxt->sax2) { |
10073 | 60.3k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
10074 | 60.3k | (!ctxt->disableSAX)) |
10075 | 36.0k | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
10076 | 60.3k | #ifdef LIBXML_SAX1_ENABLED |
10077 | 60.3k | } else { |
10078 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
10079 | 0 | (!ctxt->disableSAX)) |
10080 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
10081 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10082 | 0 | } |
10083 | 60.3k | namePop(ctxt); |
10084 | 60.3k | spacePop(ctxt); |
10085 | 60.3k | if (nsNr != ctxt->nsNr) |
10086 | 14.3k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10087 | 60.3k | if ( ret != NULL && ctxt->record_info ) { |
10088 | 0 | node_info.end_pos = ctxt->input->consumed + |
10089 | 0 | (CUR_PTR - ctxt->input->base); |
10090 | 0 | node_info.end_line = ctxt->input->line; |
10091 | 0 | node_info.node = ret; |
10092 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10093 | 0 | } |
10094 | 60.3k | return(1); |
10095 | 60.3k | } |
10096 | 5.60M | if (RAW == '>') { |
10097 | 5.38M | NEXT1; |
10098 | 5.38M | } else { |
10099 | 215k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
10100 | 215k | "Couldn't find end of Start Tag %s line %d\n", |
10101 | 215k | name, line, NULL); |
10102 | | |
10103 | | /* |
10104 | | * end of parsing of this node. |
10105 | | */ |
10106 | 215k | nodePop(ctxt); |
10107 | 215k | namePop(ctxt); |
10108 | 215k | spacePop(ctxt); |
10109 | 215k | if (nsNr != ctxt->nsNr) |
10110 | 92.6k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10111 | | |
10112 | | /* |
10113 | | * Capture end position and add node |
10114 | | */ |
10115 | 215k | if ( ret != NULL && ctxt->record_info ) { |
10116 | 0 | node_info.end_pos = ctxt->input->consumed + |
10117 | 0 | (CUR_PTR - ctxt->input->base); |
10118 | 0 | node_info.end_line = ctxt->input->line; |
10119 | 0 | node_info.node = ret; |
10120 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10121 | 0 | } |
10122 | 215k | return(-1); |
10123 | 215k | } |
10124 | | |
10125 | 5.38M | return(0); |
10126 | 5.60M | } |
10127 | | |
10128 | | /** |
10129 | | * xmlParseElementEnd: |
10130 | | * @ctxt: an XML parser context |
10131 | | * |
10132 | | * Parse the end of an XML element. |
10133 | | */ |
10134 | | static void |
10135 | 201k | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
10136 | 201k | xmlParserNodeInfo node_info; |
10137 | 201k | xmlNodePtr ret = ctxt->node; |
10138 | | |
10139 | 201k | if (ctxt->nameNr <= 0) |
10140 | 0 | return; |
10141 | | |
10142 | | /* |
10143 | | * parse the end of tag: '</' should be here. |
10144 | | */ |
10145 | 201k | if (ctxt->sax2) { |
10146 | 201k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
10147 | 201k | namePop(ctxt); |
10148 | 201k | } |
10149 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10150 | 0 | else |
10151 | 0 | xmlParseEndTag1(ctxt, 0); |
10152 | 201k | #endif /* LIBXML_SAX1_ENABLED */ |
10153 | | |
10154 | | /* |
10155 | | * Capture end position and add node |
10156 | | */ |
10157 | 201k | if ( ret != NULL && ctxt->record_info ) { |
10158 | 0 | node_info.end_pos = ctxt->input->consumed + |
10159 | 0 | (CUR_PTR - ctxt->input->base); |
10160 | 0 | node_info.end_line = ctxt->input->line; |
10161 | 0 | node_info.node = ret; |
10162 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10163 | 0 | } |
10164 | 201k | } |
10165 | | |
10166 | | /** |
10167 | | * xmlParseVersionNum: |
10168 | | * @ctxt: an XML parser context |
10169 | | * |
10170 | | * parse the XML version value. |
10171 | | * |
10172 | | * [26] VersionNum ::= '1.' [0-9]+ |
10173 | | * |
10174 | | * In practice allow [0-9].[0-9]+ at that level |
10175 | | * |
10176 | | * Returns the string giving the XML version number, or NULL |
10177 | | */ |
10178 | | xmlChar * |
10179 | 2.58k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10180 | 2.58k | xmlChar *buf = NULL; |
10181 | 2.58k | int len = 0; |
10182 | 2.58k | int size = 10; |
10183 | 2.58k | xmlChar cur; |
10184 | | |
10185 | 2.58k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
10186 | 2.58k | if (buf == NULL) { |
10187 | 0 | xmlErrMemory(ctxt, NULL); |
10188 | 0 | return(NULL); |
10189 | 0 | } |
10190 | 2.58k | cur = CUR; |
10191 | 2.58k | if (!((cur >= '0') && (cur <= '9'))) { |
10192 | 19 | xmlFree(buf); |
10193 | 19 | return(NULL); |
10194 | 19 | } |
10195 | 2.56k | buf[len++] = cur; |
10196 | 2.56k | NEXT; |
10197 | 2.56k | cur=CUR; |
10198 | 2.56k | if (cur != '.') { |
10199 | 5 | xmlFree(buf); |
10200 | 5 | return(NULL); |
10201 | 5 | } |
10202 | 2.55k | buf[len++] = cur; |
10203 | 2.55k | NEXT; |
10204 | 2.55k | cur=CUR; |
10205 | 159k | while ((cur >= '0') && (cur <= '9')) { |
10206 | 157k | if (len + 1 >= size) { |
10207 | 163 | xmlChar *tmp; |
10208 | | |
10209 | 163 | size *= 2; |
10210 | 163 | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
10211 | 163 | if (tmp == NULL) { |
10212 | 0 | xmlFree(buf); |
10213 | 0 | xmlErrMemory(ctxt, NULL); |
10214 | 0 | return(NULL); |
10215 | 0 | } |
10216 | 163 | buf = tmp; |
10217 | 163 | } |
10218 | 157k | buf[len++] = cur; |
10219 | 157k | NEXT; |
10220 | 157k | cur=CUR; |
10221 | 157k | } |
10222 | 2.55k | buf[len] = 0; |
10223 | 2.55k | return(buf); |
10224 | 2.55k | } |
10225 | | |
10226 | | /** |
10227 | | * xmlParseVersionInfo: |
10228 | | * @ctxt: an XML parser context |
10229 | | * |
10230 | | * parse the XML version. |
10231 | | * |
10232 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10233 | | * |
10234 | | * [25] Eq ::= S? '=' S? |
10235 | | * |
10236 | | * Returns the version string, e.g. "1.0" |
10237 | | */ |
10238 | | |
10239 | | xmlChar * |
10240 | 4.65k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10241 | 4.65k | xmlChar *version = NULL; |
10242 | | |
10243 | 4.65k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10244 | 2.62k | SKIP(7); |
10245 | 2.62k | SKIP_BLANKS; |
10246 | 2.62k | if (RAW != '=') { |
10247 | 13 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10248 | 13 | return(NULL); |
10249 | 13 | } |
10250 | 2.61k | NEXT; |
10251 | 2.61k | SKIP_BLANKS; |
10252 | 2.61k | if (RAW == '"') { |
10253 | 2.11k | NEXT; |
10254 | 2.11k | version = xmlParseVersionNum(ctxt); |
10255 | 2.11k | if (RAW != '"') { |
10256 | 47 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10257 | 47 | } else |
10258 | 2.07k | NEXT; |
10259 | 2.11k | } else if (RAW == '\''){ |
10260 | 464 | NEXT; |
10261 | 464 | version = xmlParseVersionNum(ctxt); |
10262 | 464 | if (RAW != '\'') { |
10263 | 6 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10264 | 6 | } else |
10265 | 458 | NEXT; |
10266 | 464 | } else { |
10267 | 33 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10268 | 33 | } |
10269 | 2.61k | } |
10270 | 4.63k | return(version); |
10271 | 4.65k | } |
10272 | | |
10273 | | /** |
10274 | | * xmlParseEncName: |
10275 | | * @ctxt: an XML parser context |
10276 | | * |
10277 | | * parse the XML encoding name |
10278 | | * |
10279 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10280 | | * |
10281 | | * Returns the encoding name value or NULL |
10282 | | */ |
10283 | | xmlChar * |
10284 | 4.16k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10285 | 4.16k | xmlChar *buf = NULL; |
10286 | 4.16k | int len = 0; |
10287 | 4.16k | int size = 10; |
10288 | 4.16k | xmlChar cur; |
10289 | | |
10290 | 4.16k | cur = CUR; |
10291 | 4.16k | if (((cur >= 'a') && (cur <= 'z')) || |
10292 | 4.16k | ((cur >= 'A') && (cur <= 'Z'))) { |
10293 | 4.15k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
10294 | 4.15k | if (buf == NULL) { |
10295 | 0 | xmlErrMemory(ctxt, NULL); |
10296 | 0 | return(NULL); |
10297 | 0 | } |
10298 | | |
10299 | 4.15k | buf[len++] = cur; |
10300 | 4.15k | NEXT; |
10301 | 4.15k | cur = CUR; |
10302 | 696k | while (((cur >= 'a') && (cur <= 'z')) || |
10303 | 696k | ((cur >= 'A') && (cur <= 'Z')) || |
10304 | 696k | ((cur >= '0') && (cur <= '9')) || |
10305 | 696k | (cur == '.') || (cur == '_') || |
10306 | 696k | (cur == '-')) { |
10307 | 692k | if (len + 1 >= size) { |
10308 | 575 | xmlChar *tmp; |
10309 | | |
10310 | 575 | size *= 2; |
10311 | 575 | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
10312 | 575 | if (tmp == NULL) { |
10313 | 0 | xmlErrMemory(ctxt, NULL); |
10314 | 0 | xmlFree(buf); |
10315 | 0 | return(NULL); |
10316 | 0 | } |
10317 | 575 | buf = tmp; |
10318 | 575 | } |
10319 | 692k | buf[len++] = cur; |
10320 | 692k | NEXT; |
10321 | 692k | cur = CUR; |
10322 | 692k | if (cur == 0) { |
10323 | 64 | SHRINK; |
10324 | 64 | GROW; |
10325 | 64 | cur = CUR; |
10326 | 64 | } |
10327 | 692k | } |
10328 | 4.15k | buf[len] = 0; |
10329 | 4.15k | } else { |
10330 | 5 | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10331 | 5 | } |
10332 | 4.16k | return(buf); |
10333 | 4.16k | } |
10334 | | |
10335 | | /** |
10336 | | * xmlParseEncodingDecl: |
10337 | | * @ctxt: an XML parser context |
10338 | | * |
10339 | | * parse the XML encoding declaration |
10340 | | * |
10341 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10342 | | * |
10343 | | * this setups the conversion filters. |
10344 | | * |
10345 | | * Returns the encoding value or NULL |
10346 | | */ |
10347 | | |
10348 | | const xmlChar * |
10349 | 4.63k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10350 | 4.63k | xmlChar *encoding = NULL; |
10351 | | |
10352 | 4.63k | SKIP_BLANKS; |
10353 | 4.63k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g')) { |
10354 | 4.18k | SKIP(8); |
10355 | 4.18k | SKIP_BLANKS; |
10356 | 4.18k | if (RAW != '=') { |
10357 | 9 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10358 | 9 | return(NULL); |
10359 | 9 | } |
10360 | 4.17k | NEXT; |
10361 | 4.17k | SKIP_BLANKS; |
10362 | 4.17k | if (RAW == '"') { |
10363 | 3.93k | NEXT; |
10364 | 3.93k | encoding = xmlParseEncName(ctxt); |
10365 | 3.93k | if (RAW != '"') { |
10366 | 86 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10367 | 86 | xmlFree((xmlChar *) encoding); |
10368 | 86 | return(NULL); |
10369 | 86 | } else |
10370 | 3.84k | NEXT; |
10371 | 3.93k | } else if (RAW == '\''){ |
10372 | 233 | NEXT; |
10373 | 233 | encoding = xmlParseEncName(ctxt); |
10374 | 233 | if (RAW != '\'') { |
10375 | 6 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10376 | 6 | xmlFree((xmlChar *) encoding); |
10377 | 6 | return(NULL); |
10378 | 6 | } else |
10379 | 227 | NEXT; |
10380 | 233 | } else { |
10381 | 13 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10382 | 13 | } |
10383 | | |
10384 | | /* |
10385 | | * Non standard parsing, allowing the user to ignore encoding |
10386 | | */ |
10387 | 4.08k | if (ctxt->options & XML_PARSE_IGNORE_ENC) { |
10388 | 0 | xmlFree((xmlChar *) encoding); |
10389 | 0 | return(NULL); |
10390 | 0 | } |
10391 | | |
10392 | | /* |
10393 | | * UTF-16 encoding switch has already taken place at this stage, |
10394 | | * more over the little-endian/big-endian selection is already done |
10395 | | */ |
10396 | 4.08k | if ((encoding != NULL) && |
10397 | 4.08k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-16")) || |
10398 | 4.07k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF16")))) { |
10399 | | /* |
10400 | | * If no encoding was passed to the parser, that we are |
10401 | | * using UTF-16 and no decoder is present i.e. the |
10402 | | * document is apparently UTF-8 compatible, then raise an |
10403 | | * encoding mismatch fatal error |
10404 | | */ |
10405 | 2 | if ((ctxt->encoding == NULL) && |
10406 | 2 | (ctxt->input->buf != NULL) && |
10407 | 2 | (ctxt->input->buf->encoder == NULL)) { |
10408 | 2 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_ENCODING, |
10409 | 2 | "Document labelled UTF-16 but has UTF-8 content\n"); |
10410 | 2 | } |
10411 | 2 | if (ctxt->encoding != NULL) |
10412 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10413 | 2 | ctxt->encoding = encoding; |
10414 | 2 | } |
10415 | | /* |
10416 | | * UTF-8 encoding is handled natively |
10417 | | */ |
10418 | 4.08k | else if ((encoding != NULL) && |
10419 | 4.08k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-8")) || |
10420 | 4.06k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF8")))) { |
10421 | 6 | if (ctxt->encoding != NULL) |
10422 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10423 | 6 | ctxt->encoding = encoding; |
10424 | 6 | } |
10425 | 4.07k | else if (encoding != NULL) { |
10426 | 4.06k | xmlCharEncodingHandlerPtr handler; |
10427 | | |
10428 | 4.06k | if (ctxt->input->encoding != NULL) |
10429 | 0 | xmlFree((xmlChar *) ctxt->input->encoding); |
10430 | 4.06k | ctxt->input->encoding = encoding; |
10431 | | |
10432 | 4.06k | handler = xmlFindCharEncodingHandler((const char *) encoding); |
10433 | 4.06k | if (handler != NULL) { |
10434 | 3.65k | if (xmlSwitchToEncoding(ctxt, handler) < 0) { |
10435 | | /* failed to convert */ |
10436 | 1 | ctxt->errNo = XML_ERR_UNSUPPORTED_ENCODING; |
10437 | 1 | return(NULL); |
10438 | 1 | } |
10439 | 3.65k | } else { |
10440 | 412 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
10441 | 412 | "Unsupported encoding %s\n", encoding); |
10442 | 412 | return(NULL); |
10443 | 412 | } |
10444 | 4.06k | } |
10445 | 4.08k | } |
10446 | 4.11k | return(encoding); |
10447 | 4.63k | } |
10448 | | |
10449 | | /** |
10450 | | * xmlParseSDDecl: |
10451 | | * @ctxt: an XML parser context |
10452 | | * |
10453 | | * parse the XML standalone declaration |
10454 | | * |
10455 | | * [32] SDDecl ::= S 'standalone' Eq |
10456 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10457 | | * |
10458 | | * [ VC: Standalone Document Declaration ] |
10459 | | * TODO The standalone document declaration must have the value "no" |
10460 | | * if any external markup declarations contain declarations of: |
10461 | | * - attributes with default values, if elements to which these |
10462 | | * attributes apply appear in the document without specifications |
10463 | | * of values for these attributes, or |
10464 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10465 | | * to those entities appear in the document, or |
10466 | | * - attributes with values subject to normalization, where the |
10467 | | * attribute appears in the document with a value which will change |
10468 | | * as a result of normalization, or |
10469 | | * - element types with element content, if white space occurs directly |
10470 | | * within any instance of those types. |
10471 | | * |
10472 | | * Returns: |
10473 | | * 1 if standalone="yes" |
10474 | | * 0 if standalone="no" |
10475 | | * -2 if standalone attribute is missing or invalid |
10476 | | * (A standalone value of -2 means that the XML declaration was found, |
10477 | | * but no value was specified for the standalone attribute). |
10478 | | */ |
10479 | | |
10480 | | int |
10481 | 1.90k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10482 | 1.90k | int standalone = -2; |
10483 | | |
10484 | 1.90k | SKIP_BLANKS; |
10485 | 1.90k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10486 | 210 | SKIP(10); |
10487 | 210 | SKIP_BLANKS; |
10488 | 210 | if (RAW != '=') { |
10489 | 1 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10490 | 1 | return(standalone); |
10491 | 1 | } |
10492 | 209 | NEXT; |
10493 | 209 | SKIP_BLANKS; |
10494 | 209 | if (RAW == '\''){ |
10495 | 164 | NEXT; |
10496 | 164 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10497 | 138 | standalone = 0; |
10498 | 138 | SKIP(2); |
10499 | 138 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10500 | 26 | (NXT(2) == 's')) { |
10501 | 16 | standalone = 1; |
10502 | 16 | SKIP(3); |
10503 | 16 | } else { |
10504 | 10 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10505 | 10 | } |
10506 | 164 | if (RAW != '\'') { |
10507 | 27 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10508 | 27 | } else |
10509 | 137 | NEXT; |
10510 | 164 | } else if (RAW == '"'){ |
10511 | 44 | NEXT; |
10512 | 44 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10513 | 2 | standalone = 0; |
10514 | 2 | SKIP(2); |
10515 | 42 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10516 | 42 | (NXT(2) == 's')) { |
10517 | 30 | standalone = 1; |
10518 | 30 | SKIP(3); |
10519 | 30 | } else { |
10520 | 12 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10521 | 12 | } |
10522 | 44 | if (RAW != '"') { |
10523 | 18 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10524 | 18 | } else |
10525 | 26 | NEXT; |
10526 | 44 | } else { |
10527 | 1 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10528 | 1 | } |
10529 | 209 | } |
10530 | 1.89k | return(standalone); |
10531 | 1.90k | } |
10532 | | |
10533 | | /** |
10534 | | * xmlParseXMLDecl: |
10535 | | * @ctxt: an XML parser context |
10536 | | * |
10537 | | * parse an XML declaration header |
10538 | | * |
10539 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10540 | | */ |
10541 | | |
10542 | | void |
10543 | 4.65k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10544 | 4.65k | xmlChar *version; |
10545 | | |
10546 | | /* |
10547 | | * This value for standalone indicates that the document has an |
10548 | | * XML declaration but it does not have a standalone attribute. |
10549 | | * It will be overwritten later if a standalone attribute is found. |
10550 | | */ |
10551 | 4.65k | ctxt->input->standalone = -2; |
10552 | | |
10553 | | /* |
10554 | | * We know that '<?xml' is here. |
10555 | | */ |
10556 | 4.65k | SKIP(5); |
10557 | | |
10558 | 4.65k | if (!IS_BLANK_CH(RAW)) { |
10559 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10560 | 0 | "Blank needed after '<?xml'\n"); |
10561 | 0 | } |
10562 | 4.65k | SKIP_BLANKS; |
10563 | | |
10564 | | /* |
10565 | | * We must have the VersionInfo here. |
10566 | | */ |
10567 | 4.65k | version = xmlParseVersionInfo(ctxt); |
10568 | 4.65k | if (version == NULL) { |
10569 | 2.09k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10570 | 2.55k | } else { |
10571 | 2.55k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10572 | | /* |
10573 | | * Changed here for XML-1.0 5th edition |
10574 | | */ |
10575 | 2.01k | if (ctxt->options & XML_PARSE_OLD10) { |
10576 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10577 | 0 | "Unsupported version '%s'\n", |
10578 | 0 | version); |
10579 | 2.01k | } else { |
10580 | 2.01k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10581 | 1.90k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10582 | 1.90k | "Unsupported version '%s'\n", |
10583 | 1.90k | version, NULL); |
10584 | 1.90k | } else { |
10585 | 105 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10586 | 105 | "Unsupported version '%s'\n", |
10587 | 105 | version); |
10588 | 105 | } |
10589 | 2.01k | } |
10590 | 2.01k | } |
10591 | 2.55k | if (ctxt->version != NULL) |
10592 | 0 | xmlFree((void *) ctxt->version); |
10593 | 2.55k | ctxt->version = version; |
10594 | 2.55k | } |
10595 | | |
10596 | | /* |
10597 | | * We may have the encoding declaration |
10598 | | */ |
10599 | 4.65k | if (!IS_BLANK_CH(RAW)) { |
10600 | 2.13k | if ((RAW == '?') && (NXT(1) == '>')) { |
10601 | 19 | SKIP(2); |
10602 | 19 | return; |
10603 | 19 | } |
10604 | 2.11k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10605 | 2.11k | } |
10606 | 4.63k | xmlParseEncodingDecl(ctxt); |
10607 | 4.63k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10608 | 4.63k | (ctxt->instate == XML_PARSER_EOF)) { |
10609 | | /* |
10610 | | * The XML REC instructs us to stop parsing right here |
10611 | | */ |
10612 | 413 | return; |
10613 | 413 | } |
10614 | | |
10615 | | /* |
10616 | | * We may have the standalone status. |
10617 | | */ |
10618 | 4.21k | if ((ctxt->input->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10619 | 3.48k | if ((RAW == '?') && (NXT(1) == '>')) { |
10620 | 2.31k | SKIP(2); |
10621 | 2.31k | return; |
10622 | 2.31k | } |
10623 | 1.16k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10624 | 1.16k | } |
10625 | | |
10626 | | /* |
10627 | | * We can grow the input buffer freely at that point |
10628 | | */ |
10629 | 1.90k | GROW; |
10630 | | |
10631 | 1.90k | SKIP_BLANKS; |
10632 | 1.90k | ctxt->input->standalone = xmlParseSDDecl(ctxt); |
10633 | | |
10634 | 1.90k | SKIP_BLANKS; |
10635 | 1.90k | if ((RAW == '?') && (NXT(1) == '>')) { |
10636 | 163 | SKIP(2); |
10637 | 1.73k | } else if (RAW == '>') { |
10638 | | /* Deprecated old WD ... */ |
10639 | 637 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10640 | 637 | NEXT; |
10641 | 1.10k | } else { |
10642 | 1.10k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10643 | 1.10k | MOVETO_ENDTAG(CUR_PTR); |
10644 | 1.10k | NEXT; |
10645 | 1.10k | } |
10646 | 1.90k | } |
10647 | | |
10648 | | /** |
10649 | | * xmlParseMisc: |
10650 | | * @ctxt: an XML parser context |
10651 | | * |
10652 | | * parse an XML Misc* optional field. |
10653 | | * |
10654 | | * [27] Misc ::= Comment | PI | S |
10655 | | */ |
10656 | | |
10657 | | void |
10658 | 37.3k | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10659 | 45.6k | while (ctxt->instate != XML_PARSER_EOF) { |
10660 | 45.6k | SKIP_BLANKS; |
10661 | 45.6k | GROW; |
10662 | 45.6k | if ((RAW == '<') && (NXT(1) == '?')) { |
10663 | 7.06k | xmlParsePI(ctxt); |
10664 | 38.5k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10665 | 1.16k | xmlParseComment(ctxt); |
10666 | 37.3k | } else { |
10667 | 37.3k | break; |
10668 | 37.3k | } |
10669 | 45.6k | } |
10670 | 37.3k | } |
10671 | | |
10672 | | /** |
10673 | | * xmlParseDocument: |
10674 | | * @ctxt: an XML parser context |
10675 | | * |
10676 | | * parse an XML document (and build a tree if using the standard SAX |
10677 | | * interface). |
10678 | | * |
10679 | | * [1] document ::= prolog element Misc* |
10680 | | * |
10681 | | * [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? |
10682 | | * |
10683 | | * Returns 0, -1 in case of error. the parser context is augmented |
10684 | | * as a result of the parsing. |
10685 | | */ |
10686 | | |
10687 | | int |
10688 | 18.9k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10689 | 18.9k | xmlChar start[4]; |
10690 | 18.9k | xmlCharEncoding enc; |
10691 | | |
10692 | 18.9k | xmlInitParser(); |
10693 | | |
10694 | 18.9k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10695 | 0 | return(-1); |
10696 | | |
10697 | 18.9k | GROW; |
10698 | | |
10699 | | /* |
10700 | | * SAX: detecting the level. |
10701 | | */ |
10702 | 18.9k | xmlDetectSAX2(ctxt); |
10703 | | |
10704 | | /* |
10705 | | * SAX: beginning of the document processing. |
10706 | | */ |
10707 | 18.9k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10708 | 18.9k | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10709 | 18.9k | if (ctxt->instate == XML_PARSER_EOF) |
10710 | 0 | return(-1); |
10711 | | |
10712 | 18.9k | if ((ctxt->encoding == NULL) && |
10713 | 18.9k | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
10714 | | /* |
10715 | | * Get the 4 first bytes and decode the charset |
10716 | | * if enc != XML_CHAR_ENCODING_NONE |
10717 | | * plug some encoding conversion routines. |
10718 | | */ |
10719 | 18.6k | start[0] = RAW; |
10720 | 18.6k | start[1] = NXT(1); |
10721 | 18.6k | start[2] = NXT(2); |
10722 | 18.6k | start[3] = NXT(3); |
10723 | 18.6k | enc = xmlDetectCharEncoding(&start[0], 4); |
10724 | 18.6k | if (enc != XML_CHAR_ENCODING_NONE) { |
10725 | 5.39k | xmlSwitchEncoding(ctxt, enc); |
10726 | 5.39k | } |
10727 | 18.6k | } |
10728 | | |
10729 | | |
10730 | 18.9k | if (CUR == 0) { |
10731 | 60 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10732 | 60 | return(-1); |
10733 | 60 | } |
10734 | | |
10735 | | /* |
10736 | | * Check for the XMLDecl in the Prolog. |
10737 | | * do not GROW here to avoid the detected encoder to decode more |
10738 | | * than just the first line, unless the amount of data is really |
10739 | | * too small to hold "<?xml version="1.0" encoding="foo" |
10740 | | */ |
10741 | 18.9k | if ((ctxt->input->end - ctxt->input->cur) < 35) { |
10742 | 6.93k | GROW; |
10743 | 6.93k | } |
10744 | 18.9k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10745 | | |
10746 | | /* |
10747 | | * Note that we will switch encoding on the fly. |
10748 | | */ |
10749 | 4.65k | xmlParseXMLDecl(ctxt); |
10750 | 4.65k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10751 | 4.65k | (ctxt->instate == XML_PARSER_EOF)) { |
10752 | | /* |
10753 | | * The XML REC instructs us to stop parsing right here |
10754 | | */ |
10755 | 413 | return(-1); |
10756 | 413 | } |
10757 | 4.23k | ctxt->standalone = ctxt->input->standalone; |
10758 | 4.23k | SKIP_BLANKS; |
10759 | 14.2k | } else { |
10760 | 14.2k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10761 | 14.2k | } |
10762 | 18.5k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10763 | 16.6k | ctxt->sax->startDocument(ctxt->userData); |
10764 | 18.5k | if (ctxt->instate == XML_PARSER_EOF) |
10765 | 0 | return(-1); |
10766 | 18.5k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10767 | 18.5k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10768 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10769 | 0 | } |
10770 | | |
10771 | | /* |
10772 | | * The Misc part of the Prolog |
10773 | | */ |
10774 | 18.5k | xmlParseMisc(ctxt); |
10775 | | |
10776 | | /* |
10777 | | * Then possibly doc type declaration(s) and more Misc |
10778 | | * (doctypedecl Misc*)? |
10779 | | */ |
10780 | 18.5k | GROW; |
10781 | 18.5k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10782 | | |
10783 | 9.57k | ctxt->inSubset = 1; |
10784 | 9.57k | xmlParseDocTypeDecl(ctxt); |
10785 | 9.57k | if (RAW == '[') { |
10786 | 9.13k | ctxt->instate = XML_PARSER_DTD; |
10787 | 9.13k | xmlParseInternalSubset(ctxt); |
10788 | 9.13k | if (ctxt->instate == XML_PARSER_EOF) |
10789 | 984 | return(-1); |
10790 | 9.13k | } |
10791 | | |
10792 | | /* |
10793 | | * Create and update the external subset. |
10794 | | */ |
10795 | 8.59k | ctxt->inSubset = 2; |
10796 | 8.59k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10797 | 8.59k | (!ctxt->disableSAX)) |
10798 | 1.94k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10799 | 1.94k | ctxt->extSubSystem, ctxt->extSubURI); |
10800 | 8.59k | if (ctxt->instate == XML_PARSER_EOF) |
10801 | 0 | return(-1); |
10802 | 8.59k | ctxt->inSubset = 0; |
10803 | | |
10804 | 8.59k | xmlCleanSpecialAttr(ctxt); |
10805 | | |
10806 | 8.59k | ctxt->instate = XML_PARSER_PROLOG; |
10807 | 8.59k | xmlParseMisc(ctxt); |
10808 | 8.59k | } |
10809 | | |
10810 | | /* |
10811 | | * Time to start parsing the tree itself |
10812 | | */ |
10813 | 17.5k | GROW; |
10814 | 17.5k | if (RAW != '<') { |
10815 | 7.26k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10816 | 7.26k | "Start tag expected, '<' not found\n"); |
10817 | 10.2k | } else { |
10818 | 10.2k | ctxt->instate = XML_PARSER_CONTENT; |
10819 | 10.2k | xmlParseElement(ctxt); |
10820 | 10.2k | ctxt->instate = XML_PARSER_EPILOG; |
10821 | | |
10822 | | |
10823 | | /* |
10824 | | * The Misc part at the end |
10825 | | */ |
10826 | 10.2k | xmlParseMisc(ctxt); |
10827 | | |
10828 | 10.2k | if (RAW != 0) { |
10829 | 1.27k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10830 | 1.27k | } |
10831 | 10.2k | ctxt->instate = XML_PARSER_EOF; |
10832 | 10.2k | } |
10833 | | |
10834 | | /* |
10835 | | * SAX: end of the document processing. |
10836 | | */ |
10837 | 17.5k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10838 | 17.5k | ctxt->sax->endDocument(ctxt->userData); |
10839 | | |
10840 | | /* |
10841 | | * Remove locally kept entity definitions if the tree was not built |
10842 | | */ |
10843 | 17.5k | if ((ctxt->myDoc != NULL) && |
10844 | 17.5k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
10845 | 177 | xmlFreeDoc(ctxt->myDoc); |
10846 | 177 | ctxt->myDoc = NULL; |
10847 | 177 | } |
10848 | | |
10849 | 17.5k | if ((ctxt->wellFormed) && (ctxt->myDoc != NULL)) { |
10850 | 9 | ctxt->myDoc->properties |= XML_DOC_WELLFORMED; |
10851 | 9 | if (ctxt->valid) |
10852 | 8 | ctxt->myDoc->properties |= XML_DOC_DTDVALID; |
10853 | 9 | if (ctxt->nsWellFormed) |
10854 | 7 | ctxt->myDoc->properties |= XML_DOC_NSVALID; |
10855 | 9 | if (ctxt->options & XML_PARSE_OLD10) |
10856 | 0 | ctxt->myDoc->properties |= XML_DOC_OLD10; |
10857 | 9 | } |
10858 | 17.5k | if (! ctxt->wellFormed) { |
10859 | 17.5k | ctxt->valid = 0; |
10860 | 17.5k | return(-1); |
10861 | 17.5k | } |
10862 | 9 | return(0); |
10863 | 17.5k | } |
10864 | | |
10865 | | /** |
10866 | | * xmlParseExtParsedEnt: |
10867 | | * @ctxt: an XML parser context |
10868 | | * |
10869 | | * parse a general parsed entity |
10870 | | * An external general parsed entity is well-formed if it matches the |
10871 | | * production labeled extParsedEnt. |
10872 | | * |
10873 | | * [78] extParsedEnt ::= TextDecl? content |
10874 | | * |
10875 | | * Returns 0, -1 in case of error. the parser context is augmented |
10876 | | * as a result of the parsing. |
10877 | | */ |
10878 | | |
10879 | | int |
10880 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10881 | 0 | xmlChar start[4]; |
10882 | 0 | xmlCharEncoding enc; |
10883 | |
|
10884 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10885 | 0 | return(-1); |
10886 | | |
10887 | 0 | xmlDetectSAX2(ctxt); |
10888 | |
|
10889 | 0 | GROW; |
10890 | | |
10891 | | /* |
10892 | | * SAX: beginning of the document processing. |
10893 | | */ |
10894 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10895 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10896 | | |
10897 | | /* |
10898 | | * Get the 4 first bytes and decode the charset |
10899 | | * if enc != XML_CHAR_ENCODING_NONE |
10900 | | * plug some encoding conversion routines. |
10901 | | */ |
10902 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
10903 | 0 | start[0] = RAW; |
10904 | 0 | start[1] = NXT(1); |
10905 | 0 | start[2] = NXT(2); |
10906 | 0 | start[3] = NXT(3); |
10907 | 0 | enc = xmlDetectCharEncoding(start, 4); |
10908 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
10909 | 0 | xmlSwitchEncoding(ctxt, enc); |
10910 | 0 | } |
10911 | 0 | } |
10912 | | |
10913 | |
|
10914 | 0 | if (CUR == 0) { |
10915 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10916 | 0 | } |
10917 | | |
10918 | | /* |
10919 | | * Check for the XMLDecl in the Prolog. |
10920 | | */ |
10921 | 0 | GROW; |
10922 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10923 | | |
10924 | | /* |
10925 | | * Note that we will switch encoding on the fly. |
10926 | | */ |
10927 | 0 | xmlParseXMLDecl(ctxt); |
10928 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
10929 | | /* |
10930 | | * The XML REC instructs us to stop parsing right here |
10931 | | */ |
10932 | 0 | return(-1); |
10933 | 0 | } |
10934 | 0 | SKIP_BLANKS; |
10935 | 0 | } else { |
10936 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10937 | 0 | } |
10938 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10939 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10940 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10941 | 0 | return(-1); |
10942 | | |
10943 | | /* |
10944 | | * Doing validity checking on chunk doesn't make sense |
10945 | | */ |
10946 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
10947 | 0 | ctxt->validate = 0; |
10948 | 0 | ctxt->loadsubset = 0; |
10949 | 0 | ctxt->depth = 0; |
10950 | |
|
10951 | 0 | xmlParseContent(ctxt); |
10952 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10953 | 0 | return(-1); |
10954 | | |
10955 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
10956 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10957 | 0 | } else if (RAW != 0) { |
10958 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
10959 | 0 | } |
10960 | | |
10961 | | /* |
10962 | | * SAX: end of the document processing. |
10963 | | */ |
10964 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10965 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10966 | |
|
10967 | 0 | if (! ctxt->wellFormed) return(-1); |
10968 | 0 | return(0); |
10969 | 0 | } |
10970 | | |
10971 | | #ifdef LIBXML_PUSH_ENABLED |
10972 | | /************************************************************************ |
10973 | | * * |
10974 | | * Progressive parsing interfaces * |
10975 | | * * |
10976 | | ************************************************************************/ |
10977 | | |
10978 | | /** |
10979 | | * xmlParseLookupSequence: |
10980 | | * @ctxt: an XML parser context |
10981 | | * @first: the first char to lookup |
10982 | | * @next: the next char to lookup or zero |
10983 | | * @third: the next char to lookup or zero |
10984 | | * |
10985 | | * Try to find if a sequence (first, next, third) or just (first next) or |
10986 | | * (first) is available in the input stream. |
10987 | | * This function has a side effect of (possibly) incrementing ctxt->checkIndex |
10988 | | * to avoid rescanning sequences of bytes, it DOES change the state of the |
10989 | | * parser, do not use liberally. |
10990 | | * |
10991 | | * Returns the index to the current parsing point if the full sequence |
10992 | | * is available, -1 otherwise. |
10993 | | */ |
10994 | | static int |
10995 | | xmlParseLookupSequence(xmlParserCtxtPtr ctxt, xmlChar first, |
10996 | 0 | xmlChar next, xmlChar third) { |
10997 | 0 | int base, len; |
10998 | 0 | xmlParserInputPtr in; |
10999 | 0 | const xmlChar *buf; |
11000 | |
|
11001 | 0 | in = ctxt->input; |
11002 | 0 | if (in == NULL) return(-1); |
11003 | 0 | base = in->cur - in->base; |
11004 | 0 | if (base < 0) return(-1); |
11005 | 0 | if (ctxt->checkIndex > base) |
11006 | 0 | base = ctxt->checkIndex; |
11007 | 0 | if (in->buf == NULL) { |
11008 | 0 | buf = in->base; |
11009 | 0 | len = in->length; |
11010 | 0 | } else { |
11011 | 0 | buf = xmlBufContent(in->buf->buffer); |
11012 | 0 | len = xmlBufUse(in->buf->buffer); |
11013 | 0 | } |
11014 | | /* take into account the sequence length */ |
11015 | 0 | if (third) len -= 2; |
11016 | 0 | else if (next) len --; |
11017 | 0 | for (;base < len;base++) { |
11018 | 0 | if (buf[base] == first) { |
11019 | 0 | if (third != 0) { |
11020 | 0 | if ((buf[base + 1] != next) || |
11021 | 0 | (buf[base + 2] != third)) continue; |
11022 | 0 | } else if (next != 0) { |
11023 | 0 | if (buf[base + 1] != next) continue; |
11024 | 0 | } |
11025 | 0 | ctxt->checkIndex = 0; |
11026 | | #ifdef DEBUG_PUSH |
11027 | | if (next == 0) |
11028 | | xmlGenericError(xmlGenericErrorContext, |
11029 | | "PP: lookup '%c' found at %d\n", |
11030 | | first, base); |
11031 | | else if (third == 0) |
11032 | | xmlGenericError(xmlGenericErrorContext, |
11033 | | "PP: lookup '%c%c' found at %d\n", |
11034 | | first, next, base); |
11035 | | else |
11036 | | xmlGenericError(xmlGenericErrorContext, |
11037 | | "PP: lookup '%c%c%c' found at %d\n", |
11038 | | first, next, third, base); |
11039 | | #endif |
11040 | 0 | return(base - (in->cur - in->base)); |
11041 | 0 | } |
11042 | 0 | } |
11043 | 0 | ctxt->checkIndex = base; |
11044 | | #ifdef DEBUG_PUSH |
11045 | | if (next == 0) |
11046 | | xmlGenericError(xmlGenericErrorContext, |
11047 | | "PP: lookup '%c' failed\n", first); |
11048 | | else if (third == 0) |
11049 | | xmlGenericError(xmlGenericErrorContext, |
11050 | | "PP: lookup '%c%c' failed\n", first, next); |
11051 | | else |
11052 | | xmlGenericError(xmlGenericErrorContext, |
11053 | | "PP: lookup '%c%c%c' failed\n", first, next, third); |
11054 | | #endif |
11055 | 0 | return(-1); |
11056 | 0 | } |
11057 | | |
11058 | | /** |
11059 | | * xmlParseGetLasts: |
11060 | | * @ctxt: an XML parser context |
11061 | | * @lastlt: pointer to store the last '<' from the input |
11062 | | * @lastgt: pointer to store the last '>' from the input |
11063 | | * |
11064 | | * Lookup the last < and > in the current chunk |
11065 | | */ |
11066 | | static void |
11067 | | xmlParseGetLasts(xmlParserCtxtPtr ctxt, const xmlChar **lastlt, |
11068 | 0 | const xmlChar **lastgt) { |
11069 | 0 | const xmlChar *tmp; |
11070 | |
|
11071 | 0 | if ((ctxt == NULL) || (lastlt == NULL) || (lastgt == NULL)) { |
11072 | 0 | xmlGenericError(xmlGenericErrorContext, |
11073 | 0 | "Internal error: xmlParseGetLasts\n"); |
11074 | 0 | return; |
11075 | 0 | } |
11076 | 0 | if ((ctxt->progressive != 0) && (ctxt->inputNr == 1)) { |
11077 | 0 | tmp = ctxt->input->end; |
11078 | 0 | tmp--; |
11079 | 0 | while ((tmp >= ctxt->input->base) && (*tmp != '<')) tmp--; |
11080 | 0 | if (tmp < ctxt->input->base) { |
11081 | 0 | *lastlt = NULL; |
11082 | 0 | *lastgt = NULL; |
11083 | 0 | } else { |
11084 | 0 | *lastlt = tmp; |
11085 | 0 | tmp++; |
11086 | 0 | while ((tmp < ctxt->input->end) && (*tmp != '>')) { |
11087 | 0 | if (*tmp == '\'') { |
11088 | 0 | tmp++; |
11089 | 0 | while ((tmp < ctxt->input->end) && (*tmp != '\'')) tmp++; |
11090 | 0 | if (tmp < ctxt->input->end) tmp++; |
11091 | 0 | } else if (*tmp == '"') { |
11092 | 0 | tmp++; |
11093 | 0 | while ((tmp < ctxt->input->end) && (*tmp != '"')) tmp++; |
11094 | 0 | if (tmp < ctxt->input->end) tmp++; |
11095 | 0 | } else |
11096 | 0 | tmp++; |
11097 | 0 | } |
11098 | 0 | if (tmp < ctxt->input->end) |
11099 | 0 | *lastgt = tmp; |
11100 | 0 | else { |
11101 | 0 | tmp = *lastlt; |
11102 | 0 | tmp--; |
11103 | 0 | while ((tmp >= ctxt->input->base) && (*tmp != '>')) tmp--; |
11104 | 0 | if (tmp >= ctxt->input->base) |
11105 | 0 | *lastgt = tmp; |
11106 | 0 | else |
11107 | 0 | *lastgt = NULL; |
11108 | 0 | } |
11109 | 0 | } |
11110 | 0 | } else { |
11111 | 0 | *lastlt = NULL; |
11112 | 0 | *lastgt = NULL; |
11113 | 0 | } |
11114 | 0 | } |
11115 | | /** |
11116 | | * xmlCheckCdataPush: |
11117 | | * @cur: pointer to the block of characters |
11118 | | * @len: length of the block in bytes |
11119 | | * @complete: 1 if complete CDATA block is passed in, 0 if partial block |
11120 | | * |
11121 | | * Check that the block of characters is okay as SCdata content [20] |
11122 | | * |
11123 | | * Returns the number of bytes to pass if okay, a negative index where an |
11124 | | * UTF-8 error occurred otherwise |
11125 | | */ |
11126 | | static int |
11127 | 0 | xmlCheckCdataPush(const xmlChar *utf, int len, int complete) { |
11128 | 0 | int ix; |
11129 | 0 | unsigned char c; |
11130 | 0 | int codepoint; |
11131 | |
|
11132 | 0 | if ((utf == NULL) || (len <= 0)) |
11133 | 0 | return(0); |
11134 | | |
11135 | 0 | for (ix = 0; ix < len;) { /* string is 0-terminated */ |
11136 | 0 | c = utf[ix]; |
11137 | 0 | if ((c & 0x80) == 0x00) { /* 1-byte code, starts with 10 */ |
11138 | 0 | if (c >= 0x20) |
11139 | 0 | ix++; |
11140 | 0 | else if ((c == 0xA) || (c == 0xD) || (c == 0x9)) |
11141 | 0 | ix++; |
11142 | 0 | else |
11143 | 0 | return(-ix); |
11144 | 0 | } else if ((c & 0xe0) == 0xc0) {/* 2-byte code, starts with 110 */ |
11145 | 0 | if (ix + 2 > len) return(complete ? -ix : ix); |
11146 | 0 | if ((utf[ix+1] & 0xc0 ) != 0x80) |
11147 | 0 | return(-ix); |
11148 | 0 | codepoint = (utf[ix] & 0x1f) << 6; |
11149 | 0 | codepoint |= utf[ix+1] & 0x3f; |
11150 | 0 | if (!xmlIsCharQ(codepoint)) |
11151 | 0 | return(-ix); |
11152 | 0 | ix += 2; |
11153 | 0 | } else if ((c & 0xf0) == 0xe0) {/* 3-byte code, starts with 1110 */ |
11154 | 0 | if (ix + 3 > len) return(complete ? -ix : ix); |
11155 | 0 | if (((utf[ix+1] & 0xc0) != 0x80) || |
11156 | 0 | ((utf[ix+2] & 0xc0) != 0x80)) |
11157 | 0 | return(-ix); |
11158 | 0 | codepoint = (utf[ix] & 0xf) << 12; |
11159 | 0 | codepoint |= (utf[ix+1] & 0x3f) << 6; |
11160 | 0 | codepoint |= utf[ix+2] & 0x3f; |
11161 | 0 | if (!xmlIsCharQ(codepoint)) |
11162 | 0 | return(-ix); |
11163 | 0 | ix += 3; |
11164 | 0 | } else if ((c & 0xf8) == 0xf0) {/* 4-byte code, starts with 11110 */ |
11165 | 0 | if (ix + 4 > len) return(complete ? -ix : ix); |
11166 | 0 | if (((utf[ix+1] & 0xc0) != 0x80) || |
11167 | 0 | ((utf[ix+2] & 0xc0) != 0x80) || |
11168 | 0 | ((utf[ix+3] & 0xc0) != 0x80)) |
11169 | 0 | return(-ix); |
11170 | 0 | codepoint = (utf[ix] & 0x7) << 18; |
11171 | 0 | codepoint |= (utf[ix+1] & 0x3f) << 12; |
11172 | 0 | codepoint |= (utf[ix+2] & 0x3f) << 6; |
11173 | 0 | codepoint |= utf[ix+3] & 0x3f; |
11174 | 0 | if (!xmlIsCharQ(codepoint)) |
11175 | 0 | return(-ix); |
11176 | 0 | ix += 4; |
11177 | 0 | } else /* unknown encoding */ |
11178 | 0 | return(-ix); |
11179 | 0 | } |
11180 | 0 | return(ix); |
11181 | 0 | } |
11182 | | |
11183 | | /** |
11184 | | * xmlParseTryOrFinish: |
11185 | | * @ctxt: an XML parser context |
11186 | | * @terminate: last chunk indicator |
11187 | | * |
11188 | | * Try to progress on parsing |
11189 | | * |
11190 | | * Returns zero if no parsing was possible |
11191 | | */ |
11192 | | static int |
11193 | 0 | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11194 | 0 | int ret = 0; |
11195 | 0 | int avail, tlen; |
11196 | 0 | xmlChar cur, next; |
11197 | 0 | const xmlChar *lastlt, *lastgt; |
11198 | |
|
11199 | 0 | if (ctxt->input == NULL) |
11200 | 0 | return(0); |
11201 | | |
11202 | | #ifdef DEBUG_PUSH |
11203 | | switch (ctxt->instate) { |
11204 | | case XML_PARSER_EOF: |
11205 | | xmlGenericError(xmlGenericErrorContext, |
11206 | | "PP: try EOF\n"); break; |
11207 | | case XML_PARSER_START: |
11208 | | xmlGenericError(xmlGenericErrorContext, |
11209 | | "PP: try START\n"); break; |
11210 | | case XML_PARSER_MISC: |
11211 | | xmlGenericError(xmlGenericErrorContext, |
11212 | | "PP: try MISC\n");break; |
11213 | | case XML_PARSER_COMMENT: |
11214 | | xmlGenericError(xmlGenericErrorContext, |
11215 | | "PP: try COMMENT\n");break; |
11216 | | case XML_PARSER_PROLOG: |
11217 | | xmlGenericError(xmlGenericErrorContext, |
11218 | | "PP: try PROLOG\n");break; |
11219 | | case XML_PARSER_START_TAG: |
11220 | | xmlGenericError(xmlGenericErrorContext, |
11221 | | "PP: try START_TAG\n");break; |
11222 | | case XML_PARSER_CONTENT: |
11223 | | xmlGenericError(xmlGenericErrorContext, |
11224 | | "PP: try CONTENT\n");break; |
11225 | | case XML_PARSER_CDATA_SECTION: |
11226 | | xmlGenericError(xmlGenericErrorContext, |
11227 | | "PP: try CDATA_SECTION\n");break; |
11228 | | case XML_PARSER_END_TAG: |
11229 | | xmlGenericError(xmlGenericErrorContext, |
11230 | | "PP: try END_TAG\n");break; |
11231 | | case XML_PARSER_ENTITY_DECL: |
11232 | | xmlGenericError(xmlGenericErrorContext, |
11233 | | "PP: try ENTITY_DECL\n");break; |
11234 | | case XML_PARSER_ENTITY_VALUE: |
11235 | | xmlGenericError(xmlGenericErrorContext, |
11236 | | "PP: try ENTITY_VALUE\n");break; |
11237 | | case XML_PARSER_ATTRIBUTE_VALUE: |
11238 | | xmlGenericError(xmlGenericErrorContext, |
11239 | | "PP: try ATTRIBUTE_VALUE\n");break; |
11240 | | case XML_PARSER_DTD: |
11241 | | xmlGenericError(xmlGenericErrorContext, |
11242 | | "PP: try DTD\n");break; |
11243 | | case XML_PARSER_EPILOG: |
11244 | | xmlGenericError(xmlGenericErrorContext, |
11245 | | "PP: try EPILOG\n");break; |
11246 | | case XML_PARSER_PI: |
11247 | | xmlGenericError(xmlGenericErrorContext, |
11248 | | "PP: try PI\n");break; |
11249 | | case XML_PARSER_IGNORE: |
11250 | | xmlGenericError(xmlGenericErrorContext, |
11251 | | "PP: try IGNORE\n");break; |
11252 | | } |
11253 | | #endif |
11254 | | |
11255 | 0 | if ((ctxt->input != NULL) && |
11256 | 0 | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11257 | 0 | xmlSHRINK(ctxt); |
11258 | 0 | ctxt->checkIndex = 0; |
11259 | 0 | } |
11260 | 0 | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11261 | |
|
11262 | 0 | while (ctxt->instate != XML_PARSER_EOF) { |
11263 | 0 | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
11264 | 0 | return(0); |
11265 | | |
11266 | 0 | if (ctxt->input == NULL) break; |
11267 | 0 | if (ctxt->input->buf == NULL) |
11268 | 0 | avail = ctxt->input->length - |
11269 | 0 | (ctxt->input->cur - ctxt->input->base); |
11270 | 0 | else { |
11271 | | /* |
11272 | | * If we are operating on converted input, try to flush |
11273 | | * remaining chars to avoid them stalling in the non-converted |
11274 | | * buffer. But do not do this in document start where |
11275 | | * encoding="..." may not have been read and we work on a |
11276 | | * guessed encoding. |
11277 | | */ |
11278 | 0 | if ((ctxt->instate != XML_PARSER_START) && |
11279 | 0 | (ctxt->input->buf->raw != NULL) && |
11280 | 0 | (xmlBufIsEmpty(ctxt->input->buf->raw) == 0)) { |
11281 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
11282 | 0 | ctxt->input); |
11283 | 0 | size_t current = ctxt->input->cur - ctxt->input->base; |
11284 | |
|
11285 | 0 | xmlParserInputBufferPush(ctxt->input->buf, 0, ""); |
11286 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
11287 | 0 | base, current); |
11288 | 0 | } |
11289 | 0 | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11290 | 0 | (ctxt->input->cur - ctxt->input->base); |
11291 | 0 | } |
11292 | 0 | if (avail < 1) |
11293 | 0 | goto done; |
11294 | 0 | switch (ctxt->instate) { |
11295 | 0 | case XML_PARSER_EOF: |
11296 | | /* |
11297 | | * Document parsing is done ! |
11298 | | */ |
11299 | 0 | goto done; |
11300 | 0 | case XML_PARSER_START: |
11301 | 0 | if (ctxt->charset == XML_CHAR_ENCODING_NONE) { |
11302 | 0 | xmlChar start[4]; |
11303 | 0 | xmlCharEncoding enc; |
11304 | | |
11305 | | /* |
11306 | | * Very first chars read from the document flow. |
11307 | | */ |
11308 | 0 | if (avail < 4) |
11309 | 0 | goto done; |
11310 | | |
11311 | | /* |
11312 | | * Get the 4 first bytes and decode the charset |
11313 | | * if enc != XML_CHAR_ENCODING_NONE |
11314 | | * plug some encoding conversion routines, |
11315 | | * else xmlSwitchEncoding will set to (default) |
11316 | | * UTF8. |
11317 | | */ |
11318 | 0 | start[0] = RAW; |
11319 | 0 | start[1] = NXT(1); |
11320 | 0 | start[2] = NXT(2); |
11321 | 0 | start[3] = NXT(3); |
11322 | 0 | enc = xmlDetectCharEncoding(start, 4); |
11323 | 0 | xmlSwitchEncoding(ctxt, enc); |
11324 | 0 | break; |
11325 | 0 | } |
11326 | | |
11327 | 0 | if (avail < 2) |
11328 | 0 | goto done; |
11329 | 0 | cur = ctxt->input->cur[0]; |
11330 | 0 | next = ctxt->input->cur[1]; |
11331 | 0 | if (cur == 0) { |
11332 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11333 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11334 | 0 | &xmlDefaultSAXLocator); |
11335 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11336 | 0 | xmlHaltParser(ctxt); |
11337 | | #ifdef DEBUG_PUSH |
11338 | | xmlGenericError(xmlGenericErrorContext, |
11339 | | "PP: entering EOF\n"); |
11340 | | #endif |
11341 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11342 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11343 | 0 | goto done; |
11344 | 0 | } |
11345 | 0 | if ((cur == '<') && (next == '?')) { |
11346 | | /* PI or XML decl */ |
11347 | 0 | if (avail < 5) return(ret); |
11348 | 0 | if ((!terminate) && |
11349 | 0 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
11350 | 0 | return(ret); |
11351 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11352 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11353 | 0 | &xmlDefaultSAXLocator); |
11354 | 0 | if ((ctxt->input->cur[2] == 'x') && |
11355 | 0 | (ctxt->input->cur[3] == 'm') && |
11356 | 0 | (ctxt->input->cur[4] == 'l') && |
11357 | 0 | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11358 | 0 | ret += 5; |
11359 | | #ifdef DEBUG_PUSH |
11360 | | xmlGenericError(xmlGenericErrorContext, |
11361 | | "PP: Parsing XML Decl\n"); |
11362 | | #endif |
11363 | 0 | xmlParseXMLDecl(ctxt); |
11364 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11365 | | /* |
11366 | | * The XML REC instructs us to stop parsing right |
11367 | | * here |
11368 | | */ |
11369 | 0 | xmlHaltParser(ctxt); |
11370 | 0 | return(0); |
11371 | 0 | } |
11372 | 0 | ctxt->standalone = ctxt->input->standalone; |
11373 | 0 | if ((ctxt->encoding == NULL) && |
11374 | 0 | (ctxt->input->encoding != NULL)) |
11375 | 0 | ctxt->encoding = xmlStrdup(ctxt->input->encoding); |
11376 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11377 | 0 | (!ctxt->disableSAX)) |
11378 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11379 | 0 | ctxt->instate = XML_PARSER_MISC; |
11380 | | #ifdef DEBUG_PUSH |
11381 | | xmlGenericError(xmlGenericErrorContext, |
11382 | | "PP: entering MISC\n"); |
11383 | | #endif |
11384 | 0 | } else { |
11385 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11386 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11387 | 0 | (!ctxt->disableSAX)) |
11388 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11389 | 0 | ctxt->instate = XML_PARSER_MISC; |
11390 | | #ifdef DEBUG_PUSH |
11391 | | xmlGenericError(xmlGenericErrorContext, |
11392 | | "PP: entering MISC\n"); |
11393 | | #endif |
11394 | 0 | } |
11395 | 0 | } else { |
11396 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11397 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11398 | 0 | &xmlDefaultSAXLocator); |
11399 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11400 | 0 | if (ctxt->version == NULL) { |
11401 | 0 | xmlErrMemory(ctxt, NULL); |
11402 | 0 | break; |
11403 | 0 | } |
11404 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11405 | 0 | (!ctxt->disableSAX)) |
11406 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11407 | 0 | ctxt->instate = XML_PARSER_MISC; |
11408 | | #ifdef DEBUG_PUSH |
11409 | | xmlGenericError(xmlGenericErrorContext, |
11410 | | "PP: entering MISC\n"); |
11411 | | #endif |
11412 | 0 | } |
11413 | 0 | break; |
11414 | 0 | case XML_PARSER_START_TAG: { |
11415 | 0 | const xmlChar *name; |
11416 | 0 | const xmlChar *prefix = NULL; |
11417 | 0 | const xmlChar *URI = NULL; |
11418 | 0 | int line = ctxt->input->line; |
11419 | 0 | int nsNr = ctxt->nsNr; |
11420 | |
|
11421 | 0 | if ((avail < 2) && (ctxt->inputNr == 1)) |
11422 | 0 | goto done; |
11423 | 0 | cur = ctxt->input->cur[0]; |
11424 | 0 | if (cur != '<') { |
11425 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11426 | 0 | xmlHaltParser(ctxt); |
11427 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11428 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11429 | 0 | goto done; |
11430 | 0 | } |
11431 | 0 | if (!terminate) { |
11432 | 0 | if (ctxt->progressive) { |
11433 | | /* > can be found unescaped in attribute values */ |
11434 | 0 | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11435 | 0 | goto done; |
11436 | 0 | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11437 | 0 | goto done; |
11438 | 0 | } |
11439 | 0 | } |
11440 | 0 | if (ctxt->spaceNr == 0) |
11441 | 0 | spacePush(ctxt, -1); |
11442 | 0 | else if (*ctxt->space == -2) |
11443 | 0 | spacePush(ctxt, -1); |
11444 | 0 | else |
11445 | 0 | spacePush(ctxt, *ctxt->space); |
11446 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11447 | 0 | if (ctxt->sax2) |
11448 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11449 | 0 | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
11450 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11451 | 0 | else |
11452 | 0 | name = xmlParseStartTag(ctxt); |
11453 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11454 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11455 | 0 | goto done; |
11456 | 0 | if (name == NULL) { |
11457 | 0 | spacePop(ctxt); |
11458 | 0 | xmlHaltParser(ctxt); |
11459 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11460 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11461 | 0 | goto done; |
11462 | 0 | } |
11463 | 0 | #ifdef LIBXML_VALID_ENABLED |
11464 | | /* |
11465 | | * [ VC: Root Element Type ] |
11466 | | * The Name in the document type declaration must match |
11467 | | * the element type of the root element. |
11468 | | */ |
11469 | 0 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11470 | 0 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11471 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11472 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
11473 | | |
11474 | | /* |
11475 | | * Check for an Empty Element. |
11476 | | */ |
11477 | 0 | if ((RAW == '/') && (NXT(1) == '>')) { |
11478 | 0 | SKIP(2); |
11479 | |
|
11480 | 0 | if (ctxt->sax2) { |
11481 | 0 | if ((ctxt->sax != NULL) && |
11482 | 0 | (ctxt->sax->endElementNs != NULL) && |
11483 | 0 | (!ctxt->disableSAX)) |
11484 | 0 | ctxt->sax->endElementNs(ctxt->userData, name, |
11485 | 0 | prefix, URI); |
11486 | 0 | if (ctxt->nsNr - nsNr > 0) |
11487 | 0 | nsPop(ctxt, ctxt->nsNr - nsNr); |
11488 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11489 | 0 | } else { |
11490 | 0 | if ((ctxt->sax != NULL) && |
11491 | 0 | (ctxt->sax->endElement != NULL) && |
11492 | 0 | (!ctxt->disableSAX)) |
11493 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
11494 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11495 | 0 | } |
11496 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11497 | 0 | goto done; |
11498 | 0 | spacePop(ctxt); |
11499 | 0 | if (ctxt->nameNr == 0) { |
11500 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11501 | 0 | } else { |
11502 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11503 | 0 | } |
11504 | 0 | ctxt->progressive = 1; |
11505 | 0 | break; |
11506 | 0 | } |
11507 | 0 | if (RAW == '>') { |
11508 | 0 | NEXT; |
11509 | 0 | } else { |
11510 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11511 | 0 | "Couldn't find end of Start Tag %s\n", |
11512 | 0 | name); |
11513 | 0 | nodePop(ctxt); |
11514 | 0 | spacePop(ctxt); |
11515 | 0 | } |
11516 | 0 | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
11517 | |
|
11518 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11519 | 0 | ctxt->progressive = 1; |
11520 | 0 | break; |
11521 | 0 | } |
11522 | 0 | case XML_PARSER_CONTENT: { |
11523 | 0 | int id; |
11524 | 0 | unsigned long cons; |
11525 | 0 | if ((avail < 2) && (ctxt->inputNr == 1)) |
11526 | 0 | goto done; |
11527 | 0 | cur = ctxt->input->cur[0]; |
11528 | 0 | next = ctxt->input->cur[1]; |
11529 | |
|
11530 | 0 | id = ctxt->input->id; |
11531 | 0 | cons = CUR_CONSUMED; |
11532 | 0 | if ((cur == '<') && (next == '/')) { |
11533 | 0 | ctxt->instate = XML_PARSER_END_TAG; |
11534 | 0 | break; |
11535 | 0 | } else if ((cur == '<') && (next == '?')) { |
11536 | 0 | if ((!terminate) && |
11537 | 0 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11538 | 0 | ctxt->progressive = XML_PARSER_PI; |
11539 | 0 | goto done; |
11540 | 0 | } |
11541 | 0 | xmlParsePI(ctxt); |
11542 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11543 | 0 | ctxt->progressive = 1; |
11544 | 0 | } else if ((cur == '<') && (next != '!')) { |
11545 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11546 | 0 | break; |
11547 | 0 | } else if ((cur == '<') && (next == '!') && |
11548 | 0 | (ctxt->input->cur[2] == '-') && |
11549 | 0 | (ctxt->input->cur[3] == '-')) { |
11550 | 0 | int term; |
11551 | |
|
11552 | 0 | if (avail < 4) |
11553 | 0 | goto done; |
11554 | 0 | ctxt->input->cur += 4; |
11555 | 0 | term = xmlParseLookupSequence(ctxt, '-', '-', '>'); |
11556 | 0 | ctxt->input->cur -= 4; |
11557 | 0 | if ((!terminate) && (term < 0)) { |
11558 | 0 | ctxt->progressive = XML_PARSER_COMMENT; |
11559 | 0 | goto done; |
11560 | 0 | } |
11561 | 0 | xmlParseComment(ctxt); |
11562 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11563 | 0 | ctxt->progressive = 1; |
11564 | 0 | } else if ((cur == '<') && (ctxt->input->cur[1] == '!') && |
11565 | 0 | (ctxt->input->cur[2] == '[') && |
11566 | 0 | (ctxt->input->cur[3] == 'C') && |
11567 | 0 | (ctxt->input->cur[4] == 'D') && |
11568 | 0 | (ctxt->input->cur[5] == 'A') && |
11569 | 0 | (ctxt->input->cur[6] == 'T') && |
11570 | 0 | (ctxt->input->cur[7] == 'A') && |
11571 | 0 | (ctxt->input->cur[8] == '[')) { |
11572 | 0 | SKIP(9); |
11573 | 0 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11574 | 0 | break; |
11575 | 0 | } else if ((cur == '<') && (next == '!') && |
11576 | 0 | (avail < 9)) { |
11577 | 0 | goto done; |
11578 | 0 | } else if (cur == '&') { |
11579 | 0 | if ((!terminate) && |
11580 | 0 | (xmlParseLookupSequence(ctxt, ';', 0, 0) < 0)) |
11581 | 0 | goto done; |
11582 | 0 | xmlParseReference(ctxt); |
11583 | 0 | } else { |
11584 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11585 | | /* |
11586 | | * Goal of the following test is: |
11587 | | * - minimize calls to the SAX 'character' callback |
11588 | | * when they are mergeable |
11589 | | * - handle an problem for isBlank when we only parse |
11590 | | * a sequence of blank chars and the next one is |
11591 | | * not available to check against '<' presence. |
11592 | | * - tries to homogenize the differences in SAX |
11593 | | * callbacks between the push and pull versions |
11594 | | * of the parser. |
11595 | | */ |
11596 | 0 | if ((ctxt->inputNr == 1) && |
11597 | 0 | (avail < XML_PARSER_BIG_BUFFER_SIZE)) { |
11598 | 0 | if (!terminate) { |
11599 | 0 | if (ctxt->progressive) { |
11600 | 0 | if ((lastlt == NULL) || |
11601 | 0 | (ctxt->input->cur > lastlt)) |
11602 | 0 | goto done; |
11603 | 0 | } else if (xmlParseLookupSequence(ctxt, |
11604 | 0 | '<', 0, 0) < 0) { |
11605 | 0 | goto done; |
11606 | 0 | } |
11607 | 0 | } |
11608 | 0 | } |
11609 | 0 | ctxt->checkIndex = 0; |
11610 | 0 | xmlParseCharData(ctxt, 0); |
11611 | 0 | } |
11612 | 0 | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
11613 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
11614 | 0 | "detected an error in element content\n"); |
11615 | 0 | xmlHaltParser(ctxt); |
11616 | 0 | break; |
11617 | 0 | } |
11618 | 0 | break; |
11619 | 0 | } |
11620 | 0 | case XML_PARSER_END_TAG: |
11621 | 0 | if (avail < 2) |
11622 | 0 | goto done; |
11623 | 0 | if (!terminate) { |
11624 | 0 | if (ctxt->progressive) { |
11625 | | /* > can be found unescaped in attribute values */ |
11626 | 0 | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11627 | 0 | goto done; |
11628 | 0 | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11629 | 0 | goto done; |
11630 | 0 | } |
11631 | 0 | } |
11632 | 0 | if (ctxt->sax2) { |
11633 | 0 | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11634 | 0 | nameNsPop(ctxt); |
11635 | 0 | } |
11636 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11637 | 0 | else |
11638 | 0 | xmlParseEndTag1(ctxt, 0); |
11639 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11640 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
11641 | | /* Nothing */ |
11642 | 0 | } else if (ctxt->nameNr == 0) { |
11643 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11644 | 0 | } else { |
11645 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11646 | 0 | } |
11647 | 0 | break; |
11648 | 0 | case XML_PARSER_CDATA_SECTION: { |
11649 | | /* |
11650 | | * The Push mode need to have the SAX callback for |
11651 | | * cdataBlock merge back contiguous callbacks. |
11652 | | */ |
11653 | 0 | int base; |
11654 | |
|
11655 | 0 | base = xmlParseLookupSequence(ctxt, ']', ']', '>'); |
11656 | 0 | if (base < 0) { |
11657 | 0 | if (avail >= XML_PARSER_BIG_BUFFER_SIZE + 2) { |
11658 | 0 | int tmp; |
11659 | |
|
11660 | 0 | tmp = xmlCheckCdataPush(ctxt->input->cur, |
11661 | 0 | XML_PARSER_BIG_BUFFER_SIZE, 0); |
11662 | 0 | if (tmp < 0) { |
11663 | 0 | tmp = -tmp; |
11664 | 0 | ctxt->input->cur += tmp; |
11665 | 0 | goto encoding_error; |
11666 | 0 | } |
11667 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
11668 | 0 | if (ctxt->sax->cdataBlock != NULL) |
11669 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11670 | 0 | ctxt->input->cur, tmp); |
11671 | 0 | else if (ctxt->sax->characters != NULL) |
11672 | 0 | ctxt->sax->characters(ctxt->userData, |
11673 | 0 | ctxt->input->cur, tmp); |
11674 | 0 | } |
11675 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11676 | 0 | goto done; |
11677 | 0 | SKIPL(tmp); |
11678 | 0 | ctxt->checkIndex = 0; |
11679 | 0 | } |
11680 | 0 | goto done; |
11681 | 0 | } else { |
11682 | 0 | int tmp; |
11683 | |
|
11684 | 0 | tmp = xmlCheckCdataPush(ctxt->input->cur, base, 1); |
11685 | 0 | if ((tmp < 0) || (tmp != base)) { |
11686 | 0 | tmp = -tmp; |
11687 | 0 | ctxt->input->cur += tmp; |
11688 | 0 | goto encoding_error; |
11689 | 0 | } |
11690 | 0 | if ((ctxt->sax != NULL) && (base == 0) && |
11691 | 0 | (ctxt->sax->cdataBlock != NULL) && |
11692 | 0 | (!ctxt->disableSAX)) { |
11693 | | /* |
11694 | | * Special case to provide identical behaviour |
11695 | | * between pull and push parsers on enpty CDATA |
11696 | | * sections |
11697 | | */ |
11698 | 0 | if ((ctxt->input->cur - ctxt->input->base >= 9) && |
11699 | 0 | (!strncmp((const char *)&ctxt->input->cur[-9], |
11700 | 0 | "<![CDATA[", 9))) |
11701 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11702 | 0 | BAD_CAST "", 0); |
11703 | 0 | } else if ((ctxt->sax != NULL) && (base > 0) && |
11704 | 0 | (!ctxt->disableSAX)) { |
11705 | 0 | if (ctxt->sax->cdataBlock != NULL) |
11706 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11707 | 0 | ctxt->input->cur, base); |
11708 | 0 | else if (ctxt->sax->characters != NULL) |
11709 | 0 | ctxt->sax->characters(ctxt->userData, |
11710 | 0 | ctxt->input->cur, base); |
11711 | 0 | } |
11712 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11713 | 0 | goto done; |
11714 | 0 | SKIPL(base + 3); |
11715 | 0 | ctxt->checkIndex = 0; |
11716 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11717 | | #ifdef DEBUG_PUSH |
11718 | | xmlGenericError(xmlGenericErrorContext, |
11719 | | "PP: entering CONTENT\n"); |
11720 | | #endif |
11721 | 0 | } |
11722 | 0 | break; |
11723 | 0 | } |
11724 | 0 | case XML_PARSER_MISC: |
11725 | 0 | SKIP_BLANKS; |
11726 | 0 | if (ctxt->input->buf == NULL) |
11727 | 0 | avail = ctxt->input->length - |
11728 | 0 | (ctxt->input->cur - ctxt->input->base); |
11729 | 0 | else |
11730 | 0 | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11731 | 0 | (ctxt->input->cur - ctxt->input->base); |
11732 | 0 | if (avail < 2) |
11733 | 0 | goto done; |
11734 | 0 | cur = ctxt->input->cur[0]; |
11735 | 0 | next = ctxt->input->cur[1]; |
11736 | 0 | if ((cur == '<') && (next == '?')) { |
11737 | 0 | if ((!terminate) && |
11738 | 0 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11739 | 0 | ctxt->progressive = XML_PARSER_PI; |
11740 | 0 | goto done; |
11741 | 0 | } |
11742 | | #ifdef DEBUG_PUSH |
11743 | | xmlGenericError(xmlGenericErrorContext, |
11744 | | "PP: Parsing PI\n"); |
11745 | | #endif |
11746 | 0 | xmlParsePI(ctxt); |
11747 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11748 | 0 | goto done; |
11749 | 0 | ctxt->instate = XML_PARSER_MISC; |
11750 | 0 | ctxt->progressive = 1; |
11751 | 0 | ctxt->checkIndex = 0; |
11752 | 0 | } else if ((cur == '<') && (next == '!') && |
11753 | 0 | (ctxt->input->cur[2] == '-') && |
11754 | 0 | (ctxt->input->cur[3] == '-')) { |
11755 | 0 | if ((!terminate) && |
11756 | 0 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11757 | 0 | ctxt->progressive = XML_PARSER_COMMENT; |
11758 | 0 | goto done; |
11759 | 0 | } |
11760 | | #ifdef DEBUG_PUSH |
11761 | | xmlGenericError(xmlGenericErrorContext, |
11762 | | "PP: Parsing Comment\n"); |
11763 | | #endif |
11764 | 0 | xmlParseComment(ctxt); |
11765 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11766 | 0 | goto done; |
11767 | 0 | ctxt->instate = XML_PARSER_MISC; |
11768 | 0 | ctxt->progressive = 1; |
11769 | 0 | ctxt->checkIndex = 0; |
11770 | 0 | } else if ((cur == '<') && (next == '!') && |
11771 | 0 | (ctxt->input->cur[2] == 'D') && |
11772 | 0 | (ctxt->input->cur[3] == 'O') && |
11773 | 0 | (ctxt->input->cur[4] == 'C') && |
11774 | 0 | (ctxt->input->cur[5] == 'T') && |
11775 | 0 | (ctxt->input->cur[6] == 'Y') && |
11776 | 0 | (ctxt->input->cur[7] == 'P') && |
11777 | 0 | (ctxt->input->cur[8] == 'E')) { |
11778 | 0 | if ((!terminate) && |
11779 | 0 | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) { |
11780 | 0 | ctxt->progressive = XML_PARSER_DTD; |
11781 | 0 | goto done; |
11782 | 0 | } |
11783 | | #ifdef DEBUG_PUSH |
11784 | | xmlGenericError(xmlGenericErrorContext, |
11785 | | "PP: Parsing internal subset\n"); |
11786 | | #endif |
11787 | 0 | ctxt->inSubset = 1; |
11788 | 0 | ctxt->progressive = 0; |
11789 | 0 | ctxt->checkIndex = 0; |
11790 | 0 | xmlParseDocTypeDecl(ctxt); |
11791 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11792 | 0 | goto done; |
11793 | 0 | if (RAW == '[') { |
11794 | 0 | ctxt->instate = XML_PARSER_DTD; |
11795 | | #ifdef DEBUG_PUSH |
11796 | | xmlGenericError(xmlGenericErrorContext, |
11797 | | "PP: entering DTD\n"); |
11798 | | #endif |
11799 | 0 | } else { |
11800 | | /* |
11801 | | * Create and update the external subset. |
11802 | | */ |
11803 | 0 | ctxt->inSubset = 2; |
11804 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11805 | 0 | (ctxt->sax->externalSubset != NULL)) |
11806 | 0 | ctxt->sax->externalSubset(ctxt->userData, |
11807 | 0 | ctxt->intSubName, ctxt->extSubSystem, |
11808 | 0 | ctxt->extSubURI); |
11809 | 0 | ctxt->inSubset = 0; |
11810 | 0 | xmlCleanSpecialAttr(ctxt); |
11811 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11812 | | #ifdef DEBUG_PUSH |
11813 | | xmlGenericError(xmlGenericErrorContext, |
11814 | | "PP: entering PROLOG\n"); |
11815 | | #endif |
11816 | 0 | } |
11817 | 0 | } else if ((cur == '<') && (next == '!') && |
11818 | 0 | (avail < 9)) { |
11819 | 0 | goto done; |
11820 | 0 | } else { |
11821 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11822 | 0 | ctxt->progressive = XML_PARSER_START_TAG; |
11823 | 0 | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11824 | | #ifdef DEBUG_PUSH |
11825 | | xmlGenericError(xmlGenericErrorContext, |
11826 | | "PP: entering START_TAG\n"); |
11827 | | #endif |
11828 | 0 | } |
11829 | 0 | break; |
11830 | 0 | case XML_PARSER_PROLOG: |
11831 | 0 | SKIP_BLANKS; |
11832 | 0 | if (ctxt->input->buf == NULL) |
11833 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11834 | 0 | else |
11835 | 0 | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11836 | 0 | (ctxt->input->cur - ctxt->input->base); |
11837 | 0 | if (avail < 2) |
11838 | 0 | goto done; |
11839 | 0 | cur = ctxt->input->cur[0]; |
11840 | 0 | next = ctxt->input->cur[1]; |
11841 | 0 | if ((cur == '<') && (next == '?')) { |
11842 | 0 | if ((!terminate) && |
11843 | 0 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11844 | 0 | ctxt->progressive = XML_PARSER_PI; |
11845 | 0 | goto done; |
11846 | 0 | } |
11847 | | #ifdef DEBUG_PUSH |
11848 | | xmlGenericError(xmlGenericErrorContext, |
11849 | | "PP: Parsing PI\n"); |
11850 | | #endif |
11851 | 0 | xmlParsePI(ctxt); |
11852 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11853 | 0 | goto done; |
11854 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11855 | 0 | ctxt->progressive = 1; |
11856 | 0 | } else if ((cur == '<') && (next == '!') && |
11857 | 0 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
11858 | 0 | if ((!terminate) && |
11859 | 0 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11860 | 0 | ctxt->progressive = XML_PARSER_COMMENT; |
11861 | 0 | goto done; |
11862 | 0 | } |
11863 | | #ifdef DEBUG_PUSH |
11864 | | xmlGenericError(xmlGenericErrorContext, |
11865 | | "PP: Parsing Comment\n"); |
11866 | | #endif |
11867 | 0 | xmlParseComment(ctxt); |
11868 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11869 | 0 | goto done; |
11870 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
11871 | 0 | ctxt->progressive = 1; |
11872 | 0 | } else if ((cur == '<') && (next == '!') && |
11873 | 0 | (avail < 4)) { |
11874 | 0 | goto done; |
11875 | 0 | } else { |
11876 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
11877 | 0 | if (ctxt->progressive == 0) |
11878 | 0 | ctxt->progressive = XML_PARSER_START_TAG; |
11879 | 0 | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11880 | | #ifdef DEBUG_PUSH |
11881 | | xmlGenericError(xmlGenericErrorContext, |
11882 | | "PP: entering START_TAG\n"); |
11883 | | #endif |
11884 | 0 | } |
11885 | 0 | break; |
11886 | 0 | case XML_PARSER_EPILOG: |
11887 | 0 | SKIP_BLANKS; |
11888 | 0 | if (ctxt->input->buf == NULL) |
11889 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11890 | 0 | else |
11891 | 0 | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11892 | 0 | (ctxt->input->cur - ctxt->input->base); |
11893 | 0 | if (avail < 2) |
11894 | 0 | goto done; |
11895 | 0 | cur = ctxt->input->cur[0]; |
11896 | 0 | next = ctxt->input->cur[1]; |
11897 | 0 | if ((cur == '<') && (next == '?')) { |
11898 | 0 | if ((!terminate) && |
11899 | 0 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11900 | 0 | ctxt->progressive = XML_PARSER_PI; |
11901 | 0 | goto done; |
11902 | 0 | } |
11903 | | #ifdef DEBUG_PUSH |
11904 | | xmlGenericError(xmlGenericErrorContext, |
11905 | | "PP: Parsing PI\n"); |
11906 | | #endif |
11907 | 0 | xmlParsePI(ctxt); |
11908 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11909 | 0 | goto done; |
11910 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11911 | 0 | ctxt->progressive = 1; |
11912 | 0 | } else if ((cur == '<') && (next == '!') && |
11913 | 0 | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
11914 | 0 | if ((!terminate) && |
11915 | 0 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11916 | 0 | ctxt->progressive = XML_PARSER_COMMENT; |
11917 | 0 | goto done; |
11918 | 0 | } |
11919 | | #ifdef DEBUG_PUSH |
11920 | | xmlGenericError(xmlGenericErrorContext, |
11921 | | "PP: Parsing Comment\n"); |
11922 | | #endif |
11923 | 0 | xmlParseComment(ctxt); |
11924 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11925 | 0 | goto done; |
11926 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
11927 | 0 | ctxt->progressive = 1; |
11928 | 0 | } else if ((cur == '<') && (next == '!') && |
11929 | 0 | (avail < 4)) { |
11930 | 0 | goto done; |
11931 | 0 | } else { |
11932 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11933 | 0 | xmlHaltParser(ctxt); |
11934 | | #ifdef DEBUG_PUSH |
11935 | | xmlGenericError(xmlGenericErrorContext, |
11936 | | "PP: entering EOF\n"); |
11937 | | #endif |
11938 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11939 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11940 | 0 | goto done; |
11941 | 0 | } |
11942 | 0 | break; |
11943 | 0 | case XML_PARSER_DTD: { |
11944 | | /* |
11945 | | * Sorry but progressive parsing of the internal subset |
11946 | | * is not expected to be supported. We first check that |
11947 | | * the full content of the internal subset is available and |
11948 | | * the parsing is launched only at that point. |
11949 | | * Internal subset ends up with "']' S? '>'" in an unescaped |
11950 | | * section and not in a ']]>' sequence which are conditional |
11951 | | * sections (whoever argued to keep that crap in XML deserve |
11952 | | * a place in hell !). |
11953 | | */ |
11954 | 0 | int base, i; |
11955 | 0 | xmlChar *buf; |
11956 | 0 | xmlChar quote = 0; |
11957 | 0 | size_t use; |
11958 | |
|
11959 | 0 | base = ctxt->input->cur - ctxt->input->base; |
11960 | 0 | if (base < 0) return(0); |
11961 | 0 | if (ctxt->checkIndex > base) |
11962 | 0 | base = ctxt->checkIndex; |
11963 | 0 | buf = xmlBufContent(ctxt->input->buf->buffer); |
11964 | 0 | use = xmlBufUse(ctxt->input->buf->buffer); |
11965 | 0 | for (;(unsigned int) base < use; base++) { |
11966 | 0 | if (quote != 0) { |
11967 | 0 | if (buf[base] == quote) |
11968 | 0 | quote = 0; |
11969 | 0 | continue; |
11970 | 0 | } |
11971 | 0 | if ((quote == 0) && (buf[base] == '<')) { |
11972 | 0 | int found = 0; |
11973 | | /* special handling of comments */ |
11974 | 0 | if (((unsigned int) base + 4 < use) && |
11975 | 0 | (buf[base + 1] == '!') && |
11976 | 0 | (buf[base + 2] == '-') && |
11977 | 0 | (buf[base + 3] == '-')) { |
11978 | 0 | for (;(unsigned int) base + 3 < use; base++) { |
11979 | 0 | if ((buf[base] == '-') && |
11980 | 0 | (buf[base + 1] == '-') && |
11981 | 0 | (buf[base + 2] == '>')) { |
11982 | 0 | found = 1; |
11983 | 0 | base += 2; |
11984 | 0 | break; |
11985 | 0 | } |
11986 | 0 | } |
11987 | 0 | if (!found) { |
11988 | | #if 0 |
11989 | | fprintf(stderr, "unfinished comment\n"); |
11990 | | #endif |
11991 | 0 | break; /* for */ |
11992 | 0 | } |
11993 | 0 | continue; |
11994 | 0 | } |
11995 | 0 | } |
11996 | 0 | if (buf[base] == '"') { |
11997 | 0 | quote = '"'; |
11998 | 0 | continue; |
11999 | 0 | } |
12000 | 0 | if (buf[base] == '\'') { |
12001 | 0 | quote = '\''; |
12002 | 0 | continue; |
12003 | 0 | } |
12004 | 0 | if (buf[base] == ']') { |
12005 | | #if 0 |
12006 | | fprintf(stderr, "%c%c%c%c: ", buf[base], |
12007 | | buf[base + 1], buf[base + 2], buf[base + 3]); |
12008 | | #endif |
12009 | 0 | if ((unsigned int) base +1 >= use) |
12010 | 0 | break; |
12011 | 0 | if (buf[base + 1] == ']') { |
12012 | | /* conditional crap, skip both ']' ! */ |
12013 | 0 | base++; |
12014 | 0 | continue; |
12015 | 0 | } |
12016 | 0 | for (i = 1; (unsigned int) base + i < use; i++) { |
12017 | 0 | if (buf[base + i] == '>') { |
12018 | | #if 0 |
12019 | | fprintf(stderr, "found\n"); |
12020 | | #endif |
12021 | 0 | goto found_end_int_subset; |
12022 | 0 | } |
12023 | 0 | if (!IS_BLANK_CH(buf[base + i])) { |
12024 | | #if 0 |
12025 | | fprintf(stderr, "not found\n"); |
12026 | | #endif |
12027 | 0 | goto not_end_of_int_subset; |
12028 | 0 | } |
12029 | 0 | } |
12030 | | #if 0 |
12031 | | fprintf(stderr, "end of stream\n"); |
12032 | | #endif |
12033 | 0 | break; |
12034 | |
|
12035 | 0 | } |
12036 | 0 | not_end_of_int_subset: |
12037 | 0 | continue; /* for */ |
12038 | 0 | } |
12039 | | /* |
12040 | | * We didn't found the end of the Internal subset |
12041 | | */ |
12042 | 0 | if (quote == 0) |
12043 | 0 | ctxt->checkIndex = base; |
12044 | 0 | else |
12045 | 0 | ctxt->checkIndex = 0; |
12046 | | #ifdef DEBUG_PUSH |
12047 | | if (next == 0) |
12048 | | xmlGenericError(xmlGenericErrorContext, |
12049 | | "PP: lookup of int subset end filed\n"); |
12050 | | #endif |
12051 | 0 | goto done; |
12052 | | |
12053 | 0 | found_end_int_subset: |
12054 | 0 | ctxt->checkIndex = 0; |
12055 | 0 | xmlParseInternalSubset(ctxt); |
12056 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
12057 | 0 | goto done; |
12058 | 0 | ctxt->inSubset = 2; |
12059 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
12060 | 0 | (ctxt->sax->externalSubset != NULL)) |
12061 | 0 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
12062 | 0 | ctxt->extSubSystem, ctxt->extSubURI); |
12063 | 0 | ctxt->inSubset = 0; |
12064 | 0 | xmlCleanSpecialAttr(ctxt); |
12065 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
12066 | 0 | goto done; |
12067 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
12068 | 0 | ctxt->checkIndex = 0; |
12069 | | #ifdef DEBUG_PUSH |
12070 | | xmlGenericError(xmlGenericErrorContext, |
12071 | | "PP: entering PROLOG\n"); |
12072 | | #endif |
12073 | 0 | break; |
12074 | 0 | } |
12075 | 0 | case XML_PARSER_COMMENT: |
12076 | 0 | xmlGenericError(xmlGenericErrorContext, |
12077 | 0 | "PP: internal error, state == COMMENT\n"); |
12078 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12079 | | #ifdef DEBUG_PUSH |
12080 | | xmlGenericError(xmlGenericErrorContext, |
12081 | | "PP: entering CONTENT\n"); |
12082 | | #endif |
12083 | 0 | break; |
12084 | 0 | case XML_PARSER_IGNORE: |
12085 | 0 | xmlGenericError(xmlGenericErrorContext, |
12086 | 0 | "PP: internal error, state == IGNORE"); |
12087 | 0 | ctxt->instate = XML_PARSER_DTD; |
12088 | | #ifdef DEBUG_PUSH |
12089 | | xmlGenericError(xmlGenericErrorContext, |
12090 | | "PP: entering DTD\n"); |
12091 | | #endif |
12092 | 0 | break; |
12093 | 0 | case XML_PARSER_PI: |
12094 | 0 | xmlGenericError(xmlGenericErrorContext, |
12095 | 0 | "PP: internal error, state == PI\n"); |
12096 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12097 | | #ifdef DEBUG_PUSH |
12098 | | xmlGenericError(xmlGenericErrorContext, |
12099 | | "PP: entering CONTENT\n"); |
12100 | | #endif |
12101 | 0 | break; |
12102 | 0 | case XML_PARSER_ENTITY_DECL: |
12103 | 0 | xmlGenericError(xmlGenericErrorContext, |
12104 | 0 | "PP: internal error, state == ENTITY_DECL\n"); |
12105 | 0 | ctxt->instate = XML_PARSER_DTD; |
12106 | | #ifdef DEBUG_PUSH |
12107 | | xmlGenericError(xmlGenericErrorContext, |
12108 | | "PP: entering DTD\n"); |
12109 | | #endif |
12110 | 0 | break; |
12111 | 0 | case XML_PARSER_ENTITY_VALUE: |
12112 | 0 | xmlGenericError(xmlGenericErrorContext, |
12113 | 0 | "PP: internal error, state == ENTITY_VALUE\n"); |
12114 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12115 | | #ifdef DEBUG_PUSH |
12116 | | xmlGenericError(xmlGenericErrorContext, |
12117 | | "PP: entering DTD\n"); |
12118 | | #endif |
12119 | 0 | break; |
12120 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
12121 | 0 | xmlGenericError(xmlGenericErrorContext, |
12122 | 0 | "PP: internal error, state == ATTRIBUTE_VALUE\n"); |
12123 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12124 | | #ifdef DEBUG_PUSH |
12125 | | xmlGenericError(xmlGenericErrorContext, |
12126 | | "PP: entering START_TAG\n"); |
12127 | | #endif |
12128 | 0 | break; |
12129 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
12130 | 0 | xmlGenericError(xmlGenericErrorContext, |
12131 | 0 | "PP: internal error, state == SYSTEM_LITERAL\n"); |
12132 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12133 | | #ifdef DEBUG_PUSH |
12134 | | xmlGenericError(xmlGenericErrorContext, |
12135 | | "PP: entering START_TAG\n"); |
12136 | | #endif |
12137 | 0 | break; |
12138 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
12139 | 0 | xmlGenericError(xmlGenericErrorContext, |
12140 | 0 | "PP: internal error, state == PUBLIC_LITERAL\n"); |
12141 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12142 | | #ifdef DEBUG_PUSH |
12143 | | xmlGenericError(xmlGenericErrorContext, |
12144 | | "PP: entering START_TAG\n"); |
12145 | | #endif |
12146 | 0 | break; |
12147 | 0 | } |
12148 | 0 | } |
12149 | 0 | done: |
12150 | | #ifdef DEBUG_PUSH |
12151 | | xmlGenericError(xmlGenericErrorContext, "PP: done %d\n", ret); |
12152 | | #endif |
12153 | 0 | return(ret); |
12154 | 0 | encoding_error: |
12155 | 0 | { |
12156 | 0 | char buffer[150]; |
12157 | |
|
12158 | 0 | snprintf(buffer, 149, "Bytes: 0x%02X 0x%02X 0x%02X 0x%02X\n", |
12159 | 0 | ctxt->input->cur[0], ctxt->input->cur[1], |
12160 | 0 | ctxt->input->cur[2], ctxt->input->cur[3]); |
12161 | 0 | __xmlErrEncoding(ctxt, XML_ERR_INVALID_CHAR, |
12162 | 0 | "Input is not proper UTF-8, indicate encoding !\n%s", |
12163 | 0 | BAD_CAST buffer, NULL); |
12164 | 0 | } |
12165 | 0 | return(0); |
12166 | 0 | } |
12167 | | |
12168 | | /** |
12169 | | * xmlParseCheckTransition: |
12170 | | * @ctxt: an XML parser context |
12171 | | * @chunk: a char array |
12172 | | * @size: the size in byte of the chunk |
12173 | | * |
12174 | | * Check depending on the current parser state if the chunk given must be |
12175 | | * processed immediately or one need more data to advance on parsing. |
12176 | | * |
12177 | | * Returns -1 in case of error, 0 if the push is not needed and 1 if needed |
12178 | | */ |
12179 | | static int |
12180 | 0 | xmlParseCheckTransition(xmlParserCtxtPtr ctxt, const char *chunk, int size) { |
12181 | 0 | if ((ctxt == NULL) || (chunk == NULL) || (size < 0)) |
12182 | 0 | return(-1); |
12183 | 0 | if (ctxt->instate == XML_PARSER_START_TAG) { |
12184 | 0 | if (memchr(chunk, '>', size) != NULL) |
12185 | 0 | return(1); |
12186 | 0 | return(0); |
12187 | 0 | } |
12188 | 0 | if (ctxt->progressive == XML_PARSER_COMMENT) { |
12189 | 0 | if (memchr(chunk, '>', size) != NULL) |
12190 | 0 | return(1); |
12191 | 0 | return(0); |
12192 | 0 | } |
12193 | 0 | if (ctxt->instate == XML_PARSER_CDATA_SECTION) { |
12194 | 0 | if (memchr(chunk, '>', size) != NULL) |
12195 | 0 | return(1); |
12196 | 0 | return(0); |
12197 | 0 | } |
12198 | 0 | if (ctxt->progressive == XML_PARSER_PI) { |
12199 | 0 | if (memchr(chunk, '>', size) != NULL) |
12200 | 0 | return(1); |
12201 | 0 | return(0); |
12202 | 0 | } |
12203 | 0 | if (ctxt->instate == XML_PARSER_END_TAG) { |
12204 | 0 | if (memchr(chunk, '>', size) != NULL) |
12205 | 0 | return(1); |
12206 | 0 | return(0); |
12207 | 0 | } |
12208 | 0 | if ((ctxt->progressive == XML_PARSER_DTD) || |
12209 | 0 | (ctxt->instate == XML_PARSER_DTD)) { |
12210 | 0 | if (memchr(chunk, '>', size) != NULL) |
12211 | 0 | return(1); |
12212 | 0 | return(0); |
12213 | 0 | } |
12214 | 0 | return(1); |
12215 | 0 | } |
12216 | | |
12217 | | /** |
12218 | | * xmlParseChunk: |
12219 | | * @ctxt: an XML parser context |
12220 | | * @chunk: an char array |
12221 | | * @size: the size in byte of the chunk |
12222 | | * @terminate: last chunk indicator |
12223 | | * |
12224 | | * Parse a Chunk of memory |
12225 | | * |
12226 | | * Returns zero if no error, the xmlParserErrors otherwise. |
12227 | | */ |
12228 | | int |
12229 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
12230 | 0 | int terminate) { |
12231 | 0 | int end_in_lf = 0; |
12232 | 0 | int remain = 0; |
12233 | 0 | size_t old_avail = 0; |
12234 | 0 | size_t avail = 0; |
12235 | |
|
12236 | 0 | if (ctxt == NULL) |
12237 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12238 | 0 | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12239 | 0 | return(ctxt->errNo); |
12240 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
12241 | 0 | return(-1); |
12242 | 0 | if (ctxt->instate == XML_PARSER_START) |
12243 | 0 | xmlDetectSAX2(ctxt); |
12244 | 0 | if ((size > 0) && (chunk != NULL) && (!terminate) && |
12245 | 0 | (chunk[size - 1] == '\r')) { |
12246 | 0 | end_in_lf = 1; |
12247 | 0 | size--; |
12248 | 0 | } |
12249 | |
|
12250 | 0 | xmldecl_done: |
12251 | |
|
12252 | 0 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
12253 | 0 | (ctxt->input->buf != NULL) && (ctxt->instate != XML_PARSER_EOF)) { |
12254 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12255 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
12256 | 0 | int res; |
12257 | |
|
12258 | 0 | old_avail = xmlBufUse(ctxt->input->buf->buffer); |
12259 | | /* |
12260 | | * Specific handling if we autodetected an encoding, we should not |
12261 | | * push more than the first line ... which depend on the encoding |
12262 | | * And only push the rest once the final encoding was detected |
12263 | | */ |
12264 | 0 | if ((ctxt->instate == XML_PARSER_START) && (ctxt->input != NULL) && |
12265 | 0 | (ctxt->input->buf != NULL) && (ctxt->input->buf->encoder != NULL)) { |
12266 | 0 | unsigned int len = 45; |
12267 | |
|
12268 | 0 | if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12269 | 0 | BAD_CAST "UTF-16")) || |
12270 | 0 | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12271 | 0 | BAD_CAST "UTF16"))) |
12272 | 0 | len = 90; |
12273 | 0 | else if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12274 | 0 | BAD_CAST "UCS-4")) || |
12275 | 0 | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12276 | 0 | BAD_CAST "UCS4"))) |
12277 | 0 | len = 180; |
12278 | |
|
12279 | 0 | if (ctxt->input->buf->rawconsumed < len) |
12280 | 0 | len -= ctxt->input->buf->rawconsumed; |
12281 | | |
12282 | | /* |
12283 | | * Change size for reading the initial declaration only |
12284 | | * if size is greater than len. Otherwise, memmove in xmlBufferAdd |
12285 | | * will blindly copy extra bytes from memory. |
12286 | | */ |
12287 | 0 | if ((unsigned int) size > len) { |
12288 | 0 | remain = size - len; |
12289 | 0 | size = len; |
12290 | 0 | } else { |
12291 | 0 | remain = 0; |
12292 | 0 | } |
12293 | 0 | } |
12294 | 0 | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12295 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12296 | 0 | if (res < 0) { |
12297 | 0 | ctxt->errNo = XML_PARSER_EOF; |
12298 | 0 | xmlHaltParser(ctxt); |
12299 | 0 | return (XML_PARSER_EOF); |
12300 | 0 | } |
12301 | | #ifdef DEBUG_PUSH |
12302 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12303 | | #endif |
12304 | |
|
12305 | 0 | } else if (ctxt->instate != XML_PARSER_EOF) { |
12306 | 0 | if ((ctxt->input != NULL) && ctxt->input->buf != NULL) { |
12307 | 0 | xmlParserInputBufferPtr in = ctxt->input->buf; |
12308 | 0 | if ((in->encoder != NULL) && (in->buffer != NULL) && |
12309 | 0 | (in->raw != NULL)) { |
12310 | 0 | int nbchars; |
12311 | 0 | size_t base = xmlBufGetInputBase(in->buffer, ctxt->input); |
12312 | 0 | size_t current = ctxt->input->cur - ctxt->input->base; |
12313 | |
|
12314 | 0 | nbchars = xmlCharEncInput(in, terminate); |
12315 | 0 | xmlBufSetInputBaseCur(in->buffer, ctxt->input, base, current); |
12316 | 0 | if (nbchars < 0) { |
12317 | | /* TODO 2.6.0 */ |
12318 | 0 | xmlGenericError(xmlGenericErrorContext, |
12319 | 0 | "xmlParseChunk: encoder error\n"); |
12320 | 0 | xmlHaltParser(ctxt); |
12321 | 0 | return(XML_ERR_INVALID_ENCODING); |
12322 | 0 | } |
12323 | 0 | } |
12324 | 0 | } |
12325 | 0 | } |
12326 | 0 | if (remain != 0) { |
12327 | 0 | xmlParseTryOrFinish(ctxt, 0); |
12328 | 0 | } else { |
12329 | 0 | if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) |
12330 | 0 | avail = xmlBufUse(ctxt->input->buf->buffer); |
12331 | | /* |
12332 | | * Depending on the current state it may not be such |
12333 | | * a good idea to try parsing if there is nothing in the chunk |
12334 | | * which would be worth doing a parser state transition and we |
12335 | | * need to wait for more data |
12336 | | */ |
12337 | 0 | if ((terminate) || (avail > XML_MAX_TEXT_LENGTH) || |
12338 | 0 | (old_avail == 0) || (avail == 0) || |
12339 | 0 | (xmlParseCheckTransition(ctxt, |
12340 | 0 | (const char *)&ctxt->input->base[old_avail], |
12341 | 0 | avail - old_avail))) |
12342 | 0 | xmlParseTryOrFinish(ctxt, terminate); |
12343 | 0 | } |
12344 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
12345 | 0 | return(ctxt->errNo); |
12346 | | |
12347 | 0 | if ((ctxt->input != NULL) && |
12348 | 0 | (((ctxt->input->end - ctxt->input->cur) > XML_MAX_LOOKUP_LIMIT) || |
12349 | 0 | ((ctxt->input->cur - ctxt->input->base) > XML_MAX_LOOKUP_LIMIT)) && |
12350 | 0 | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
12351 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
12352 | 0 | xmlHaltParser(ctxt); |
12353 | 0 | } |
12354 | 0 | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12355 | 0 | return(ctxt->errNo); |
12356 | | |
12357 | 0 | if (remain != 0) { |
12358 | 0 | chunk += size; |
12359 | 0 | size = remain; |
12360 | 0 | remain = 0; |
12361 | 0 | goto xmldecl_done; |
12362 | 0 | } |
12363 | 0 | if ((end_in_lf == 1) && (ctxt->input != NULL) && |
12364 | 0 | (ctxt->input->buf != NULL)) { |
12365 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
12366 | 0 | ctxt->input); |
12367 | 0 | size_t current = ctxt->input->cur - ctxt->input->base; |
12368 | |
|
12369 | 0 | xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
12370 | |
|
12371 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
12372 | 0 | base, current); |
12373 | 0 | } |
12374 | 0 | if (terminate) { |
12375 | | /* |
12376 | | * Check for termination |
12377 | | */ |
12378 | 0 | int cur_avail = 0; |
12379 | |
|
12380 | 0 | if (ctxt->input != NULL) { |
12381 | 0 | if (ctxt->input->buf == NULL) |
12382 | 0 | cur_avail = ctxt->input->length - |
12383 | 0 | (ctxt->input->cur - ctxt->input->base); |
12384 | 0 | else |
12385 | 0 | cur_avail = xmlBufUse(ctxt->input->buf->buffer) - |
12386 | 0 | (ctxt->input->cur - ctxt->input->base); |
12387 | 0 | } |
12388 | |
|
12389 | 0 | if ((ctxt->instate != XML_PARSER_EOF) && |
12390 | 0 | (ctxt->instate != XML_PARSER_EPILOG)) { |
12391 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12392 | 0 | } |
12393 | 0 | if ((ctxt->instate == XML_PARSER_EPILOG) && (cur_avail > 0)) { |
12394 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12395 | 0 | } |
12396 | 0 | if (ctxt->instate != XML_PARSER_EOF) { |
12397 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
12398 | 0 | ctxt->sax->endDocument(ctxt->userData); |
12399 | 0 | } |
12400 | 0 | ctxt->instate = XML_PARSER_EOF; |
12401 | 0 | } |
12402 | 0 | if (ctxt->wellFormed == 0) |
12403 | 0 | return((xmlParserErrors) ctxt->errNo); |
12404 | 0 | else |
12405 | 0 | return(0); |
12406 | 0 | } |
12407 | | |
12408 | | /************************************************************************ |
12409 | | * * |
12410 | | * I/O front end functions to the parser * |
12411 | | * * |
12412 | | ************************************************************************/ |
12413 | | |
12414 | | /** |
12415 | | * xmlCreatePushParserCtxt: |
12416 | | * @sax: a SAX handler |
12417 | | * @user_data: The user data returned on SAX callbacks |
12418 | | * @chunk: a pointer to an array of chars |
12419 | | * @size: number of chars in the array |
12420 | | * @filename: an optional file name or URI |
12421 | | * |
12422 | | * Create a parser context for using the XML parser in push mode. |
12423 | | * If @buffer and @size are non-NULL, the data is used to detect |
12424 | | * the encoding. The remaining characters will be parsed so they |
12425 | | * don't need to be fed in again through xmlParseChunk. |
12426 | | * To allow content encoding detection, @size should be >= 4 |
12427 | | * The value of @filename is used for fetching external entities |
12428 | | * and error/warning reports. |
12429 | | * |
12430 | | * Returns the new parser context or NULL |
12431 | | */ |
12432 | | |
12433 | | xmlParserCtxtPtr |
12434 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12435 | 0 | const char *chunk, int size, const char *filename) { |
12436 | 0 | xmlParserCtxtPtr ctxt; |
12437 | 0 | xmlParserInputPtr inputStream; |
12438 | 0 | xmlParserInputBufferPtr buf; |
12439 | 0 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
12440 | | |
12441 | | /* |
12442 | | * plug some encoding conversion routines |
12443 | | */ |
12444 | 0 | if ((chunk != NULL) && (size >= 4)) |
12445 | 0 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
12446 | |
|
12447 | 0 | buf = xmlAllocParserInputBuffer(enc); |
12448 | 0 | if (buf == NULL) return(NULL); |
12449 | | |
12450 | 0 | ctxt = xmlNewParserCtxt(); |
12451 | 0 | if (ctxt == NULL) { |
12452 | 0 | xmlErrMemory(NULL, "creating parser: out of memory\n"); |
12453 | 0 | xmlFreeParserInputBuffer(buf); |
12454 | 0 | return(NULL); |
12455 | 0 | } |
12456 | 0 | ctxt->dictNames = 1; |
12457 | 0 | if (sax != NULL) { |
12458 | 0 | #ifdef LIBXML_SAX1_ENABLED |
12459 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
12460 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
12461 | 0 | xmlFree(ctxt->sax); |
12462 | 0 | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
12463 | 0 | if (ctxt->sax == NULL) { |
12464 | 0 | xmlErrMemory(ctxt, NULL); |
12465 | 0 | xmlFreeParserInputBuffer(buf); |
12466 | 0 | xmlFreeParserCtxt(ctxt); |
12467 | 0 | return(NULL); |
12468 | 0 | } |
12469 | 0 | memset(ctxt->sax, 0, sizeof(xmlSAXHandler)); |
12470 | 0 | if (sax->initialized == XML_SAX2_MAGIC) |
12471 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
12472 | 0 | else |
12473 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12474 | 0 | if (user_data != NULL) |
12475 | 0 | ctxt->userData = user_data; |
12476 | 0 | } |
12477 | 0 | if (filename == NULL) { |
12478 | 0 | ctxt->directory = NULL; |
12479 | 0 | } else { |
12480 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
12481 | 0 | } |
12482 | |
|
12483 | 0 | inputStream = xmlNewInputStream(ctxt); |
12484 | 0 | if (inputStream == NULL) { |
12485 | 0 | xmlFreeParserCtxt(ctxt); |
12486 | 0 | xmlFreeParserInputBuffer(buf); |
12487 | 0 | return(NULL); |
12488 | 0 | } |
12489 | | |
12490 | 0 | if (filename == NULL) |
12491 | 0 | inputStream->filename = NULL; |
12492 | 0 | else { |
12493 | 0 | inputStream->filename = (char *) |
12494 | 0 | xmlCanonicPath((const xmlChar *) filename); |
12495 | 0 | if (inputStream->filename == NULL) { |
12496 | 0 | xmlFreeParserCtxt(ctxt); |
12497 | 0 | xmlFreeParserInputBuffer(buf); |
12498 | 0 | return(NULL); |
12499 | 0 | } |
12500 | 0 | } |
12501 | 0 | inputStream->buf = buf; |
12502 | 0 | xmlBufResetInput(inputStream->buf->buffer, inputStream); |
12503 | 0 | inputPush(ctxt, inputStream); |
12504 | | |
12505 | | /* |
12506 | | * If the caller didn't provide an initial 'chunk' for determining |
12507 | | * the encoding, we set the context to XML_CHAR_ENCODING_NONE so |
12508 | | * that it can be automatically determined later |
12509 | | */ |
12510 | 0 | if ((size == 0) || (chunk == NULL)) { |
12511 | 0 | ctxt->charset = XML_CHAR_ENCODING_NONE; |
12512 | 0 | } else if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) { |
12513 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12514 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
12515 | |
|
12516 | 0 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12517 | |
|
12518 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12519 | | #ifdef DEBUG_PUSH |
12520 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12521 | | #endif |
12522 | 0 | } |
12523 | |
|
12524 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12525 | 0 | xmlSwitchEncoding(ctxt, enc); |
12526 | 0 | } |
12527 | |
|
12528 | 0 | return(ctxt); |
12529 | 0 | } |
12530 | | #endif /* LIBXML_PUSH_ENABLED */ |
12531 | | |
12532 | | /** |
12533 | | * xmlHaltParser: |
12534 | | * @ctxt: an XML parser context |
12535 | | * |
12536 | | * Blocks further parser processing don't override error |
12537 | | * for internal use |
12538 | | */ |
12539 | | static void |
12540 | 287k | xmlHaltParser(xmlParserCtxtPtr ctxt) { |
12541 | 287k | if (ctxt == NULL) |
12542 | 0 | return; |
12543 | 287k | ctxt->instate = XML_PARSER_EOF; |
12544 | 287k | ctxt->disableSAX = 1; |
12545 | 288k | while (ctxt->inputNr > 1) |
12546 | 1.14k | xmlFreeInputStream(inputPop(ctxt)); |
12547 | 287k | if (ctxt->input != NULL) { |
12548 | | /* |
12549 | | * in case there was a specific allocation deallocate before |
12550 | | * overriding base |
12551 | | */ |
12552 | 287k | if (ctxt->input->free != NULL) { |
12553 | 0 | ctxt->input->free((xmlChar *) ctxt->input->base); |
12554 | 0 | ctxt->input->free = NULL; |
12555 | 0 | } |
12556 | 287k | if (ctxt->input->buf != NULL) { |
12557 | 234k | xmlFreeParserInputBuffer(ctxt->input->buf); |
12558 | 234k | ctxt->input->buf = NULL; |
12559 | 234k | } |
12560 | 287k | ctxt->input->cur = BAD_CAST""; |
12561 | 287k | ctxt->input->length = 0; |
12562 | 287k | ctxt->input->base = ctxt->input->cur; |
12563 | 287k | ctxt->input->end = ctxt->input->cur; |
12564 | 287k | } |
12565 | 287k | } |
12566 | | |
12567 | | /** |
12568 | | * xmlStopParser: |
12569 | | * @ctxt: an XML parser context |
12570 | | * |
12571 | | * Blocks further parser processing |
12572 | | */ |
12573 | | void |
12574 | 16 | xmlStopParser(xmlParserCtxtPtr ctxt) { |
12575 | 16 | if (ctxt == NULL) |
12576 | 0 | return; |
12577 | 16 | xmlHaltParser(ctxt); |
12578 | 16 | ctxt->errNo = XML_ERR_USER_STOP; |
12579 | 16 | } |
12580 | | |
12581 | | /** |
12582 | | * xmlCreateIOParserCtxt: |
12583 | | * @sax: a SAX handler |
12584 | | * @user_data: The user data returned on SAX callbacks |
12585 | | * @ioread: an I/O read function |
12586 | | * @ioclose: an I/O close function |
12587 | | * @ioctx: an I/O handler |
12588 | | * @enc: the charset encoding if known |
12589 | | * |
12590 | | * Create a parser context for using the XML parser with an existing |
12591 | | * I/O stream |
12592 | | * |
12593 | | * Returns the new parser context or NULL |
12594 | | */ |
12595 | | xmlParserCtxtPtr |
12596 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12597 | | xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
12598 | 0 | void *ioctx, xmlCharEncoding enc) { |
12599 | 0 | xmlParserCtxtPtr ctxt; |
12600 | 0 | xmlParserInputPtr inputStream; |
12601 | 0 | xmlParserInputBufferPtr buf; |
12602 | |
|
12603 | 0 | if (ioread == NULL) return(NULL); |
12604 | | |
12605 | 0 | buf = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, enc); |
12606 | 0 | if (buf == NULL) { |
12607 | 0 | if (ioclose != NULL) |
12608 | 0 | ioclose(ioctx); |
12609 | 0 | return (NULL); |
12610 | 0 | } |
12611 | | |
12612 | 0 | ctxt = xmlNewParserCtxt(); |
12613 | 0 | if (ctxt == NULL) { |
12614 | 0 | xmlFreeParserInputBuffer(buf); |
12615 | 0 | return(NULL); |
12616 | 0 | } |
12617 | 0 | if (sax != NULL) { |
12618 | 0 | #ifdef LIBXML_SAX1_ENABLED |
12619 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
12620 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
12621 | 0 | xmlFree(ctxt->sax); |
12622 | 0 | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
12623 | 0 | if (ctxt->sax == NULL) { |
12624 | 0 | xmlFreeParserInputBuffer(buf); |
12625 | 0 | xmlErrMemory(ctxt, NULL); |
12626 | 0 | xmlFreeParserCtxt(ctxt); |
12627 | 0 | return(NULL); |
12628 | 0 | } |
12629 | 0 | memset(ctxt->sax, 0, sizeof(xmlSAXHandler)); |
12630 | 0 | if (sax->initialized == XML_SAX2_MAGIC) |
12631 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
12632 | 0 | else |
12633 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12634 | 0 | if (user_data != NULL) |
12635 | 0 | ctxt->userData = user_data; |
12636 | 0 | } |
12637 | | |
12638 | 0 | inputStream = xmlNewIOInputStream(ctxt, buf, enc); |
12639 | 0 | if (inputStream == NULL) { |
12640 | 0 | xmlFreeParserCtxt(ctxt); |
12641 | 0 | return(NULL); |
12642 | 0 | } |
12643 | 0 | inputPush(ctxt, inputStream); |
12644 | |
|
12645 | 0 | return(ctxt); |
12646 | 0 | } |
12647 | | |
12648 | | #ifdef LIBXML_VALID_ENABLED |
12649 | | /************************************************************************ |
12650 | | * * |
12651 | | * Front ends when parsing a DTD * |
12652 | | * * |
12653 | | ************************************************************************/ |
12654 | | |
12655 | | /** |
12656 | | * xmlIOParseDTD: |
12657 | | * @sax: the SAX handler block or NULL |
12658 | | * @input: an Input Buffer |
12659 | | * @enc: the charset encoding if known |
12660 | | * |
12661 | | * Load and parse a DTD |
12662 | | * |
12663 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12664 | | * @input will be freed by the function in any case. |
12665 | | */ |
12666 | | |
12667 | | xmlDtdPtr |
12668 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
12669 | 0 | xmlCharEncoding enc) { |
12670 | 0 | xmlDtdPtr ret = NULL; |
12671 | 0 | xmlParserCtxtPtr ctxt; |
12672 | 0 | xmlParserInputPtr pinput = NULL; |
12673 | 0 | xmlChar start[4]; |
12674 | |
|
12675 | 0 | if (input == NULL) |
12676 | 0 | return(NULL); |
12677 | | |
12678 | 0 | ctxt = xmlNewParserCtxt(); |
12679 | 0 | if (ctxt == NULL) { |
12680 | 0 | xmlFreeParserInputBuffer(input); |
12681 | 0 | return(NULL); |
12682 | 0 | } |
12683 | | |
12684 | | /* We are loading a DTD */ |
12685 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12686 | | |
12687 | | /* |
12688 | | * Set-up the SAX context |
12689 | | */ |
12690 | 0 | if (sax != NULL) { |
12691 | 0 | if (ctxt->sax != NULL) |
12692 | 0 | xmlFree(ctxt->sax); |
12693 | 0 | ctxt->sax = sax; |
12694 | 0 | ctxt->userData = ctxt; |
12695 | 0 | } |
12696 | 0 | xmlDetectSAX2(ctxt); |
12697 | | |
12698 | | /* |
12699 | | * generate a parser input from the I/O handler |
12700 | | */ |
12701 | |
|
12702 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
12703 | 0 | if (pinput == NULL) { |
12704 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12705 | 0 | xmlFreeParserInputBuffer(input); |
12706 | 0 | xmlFreeParserCtxt(ctxt); |
12707 | 0 | return(NULL); |
12708 | 0 | } |
12709 | | |
12710 | | /* |
12711 | | * plug some encoding conversion routines here. |
12712 | | */ |
12713 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
12714 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12715 | 0 | xmlFreeParserCtxt(ctxt); |
12716 | 0 | return(NULL); |
12717 | 0 | } |
12718 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12719 | 0 | xmlSwitchEncoding(ctxt, enc); |
12720 | 0 | } |
12721 | |
|
12722 | 0 | pinput->filename = NULL; |
12723 | 0 | pinput->line = 1; |
12724 | 0 | pinput->col = 1; |
12725 | 0 | pinput->base = ctxt->input->cur; |
12726 | 0 | pinput->cur = ctxt->input->cur; |
12727 | 0 | pinput->free = NULL; |
12728 | | |
12729 | | /* |
12730 | | * let's parse that entity knowing it's an external subset. |
12731 | | */ |
12732 | 0 | ctxt->inSubset = 2; |
12733 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12734 | 0 | if (ctxt->myDoc == NULL) { |
12735 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12736 | 0 | return(NULL); |
12737 | 0 | } |
12738 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12739 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12740 | 0 | BAD_CAST "none", BAD_CAST "none"); |
12741 | |
|
12742 | 0 | if ((enc == XML_CHAR_ENCODING_NONE) && |
12743 | 0 | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
12744 | | /* |
12745 | | * Get the 4 first bytes and decode the charset |
12746 | | * if enc != XML_CHAR_ENCODING_NONE |
12747 | | * plug some encoding conversion routines. |
12748 | | */ |
12749 | 0 | start[0] = RAW; |
12750 | 0 | start[1] = NXT(1); |
12751 | 0 | start[2] = NXT(2); |
12752 | 0 | start[3] = NXT(3); |
12753 | 0 | enc = xmlDetectCharEncoding(start, 4); |
12754 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12755 | 0 | xmlSwitchEncoding(ctxt, enc); |
12756 | 0 | } |
12757 | 0 | } |
12758 | |
|
12759 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
12760 | |
|
12761 | 0 | if (ctxt->myDoc != NULL) { |
12762 | 0 | if (ctxt->wellFormed) { |
12763 | 0 | ret = ctxt->myDoc->extSubset; |
12764 | 0 | ctxt->myDoc->extSubset = NULL; |
12765 | 0 | if (ret != NULL) { |
12766 | 0 | xmlNodePtr tmp; |
12767 | |
|
12768 | 0 | ret->doc = NULL; |
12769 | 0 | tmp = ret->children; |
12770 | 0 | while (tmp != NULL) { |
12771 | 0 | tmp->doc = NULL; |
12772 | 0 | tmp = tmp->next; |
12773 | 0 | } |
12774 | 0 | } |
12775 | 0 | } else { |
12776 | 0 | ret = NULL; |
12777 | 0 | } |
12778 | 0 | xmlFreeDoc(ctxt->myDoc); |
12779 | 0 | ctxt->myDoc = NULL; |
12780 | 0 | } |
12781 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12782 | 0 | xmlFreeParserCtxt(ctxt); |
12783 | |
|
12784 | 0 | return(ret); |
12785 | 0 | } |
12786 | | |
12787 | | /** |
12788 | | * xmlSAXParseDTD: |
12789 | | * @sax: the SAX handler block |
12790 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12791 | | * @SystemID: a NAME* containing the URL to the DTD |
12792 | | * |
12793 | | * Load and parse an external subset. |
12794 | | * |
12795 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12796 | | */ |
12797 | | |
12798 | | xmlDtdPtr |
12799 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
12800 | 0 | const xmlChar *SystemID) { |
12801 | 0 | xmlDtdPtr ret = NULL; |
12802 | 0 | xmlParserCtxtPtr ctxt; |
12803 | 0 | xmlParserInputPtr input = NULL; |
12804 | 0 | xmlCharEncoding enc; |
12805 | 0 | xmlChar* systemIdCanonic; |
12806 | |
|
12807 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
12808 | | |
12809 | 0 | ctxt = xmlNewParserCtxt(); |
12810 | 0 | if (ctxt == NULL) { |
12811 | 0 | return(NULL); |
12812 | 0 | } |
12813 | | |
12814 | | /* We are loading a DTD */ |
12815 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12816 | | |
12817 | | /* |
12818 | | * Set-up the SAX context |
12819 | | */ |
12820 | 0 | if (sax != NULL) { |
12821 | 0 | if (ctxt->sax != NULL) |
12822 | 0 | xmlFree(ctxt->sax); |
12823 | 0 | ctxt->sax = sax; |
12824 | 0 | ctxt->userData = ctxt; |
12825 | 0 | } |
12826 | | |
12827 | | /* |
12828 | | * Canonicalise the system ID |
12829 | | */ |
12830 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
12831 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
12832 | 0 | xmlFreeParserCtxt(ctxt); |
12833 | 0 | return(NULL); |
12834 | 0 | } |
12835 | | |
12836 | | /* |
12837 | | * Ask the Entity resolver to load the damn thing |
12838 | | */ |
12839 | | |
12840 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
12841 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
12842 | 0 | systemIdCanonic); |
12843 | 0 | if (input == NULL) { |
12844 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12845 | 0 | xmlFreeParserCtxt(ctxt); |
12846 | 0 | if (systemIdCanonic != NULL) |
12847 | 0 | xmlFree(systemIdCanonic); |
12848 | 0 | return(NULL); |
12849 | 0 | } |
12850 | | |
12851 | | /* |
12852 | | * plug some encoding conversion routines here. |
12853 | | */ |
12854 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
12855 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12856 | 0 | xmlFreeParserCtxt(ctxt); |
12857 | 0 | if (systemIdCanonic != NULL) |
12858 | 0 | xmlFree(systemIdCanonic); |
12859 | 0 | return(NULL); |
12860 | 0 | } |
12861 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
12862 | 0 | enc = xmlDetectCharEncoding(ctxt->input->cur, 4); |
12863 | 0 | xmlSwitchEncoding(ctxt, enc); |
12864 | 0 | } |
12865 | |
|
12866 | 0 | if (input->filename == NULL) |
12867 | 0 | input->filename = (char *) systemIdCanonic; |
12868 | 0 | else |
12869 | 0 | xmlFree(systemIdCanonic); |
12870 | 0 | input->line = 1; |
12871 | 0 | input->col = 1; |
12872 | 0 | input->base = ctxt->input->cur; |
12873 | 0 | input->cur = ctxt->input->cur; |
12874 | 0 | input->free = NULL; |
12875 | | |
12876 | | /* |
12877 | | * let's parse that entity knowing it's an external subset. |
12878 | | */ |
12879 | 0 | ctxt->inSubset = 2; |
12880 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12881 | 0 | if (ctxt->myDoc == NULL) { |
12882 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12883 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12884 | 0 | xmlFreeParserCtxt(ctxt); |
12885 | 0 | return(NULL); |
12886 | 0 | } |
12887 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12888 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12889 | 0 | ExternalID, SystemID); |
12890 | 0 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
12891 | |
|
12892 | 0 | if (ctxt->myDoc != NULL) { |
12893 | 0 | if (ctxt->wellFormed) { |
12894 | 0 | ret = ctxt->myDoc->extSubset; |
12895 | 0 | ctxt->myDoc->extSubset = NULL; |
12896 | 0 | if (ret != NULL) { |
12897 | 0 | xmlNodePtr tmp; |
12898 | |
|
12899 | 0 | ret->doc = NULL; |
12900 | 0 | tmp = ret->children; |
12901 | 0 | while (tmp != NULL) { |
12902 | 0 | tmp->doc = NULL; |
12903 | 0 | tmp = tmp->next; |
12904 | 0 | } |
12905 | 0 | } |
12906 | 0 | } else { |
12907 | 0 | ret = NULL; |
12908 | 0 | } |
12909 | 0 | xmlFreeDoc(ctxt->myDoc); |
12910 | 0 | ctxt->myDoc = NULL; |
12911 | 0 | } |
12912 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12913 | 0 | xmlFreeParserCtxt(ctxt); |
12914 | |
|
12915 | 0 | return(ret); |
12916 | 0 | } |
12917 | | |
12918 | | |
12919 | | /** |
12920 | | * xmlParseDTD: |
12921 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12922 | | * @SystemID: a NAME* containing the URL to the DTD |
12923 | | * |
12924 | | * Load and parse an external subset. |
12925 | | * |
12926 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12927 | | */ |
12928 | | |
12929 | | xmlDtdPtr |
12930 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
12931 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
12932 | 0 | } |
12933 | | #endif /* LIBXML_VALID_ENABLED */ |
12934 | | |
12935 | | /************************************************************************ |
12936 | | * * |
12937 | | * Front ends when parsing an Entity * |
12938 | | * * |
12939 | | ************************************************************************/ |
12940 | | |
12941 | | /** |
12942 | | * xmlParseCtxtExternalEntity: |
12943 | | * @ctx: the existing parsing context |
12944 | | * @URL: the URL for the entity to load |
12945 | | * @ID: the System ID for the entity to load |
12946 | | * @lst: the return value for the set of parsed nodes |
12947 | | * |
12948 | | * Parse an external general entity within an existing parsing context |
12949 | | * An external general parsed entity is well-formed if it matches the |
12950 | | * production labeled extParsedEnt. |
12951 | | * |
12952 | | * [78] extParsedEnt ::= TextDecl? content |
12953 | | * |
12954 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12955 | | * the parser error code otherwise |
12956 | | */ |
12957 | | |
12958 | | int |
12959 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx, const xmlChar *URL, |
12960 | 0 | const xmlChar *ID, xmlNodePtr *lst) { |
12961 | 0 | void *userData; |
12962 | |
|
12963 | 0 | if (ctx == NULL) return(-1); |
12964 | | /* |
12965 | | * If the user provided their own SAX callbacks, then reuse the |
12966 | | * userData callback field, otherwise the expected setup in a |
12967 | | * DOM builder is to have userData == ctxt |
12968 | | */ |
12969 | 0 | if (ctx->userData == ctx) |
12970 | 0 | userData = NULL; |
12971 | 0 | else |
12972 | 0 | userData = ctx->userData; |
12973 | 0 | return xmlParseExternalEntityPrivate(ctx->myDoc, ctx, ctx->sax, |
12974 | 0 | userData, ctx->depth + 1, |
12975 | 0 | URL, ID, lst); |
12976 | 0 | } |
12977 | | |
12978 | | /** |
12979 | | * xmlParseExternalEntityPrivate: |
12980 | | * @doc: the document the chunk pertains to |
12981 | | * @oldctxt: the previous parser context if available |
12982 | | * @sax: the SAX handler block (possibly NULL) |
12983 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12984 | | * @depth: Used for loop detection, use 0 |
12985 | | * @URL: the URL for the entity to load |
12986 | | * @ID: the System ID for the entity to load |
12987 | | * @list: the return value for the set of parsed nodes |
12988 | | * |
12989 | | * Private version of xmlParseExternalEntity() |
12990 | | * |
12991 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12992 | | * the parser error code otherwise |
12993 | | */ |
12994 | | |
12995 | | static xmlParserErrors |
12996 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
12997 | | xmlSAXHandlerPtr sax, |
12998 | | void *user_data, int depth, const xmlChar *URL, |
12999 | 0 | const xmlChar *ID, xmlNodePtr *list) { |
13000 | 0 | xmlParserCtxtPtr ctxt; |
13001 | 0 | xmlDocPtr newDoc; |
13002 | 0 | xmlNodePtr newRoot; |
13003 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13004 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13005 | 0 | xmlChar start[4]; |
13006 | 0 | xmlCharEncoding enc; |
13007 | |
|
13008 | 0 | if (((depth > 40) && |
13009 | 0 | ((oldctxt == NULL) || (oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13010 | 0 | (depth > 1024)) { |
13011 | 0 | return(XML_ERR_ENTITY_LOOP); |
13012 | 0 | } |
13013 | | |
13014 | 0 | if (list != NULL) |
13015 | 0 | *list = NULL; |
13016 | 0 | if ((URL == NULL) && (ID == NULL)) |
13017 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13018 | 0 | if (doc == NULL) |
13019 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13020 | | |
13021 | | |
13022 | 0 | ctxt = xmlCreateEntityParserCtxtInternal(URL, ID, NULL, oldctxt); |
13023 | 0 | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13024 | 0 | ctxt->userData = ctxt; |
13025 | 0 | if (sax != NULL) { |
13026 | 0 | oldsax = ctxt->sax; |
13027 | 0 | ctxt->sax = sax; |
13028 | 0 | if (user_data != NULL) |
13029 | 0 | ctxt->userData = user_data; |
13030 | 0 | } |
13031 | 0 | xmlDetectSAX2(ctxt); |
13032 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13033 | 0 | if (newDoc == NULL) { |
13034 | 0 | xmlFreeParserCtxt(ctxt); |
13035 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13036 | 0 | } |
13037 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13038 | 0 | if (doc) { |
13039 | 0 | newDoc->intSubset = doc->intSubset; |
13040 | 0 | newDoc->extSubset = doc->extSubset; |
13041 | 0 | if (doc->dict) { |
13042 | 0 | newDoc->dict = doc->dict; |
13043 | 0 | xmlDictReference(newDoc->dict); |
13044 | 0 | } |
13045 | 0 | if (doc->URL != NULL) { |
13046 | 0 | newDoc->URL = xmlStrdup(doc->URL); |
13047 | 0 | } |
13048 | 0 | } |
13049 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13050 | 0 | if (newRoot == NULL) { |
13051 | 0 | if (sax != NULL) |
13052 | 0 | ctxt->sax = oldsax; |
13053 | 0 | xmlFreeParserCtxt(ctxt); |
13054 | 0 | newDoc->intSubset = NULL; |
13055 | 0 | newDoc->extSubset = NULL; |
13056 | 0 | xmlFreeDoc(newDoc); |
13057 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13058 | 0 | } |
13059 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13060 | 0 | nodePush(ctxt, newDoc->children); |
13061 | 0 | if (doc == NULL) { |
13062 | 0 | ctxt->myDoc = newDoc; |
13063 | 0 | } else { |
13064 | 0 | ctxt->myDoc = doc; |
13065 | 0 | newRoot->doc = doc; |
13066 | 0 | } |
13067 | | |
13068 | | /* |
13069 | | * Get the 4 first bytes and decode the charset |
13070 | | * if enc != XML_CHAR_ENCODING_NONE |
13071 | | * plug some encoding conversion routines. |
13072 | | */ |
13073 | 0 | GROW; |
13074 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
13075 | 0 | start[0] = RAW; |
13076 | 0 | start[1] = NXT(1); |
13077 | 0 | start[2] = NXT(2); |
13078 | 0 | start[3] = NXT(3); |
13079 | 0 | enc = xmlDetectCharEncoding(start, 4); |
13080 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
13081 | 0 | xmlSwitchEncoding(ctxt, enc); |
13082 | 0 | } |
13083 | 0 | } |
13084 | | |
13085 | | /* |
13086 | | * Parse a possible text declaration first |
13087 | | */ |
13088 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
13089 | 0 | xmlParseTextDecl(ctxt); |
13090 | | /* |
13091 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
13092 | | */ |
13093 | 0 | if ((xmlStrEqual(oldctxt->version, BAD_CAST "1.0")) && |
13094 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
13095 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
13096 | 0 | "Version mismatch between document and entity\n"); |
13097 | 0 | } |
13098 | 0 | } |
13099 | |
|
13100 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13101 | 0 | ctxt->depth = depth; |
13102 | 0 | if (oldctxt != NULL) { |
13103 | 0 | ctxt->_private = oldctxt->_private; |
13104 | 0 | ctxt->loadsubset = oldctxt->loadsubset; |
13105 | 0 | ctxt->validate = oldctxt->validate; |
13106 | 0 | ctxt->valid = oldctxt->valid; |
13107 | 0 | ctxt->replaceEntities = oldctxt->replaceEntities; |
13108 | 0 | if (oldctxt->validate) { |
13109 | 0 | ctxt->vctxt.error = oldctxt->vctxt.error; |
13110 | 0 | ctxt->vctxt.warning = oldctxt->vctxt.warning; |
13111 | 0 | ctxt->vctxt.userData = oldctxt->vctxt.userData; |
13112 | 0 | } |
13113 | 0 | ctxt->external = oldctxt->external; |
13114 | 0 | if (ctxt->dict) xmlDictFree(ctxt->dict); |
13115 | 0 | ctxt->dict = oldctxt->dict; |
13116 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13117 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13118 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13119 | 0 | ctxt->dictNames = oldctxt->dictNames; |
13120 | 0 | ctxt->attsDefault = oldctxt->attsDefault; |
13121 | 0 | ctxt->attsSpecial = oldctxt->attsSpecial; |
13122 | 0 | ctxt->linenumbers = oldctxt->linenumbers; |
13123 | 0 | ctxt->record_info = oldctxt->record_info; |
13124 | 0 | ctxt->node_seq.maximum = oldctxt->node_seq.maximum; |
13125 | 0 | ctxt->node_seq.length = oldctxt->node_seq.length; |
13126 | 0 | ctxt->node_seq.buffer = oldctxt->node_seq.buffer; |
13127 | 0 | } else { |
13128 | | /* |
13129 | | * Doing validity checking on chunk without context |
13130 | | * doesn't make sense |
13131 | | */ |
13132 | 0 | ctxt->_private = NULL; |
13133 | 0 | ctxt->validate = 0; |
13134 | 0 | ctxt->external = 2; |
13135 | 0 | ctxt->loadsubset = 0; |
13136 | 0 | } |
13137 | |
|
13138 | 0 | xmlParseContent(ctxt); |
13139 | |
|
13140 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13141 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13142 | 0 | } else if (RAW != 0) { |
13143 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13144 | 0 | } |
13145 | 0 | if (ctxt->node != newDoc->children) { |
13146 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13147 | 0 | } |
13148 | |
|
13149 | 0 | if (!ctxt->wellFormed) { |
13150 | 0 | if (ctxt->errNo == 0) |
13151 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13152 | 0 | else |
13153 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13154 | 0 | } else { |
13155 | 0 | if (list != NULL) { |
13156 | 0 | xmlNodePtr cur; |
13157 | | |
13158 | | /* |
13159 | | * Return the newly created nodeset after unlinking it from |
13160 | | * they pseudo parent. |
13161 | | */ |
13162 | 0 | cur = newDoc->children->children; |
13163 | 0 | *list = cur; |
13164 | 0 | while (cur != NULL) { |
13165 | 0 | cur->parent = NULL; |
13166 | 0 | cur = cur->next; |
13167 | 0 | } |
13168 | 0 | newDoc->children->children = NULL; |
13169 | 0 | } |
13170 | 0 | ret = XML_ERR_OK; |
13171 | 0 | } |
13172 | | |
13173 | | /* |
13174 | | * Record in the parent context the number of entities replacement |
13175 | | * done when parsing that reference. |
13176 | | */ |
13177 | 0 | if (oldctxt != NULL) |
13178 | 0 | oldctxt->nbentities += ctxt->nbentities; |
13179 | | |
13180 | | /* |
13181 | | * Also record the size of the entity parsed |
13182 | | */ |
13183 | 0 | if (ctxt->input != NULL && oldctxt != NULL) { |
13184 | 0 | oldctxt->sizeentities += ctxt->input->consumed; |
13185 | 0 | oldctxt->sizeentities += (ctxt->input->cur - ctxt->input->base); |
13186 | 0 | } |
13187 | | /* |
13188 | | * And record the last error if any |
13189 | | */ |
13190 | 0 | if ((oldctxt != NULL) && (ctxt->lastError.code != XML_ERR_OK)) |
13191 | 0 | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13192 | |
|
13193 | 0 | if (sax != NULL) |
13194 | 0 | ctxt->sax = oldsax; |
13195 | 0 | if (oldctxt != NULL) { |
13196 | 0 | ctxt->dict = NULL; |
13197 | 0 | ctxt->attsDefault = NULL; |
13198 | 0 | ctxt->attsSpecial = NULL; |
13199 | 0 | oldctxt->validate = ctxt->validate; |
13200 | 0 | oldctxt->valid = ctxt->valid; |
13201 | 0 | oldctxt->node_seq.maximum = ctxt->node_seq.maximum; |
13202 | 0 | oldctxt->node_seq.length = ctxt->node_seq.length; |
13203 | 0 | oldctxt->node_seq.buffer = ctxt->node_seq.buffer; |
13204 | 0 | } |
13205 | 0 | ctxt->node_seq.maximum = 0; |
13206 | 0 | ctxt->node_seq.length = 0; |
13207 | 0 | ctxt->node_seq.buffer = NULL; |
13208 | 0 | xmlFreeParserCtxt(ctxt); |
13209 | 0 | newDoc->intSubset = NULL; |
13210 | 0 | newDoc->extSubset = NULL; |
13211 | 0 | xmlFreeDoc(newDoc); |
13212 | |
|
13213 | 0 | return(ret); |
13214 | 0 | } |
13215 | | |
13216 | | #ifdef LIBXML_SAX1_ENABLED |
13217 | | /** |
13218 | | * xmlParseExternalEntity: |
13219 | | * @doc: the document the chunk pertains to |
13220 | | * @sax: the SAX handler block (possibly NULL) |
13221 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13222 | | * @depth: Used for loop detection, use 0 |
13223 | | * @URL: the URL for the entity to load |
13224 | | * @ID: the System ID for the entity to load |
13225 | | * @lst: the return value for the set of parsed nodes |
13226 | | * |
13227 | | * Parse an external general entity |
13228 | | * An external general parsed entity is well-formed if it matches the |
13229 | | * production labeled extParsedEnt. |
13230 | | * |
13231 | | * [78] extParsedEnt ::= TextDecl? content |
13232 | | * |
13233 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13234 | | * the parser error code otherwise |
13235 | | */ |
13236 | | |
13237 | | int |
13238 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
13239 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *lst) { |
13240 | 0 | return(xmlParseExternalEntityPrivate(doc, NULL, sax, user_data, depth, URL, |
13241 | 0 | ID, lst)); |
13242 | 0 | } |
13243 | | |
13244 | | /** |
13245 | | * xmlParseBalancedChunkMemory: |
13246 | | * @doc: the document the chunk pertains to (must not be NULL) |
13247 | | * @sax: the SAX handler block (possibly NULL) |
13248 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13249 | | * @depth: Used for loop detection, use 0 |
13250 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13251 | | * @lst: the return value for the set of parsed nodes |
13252 | | * |
13253 | | * Parse a well-balanced chunk of an XML document |
13254 | | * called by the parser |
13255 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13256 | | * the content production in the XML grammar: |
13257 | | * |
13258 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13259 | | * |
13260 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13261 | | * the parser error code otherwise |
13262 | | */ |
13263 | | |
13264 | | int |
13265 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13266 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
13267 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
13268 | 0 | depth, string, lst, 0 ); |
13269 | 0 | } |
13270 | | #endif /* LIBXML_SAX1_ENABLED */ |
13271 | | |
13272 | | /** |
13273 | | * xmlParseBalancedChunkMemoryInternal: |
13274 | | * @oldctxt: the existing parsing context |
13275 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13276 | | * @user_data: the user data field for the parser context |
13277 | | * @lst: the return value for the set of parsed nodes |
13278 | | * |
13279 | | * |
13280 | | * Parse a well-balanced chunk of an XML document |
13281 | | * called by the parser |
13282 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13283 | | * the content production in the XML grammar: |
13284 | | * |
13285 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13286 | | * |
13287 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13288 | | * error code otherwise |
13289 | | * |
13290 | | * In case recover is set to 1, the nodelist will not be empty even if |
13291 | | * the parsed chunk is not well balanced. |
13292 | | */ |
13293 | | static xmlParserErrors |
13294 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
13295 | 433k | const xmlChar *string, void *user_data, xmlNodePtr *lst) { |
13296 | 433k | xmlParserCtxtPtr ctxt; |
13297 | 433k | xmlDocPtr newDoc = NULL; |
13298 | 433k | xmlNodePtr newRoot; |
13299 | 433k | xmlSAXHandlerPtr oldsax = NULL; |
13300 | 433k | xmlNodePtr content = NULL; |
13301 | 433k | xmlNodePtr last = NULL; |
13302 | 433k | int size; |
13303 | 433k | xmlParserErrors ret = XML_ERR_OK; |
13304 | 433k | #ifdef SAX2 |
13305 | 433k | int i; |
13306 | 433k | #endif |
13307 | | |
13308 | 433k | if (((oldctxt->depth > 40) && ((oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13309 | 433k | (oldctxt->depth > 1024)) { |
13310 | 582 | return(XML_ERR_ENTITY_LOOP); |
13311 | 582 | } |
13312 | | |
13313 | | |
13314 | 432k | if (lst != NULL) |
13315 | 299k | *lst = NULL; |
13316 | 432k | if (string == NULL) |
13317 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13318 | | |
13319 | 432k | size = xmlStrlen(string); |
13320 | | |
13321 | 432k | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13322 | 432k | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13323 | 300k | if (user_data != NULL) |
13324 | 0 | ctxt->userData = user_data; |
13325 | 300k | else |
13326 | 300k | ctxt->userData = ctxt; |
13327 | 300k | if (ctxt->dict != NULL) xmlDictFree(ctxt->dict); |
13328 | 300k | ctxt->dict = oldctxt->dict; |
13329 | 300k | ctxt->input_id = oldctxt->input_id + 1; |
13330 | 300k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13331 | 300k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13332 | 300k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13333 | | |
13334 | 300k | #ifdef SAX2 |
13335 | | /* propagate namespaces down the entity */ |
13336 | 29.2M | for (i = 0;i < oldctxt->nsNr;i += 2) { |
13337 | 28.9M | nsPush(ctxt, oldctxt->nsTab[i], oldctxt->nsTab[i+1]); |
13338 | 28.9M | } |
13339 | 300k | #endif |
13340 | | |
13341 | 300k | oldsax = ctxt->sax; |
13342 | 300k | ctxt->sax = oldctxt->sax; |
13343 | 300k | xmlDetectSAX2(ctxt); |
13344 | 300k | ctxt->replaceEntities = oldctxt->replaceEntities; |
13345 | 300k | ctxt->options = oldctxt->options; |
13346 | | |
13347 | 300k | ctxt->_private = oldctxt->_private; |
13348 | 300k | if (oldctxt->myDoc == NULL) { |
13349 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13350 | 0 | if (newDoc == NULL) { |
13351 | 0 | ctxt->sax = oldsax; |
13352 | 0 | ctxt->dict = NULL; |
13353 | 0 | xmlFreeParserCtxt(ctxt); |
13354 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13355 | 0 | } |
13356 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13357 | 0 | newDoc->dict = ctxt->dict; |
13358 | 0 | xmlDictReference(newDoc->dict); |
13359 | 0 | ctxt->myDoc = newDoc; |
13360 | 300k | } else { |
13361 | 300k | ctxt->myDoc = oldctxt->myDoc; |
13362 | 300k | content = ctxt->myDoc->children; |
13363 | 300k | last = ctxt->myDoc->last; |
13364 | 300k | } |
13365 | 300k | newRoot = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13366 | 300k | if (newRoot == NULL) { |
13367 | 0 | ctxt->sax = oldsax; |
13368 | 0 | ctxt->dict = NULL; |
13369 | 0 | xmlFreeParserCtxt(ctxt); |
13370 | 0 | if (newDoc != NULL) { |
13371 | 0 | xmlFreeDoc(newDoc); |
13372 | 0 | } |
13373 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13374 | 0 | } |
13375 | 300k | ctxt->myDoc->children = NULL; |
13376 | 300k | ctxt->myDoc->last = NULL; |
13377 | 300k | xmlAddChild((xmlNodePtr) ctxt->myDoc, newRoot); |
13378 | 300k | nodePush(ctxt, ctxt->myDoc->children); |
13379 | 300k | ctxt->instate = XML_PARSER_CONTENT; |
13380 | 300k | ctxt->depth = oldctxt->depth + 1; |
13381 | | |
13382 | 300k | ctxt->validate = 0; |
13383 | 300k | ctxt->loadsubset = oldctxt->loadsubset; |
13384 | 300k | if ((oldctxt->validate) || (oldctxt->replaceEntities != 0)) { |
13385 | | /* |
13386 | | * ID/IDREF registration will be done in xmlValidateElement below |
13387 | | */ |
13388 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13389 | 0 | } |
13390 | 300k | ctxt->dictNames = oldctxt->dictNames; |
13391 | 300k | ctxt->attsDefault = oldctxt->attsDefault; |
13392 | 300k | ctxt->attsSpecial = oldctxt->attsSpecial; |
13393 | | |
13394 | 300k | xmlParseContent(ctxt); |
13395 | 300k | if ((RAW == '<') && (NXT(1) == '/')) { |
13396 | 1.56k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13397 | 299k | } else if (RAW != 0) { |
13398 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13399 | 0 | } |
13400 | 300k | if (ctxt->node != ctxt->myDoc->children) { |
13401 | 125k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13402 | 125k | } |
13403 | | |
13404 | 300k | if (!ctxt->wellFormed) { |
13405 | 299k | if (ctxt->errNo == 0) |
13406 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13407 | 299k | else |
13408 | 299k | ret = (xmlParserErrors)ctxt->errNo; |
13409 | 299k | } else { |
13410 | 1.35k | ret = XML_ERR_OK; |
13411 | 1.35k | } |
13412 | | |
13413 | 300k | if ((lst != NULL) && (ret == XML_ERR_OK)) { |
13414 | 255 | xmlNodePtr cur; |
13415 | | |
13416 | | /* |
13417 | | * Return the newly created nodeset after unlinking it from |
13418 | | * they pseudo parent. |
13419 | | */ |
13420 | 255 | cur = ctxt->myDoc->children->children; |
13421 | 255 | *lst = cur; |
13422 | 2.64k | while (cur != NULL) { |
13423 | 2.38k | #ifdef LIBXML_VALID_ENABLED |
13424 | 2.38k | if ((oldctxt->validate) && (oldctxt->wellFormed) && |
13425 | 2.38k | (oldctxt->myDoc) && (oldctxt->myDoc->intSubset) && |
13426 | 2.38k | (cur->type == XML_ELEMENT_NODE)) { |
13427 | 0 | oldctxt->valid &= xmlValidateElement(&oldctxt->vctxt, |
13428 | 0 | oldctxt->myDoc, cur); |
13429 | 0 | } |
13430 | 2.38k | #endif /* LIBXML_VALID_ENABLED */ |
13431 | 2.38k | cur->parent = NULL; |
13432 | 2.38k | cur = cur->next; |
13433 | 2.38k | } |
13434 | 255 | ctxt->myDoc->children->children = NULL; |
13435 | 255 | } |
13436 | 300k | if (ctxt->myDoc != NULL) { |
13437 | 300k | xmlFreeNode(ctxt->myDoc->children); |
13438 | 300k | ctxt->myDoc->children = content; |
13439 | 300k | ctxt->myDoc->last = last; |
13440 | 300k | } |
13441 | | |
13442 | | /* |
13443 | | * Record in the parent context the number of entities replacement |
13444 | | * done when parsing that reference. |
13445 | | */ |
13446 | 300k | if (oldctxt != NULL) |
13447 | 300k | oldctxt->nbentities += ctxt->nbentities; |
13448 | | |
13449 | | /* |
13450 | | * Also record the last error if any |
13451 | | */ |
13452 | 300k | if (ctxt->lastError.code != XML_ERR_OK) |
13453 | 299k | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13454 | | |
13455 | 300k | ctxt->sax = oldsax; |
13456 | 300k | ctxt->dict = NULL; |
13457 | 300k | ctxt->attsDefault = NULL; |
13458 | 300k | ctxt->attsSpecial = NULL; |
13459 | 300k | xmlFreeParserCtxt(ctxt); |
13460 | 300k | if (newDoc != NULL) { |
13461 | 0 | xmlFreeDoc(newDoc); |
13462 | 0 | } |
13463 | | |
13464 | 300k | return(ret); |
13465 | 300k | } |
13466 | | |
13467 | | /** |
13468 | | * xmlParseInNodeContext: |
13469 | | * @node: the context node |
13470 | | * @data: the input string |
13471 | | * @datalen: the input string length in bytes |
13472 | | * @options: a combination of xmlParserOption |
13473 | | * @lst: the return value for the set of parsed nodes |
13474 | | * |
13475 | | * Parse a well-balanced chunk of an XML document |
13476 | | * within the context (DTD, namespaces, etc ...) of the given node. |
13477 | | * |
13478 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
13479 | | * the content production in the XML grammar: |
13480 | | * |
13481 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13482 | | * |
13483 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13484 | | * error code otherwise |
13485 | | */ |
13486 | | xmlParserErrors |
13487 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
13488 | 0 | int options, xmlNodePtr *lst) { |
13489 | 0 | #ifdef SAX2 |
13490 | 0 | xmlParserCtxtPtr ctxt; |
13491 | 0 | xmlDocPtr doc = NULL; |
13492 | 0 | xmlNodePtr fake, cur; |
13493 | 0 | int nsnr = 0; |
13494 | |
|
13495 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13496 | | |
13497 | | /* |
13498 | | * check all input parameters, grab the document |
13499 | | */ |
13500 | 0 | if ((lst == NULL) || (node == NULL) || (data == NULL) || (datalen < 0)) |
13501 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13502 | 0 | switch (node->type) { |
13503 | 0 | case XML_ELEMENT_NODE: |
13504 | 0 | case XML_ATTRIBUTE_NODE: |
13505 | 0 | case XML_TEXT_NODE: |
13506 | 0 | case XML_CDATA_SECTION_NODE: |
13507 | 0 | case XML_ENTITY_REF_NODE: |
13508 | 0 | case XML_PI_NODE: |
13509 | 0 | case XML_COMMENT_NODE: |
13510 | 0 | case XML_DOCUMENT_NODE: |
13511 | 0 | case XML_HTML_DOCUMENT_NODE: |
13512 | 0 | break; |
13513 | 0 | default: |
13514 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13515 | |
|
13516 | 0 | } |
13517 | 0 | while ((node != NULL) && (node->type != XML_ELEMENT_NODE) && |
13518 | 0 | (node->type != XML_DOCUMENT_NODE) && |
13519 | 0 | (node->type != XML_HTML_DOCUMENT_NODE)) |
13520 | 0 | node = node->parent; |
13521 | 0 | if (node == NULL) |
13522 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13523 | 0 | if (node->type == XML_ELEMENT_NODE) |
13524 | 0 | doc = node->doc; |
13525 | 0 | else |
13526 | 0 | doc = (xmlDocPtr) node; |
13527 | 0 | if (doc == NULL) |
13528 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13529 | | |
13530 | | /* |
13531 | | * allocate a context and set-up everything not related to the |
13532 | | * node position in the tree |
13533 | | */ |
13534 | 0 | if (doc->type == XML_DOCUMENT_NODE) |
13535 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) data, datalen); |
13536 | 0 | #ifdef LIBXML_HTML_ENABLED |
13537 | 0 | else if (doc->type == XML_HTML_DOCUMENT_NODE) { |
13538 | 0 | ctxt = htmlCreateMemoryParserCtxt((char *) data, datalen); |
13539 | | /* |
13540 | | * When parsing in context, it makes no sense to add implied |
13541 | | * elements like html/body/etc... |
13542 | | */ |
13543 | 0 | options |= HTML_PARSE_NOIMPLIED; |
13544 | 0 | } |
13545 | 0 | #endif |
13546 | 0 | else |
13547 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13548 | | |
13549 | 0 | if (ctxt == NULL) |
13550 | 0 | return(XML_ERR_NO_MEMORY); |
13551 | | |
13552 | | /* |
13553 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
13554 | | * We need a dictionary for xmlDetectSAX2, so if there's no doc dict |
13555 | | * we must wait until the last moment to free the original one. |
13556 | | */ |
13557 | 0 | if (doc->dict != NULL) { |
13558 | 0 | if (ctxt->dict != NULL) |
13559 | 0 | xmlDictFree(ctxt->dict); |
13560 | 0 | ctxt->dict = doc->dict; |
13561 | 0 | } else |
13562 | 0 | options |= XML_PARSE_NODICT; |
13563 | |
|
13564 | 0 | if (doc->encoding != NULL) { |
13565 | 0 | xmlCharEncodingHandlerPtr hdlr; |
13566 | |
|
13567 | 0 | if (ctxt->encoding != NULL) |
13568 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
13569 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) doc->encoding); |
13570 | |
|
13571 | 0 | hdlr = xmlFindCharEncodingHandler((const char *) doc->encoding); |
13572 | 0 | if (hdlr != NULL) { |
13573 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
13574 | 0 | } else { |
13575 | 0 | return(XML_ERR_UNSUPPORTED_ENCODING); |
13576 | 0 | } |
13577 | 0 | } |
13578 | | |
13579 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
13580 | 0 | xmlDetectSAX2(ctxt); |
13581 | 0 | ctxt->myDoc = doc; |
13582 | | /* parsing in context, i.e. as within existing content */ |
13583 | 0 | ctxt->input_id = 2; |
13584 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13585 | |
|
13586 | 0 | fake = xmlNewComment(NULL); |
13587 | 0 | if (fake == NULL) { |
13588 | 0 | xmlFreeParserCtxt(ctxt); |
13589 | 0 | return(XML_ERR_NO_MEMORY); |
13590 | 0 | } |
13591 | 0 | xmlAddChild(node, fake); |
13592 | |
|
13593 | 0 | if (node->type == XML_ELEMENT_NODE) { |
13594 | 0 | nodePush(ctxt, node); |
13595 | | /* |
13596 | | * initialize the SAX2 namespaces stack |
13597 | | */ |
13598 | 0 | cur = node; |
13599 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
13600 | 0 | xmlNsPtr ns = cur->nsDef; |
13601 | 0 | const xmlChar *iprefix, *ihref; |
13602 | |
|
13603 | 0 | while (ns != NULL) { |
13604 | 0 | if (ctxt->dict) { |
13605 | 0 | iprefix = xmlDictLookup(ctxt->dict, ns->prefix, -1); |
13606 | 0 | ihref = xmlDictLookup(ctxt->dict, ns->href, -1); |
13607 | 0 | } else { |
13608 | 0 | iprefix = ns->prefix; |
13609 | 0 | ihref = ns->href; |
13610 | 0 | } |
13611 | |
|
13612 | 0 | if (xmlGetNamespace(ctxt, iprefix) == NULL) { |
13613 | 0 | nsPush(ctxt, iprefix, ihref); |
13614 | 0 | nsnr++; |
13615 | 0 | } |
13616 | 0 | ns = ns->next; |
13617 | 0 | } |
13618 | 0 | cur = cur->parent; |
13619 | 0 | } |
13620 | 0 | } |
13621 | |
|
13622 | 0 | if ((ctxt->validate) || (ctxt->replaceEntities != 0)) { |
13623 | | /* |
13624 | | * ID/IDREF registration will be done in xmlValidateElement below |
13625 | | */ |
13626 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13627 | 0 | } |
13628 | |
|
13629 | 0 | #ifdef LIBXML_HTML_ENABLED |
13630 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) |
13631 | 0 | __htmlParseContent(ctxt); |
13632 | 0 | else |
13633 | 0 | #endif |
13634 | 0 | xmlParseContent(ctxt); |
13635 | |
|
13636 | 0 | nsPop(ctxt, nsnr); |
13637 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13638 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13639 | 0 | } else if (RAW != 0) { |
13640 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13641 | 0 | } |
13642 | 0 | if ((ctxt->node != NULL) && (ctxt->node != node)) { |
13643 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13644 | 0 | ctxt->wellFormed = 0; |
13645 | 0 | } |
13646 | |
|
13647 | 0 | if (!ctxt->wellFormed) { |
13648 | 0 | if (ctxt->errNo == 0) |
13649 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13650 | 0 | else |
13651 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13652 | 0 | } else { |
13653 | 0 | ret = XML_ERR_OK; |
13654 | 0 | } |
13655 | | |
13656 | | /* |
13657 | | * Return the newly created nodeset after unlinking it from |
13658 | | * the pseudo sibling. |
13659 | | */ |
13660 | |
|
13661 | 0 | cur = fake->next; |
13662 | 0 | fake->next = NULL; |
13663 | 0 | node->last = fake; |
13664 | |
|
13665 | 0 | if (cur != NULL) { |
13666 | 0 | cur->prev = NULL; |
13667 | 0 | } |
13668 | |
|
13669 | 0 | *lst = cur; |
13670 | |
|
13671 | 0 | while (cur != NULL) { |
13672 | 0 | cur->parent = NULL; |
13673 | 0 | cur = cur->next; |
13674 | 0 | } |
13675 | |
|
13676 | 0 | xmlUnlinkNode(fake); |
13677 | 0 | xmlFreeNode(fake); |
13678 | | |
13679 | |
|
13680 | 0 | if (ret != XML_ERR_OK) { |
13681 | 0 | xmlFreeNodeList(*lst); |
13682 | 0 | *lst = NULL; |
13683 | 0 | } |
13684 | |
|
13685 | 0 | if (doc->dict != NULL) |
13686 | 0 | ctxt->dict = NULL; |
13687 | 0 | xmlFreeParserCtxt(ctxt); |
13688 | |
|
13689 | 0 | return(ret); |
13690 | | #else /* !SAX2 */ |
13691 | | return(XML_ERR_INTERNAL_ERROR); |
13692 | | #endif |
13693 | 0 | } |
13694 | | |
13695 | | #ifdef LIBXML_SAX1_ENABLED |
13696 | | /** |
13697 | | * xmlParseBalancedChunkMemoryRecover: |
13698 | | * @doc: the document the chunk pertains to (must not be NULL) |
13699 | | * @sax: the SAX handler block (possibly NULL) |
13700 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13701 | | * @depth: Used for loop detection, use 0 |
13702 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13703 | | * @lst: the return value for the set of parsed nodes |
13704 | | * @recover: return nodes even if the data is broken (use 0) |
13705 | | * |
13706 | | * |
13707 | | * Parse a well-balanced chunk of an XML document |
13708 | | * called by the parser |
13709 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13710 | | * the content production in the XML grammar: |
13711 | | * |
13712 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13713 | | * |
13714 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13715 | | * the parser error code otherwise |
13716 | | * |
13717 | | * In case recover is set to 1, the nodelist will not be empty even if |
13718 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
13719 | | * some extent. |
13720 | | */ |
13721 | | int |
13722 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13723 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst, |
13724 | 0 | int recover) { |
13725 | 0 | xmlParserCtxtPtr ctxt; |
13726 | 0 | xmlDocPtr newDoc; |
13727 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13728 | 0 | xmlNodePtr content, newRoot; |
13729 | 0 | int size; |
13730 | 0 | int ret = 0; |
13731 | |
|
13732 | 0 | if (depth > 40) { |
13733 | 0 | return(XML_ERR_ENTITY_LOOP); |
13734 | 0 | } |
13735 | | |
13736 | | |
13737 | 0 | if (lst != NULL) |
13738 | 0 | *lst = NULL; |
13739 | 0 | if (string == NULL) |
13740 | 0 | return(-1); |
13741 | | |
13742 | 0 | size = xmlStrlen(string); |
13743 | |
|
13744 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13745 | 0 | if (ctxt == NULL) return(-1); |
13746 | 0 | ctxt->userData = ctxt; |
13747 | 0 | if (sax != NULL) { |
13748 | 0 | oldsax = ctxt->sax; |
13749 | 0 | ctxt->sax = sax; |
13750 | 0 | if (user_data != NULL) |
13751 | 0 | ctxt->userData = user_data; |
13752 | 0 | } |
13753 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13754 | 0 | if (newDoc == NULL) { |
13755 | 0 | xmlFreeParserCtxt(ctxt); |
13756 | 0 | return(-1); |
13757 | 0 | } |
13758 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13759 | 0 | if ((doc != NULL) && (doc->dict != NULL)) { |
13760 | 0 | xmlDictFree(ctxt->dict); |
13761 | 0 | ctxt->dict = doc->dict; |
13762 | 0 | xmlDictReference(ctxt->dict); |
13763 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13764 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13765 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13766 | 0 | ctxt->dictNames = 1; |
13767 | 0 | } else { |
13768 | 0 | xmlCtxtUseOptionsInternal(ctxt, XML_PARSE_NODICT, NULL); |
13769 | 0 | } |
13770 | | /* doc == NULL is only supported for historic reasons */ |
13771 | 0 | if (doc != NULL) { |
13772 | 0 | newDoc->intSubset = doc->intSubset; |
13773 | 0 | newDoc->extSubset = doc->extSubset; |
13774 | 0 | } |
13775 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13776 | 0 | if (newRoot == NULL) { |
13777 | 0 | if (sax != NULL) |
13778 | 0 | ctxt->sax = oldsax; |
13779 | 0 | xmlFreeParserCtxt(ctxt); |
13780 | 0 | newDoc->intSubset = NULL; |
13781 | 0 | newDoc->extSubset = NULL; |
13782 | 0 | xmlFreeDoc(newDoc); |
13783 | 0 | return(-1); |
13784 | 0 | } |
13785 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13786 | 0 | nodePush(ctxt, newRoot); |
13787 | | /* doc == NULL is only supported for historic reasons */ |
13788 | 0 | if (doc == NULL) { |
13789 | 0 | ctxt->myDoc = newDoc; |
13790 | 0 | } else { |
13791 | 0 | ctxt->myDoc = newDoc; |
13792 | 0 | newDoc->children->doc = doc; |
13793 | | /* Ensure that doc has XML spec namespace */ |
13794 | 0 | xmlSearchNsByHref(doc, (xmlNodePtr)doc, XML_XML_NAMESPACE); |
13795 | 0 | newDoc->oldNs = doc->oldNs; |
13796 | 0 | } |
13797 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13798 | 0 | ctxt->input_id = 2; |
13799 | 0 | ctxt->depth = depth; |
13800 | | |
13801 | | /* |
13802 | | * Doing validity checking on chunk doesn't make sense |
13803 | | */ |
13804 | 0 | ctxt->validate = 0; |
13805 | 0 | ctxt->loadsubset = 0; |
13806 | 0 | xmlDetectSAX2(ctxt); |
13807 | |
|
13808 | 0 | if ( doc != NULL ){ |
13809 | 0 | content = doc->children; |
13810 | 0 | doc->children = NULL; |
13811 | 0 | xmlParseContent(ctxt); |
13812 | 0 | doc->children = content; |
13813 | 0 | } |
13814 | 0 | else { |
13815 | 0 | xmlParseContent(ctxt); |
13816 | 0 | } |
13817 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13818 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13819 | 0 | } else if (RAW != 0) { |
13820 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13821 | 0 | } |
13822 | 0 | if (ctxt->node != newDoc->children) { |
13823 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13824 | 0 | } |
13825 | |
|
13826 | 0 | if (!ctxt->wellFormed) { |
13827 | 0 | if (ctxt->errNo == 0) |
13828 | 0 | ret = 1; |
13829 | 0 | else |
13830 | 0 | ret = ctxt->errNo; |
13831 | 0 | } else { |
13832 | 0 | ret = 0; |
13833 | 0 | } |
13834 | |
|
13835 | 0 | if ((lst != NULL) && ((ret == 0) || (recover == 1))) { |
13836 | 0 | xmlNodePtr cur; |
13837 | | |
13838 | | /* |
13839 | | * Return the newly created nodeset after unlinking it from |
13840 | | * they pseudo parent. |
13841 | | */ |
13842 | 0 | cur = newDoc->children->children; |
13843 | 0 | *lst = cur; |
13844 | 0 | while (cur != NULL) { |
13845 | 0 | xmlSetTreeDoc(cur, doc); |
13846 | 0 | cur->parent = NULL; |
13847 | 0 | cur = cur->next; |
13848 | 0 | } |
13849 | 0 | newDoc->children->children = NULL; |
13850 | 0 | } |
13851 | |
|
13852 | 0 | if (sax != NULL) |
13853 | 0 | ctxt->sax = oldsax; |
13854 | 0 | xmlFreeParserCtxt(ctxt); |
13855 | 0 | newDoc->intSubset = NULL; |
13856 | 0 | newDoc->extSubset = NULL; |
13857 | | /* This leaks the namespace list if doc == NULL */ |
13858 | 0 | newDoc->oldNs = NULL; |
13859 | 0 | xmlFreeDoc(newDoc); |
13860 | |
|
13861 | 0 | return(ret); |
13862 | 0 | } |
13863 | | |
13864 | | /** |
13865 | | * xmlSAXParseEntity: |
13866 | | * @sax: the SAX handler block |
13867 | | * @filename: the filename |
13868 | | * |
13869 | | * parse an XML external entity out of context and build a tree. |
13870 | | * It use the given SAX function block to handle the parsing callback. |
13871 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13872 | | * |
13873 | | * [78] extParsedEnt ::= TextDecl? content |
13874 | | * |
13875 | | * This correspond to a "Well Balanced" chunk |
13876 | | * |
13877 | | * Returns the resulting document tree |
13878 | | */ |
13879 | | |
13880 | | xmlDocPtr |
13881 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
13882 | 0 | xmlDocPtr ret; |
13883 | 0 | xmlParserCtxtPtr ctxt; |
13884 | |
|
13885 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13886 | 0 | if (ctxt == NULL) { |
13887 | 0 | return(NULL); |
13888 | 0 | } |
13889 | 0 | if (sax != NULL) { |
13890 | 0 | if (ctxt->sax != NULL) |
13891 | 0 | xmlFree(ctxt->sax); |
13892 | 0 | ctxt->sax = sax; |
13893 | 0 | ctxt->userData = NULL; |
13894 | 0 | } |
13895 | |
|
13896 | 0 | xmlParseExtParsedEnt(ctxt); |
13897 | |
|
13898 | 0 | if (ctxt->wellFormed) |
13899 | 0 | ret = ctxt->myDoc; |
13900 | 0 | else { |
13901 | 0 | ret = NULL; |
13902 | 0 | xmlFreeDoc(ctxt->myDoc); |
13903 | 0 | ctxt->myDoc = NULL; |
13904 | 0 | } |
13905 | 0 | if (sax != NULL) |
13906 | 0 | ctxt->sax = NULL; |
13907 | 0 | xmlFreeParserCtxt(ctxt); |
13908 | |
|
13909 | 0 | return(ret); |
13910 | 0 | } |
13911 | | |
13912 | | /** |
13913 | | * xmlParseEntity: |
13914 | | * @filename: the filename |
13915 | | * |
13916 | | * parse an XML external entity out of context and build a tree. |
13917 | | * |
13918 | | * [78] extParsedEnt ::= TextDecl? content |
13919 | | * |
13920 | | * This correspond to a "Well Balanced" chunk |
13921 | | * |
13922 | | * Returns the resulting document tree |
13923 | | */ |
13924 | | |
13925 | | xmlDocPtr |
13926 | 0 | xmlParseEntity(const char *filename) { |
13927 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
13928 | 0 | } |
13929 | | #endif /* LIBXML_SAX1_ENABLED */ |
13930 | | |
13931 | | /** |
13932 | | * xmlCreateEntityParserCtxtInternal: |
13933 | | * @URL: the entity URL |
13934 | | * @ID: the entity PUBLIC ID |
13935 | | * @base: a possible base for the target URI |
13936 | | * @pctx: parser context used to set options on new context |
13937 | | * |
13938 | | * Create a parser context for an external entity |
13939 | | * Automatic support for ZLIB/Compress compressed document is provided |
13940 | | * by default if found at compile-time. |
13941 | | * |
13942 | | * Returns the new parser context or NULL |
13943 | | */ |
13944 | | static xmlParserCtxtPtr |
13945 | | xmlCreateEntityParserCtxtInternal(const xmlChar *URL, const xmlChar *ID, |
13946 | 0 | const xmlChar *base, xmlParserCtxtPtr pctx) { |
13947 | 0 | xmlParserCtxtPtr ctxt; |
13948 | 0 | xmlParserInputPtr inputStream; |
13949 | 0 | char *directory = NULL; |
13950 | 0 | xmlChar *uri; |
13951 | |
|
13952 | 0 | ctxt = xmlNewParserCtxt(); |
13953 | 0 | if (ctxt == NULL) { |
13954 | 0 | return(NULL); |
13955 | 0 | } |
13956 | | |
13957 | 0 | if (pctx != NULL) { |
13958 | 0 | ctxt->options = pctx->options; |
13959 | 0 | ctxt->_private = pctx->_private; |
13960 | | /* |
13961 | | * this is a subparser of pctx, so the input_id should be |
13962 | | * incremented to distinguish from main entity |
13963 | | */ |
13964 | 0 | ctxt->input_id = pctx->input_id + 1; |
13965 | 0 | } |
13966 | | |
13967 | | /* Don't read from stdin. */ |
13968 | 0 | if (xmlStrcmp(URL, BAD_CAST "-") == 0) |
13969 | 0 | URL = BAD_CAST "./-"; |
13970 | |
|
13971 | 0 | uri = xmlBuildURI(URL, base); |
13972 | |
|
13973 | 0 | if (uri == NULL) { |
13974 | 0 | inputStream = xmlLoadExternalEntity((char *)URL, (char *)ID, ctxt); |
13975 | 0 | if (inputStream == NULL) { |
13976 | 0 | xmlFreeParserCtxt(ctxt); |
13977 | 0 | return(NULL); |
13978 | 0 | } |
13979 | | |
13980 | 0 | inputPush(ctxt, inputStream); |
13981 | |
|
13982 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
13983 | 0 | directory = xmlParserGetDirectory((char *)URL); |
13984 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
13985 | 0 | ctxt->directory = directory; |
13986 | 0 | } else { |
13987 | 0 | inputStream = xmlLoadExternalEntity((char *)uri, (char *)ID, ctxt); |
13988 | 0 | if (inputStream == NULL) { |
13989 | 0 | xmlFree(uri); |
13990 | 0 | xmlFreeParserCtxt(ctxt); |
13991 | 0 | return(NULL); |
13992 | 0 | } |
13993 | | |
13994 | 0 | inputPush(ctxt, inputStream); |
13995 | |
|
13996 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
13997 | 0 | directory = xmlParserGetDirectory((char *)uri); |
13998 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
13999 | 0 | ctxt->directory = directory; |
14000 | 0 | xmlFree(uri); |
14001 | 0 | } |
14002 | 0 | return(ctxt); |
14003 | 0 | } |
14004 | | |
14005 | | /** |
14006 | | * xmlCreateEntityParserCtxt: |
14007 | | * @URL: the entity URL |
14008 | | * @ID: the entity PUBLIC ID |
14009 | | * @base: a possible base for the target URI |
14010 | | * |
14011 | | * Create a parser context for an external entity |
14012 | | * Automatic support for ZLIB/Compress compressed document is provided |
14013 | | * by default if found at compile-time. |
14014 | | * |
14015 | | * Returns the new parser context or NULL |
14016 | | */ |
14017 | | xmlParserCtxtPtr |
14018 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
14019 | 0 | const xmlChar *base) { |
14020 | 0 | return xmlCreateEntityParserCtxtInternal(URL, ID, base, NULL); |
14021 | |
|
14022 | 0 | } |
14023 | | |
14024 | | /************************************************************************ |
14025 | | * * |
14026 | | * Front ends when parsing from a file * |
14027 | | * * |
14028 | | ************************************************************************/ |
14029 | | |
14030 | | /** |
14031 | | * xmlCreateURLParserCtxt: |
14032 | | * @filename: the filename or URL |
14033 | | * @options: a combination of xmlParserOption |
14034 | | * |
14035 | | * Create a parser context for a file or URL content. |
14036 | | * Automatic support for ZLIB/Compress compressed document is provided |
14037 | | * by default if found at compile-time and for file accesses |
14038 | | * |
14039 | | * Returns the new parser context or NULL |
14040 | | */ |
14041 | | xmlParserCtxtPtr |
14042 | | xmlCreateURLParserCtxt(const char *filename, int options) |
14043 | 0 | { |
14044 | 0 | xmlParserCtxtPtr ctxt; |
14045 | 0 | xmlParserInputPtr inputStream; |
14046 | 0 | char *directory = NULL; |
14047 | |
|
14048 | 0 | ctxt = xmlNewParserCtxt(); |
14049 | 0 | if (ctxt == NULL) { |
14050 | 0 | xmlErrMemory(NULL, "cannot allocate parser context"); |
14051 | 0 | return(NULL); |
14052 | 0 | } |
14053 | | |
14054 | 0 | if (options) |
14055 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
14056 | 0 | ctxt->linenumbers = 1; |
14057 | |
|
14058 | 0 | inputStream = xmlLoadExternalEntity(filename, NULL, ctxt); |
14059 | 0 | if (inputStream == NULL) { |
14060 | 0 | xmlFreeParserCtxt(ctxt); |
14061 | 0 | return(NULL); |
14062 | 0 | } |
14063 | | |
14064 | 0 | inputPush(ctxt, inputStream); |
14065 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
14066 | 0 | directory = xmlParserGetDirectory(filename); |
14067 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
14068 | 0 | ctxt->directory = directory; |
14069 | |
|
14070 | 0 | return(ctxt); |
14071 | 0 | } |
14072 | | |
14073 | | /** |
14074 | | * xmlCreateFileParserCtxt: |
14075 | | * @filename: the filename |
14076 | | * |
14077 | | * Create a parser context for a file content. |
14078 | | * Automatic support for ZLIB/Compress compressed document is provided |
14079 | | * by default if found at compile-time. |
14080 | | * |
14081 | | * Returns the new parser context or NULL |
14082 | | */ |
14083 | | xmlParserCtxtPtr |
14084 | | xmlCreateFileParserCtxt(const char *filename) |
14085 | 0 | { |
14086 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
14087 | 0 | } |
14088 | | |
14089 | | #ifdef LIBXML_SAX1_ENABLED |
14090 | | /** |
14091 | | * xmlSAXParseFileWithData: |
14092 | | * @sax: the SAX handler block |
14093 | | * @filename: the filename |
14094 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14095 | | * documents |
14096 | | * @data: the userdata |
14097 | | * |
14098 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14099 | | * compressed document is provided by default if found at compile-time. |
14100 | | * It use the given SAX function block to handle the parsing callback. |
14101 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14102 | | * |
14103 | | * User data (void *) is stored within the parser context in the |
14104 | | * context's _private member, so it is available nearly everywhere in libxml |
14105 | | * |
14106 | | * Returns the resulting document tree |
14107 | | */ |
14108 | | |
14109 | | xmlDocPtr |
14110 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
14111 | 0 | int recovery, void *data) { |
14112 | 0 | xmlDocPtr ret; |
14113 | 0 | xmlParserCtxtPtr ctxt; |
14114 | |
|
14115 | 0 | xmlInitParser(); |
14116 | |
|
14117 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14118 | 0 | if (ctxt == NULL) { |
14119 | 0 | return(NULL); |
14120 | 0 | } |
14121 | 0 | if (sax != NULL) { |
14122 | 0 | if (ctxt->sax != NULL) |
14123 | 0 | xmlFree(ctxt->sax); |
14124 | 0 | ctxt->sax = sax; |
14125 | 0 | } |
14126 | 0 | xmlDetectSAX2(ctxt); |
14127 | 0 | if (data!=NULL) { |
14128 | 0 | ctxt->_private = data; |
14129 | 0 | } |
14130 | |
|
14131 | 0 | if (ctxt->directory == NULL) |
14132 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14133 | |
|
14134 | 0 | ctxt->recovery = recovery; |
14135 | |
|
14136 | 0 | xmlParseDocument(ctxt); |
14137 | |
|
14138 | 0 | if ((ctxt->wellFormed) || recovery) { |
14139 | 0 | ret = ctxt->myDoc; |
14140 | 0 | if ((ret != NULL) && (ctxt->input->buf != NULL)) { |
14141 | 0 | if (ctxt->input->buf->compressed > 0) |
14142 | 0 | ret->compression = 9; |
14143 | 0 | else |
14144 | 0 | ret->compression = ctxt->input->buf->compressed; |
14145 | 0 | } |
14146 | 0 | } |
14147 | 0 | else { |
14148 | 0 | ret = NULL; |
14149 | 0 | xmlFreeDoc(ctxt->myDoc); |
14150 | 0 | ctxt->myDoc = NULL; |
14151 | 0 | } |
14152 | 0 | if (sax != NULL) |
14153 | 0 | ctxt->sax = NULL; |
14154 | 0 | xmlFreeParserCtxt(ctxt); |
14155 | |
|
14156 | 0 | return(ret); |
14157 | 0 | } |
14158 | | |
14159 | | /** |
14160 | | * xmlSAXParseFile: |
14161 | | * @sax: the SAX handler block |
14162 | | * @filename: the filename |
14163 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14164 | | * documents |
14165 | | * |
14166 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14167 | | * compressed document is provided by default if found at compile-time. |
14168 | | * It use the given SAX function block to handle the parsing callback. |
14169 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14170 | | * |
14171 | | * Returns the resulting document tree |
14172 | | */ |
14173 | | |
14174 | | xmlDocPtr |
14175 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
14176 | 0 | int recovery) { |
14177 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
14178 | 0 | } |
14179 | | |
14180 | | /** |
14181 | | * xmlRecoverDoc: |
14182 | | * @cur: a pointer to an array of xmlChar |
14183 | | * |
14184 | | * parse an XML in-memory document and build a tree. |
14185 | | * In the case the document is not Well Formed, a attempt to build a |
14186 | | * tree is tried anyway |
14187 | | * |
14188 | | * Returns the resulting document tree or NULL in case of failure |
14189 | | */ |
14190 | | |
14191 | | xmlDocPtr |
14192 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
14193 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
14194 | 0 | } |
14195 | | |
14196 | | /** |
14197 | | * xmlParseFile: |
14198 | | * @filename: the filename |
14199 | | * |
14200 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14201 | | * compressed document is provided by default if found at compile-time. |
14202 | | * |
14203 | | * Returns the resulting document tree if the file was wellformed, |
14204 | | * NULL otherwise. |
14205 | | */ |
14206 | | |
14207 | | xmlDocPtr |
14208 | 0 | xmlParseFile(const char *filename) { |
14209 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
14210 | 0 | } |
14211 | | |
14212 | | /** |
14213 | | * xmlRecoverFile: |
14214 | | * @filename: the filename |
14215 | | * |
14216 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14217 | | * compressed document is provided by default if found at compile-time. |
14218 | | * In the case the document is not Well Formed, it attempts to build |
14219 | | * a tree anyway |
14220 | | * |
14221 | | * Returns the resulting document tree or NULL in case of failure |
14222 | | */ |
14223 | | |
14224 | | xmlDocPtr |
14225 | 0 | xmlRecoverFile(const char *filename) { |
14226 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
14227 | 0 | } |
14228 | | |
14229 | | |
14230 | | /** |
14231 | | * xmlSetupParserForBuffer: |
14232 | | * @ctxt: an XML parser context |
14233 | | * @buffer: a xmlChar * buffer |
14234 | | * @filename: a file name |
14235 | | * |
14236 | | * Setup the parser context to parse a new buffer; Clears any prior |
14237 | | * contents from the parser context. The buffer parameter must not be |
14238 | | * NULL, but the filename parameter can be |
14239 | | */ |
14240 | | void |
14241 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
14242 | | const char* filename) |
14243 | 0 | { |
14244 | 0 | xmlParserInputPtr input; |
14245 | |
|
14246 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
14247 | 0 | return; |
14248 | | |
14249 | 0 | input = xmlNewInputStream(ctxt); |
14250 | 0 | if (input == NULL) { |
14251 | 0 | xmlErrMemory(NULL, "parsing new buffer: out of memory\n"); |
14252 | 0 | xmlClearParserCtxt(ctxt); |
14253 | 0 | return; |
14254 | 0 | } |
14255 | | |
14256 | 0 | xmlClearParserCtxt(ctxt); |
14257 | 0 | if (filename != NULL) |
14258 | 0 | input->filename = (char *) xmlCanonicPath((const xmlChar *)filename); |
14259 | 0 | input->base = buffer; |
14260 | 0 | input->cur = buffer; |
14261 | 0 | input->end = &buffer[xmlStrlen(buffer)]; |
14262 | 0 | inputPush(ctxt, input); |
14263 | 0 | } |
14264 | | |
14265 | | /** |
14266 | | * xmlSAXUserParseFile: |
14267 | | * @sax: a SAX handler |
14268 | | * @user_data: The user data returned on SAX callbacks |
14269 | | * @filename: a file name |
14270 | | * |
14271 | | * parse an XML file and call the given SAX handler routines. |
14272 | | * Automatic support for ZLIB/Compress compressed document is provided |
14273 | | * |
14274 | | * Returns 0 in case of success or a error number otherwise |
14275 | | */ |
14276 | | int |
14277 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
14278 | 0 | const char *filename) { |
14279 | 0 | int ret = 0; |
14280 | 0 | xmlParserCtxtPtr ctxt; |
14281 | |
|
14282 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14283 | 0 | if (ctxt == NULL) return -1; |
14284 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14285 | 0 | xmlFree(ctxt->sax); |
14286 | 0 | ctxt->sax = sax; |
14287 | 0 | xmlDetectSAX2(ctxt); |
14288 | |
|
14289 | 0 | if (user_data != NULL) |
14290 | 0 | ctxt->userData = user_data; |
14291 | |
|
14292 | 0 | xmlParseDocument(ctxt); |
14293 | |
|
14294 | 0 | if (ctxt->wellFormed) |
14295 | 0 | ret = 0; |
14296 | 0 | else { |
14297 | 0 | if (ctxt->errNo != 0) |
14298 | 0 | ret = ctxt->errNo; |
14299 | 0 | else |
14300 | 0 | ret = -1; |
14301 | 0 | } |
14302 | 0 | if (sax != NULL) |
14303 | 0 | ctxt->sax = NULL; |
14304 | 0 | if (ctxt->myDoc != NULL) { |
14305 | 0 | xmlFreeDoc(ctxt->myDoc); |
14306 | 0 | ctxt->myDoc = NULL; |
14307 | 0 | } |
14308 | 0 | xmlFreeParserCtxt(ctxt); |
14309 | |
|
14310 | 0 | return ret; |
14311 | 0 | } |
14312 | | #endif /* LIBXML_SAX1_ENABLED */ |
14313 | | |
14314 | | /************************************************************************ |
14315 | | * * |
14316 | | * Front ends when parsing from memory * |
14317 | | * * |
14318 | | ************************************************************************/ |
14319 | | |
14320 | | /** |
14321 | | * xmlCreateMemoryParserCtxt: |
14322 | | * @buffer: a pointer to a char array |
14323 | | * @size: the size of the array |
14324 | | * |
14325 | | * Create a parser context for an XML in-memory document. |
14326 | | * |
14327 | | * Returns the new parser context or NULL |
14328 | | */ |
14329 | | xmlParserCtxtPtr |
14330 | 451k | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
14331 | 451k | xmlParserCtxtPtr ctxt; |
14332 | 451k | xmlParserInputPtr input; |
14333 | 451k | xmlParserInputBufferPtr buf; |
14334 | | |
14335 | 451k | if (buffer == NULL) |
14336 | 0 | return(NULL); |
14337 | 451k | if (size <= 0) |
14338 | 131k | return(NULL); |
14339 | | |
14340 | 319k | ctxt = xmlNewParserCtxt(); |
14341 | 319k | if (ctxt == NULL) |
14342 | 0 | return(NULL); |
14343 | | |
14344 | | /* TODO: xmlParserInputBufferCreateStatic, requires some serious changes */ |
14345 | 319k | buf = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
14346 | 319k | if (buf == NULL) { |
14347 | 0 | xmlFreeParserCtxt(ctxt); |
14348 | 0 | return(NULL); |
14349 | 0 | } |
14350 | | |
14351 | 319k | input = xmlNewInputStream(ctxt); |
14352 | 319k | if (input == NULL) { |
14353 | 0 | xmlFreeParserInputBuffer(buf); |
14354 | 0 | xmlFreeParserCtxt(ctxt); |
14355 | 0 | return(NULL); |
14356 | 0 | } |
14357 | | |
14358 | 319k | input->filename = NULL; |
14359 | 319k | input->buf = buf; |
14360 | 319k | xmlBufResetInput(input->buf->buffer, input); |
14361 | | |
14362 | 319k | inputPush(ctxt, input); |
14363 | 319k | return(ctxt); |
14364 | 319k | } |
14365 | | |
14366 | | #ifdef LIBXML_SAX1_ENABLED |
14367 | | /** |
14368 | | * xmlSAXParseMemoryWithData: |
14369 | | * @sax: the SAX handler block |
14370 | | * @buffer: an pointer to a char array |
14371 | | * @size: the size of the array |
14372 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14373 | | * documents |
14374 | | * @data: the userdata |
14375 | | * |
14376 | | * parse an XML in-memory block and use the given SAX function block |
14377 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14378 | | * DOM tree building routines. |
14379 | | * |
14380 | | * User data (void *) is stored within the parser context in the |
14381 | | * context's _private member, so it is available nearly everywhere in libxml |
14382 | | * |
14383 | | * Returns the resulting document tree |
14384 | | */ |
14385 | | |
14386 | | xmlDocPtr |
14387 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
14388 | 0 | int size, int recovery, void *data) { |
14389 | 0 | xmlDocPtr ret; |
14390 | 0 | xmlParserCtxtPtr ctxt; |
14391 | |
|
14392 | 0 | xmlInitParser(); |
14393 | |
|
14394 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14395 | 0 | if (ctxt == NULL) return(NULL); |
14396 | 0 | if (sax != NULL) { |
14397 | 0 | if (ctxt->sax != NULL) |
14398 | 0 | xmlFree(ctxt->sax); |
14399 | 0 | ctxt->sax = sax; |
14400 | 0 | } |
14401 | 0 | xmlDetectSAX2(ctxt); |
14402 | 0 | if (data!=NULL) { |
14403 | 0 | ctxt->_private=data; |
14404 | 0 | } |
14405 | |
|
14406 | 0 | ctxt->recovery = recovery; |
14407 | |
|
14408 | 0 | xmlParseDocument(ctxt); |
14409 | |
|
14410 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14411 | 0 | else { |
14412 | 0 | ret = NULL; |
14413 | 0 | xmlFreeDoc(ctxt->myDoc); |
14414 | 0 | ctxt->myDoc = NULL; |
14415 | 0 | } |
14416 | 0 | if (sax != NULL) |
14417 | 0 | ctxt->sax = NULL; |
14418 | 0 | xmlFreeParserCtxt(ctxt); |
14419 | |
|
14420 | 0 | return(ret); |
14421 | 0 | } |
14422 | | |
14423 | | /** |
14424 | | * xmlSAXParseMemory: |
14425 | | * @sax: the SAX handler block |
14426 | | * @buffer: an pointer to a char array |
14427 | | * @size: the size of the array |
14428 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
14429 | | * documents |
14430 | | * |
14431 | | * parse an XML in-memory block and use the given SAX function block |
14432 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14433 | | * DOM tree building routines. |
14434 | | * |
14435 | | * Returns the resulting document tree |
14436 | | */ |
14437 | | xmlDocPtr |
14438 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
14439 | 0 | int size, int recovery) { |
14440 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
14441 | 0 | } |
14442 | | |
14443 | | /** |
14444 | | * xmlParseMemory: |
14445 | | * @buffer: an pointer to a char array |
14446 | | * @size: the size of the array |
14447 | | * |
14448 | | * parse an XML in-memory block and build a tree. |
14449 | | * |
14450 | | * Returns the resulting document tree |
14451 | | */ |
14452 | | |
14453 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
14454 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
14455 | 0 | } |
14456 | | |
14457 | | /** |
14458 | | * xmlRecoverMemory: |
14459 | | * @buffer: an pointer to a char array |
14460 | | * @size: the size of the array |
14461 | | * |
14462 | | * parse an XML in-memory block and build a tree. |
14463 | | * In the case the document is not Well Formed, an attempt to |
14464 | | * build a tree is tried anyway |
14465 | | * |
14466 | | * Returns the resulting document tree or NULL in case of error |
14467 | | */ |
14468 | | |
14469 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
14470 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
14471 | 0 | } |
14472 | | |
14473 | | /** |
14474 | | * xmlSAXUserParseMemory: |
14475 | | * @sax: a SAX handler |
14476 | | * @user_data: The user data returned on SAX callbacks |
14477 | | * @buffer: an in-memory XML document input |
14478 | | * @size: the length of the XML document in bytes |
14479 | | * |
14480 | | * A better SAX parsing routine. |
14481 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
14482 | | * |
14483 | | * Returns 0 in case of success or a error number otherwise |
14484 | | */ |
14485 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
14486 | 0 | const char *buffer, int size) { |
14487 | 0 | int ret = 0; |
14488 | 0 | xmlParserCtxtPtr ctxt; |
14489 | |
|
14490 | 0 | xmlInitParser(); |
14491 | |
|
14492 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14493 | 0 | if (ctxt == NULL) return -1; |
14494 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14495 | 0 | xmlFree(ctxt->sax); |
14496 | 0 | ctxt->sax = sax; |
14497 | 0 | xmlDetectSAX2(ctxt); |
14498 | |
|
14499 | 0 | if (user_data != NULL) |
14500 | 0 | ctxt->userData = user_data; |
14501 | |
|
14502 | 0 | xmlParseDocument(ctxt); |
14503 | |
|
14504 | 0 | if (ctxt->wellFormed) |
14505 | 0 | ret = 0; |
14506 | 0 | else { |
14507 | 0 | if (ctxt->errNo != 0) |
14508 | 0 | ret = ctxt->errNo; |
14509 | 0 | else |
14510 | 0 | ret = -1; |
14511 | 0 | } |
14512 | 0 | if (sax != NULL) |
14513 | 0 | ctxt->sax = NULL; |
14514 | 0 | if (ctxt->myDoc != NULL) { |
14515 | 0 | xmlFreeDoc(ctxt->myDoc); |
14516 | 0 | ctxt->myDoc = NULL; |
14517 | 0 | } |
14518 | 0 | xmlFreeParserCtxt(ctxt); |
14519 | |
|
14520 | 0 | return ret; |
14521 | 0 | } |
14522 | | #endif /* LIBXML_SAX1_ENABLED */ |
14523 | | |
14524 | | /** |
14525 | | * xmlCreateDocParserCtxt: |
14526 | | * @cur: a pointer to an array of xmlChar |
14527 | | * |
14528 | | * Creates a parser context for an XML in-memory document. |
14529 | | * |
14530 | | * Returns the new parser context or NULL |
14531 | | */ |
14532 | | xmlParserCtxtPtr |
14533 | 0 | xmlCreateDocParserCtxt(const xmlChar *cur) { |
14534 | 0 | int len; |
14535 | |
|
14536 | 0 | if (cur == NULL) |
14537 | 0 | return(NULL); |
14538 | 0 | len = xmlStrlen(cur); |
14539 | 0 | return(xmlCreateMemoryParserCtxt((const char *)cur, len)); |
14540 | 0 | } |
14541 | | |
14542 | | #ifdef LIBXML_SAX1_ENABLED |
14543 | | /** |
14544 | | * xmlSAXParseDoc: |
14545 | | * @sax: the SAX handler block |
14546 | | * @cur: a pointer to an array of xmlChar |
14547 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14548 | | * documents |
14549 | | * |
14550 | | * parse an XML in-memory document and build a tree. |
14551 | | * It use the given SAX function block to handle the parsing callback. |
14552 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14553 | | * |
14554 | | * Returns the resulting document tree |
14555 | | */ |
14556 | | |
14557 | | xmlDocPtr |
14558 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
14559 | 0 | xmlDocPtr ret; |
14560 | 0 | xmlParserCtxtPtr ctxt; |
14561 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
14562 | |
|
14563 | 0 | if (cur == NULL) return(NULL); |
14564 | | |
14565 | | |
14566 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
14567 | 0 | if (ctxt == NULL) return(NULL); |
14568 | 0 | if (sax != NULL) { |
14569 | 0 | oldsax = ctxt->sax; |
14570 | 0 | ctxt->sax = sax; |
14571 | 0 | ctxt->userData = NULL; |
14572 | 0 | } |
14573 | 0 | xmlDetectSAX2(ctxt); |
14574 | |
|
14575 | 0 | xmlParseDocument(ctxt); |
14576 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14577 | 0 | else { |
14578 | 0 | ret = NULL; |
14579 | 0 | xmlFreeDoc(ctxt->myDoc); |
14580 | 0 | ctxt->myDoc = NULL; |
14581 | 0 | } |
14582 | 0 | if (sax != NULL) |
14583 | 0 | ctxt->sax = oldsax; |
14584 | 0 | xmlFreeParserCtxt(ctxt); |
14585 | |
|
14586 | 0 | return(ret); |
14587 | 0 | } |
14588 | | |
14589 | | /** |
14590 | | * xmlParseDoc: |
14591 | | * @cur: a pointer to an array of xmlChar |
14592 | | * |
14593 | | * parse an XML in-memory document and build a tree. |
14594 | | * |
14595 | | * Returns the resulting document tree |
14596 | | */ |
14597 | | |
14598 | | xmlDocPtr |
14599 | 0 | xmlParseDoc(const xmlChar *cur) { |
14600 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
14601 | 0 | } |
14602 | | #endif /* LIBXML_SAX1_ENABLED */ |
14603 | | |
14604 | | #ifdef LIBXML_LEGACY_ENABLED |
14605 | | /************************************************************************ |
14606 | | * * |
14607 | | * Specific function to keep track of entities references * |
14608 | | * and used by the XSLT debugger * |
14609 | | * * |
14610 | | ************************************************************************/ |
14611 | | |
14612 | | static xmlEntityReferenceFunc xmlEntityRefFunc = NULL; |
14613 | | |
14614 | | /** |
14615 | | * xmlAddEntityReference: |
14616 | | * @ent : A valid entity |
14617 | | * @firstNode : A valid first node for children of entity |
14618 | | * @lastNode : A valid last node of children entity |
14619 | | * |
14620 | | * Notify of a reference to an entity of type XML_EXTERNAL_GENERAL_PARSED_ENTITY |
14621 | | */ |
14622 | | static void |
14623 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
14624 | | xmlNodePtr lastNode) |
14625 | | { |
14626 | | if (xmlEntityRefFunc != NULL) { |
14627 | | (*xmlEntityRefFunc) (ent, firstNode, lastNode); |
14628 | | } |
14629 | | } |
14630 | | |
14631 | | |
14632 | | /** |
14633 | | * xmlSetEntityReferenceFunc: |
14634 | | * @func: A valid function |
14635 | | * |
14636 | | * Set the function to call call back when a xml reference has been made |
14637 | | */ |
14638 | | void |
14639 | | xmlSetEntityReferenceFunc(xmlEntityReferenceFunc func) |
14640 | | { |
14641 | | xmlEntityRefFunc = func; |
14642 | | } |
14643 | | #endif /* LIBXML_LEGACY_ENABLED */ |
14644 | | |
14645 | | /************************************************************************ |
14646 | | * * |
14647 | | * Miscellaneous * |
14648 | | * * |
14649 | | ************************************************************************/ |
14650 | | |
14651 | | #ifdef LIBXML_XPATH_ENABLED |
14652 | | #include <libxml/xpath.h> |
14653 | | #endif |
14654 | | |
14655 | | extern void XMLCDECL xmlGenericErrorDefaultFunc(void *ctx, const char *msg, ...); |
14656 | | static int xmlParserInitialized = 0; |
14657 | | |
14658 | | /** |
14659 | | * xmlInitParser: |
14660 | | * |
14661 | | * Initialization function for the XML parser. |
14662 | | * This is not reentrant. Call once before processing in case of |
14663 | | * use in multithreaded programs. |
14664 | | */ |
14665 | | |
14666 | | void |
14667 | 338k | xmlInitParser(void) { |
14668 | 338k | if (xmlParserInitialized != 0) |
14669 | 338k | return; |
14670 | | |
14671 | | #if defined(_WIN32) && (!defined(LIBXML_STATIC) || defined(LIBXML_STATIC_FOR_DLL)) |
14672 | | if (xmlFree == free) |
14673 | | atexit(xmlCleanupParser); |
14674 | | #endif |
14675 | | |
14676 | 1 | #ifdef LIBXML_THREAD_ENABLED |
14677 | 1 | __xmlGlobalInitMutexLock(); |
14678 | 1 | if (xmlParserInitialized == 0) { |
14679 | 1 | #endif |
14680 | 1 | xmlInitThreads(); |
14681 | 1 | xmlInitGlobals(); |
14682 | 1 | if ((xmlGenericError == xmlGenericErrorDefaultFunc) || |
14683 | 1 | (xmlGenericError == NULL)) |
14684 | 0 | initGenericErrorDefaultFunc(NULL); |
14685 | 1 | xmlInitMemory(); |
14686 | 1 | xmlInitializeDict(); |
14687 | 1 | xmlInitCharEncodingHandlers(); |
14688 | 1 | xmlDefaultSAXHandlerInit(); |
14689 | 1 | xmlRegisterDefaultInputCallbacks(); |
14690 | 1 | #ifdef LIBXML_OUTPUT_ENABLED |
14691 | 1 | xmlRegisterDefaultOutputCallbacks(); |
14692 | 1 | #endif /* LIBXML_OUTPUT_ENABLED */ |
14693 | 1 | #ifdef LIBXML_HTML_ENABLED |
14694 | 1 | htmlInitAutoClose(); |
14695 | 1 | htmlDefaultSAXHandlerInit(); |
14696 | 1 | #endif |
14697 | 1 | #ifdef LIBXML_XPATH_ENABLED |
14698 | 1 | xmlXPathInit(); |
14699 | 1 | #endif |
14700 | 1 | xmlParserInitialized = 1; |
14701 | 1 | #ifdef LIBXML_THREAD_ENABLED |
14702 | 1 | } |
14703 | 1 | __xmlGlobalInitMutexUnlock(); |
14704 | 1 | #endif |
14705 | 1 | } |
14706 | | |
14707 | | /** |
14708 | | * xmlCleanupParser: |
14709 | | * |
14710 | | * This function name is somewhat misleading. It does not clean up |
14711 | | * parser state, it cleans up memory allocated by the library itself. |
14712 | | * It is a cleanup function for the XML library. It tries to reclaim all |
14713 | | * related global memory allocated for the library processing. |
14714 | | * It doesn't deallocate any document related memory. One should |
14715 | | * call xmlCleanupParser() only when the process has finished using |
14716 | | * the library and all XML/HTML documents built with it. |
14717 | | * See also xmlInitParser() which has the opposite function of preparing |
14718 | | * the library for operations. |
14719 | | * |
14720 | | * WARNING: if your application is multithreaded or has plugin support |
14721 | | * calling this may crash the application if another thread or |
14722 | | * a plugin is still using libxml2. It's sometimes very hard to |
14723 | | * guess if libxml2 is in use in the application, some libraries |
14724 | | * or plugins may use it without notice. In case of doubt abstain |
14725 | | * from calling this function or do it just before calling exit() |
14726 | | * to avoid leak reports from valgrind ! |
14727 | | */ |
14728 | | |
14729 | | void |
14730 | 0 | xmlCleanupParser(void) { |
14731 | 0 | if (!xmlParserInitialized) |
14732 | 0 | return; |
14733 | | |
14734 | 0 | xmlCleanupCharEncodingHandlers(); |
14735 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14736 | 0 | xmlCatalogCleanup(); |
14737 | 0 | #endif |
14738 | 0 | xmlDictCleanup(); |
14739 | 0 | xmlCleanupInputCallbacks(); |
14740 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
14741 | 0 | xmlCleanupOutputCallbacks(); |
14742 | 0 | #endif |
14743 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
14744 | 0 | xmlSchemaCleanupTypes(); |
14745 | 0 | xmlRelaxNGCleanupTypes(); |
14746 | 0 | #endif |
14747 | 0 | xmlCleanupGlobals(); |
14748 | 0 | xmlCleanupThreads(); /* must be last if called not from the main thread */ |
14749 | 0 | xmlCleanupMemory(); |
14750 | 0 | xmlParserInitialized = 0; |
14751 | 0 | } |
14752 | | |
14753 | | #if defined(HAVE_ATTRIBUTE_DESTRUCTOR) && !defined(LIBXML_STATIC) && \ |
14754 | | !defined(_WIN32) |
14755 | | static void |
14756 | | ATTRIBUTE_DESTRUCTOR |
14757 | | xmlDestructor(void) { |
14758 | | /* |
14759 | | * Calling custom deallocation functions in a destructor can cause |
14760 | | * problems, for example with Nokogiri. |
14761 | | */ |
14762 | | if (xmlFree == free) |
14763 | | xmlCleanupParser(); |
14764 | | } |
14765 | | #endif |
14766 | | |
14767 | | /************************************************************************ |
14768 | | * * |
14769 | | * New set (2.6.0) of simpler and more flexible APIs * |
14770 | | * * |
14771 | | ************************************************************************/ |
14772 | | |
14773 | | /** |
14774 | | * DICT_FREE: |
14775 | | * @str: a string |
14776 | | * |
14777 | | * Free a string if it is not owned by the "dict" dictionary in the |
14778 | | * current scope |
14779 | | */ |
14780 | | #define DICT_FREE(str) \ |
14781 | 0 | if ((str) && ((!dict) || \ |
14782 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
14783 | 0 | xmlFree((char *)(str)); |
14784 | | |
14785 | | /** |
14786 | | * xmlCtxtReset: |
14787 | | * @ctxt: an XML parser context |
14788 | | * |
14789 | | * Reset a parser context |
14790 | | */ |
14791 | | void |
14792 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
14793 | 0 | { |
14794 | 0 | xmlParserInputPtr input; |
14795 | 0 | xmlDictPtr dict; |
14796 | |
|
14797 | 0 | if (ctxt == NULL) |
14798 | 0 | return; |
14799 | | |
14800 | 0 | dict = ctxt->dict; |
14801 | |
|
14802 | 0 | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
14803 | 0 | xmlFreeInputStream(input); |
14804 | 0 | } |
14805 | 0 | ctxt->inputNr = 0; |
14806 | 0 | ctxt->input = NULL; |
14807 | |
|
14808 | 0 | ctxt->spaceNr = 0; |
14809 | 0 | if (ctxt->spaceTab != NULL) { |
14810 | 0 | ctxt->spaceTab[0] = -1; |
14811 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
14812 | 0 | } else { |
14813 | 0 | ctxt->space = NULL; |
14814 | 0 | } |
14815 | | |
14816 | |
|
14817 | 0 | ctxt->nodeNr = 0; |
14818 | 0 | ctxt->node = NULL; |
14819 | |
|
14820 | 0 | ctxt->nameNr = 0; |
14821 | 0 | ctxt->name = NULL; |
14822 | |
|
14823 | 0 | DICT_FREE(ctxt->version); |
14824 | 0 | ctxt->version = NULL; |
14825 | 0 | DICT_FREE(ctxt->encoding); |
14826 | 0 | ctxt->encoding = NULL; |
14827 | 0 | DICT_FREE(ctxt->directory); |
14828 | 0 | ctxt->directory = NULL; |
14829 | 0 | DICT_FREE(ctxt->extSubURI); |
14830 | 0 | ctxt->extSubURI = NULL; |
14831 | 0 | DICT_FREE(ctxt->extSubSystem); |
14832 | 0 | ctxt->extSubSystem = NULL; |
14833 | 0 | if (ctxt->myDoc != NULL) |
14834 | 0 | xmlFreeDoc(ctxt->myDoc); |
14835 | 0 | ctxt->myDoc = NULL; |
14836 | |
|
14837 | 0 | ctxt->standalone = -1; |
14838 | 0 | ctxt->hasExternalSubset = 0; |
14839 | 0 | ctxt->hasPErefs = 0; |
14840 | 0 | ctxt->html = 0; |
14841 | 0 | ctxt->external = 0; |
14842 | 0 | ctxt->instate = XML_PARSER_START; |
14843 | 0 | ctxt->token = 0; |
14844 | |
|
14845 | 0 | ctxt->wellFormed = 1; |
14846 | 0 | ctxt->nsWellFormed = 1; |
14847 | 0 | ctxt->disableSAX = 0; |
14848 | 0 | ctxt->valid = 1; |
14849 | | #if 0 |
14850 | | ctxt->vctxt.userData = ctxt; |
14851 | | ctxt->vctxt.error = xmlParserValidityError; |
14852 | | ctxt->vctxt.warning = xmlParserValidityWarning; |
14853 | | #endif |
14854 | 0 | ctxt->record_info = 0; |
14855 | 0 | ctxt->checkIndex = 0; |
14856 | 0 | ctxt->inSubset = 0; |
14857 | 0 | ctxt->errNo = XML_ERR_OK; |
14858 | 0 | ctxt->depth = 0; |
14859 | 0 | ctxt->charset = XML_CHAR_ENCODING_UTF8; |
14860 | 0 | ctxt->catalogs = NULL; |
14861 | 0 | ctxt->nbentities = 0; |
14862 | 0 | ctxt->sizeentities = 0; |
14863 | 0 | ctxt->sizeentcopy = 0; |
14864 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
14865 | |
|
14866 | 0 | if (ctxt->attsDefault != NULL) { |
14867 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
14868 | 0 | ctxt->attsDefault = NULL; |
14869 | 0 | } |
14870 | 0 | if (ctxt->attsSpecial != NULL) { |
14871 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
14872 | 0 | ctxt->attsSpecial = NULL; |
14873 | 0 | } |
14874 | |
|
14875 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14876 | 0 | if (ctxt->catalogs != NULL) |
14877 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
14878 | 0 | #endif |
14879 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
14880 | 0 | xmlResetError(&ctxt->lastError); |
14881 | 0 | } |
14882 | | |
14883 | | /** |
14884 | | * xmlCtxtResetPush: |
14885 | | * @ctxt: an XML parser context |
14886 | | * @chunk: a pointer to an array of chars |
14887 | | * @size: number of chars in the array |
14888 | | * @filename: an optional file name or URI |
14889 | | * @encoding: the document encoding, or NULL |
14890 | | * |
14891 | | * Reset a push parser context |
14892 | | * |
14893 | | * Returns 0 in case of success and 1 in case of error |
14894 | | */ |
14895 | | int |
14896 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
14897 | | int size, const char *filename, const char *encoding) |
14898 | 0 | { |
14899 | 0 | xmlParserInputPtr inputStream; |
14900 | 0 | xmlParserInputBufferPtr buf; |
14901 | 0 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
14902 | |
|
14903 | 0 | if (ctxt == NULL) |
14904 | 0 | return(1); |
14905 | | |
14906 | 0 | if ((encoding == NULL) && (chunk != NULL) && (size >= 4)) |
14907 | 0 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
14908 | |
|
14909 | 0 | buf = xmlAllocParserInputBuffer(enc); |
14910 | 0 | if (buf == NULL) |
14911 | 0 | return(1); |
14912 | | |
14913 | 0 | if (ctxt == NULL) { |
14914 | 0 | xmlFreeParserInputBuffer(buf); |
14915 | 0 | return(1); |
14916 | 0 | } |
14917 | | |
14918 | 0 | xmlCtxtReset(ctxt); |
14919 | |
|
14920 | 0 | if (filename == NULL) { |
14921 | 0 | ctxt->directory = NULL; |
14922 | 0 | } else { |
14923 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14924 | 0 | } |
14925 | |
|
14926 | 0 | inputStream = xmlNewInputStream(ctxt); |
14927 | 0 | if (inputStream == NULL) { |
14928 | 0 | xmlFreeParserInputBuffer(buf); |
14929 | 0 | return(1); |
14930 | 0 | } |
14931 | | |
14932 | 0 | if (filename == NULL) |
14933 | 0 | inputStream->filename = NULL; |
14934 | 0 | else |
14935 | 0 | inputStream->filename = (char *) |
14936 | 0 | xmlCanonicPath((const xmlChar *) filename); |
14937 | 0 | inputStream->buf = buf; |
14938 | 0 | xmlBufResetInput(buf->buffer, inputStream); |
14939 | |
|
14940 | 0 | inputPush(ctxt, inputStream); |
14941 | |
|
14942 | 0 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
14943 | 0 | (ctxt->input->buf != NULL)) { |
14944 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
14945 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
14946 | |
|
14947 | 0 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
14948 | |
|
14949 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
14950 | | #ifdef DEBUG_PUSH |
14951 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
14952 | | #endif |
14953 | 0 | } |
14954 | |
|
14955 | 0 | if (encoding != NULL) { |
14956 | 0 | xmlCharEncodingHandlerPtr hdlr; |
14957 | |
|
14958 | 0 | if (ctxt->encoding != NULL) |
14959 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
14960 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
14961 | |
|
14962 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
14963 | 0 | if (hdlr != NULL) { |
14964 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
14965 | 0 | } else { |
14966 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
14967 | 0 | "Unsupported encoding %s\n", BAD_CAST encoding); |
14968 | 0 | } |
14969 | 0 | } else if (enc != XML_CHAR_ENCODING_NONE) { |
14970 | 0 | xmlSwitchEncoding(ctxt, enc); |
14971 | 0 | } |
14972 | |
|
14973 | 0 | return(0); |
14974 | 0 | } |
14975 | | |
14976 | | |
14977 | | /** |
14978 | | * xmlCtxtUseOptionsInternal: |
14979 | | * @ctxt: an XML parser context |
14980 | | * @options: a combination of xmlParserOption |
14981 | | * @encoding: the user provided encoding to use |
14982 | | * |
14983 | | * Applies the options to the parser context |
14984 | | * |
14985 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
14986 | | * in case of error. |
14987 | | */ |
14988 | | static int |
14989 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, const char *encoding) |
14990 | 18.9k | { |
14991 | 18.9k | if (ctxt == NULL) |
14992 | 0 | return(-1); |
14993 | 18.9k | if (encoding != NULL) { |
14994 | 0 | if (ctxt->encoding != NULL) |
14995 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
14996 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
14997 | 0 | } |
14998 | 18.9k | if (options & XML_PARSE_RECOVER) { |
14999 | 0 | ctxt->recovery = 1; |
15000 | 0 | options -= XML_PARSE_RECOVER; |
15001 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
15002 | 0 | } else |
15003 | 18.9k | ctxt->recovery = 0; |
15004 | 18.9k | if (options & XML_PARSE_DTDLOAD) { |
15005 | 0 | ctxt->loadsubset = XML_DETECT_IDS; |
15006 | 0 | options -= XML_PARSE_DTDLOAD; |
15007 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
15008 | 0 | } else |
15009 | 18.9k | ctxt->loadsubset = 0; |
15010 | 18.9k | if (options & XML_PARSE_DTDATTR) { |
15011 | 0 | ctxt->loadsubset |= XML_COMPLETE_ATTRS; |
15012 | 0 | options -= XML_PARSE_DTDATTR; |
15013 | 0 | ctxt->options |= XML_PARSE_DTDATTR; |
15014 | 0 | } |
15015 | 18.9k | if (options & XML_PARSE_NOENT) { |
15016 | 0 | ctxt->replaceEntities = 1; |
15017 | | /* ctxt->loadsubset |= XML_DETECT_IDS; */ |
15018 | 0 | options -= XML_PARSE_NOENT; |
15019 | 0 | ctxt->options |= XML_PARSE_NOENT; |
15020 | 0 | } else |
15021 | 18.9k | ctxt->replaceEntities = 0; |
15022 | 18.9k | if (options & XML_PARSE_PEDANTIC) { |
15023 | 0 | ctxt->pedantic = 1; |
15024 | 0 | options -= XML_PARSE_PEDANTIC; |
15025 | 0 | ctxt->options |= XML_PARSE_PEDANTIC; |
15026 | 0 | } else |
15027 | 18.9k | ctxt->pedantic = 0; |
15028 | 18.9k | if (options & XML_PARSE_NOBLANKS) { |
15029 | 0 | ctxt->keepBlanks = 0; |
15030 | 0 | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
15031 | 0 | options -= XML_PARSE_NOBLANKS; |
15032 | 0 | ctxt->options |= XML_PARSE_NOBLANKS; |
15033 | 0 | } else |
15034 | 18.9k | ctxt->keepBlanks = 1; |
15035 | 18.9k | if (options & XML_PARSE_DTDVALID) { |
15036 | 0 | ctxt->validate = 1; |
15037 | 0 | if (options & XML_PARSE_NOWARNING) |
15038 | 0 | ctxt->vctxt.warning = NULL; |
15039 | 0 | if (options & XML_PARSE_NOERROR) |
15040 | 0 | ctxt->vctxt.error = NULL; |
15041 | 0 | options -= XML_PARSE_DTDVALID; |
15042 | 0 | ctxt->options |= XML_PARSE_DTDVALID; |
15043 | 0 | } else |
15044 | 18.9k | ctxt->validate = 0; |
15045 | 18.9k | if (options & XML_PARSE_NOWARNING) { |
15046 | 0 | ctxt->sax->warning = NULL; |
15047 | 0 | options -= XML_PARSE_NOWARNING; |
15048 | 0 | } |
15049 | 18.9k | if (options & XML_PARSE_NOERROR) { |
15050 | 0 | ctxt->sax->error = NULL; |
15051 | 0 | ctxt->sax->fatalError = NULL; |
15052 | 0 | options -= XML_PARSE_NOERROR; |
15053 | 0 | } |
15054 | 18.9k | #ifdef LIBXML_SAX1_ENABLED |
15055 | 18.9k | if (options & XML_PARSE_SAX1) { |
15056 | 0 | ctxt->sax->startElement = xmlSAX2StartElement; |
15057 | 0 | ctxt->sax->endElement = xmlSAX2EndElement; |
15058 | 0 | ctxt->sax->startElementNs = NULL; |
15059 | 0 | ctxt->sax->endElementNs = NULL; |
15060 | 0 | ctxt->sax->initialized = 1; |
15061 | 0 | options -= XML_PARSE_SAX1; |
15062 | 0 | ctxt->options |= XML_PARSE_SAX1; |
15063 | 0 | } |
15064 | 18.9k | #endif /* LIBXML_SAX1_ENABLED */ |
15065 | 18.9k | if (options & XML_PARSE_NODICT) { |
15066 | 18.9k | ctxt->dictNames = 0; |
15067 | 18.9k | options -= XML_PARSE_NODICT; |
15068 | 18.9k | ctxt->options |= XML_PARSE_NODICT; |
15069 | 18.9k | } else { |
15070 | 0 | ctxt->dictNames = 1; |
15071 | 0 | } |
15072 | 18.9k | if (options & XML_PARSE_NOCDATA) { |
15073 | 0 | ctxt->sax->cdataBlock = NULL; |
15074 | 0 | options -= XML_PARSE_NOCDATA; |
15075 | 0 | ctxt->options |= XML_PARSE_NOCDATA; |
15076 | 0 | } |
15077 | 18.9k | if (options & XML_PARSE_NSCLEAN) { |
15078 | 0 | ctxt->options |= XML_PARSE_NSCLEAN; |
15079 | 0 | options -= XML_PARSE_NSCLEAN; |
15080 | 0 | } |
15081 | 18.9k | if (options & XML_PARSE_NONET) { |
15082 | 18.9k | ctxt->options |= XML_PARSE_NONET; |
15083 | 18.9k | options -= XML_PARSE_NONET; |
15084 | 18.9k | } |
15085 | 18.9k | if (options & XML_PARSE_COMPACT) { |
15086 | 0 | ctxt->options |= XML_PARSE_COMPACT; |
15087 | 0 | options -= XML_PARSE_COMPACT; |
15088 | 0 | } |
15089 | 18.9k | if (options & XML_PARSE_OLD10) { |
15090 | 0 | ctxt->options |= XML_PARSE_OLD10; |
15091 | 0 | options -= XML_PARSE_OLD10; |
15092 | 0 | } |
15093 | 18.9k | if (options & XML_PARSE_NOBASEFIX) { |
15094 | 0 | ctxt->options |= XML_PARSE_NOBASEFIX; |
15095 | 0 | options -= XML_PARSE_NOBASEFIX; |
15096 | 0 | } |
15097 | 18.9k | if (options & XML_PARSE_HUGE) { |
15098 | 18.9k | ctxt->options |= XML_PARSE_HUGE; |
15099 | 18.9k | options -= XML_PARSE_HUGE; |
15100 | 18.9k | if (ctxt->dict != NULL) |
15101 | 18.9k | xmlDictSetLimit(ctxt->dict, 0); |
15102 | 18.9k | } |
15103 | 18.9k | if (options & XML_PARSE_OLDSAX) { |
15104 | 0 | ctxt->options |= XML_PARSE_OLDSAX; |
15105 | 0 | options -= XML_PARSE_OLDSAX; |
15106 | 0 | } |
15107 | 18.9k | if (options & XML_PARSE_IGNORE_ENC) { |
15108 | 0 | ctxt->options |= XML_PARSE_IGNORE_ENC; |
15109 | 0 | options -= XML_PARSE_IGNORE_ENC; |
15110 | 0 | } |
15111 | 18.9k | if (options & XML_PARSE_BIG_LINES) { |
15112 | 0 | ctxt->options |= XML_PARSE_BIG_LINES; |
15113 | 0 | options -= XML_PARSE_BIG_LINES; |
15114 | 0 | } |
15115 | 18.9k | ctxt->linenumbers = 1; |
15116 | 18.9k | return (options); |
15117 | 18.9k | } |
15118 | | |
15119 | | /** |
15120 | | * xmlCtxtUseOptions: |
15121 | | * @ctxt: an XML parser context |
15122 | | * @options: a combination of xmlParserOption |
15123 | | * |
15124 | | * Applies the options to the parser context |
15125 | | * |
15126 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
15127 | | * in case of error. |
15128 | | */ |
15129 | | int |
15130 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
15131 | 18.9k | { |
15132 | 18.9k | return(xmlCtxtUseOptionsInternal(ctxt, options, NULL)); |
15133 | 18.9k | } |
15134 | | |
15135 | | /** |
15136 | | * xmlDoRead: |
15137 | | * @ctxt: an XML parser context |
15138 | | * @URL: the base URL to use for the document |
15139 | | * @encoding: the document encoding, or NULL |
15140 | | * @options: a combination of xmlParserOption |
15141 | | * @reuse: keep the context for reuse |
15142 | | * |
15143 | | * Common front-end for the xmlRead functions |
15144 | | * |
15145 | | * Returns the resulting document tree or NULL |
15146 | | */ |
15147 | | static xmlDocPtr |
15148 | | xmlDoRead(xmlParserCtxtPtr ctxt, const char *URL, const char *encoding, |
15149 | | int options, int reuse) |
15150 | 0 | { |
15151 | 0 | xmlDocPtr ret; |
15152 | |
|
15153 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, encoding); |
15154 | 0 | if (encoding != NULL) { |
15155 | 0 | xmlCharEncodingHandlerPtr hdlr; |
15156 | |
|
15157 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
15158 | 0 | if (hdlr != NULL) |
15159 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
15160 | 0 | } |
15161 | 0 | if ((URL != NULL) && (ctxt->input != NULL) && |
15162 | 0 | (ctxt->input->filename == NULL)) |
15163 | 0 | ctxt->input->filename = (char *) xmlStrdup((const xmlChar *) URL); |
15164 | 0 | xmlParseDocument(ctxt); |
15165 | 0 | if ((ctxt->wellFormed) || ctxt->recovery) |
15166 | 0 | ret = ctxt->myDoc; |
15167 | 0 | else { |
15168 | 0 | ret = NULL; |
15169 | 0 | if (ctxt->myDoc != NULL) { |
15170 | 0 | xmlFreeDoc(ctxt->myDoc); |
15171 | 0 | } |
15172 | 0 | } |
15173 | 0 | ctxt->myDoc = NULL; |
15174 | 0 | if (!reuse) { |
15175 | 0 | xmlFreeParserCtxt(ctxt); |
15176 | 0 | } |
15177 | |
|
15178 | 0 | return (ret); |
15179 | 0 | } |
15180 | | |
15181 | | /** |
15182 | | * xmlReadDoc: |
15183 | | * @cur: a pointer to a zero terminated string |
15184 | | * @URL: the base URL to use for the document |
15185 | | * @encoding: the document encoding, or NULL |
15186 | | * @options: a combination of xmlParserOption |
15187 | | * |
15188 | | * parse an XML in-memory document and build a tree. |
15189 | | * |
15190 | | * Returns the resulting document tree |
15191 | | */ |
15192 | | xmlDocPtr |
15193 | | xmlReadDoc(const xmlChar * cur, const char *URL, const char *encoding, int options) |
15194 | 0 | { |
15195 | 0 | xmlParserCtxtPtr ctxt; |
15196 | |
|
15197 | 0 | if (cur == NULL) |
15198 | 0 | return (NULL); |
15199 | 0 | xmlInitParser(); |
15200 | |
|
15201 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
15202 | 0 | if (ctxt == NULL) |
15203 | 0 | return (NULL); |
15204 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15205 | 0 | } |
15206 | | |
15207 | | /** |
15208 | | * xmlReadFile: |
15209 | | * @filename: a file or URL |
15210 | | * @encoding: the document encoding, or NULL |
15211 | | * @options: a combination of xmlParserOption |
15212 | | * |
15213 | | * parse an XML file from the filesystem or the network. |
15214 | | * |
15215 | | * Returns the resulting document tree |
15216 | | */ |
15217 | | xmlDocPtr |
15218 | | xmlReadFile(const char *filename, const char *encoding, int options) |
15219 | 0 | { |
15220 | 0 | xmlParserCtxtPtr ctxt; |
15221 | |
|
15222 | 0 | xmlInitParser(); |
15223 | 0 | ctxt = xmlCreateURLParserCtxt(filename, options); |
15224 | 0 | if (ctxt == NULL) |
15225 | 0 | return (NULL); |
15226 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 0)); |
15227 | 0 | } |
15228 | | |
15229 | | /** |
15230 | | * xmlReadMemory: |
15231 | | * @buffer: a pointer to a char array |
15232 | | * @size: the size of the array |
15233 | | * @URL: the base URL to use for the document |
15234 | | * @encoding: the document encoding, or NULL |
15235 | | * @options: a combination of xmlParserOption |
15236 | | * |
15237 | | * parse an XML in-memory document and build a tree. |
15238 | | * |
15239 | | * Returns the resulting document tree |
15240 | | */ |
15241 | | xmlDocPtr |
15242 | | xmlReadMemory(const char *buffer, int size, const char *URL, const char *encoding, int options) |
15243 | 0 | { |
15244 | 0 | xmlParserCtxtPtr ctxt; |
15245 | |
|
15246 | 0 | xmlInitParser(); |
15247 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
15248 | 0 | if (ctxt == NULL) |
15249 | 0 | return (NULL); |
15250 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15251 | 0 | } |
15252 | | |
15253 | | /** |
15254 | | * xmlReadFd: |
15255 | | * @fd: an open file descriptor |
15256 | | * @URL: the base URL to use for the document |
15257 | | * @encoding: the document encoding, or NULL |
15258 | | * @options: a combination of xmlParserOption |
15259 | | * |
15260 | | * parse an XML from a file descriptor and build a tree. |
15261 | | * NOTE that the file descriptor will not be closed when the |
15262 | | * reader is closed or reset. |
15263 | | * |
15264 | | * Returns the resulting document tree |
15265 | | */ |
15266 | | xmlDocPtr |
15267 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
15268 | 0 | { |
15269 | 0 | xmlParserCtxtPtr ctxt; |
15270 | 0 | xmlParserInputBufferPtr input; |
15271 | 0 | xmlParserInputPtr stream; |
15272 | |
|
15273 | 0 | if (fd < 0) |
15274 | 0 | return (NULL); |
15275 | 0 | xmlInitParser(); |
15276 | |
|
15277 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15278 | 0 | if (input == NULL) |
15279 | 0 | return (NULL); |
15280 | 0 | input->closecallback = NULL; |
15281 | 0 | ctxt = xmlNewParserCtxt(); |
15282 | 0 | if (ctxt == NULL) { |
15283 | 0 | xmlFreeParserInputBuffer(input); |
15284 | 0 | return (NULL); |
15285 | 0 | } |
15286 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15287 | 0 | if (stream == NULL) { |
15288 | 0 | xmlFreeParserInputBuffer(input); |
15289 | 0 | xmlFreeParserCtxt(ctxt); |
15290 | 0 | return (NULL); |
15291 | 0 | } |
15292 | 0 | inputPush(ctxt, stream); |
15293 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15294 | 0 | } |
15295 | | |
15296 | | /** |
15297 | | * xmlReadIO: |
15298 | | * @ioread: an I/O read function |
15299 | | * @ioclose: an I/O close function |
15300 | | * @ioctx: an I/O handler |
15301 | | * @URL: the base URL to use for the document |
15302 | | * @encoding: the document encoding, or NULL |
15303 | | * @options: a combination of xmlParserOption |
15304 | | * |
15305 | | * parse an XML document from I/O functions and source and build a tree. |
15306 | | * |
15307 | | * Returns the resulting document tree |
15308 | | */ |
15309 | | xmlDocPtr |
15310 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
15311 | | void *ioctx, const char *URL, const char *encoding, int options) |
15312 | 0 | { |
15313 | 0 | xmlParserCtxtPtr ctxt; |
15314 | 0 | xmlParserInputBufferPtr input; |
15315 | 0 | xmlParserInputPtr stream; |
15316 | |
|
15317 | 0 | if (ioread == NULL) |
15318 | 0 | return (NULL); |
15319 | 0 | xmlInitParser(); |
15320 | |
|
15321 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15322 | 0 | XML_CHAR_ENCODING_NONE); |
15323 | 0 | if (input == NULL) { |
15324 | 0 | if (ioclose != NULL) |
15325 | 0 | ioclose(ioctx); |
15326 | 0 | return (NULL); |
15327 | 0 | } |
15328 | 0 | ctxt = xmlNewParserCtxt(); |
15329 | 0 | if (ctxt == NULL) { |
15330 | 0 | xmlFreeParserInputBuffer(input); |
15331 | 0 | return (NULL); |
15332 | 0 | } |
15333 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15334 | 0 | if (stream == NULL) { |
15335 | 0 | xmlFreeParserInputBuffer(input); |
15336 | 0 | xmlFreeParserCtxt(ctxt); |
15337 | 0 | return (NULL); |
15338 | 0 | } |
15339 | 0 | inputPush(ctxt, stream); |
15340 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15341 | 0 | } |
15342 | | |
15343 | | /** |
15344 | | * xmlCtxtReadDoc: |
15345 | | * @ctxt: an XML parser context |
15346 | | * @cur: a pointer to a zero terminated string |
15347 | | * @URL: the base URL to use for the document |
15348 | | * @encoding: the document encoding, or NULL |
15349 | | * @options: a combination of xmlParserOption |
15350 | | * |
15351 | | * parse an XML in-memory document and build a tree. |
15352 | | * This reuses the existing @ctxt parser context |
15353 | | * |
15354 | | * Returns the resulting document tree |
15355 | | */ |
15356 | | xmlDocPtr |
15357 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar * cur, |
15358 | | const char *URL, const char *encoding, int options) |
15359 | 0 | { |
15360 | 0 | xmlParserInputPtr stream; |
15361 | |
|
15362 | 0 | if (cur == NULL) |
15363 | 0 | return (NULL); |
15364 | 0 | if (ctxt == NULL) |
15365 | 0 | return (NULL); |
15366 | 0 | xmlInitParser(); |
15367 | |
|
15368 | 0 | xmlCtxtReset(ctxt); |
15369 | |
|
15370 | 0 | stream = xmlNewStringInputStream(ctxt, cur); |
15371 | 0 | if (stream == NULL) { |
15372 | 0 | return (NULL); |
15373 | 0 | } |
15374 | 0 | inputPush(ctxt, stream); |
15375 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15376 | 0 | } |
15377 | | |
15378 | | /** |
15379 | | * xmlCtxtReadFile: |
15380 | | * @ctxt: an XML parser context |
15381 | | * @filename: a file or URL |
15382 | | * @encoding: the document encoding, or NULL |
15383 | | * @options: a combination of xmlParserOption |
15384 | | * |
15385 | | * parse an XML file from the filesystem or the network. |
15386 | | * This reuses the existing @ctxt parser context |
15387 | | * |
15388 | | * Returns the resulting document tree |
15389 | | */ |
15390 | | xmlDocPtr |
15391 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
15392 | | const char *encoding, int options) |
15393 | 0 | { |
15394 | 0 | xmlParserInputPtr stream; |
15395 | |
|
15396 | 0 | if (filename == NULL) |
15397 | 0 | return (NULL); |
15398 | 0 | if (ctxt == NULL) |
15399 | 0 | return (NULL); |
15400 | 0 | xmlInitParser(); |
15401 | |
|
15402 | 0 | xmlCtxtReset(ctxt); |
15403 | |
|
15404 | 0 | stream = xmlLoadExternalEntity(filename, NULL, ctxt); |
15405 | 0 | if (stream == NULL) { |
15406 | 0 | return (NULL); |
15407 | 0 | } |
15408 | 0 | inputPush(ctxt, stream); |
15409 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 1)); |
15410 | 0 | } |
15411 | | |
15412 | | /** |
15413 | | * xmlCtxtReadMemory: |
15414 | | * @ctxt: an XML parser context |
15415 | | * @buffer: a pointer to a char array |
15416 | | * @size: the size of the array |
15417 | | * @URL: the base URL to use for the document |
15418 | | * @encoding: the document encoding, or NULL |
15419 | | * @options: a combination of xmlParserOption |
15420 | | * |
15421 | | * parse an XML in-memory document and build a tree. |
15422 | | * This reuses the existing @ctxt parser context |
15423 | | * |
15424 | | * Returns the resulting document tree |
15425 | | */ |
15426 | | xmlDocPtr |
15427 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
15428 | | const char *URL, const char *encoding, int options) |
15429 | 0 | { |
15430 | 0 | xmlParserInputBufferPtr input; |
15431 | 0 | xmlParserInputPtr stream; |
15432 | |
|
15433 | 0 | if (ctxt == NULL) |
15434 | 0 | return (NULL); |
15435 | 0 | if (buffer == NULL) |
15436 | 0 | return (NULL); |
15437 | 0 | xmlInitParser(); |
15438 | |
|
15439 | 0 | xmlCtxtReset(ctxt); |
15440 | |
|
15441 | 0 | input = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
15442 | 0 | if (input == NULL) { |
15443 | 0 | return(NULL); |
15444 | 0 | } |
15445 | | |
15446 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15447 | 0 | if (stream == NULL) { |
15448 | 0 | xmlFreeParserInputBuffer(input); |
15449 | 0 | return(NULL); |
15450 | 0 | } |
15451 | | |
15452 | 0 | inputPush(ctxt, stream); |
15453 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15454 | 0 | } |
15455 | | |
15456 | | /** |
15457 | | * xmlCtxtReadFd: |
15458 | | * @ctxt: an XML parser context |
15459 | | * @fd: an open file descriptor |
15460 | | * @URL: the base URL to use for the document |
15461 | | * @encoding: the document encoding, or NULL |
15462 | | * @options: a combination of xmlParserOption |
15463 | | * |
15464 | | * parse an XML from a file descriptor and build a tree. |
15465 | | * This reuses the existing @ctxt parser context |
15466 | | * NOTE that the file descriptor will not be closed when the |
15467 | | * reader is closed or reset. |
15468 | | * |
15469 | | * Returns the resulting document tree |
15470 | | */ |
15471 | | xmlDocPtr |
15472 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
15473 | | const char *URL, const char *encoding, int options) |
15474 | 0 | { |
15475 | 0 | xmlParserInputBufferPtr input; |
15476 | 0 | xmlParserInputPtr stream; |
15477 | |
|
15478 | 0 | if (fd < 0) |
15479 | 0 | return (NULL); |
15480 | 0 | if (ctxt == NULL) |
15481 | 0 | return (NULL); |
15482 | 0 | xmlInitParser(); |
15483 | |
|
15484 | 0 | xmlCtxtReset(ctxt); |
15485 | | |
15486 | |
|
15487 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15488 | 0 | if (input == NULL) |
15489 | 0 | return (NULL); |
15490 | 0 | input->closecallback = NULL; |
15491 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15492 | 0 | if (stream == NULL) { |
15493 | 0 | xmlFreeParserInputBuffer(input); |
15494 | 0 | return (NULL); |
15495 | 0 | } |
15496 | 0 | inputPush(ctxt, stream); |
15497 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15498 | 0 | } |
15499 | | |
15500 | | /** |
15501 | | * xmlCtxtReadIO: |
15502 | | * @ctxt: an XML parser context |
15503 | | * @ioread: an I/O read function |
15504 | | * @ioclose: an I/O close function |
15505 | | * @ioctx: an I/O handler |
15506 | | * @URL: the base URL to use for the document |
15507 | | * @encoding: the document encoding, or NULL |
15508 | | * @options: a combination of xmlParserOption |
15509 | | * |
15510 | | * parse an XML document from I/O functions and source and build a tree. |
15511 | | * This reuses the existing @ctxt parser context |
15512 | | * |
15513 | | * Returns the resulting document tree |
15514 | | */ |
15515 | | xmlDocPtr |
15516 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
15517 | | xmlInputCloseCallback ioclose, void *ioctx, |
15518 | | const char *URL, |
15519 | | const char *encoding, int options) |
15520 | 0 | { |
15521 | 0 | xmlParserInputBufferPtr input; |
15522 | 0 | xmlParserInputPtr stream; |
15523 | |
|
15524 | 0 | if (ioread == NULL) |
15525 | 0 | return (NULL); |
15526 | 0 | if (ctxt == NULL) |
15527 | 0 | return (NULL); |
15528 | 0 | xmlInitParser(); |
15529 | |
|
15530 | 0 | xmlCtxtReset(ctxt); |
15531 | |
|
15532 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15533 | 0 | XML_CHAR_ENCODING_NONE); |
15534 | 0 | if (input == NULL) { |
15535 | 0 | if (ioclose != NULL) |
15536 | 0 | ioclose(ioctx); |
15537 | 0 | return (NULL); |
15538 | 0 | } |
15539 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15540 | 0 | if (stream == NULL) { |
15541 | 0 | xmlFreeParserInputBuffer(input); |
15542 | 0 | return (NULL); |
15543 | 0 | } |
15544 | 0 | inputPush(ctxt, stream); |
15545 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15546 | 0 | } |
15547 | | |