/src/libxml2-2.10.3/parser.c
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 | 968k | #define XML_MAX_HUGE_LENGTH 1000000000 |
106 | | |
107 | 0 | #define XML_PARSER_BIG_ENTITY 1000 |
108 | | #define XML_PARSER_LOT_ENTITY 5000 |
109 | | |
110 | | /* |
111 | | * XML_PARSER_NON_LINEAR is the threshold where the ratio of parsed entity |
112 | | * replacement over the size in byte of the input indicates that you have |
113 | | * and exponential behaviour. A value of 10 correspond to at least 3 entity |
114 | | * replacement per byte of input. |
115 | | */ |
116 | 0 | #define XML_PARSER_NON_LINEAR 10 |
117 | | |
118 | | /* |
119 | | * xmlParserEntityCheck |
120 | | * |
121 | | * Function to check non-linear entity expansion behaviour |
122 | | * This is here to detect and stop exponential linear entity expansion |
123 | | * This is not a limitation of the parser but a safety |
124 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
125 | | * parser option. |
126 | | */ |
127 | | static int |
128 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, size_t size, |
129 | | xmlEntityPtr ent, size_t replacement) |
130 | 448 | { |
131 | 448 | size_t consumed = 0; |
132 | 448 | int i; |
133 | | |
134 | 448 | if ((ctxt == NULL) || (ctxt->options & XML_PARSE_HUGE)) |
135 | 448 | return (0); |
136 | 0 | if (ctxt->lastError.code == XML_ERR_ENTITY_LOOP) |
137 | 0 | return (1); |
138 | | |
139 | | /* |
140 | | * This may look absurd but is needed to detect |
141 | | * entities problems |
142 | | */ |
143 | 0 | if ((ent != NULL) && (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
144 | 0 | (ent->content != NULL) && (ent->checked == 0) && |
145 | 0 | (ctxt->errNo != XML_ERR_ENTITY_LOOP)) { |
146 | 0 | unsigned long oldnbent = ctxt->nbentities, diff; |
147 | 0 | xmlChar *rep; |
148 | |
|
149 | 0 | ent->checked = 1; |
150 | |
|
151 | 0 | ++ctxt->depth; |
152 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
153 | 0 | XML_SUBSTITUTE_REF, 0, 0, 0); |
154 | 0 | --ctxt->depth; |
155 | 0 | if ((rep == NULL) || (ctxt->errNo == XML_ERR_ENTITY_LOOP)) { |
156 | 0 | ent->content[0] = 0; |
157 | 0 | } |
158 | |
|
159 | 0 | diff = ctxt->nbentities - oldnbent + 1; |
160 | 0 | if (diff > INT_MAX / 2) |
161 | 0 | diff = INT_MAX / 2; |
162 | 0 | ent->checked = diff * 2; |
163 | 0 | if (rep != NULL) { |
164 | 0 | if (xmlStrchr(rep, '<')) |
165 | 0 | ent->checked |= 1; |
166 | 0 | xmlFree(rep); |
167 | 0 | rep = NULL; |
168 | 0 | } |
169 | 0 | } |
170 | | |
171 | | /* |
172 | | * Prevent entity exponential check, not just replacement while |
173 | | * parsing the DTD |
174 | | * The check is potentially costly so do that only once in a thousand |
175 | | */ |
176 | 0 | if ((ctxt->instate == XML_PARSER_DTD) && (ctxt->nbentities > 10000) && |
177 | 0 | (ctxt->nbentities % 1024 == 0)) { |
178 | 0 | for (i = 0;i < ctxt->inputNr;i++) { |
179 | 0 | consumed += ctxt->inputTab[i]->consumed + |
180 | 0 | (ctxt->inputTab[i]->cur - ctxt->inputTab[i]->base); |
181 | 0 | } |
182 | 0 | if (ctxt->nbentities > consumed * XML_PARSER_NON_LINEAR) { |
183 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
184 | 0 | ctxt->instate = XML_PARSER_EOF; |
185 | 0 | return (1); |
186 | 0 | } |
187 | 0 | consumed = 0; |
188 | 0 | } |
189 | | |
190 | | |
191 | | |
192 | 0 | if (replacement != 0) { |
193 | 0 | if (replacement < XML_MAX_TEXT_LENGTH) |
194 | 0 | return(0); |
195 | | |
196 | | /* |
197 | | * If the volume of entity copy reaches 10 times the |
198 | | * amount of parsed data and over the large text threshold |
199 | | * then that's very likely to be an abuse. |
200 | | */ |
201 | 0 | if (ctxt->input != NULL) { |
202 | 0 | consumed = ctxt->input->consumed + |
203 | 0 | (ctxt->input->cur - ctxt->input->base); |
204 | 0 | } |
205 | 0 | consumed += ctxt->sizeentities; |
206 | |
|
207 | 0 | if (replacement < XML_PARSER_NON_LINEAR * consumed) |
208 | 0 | return(0); |
209 | 0 | } else if (size != 0) { |
210 | | /* |
211 | | * Do the check based on the replacement size of the entity |
212 | | */ |
213 | 0 | if (size < XML_PARSER_BIG_ENTITY) |
214 | 0 | return(0); |
215 | | |
216 | | /* |
217 | | * A limit on the amount of text data reasonably used |
218 | | */ |
219 | 0 | if (ctxt->input != NULL) { |
220 | 0 | consumed = ctxt->input->consumed + |
221 | 0 | (ctxt->input->cur - ctxt->input->base); |
222 | 0 | } |
223 | 0 | consumed += ctxt->sizeentities; |
224 | |
|
225 | 0 | if ((size < XML_PARSER_NON_LINEAR * consumed) && |
226 | 0 | (ctxt->nbentities * 3 < XML_PARSER_NON_LINEAR * consumed)) |
227 | 0 | return (0); |
228 | 0 | } else if (ent != NULL) { |
229 | | /* |
230 | | * use the number of parsed entities in the replacement |
231 | | */ |
232 | 0 | size = ent->checked / 2; |
233 | | |
234 | | /* |
235 | | * The amount of data parsed counting entities size only once |
236 | | */ |
237 | 0 | if (ctxt->input != NULL) { |
238 | 0 | consumed = ctxt->input->consumed + |
239 | 0 | (ctxt->input->cur - ctxt->input->base); |
240 | 0 | } |
241 | 0 | consumed += ctxt->sizeentities; |
242 | | |
243 | | /* |
244 | | * Check the density of entities for the amount of data |
245 | | * knowing an entity reference will take at least 3 bytes |
246 | | */ |
247 | 0 | if (size * 3 < consumed * XML_PARSER_NON_LINEAR) |
248 | 0 | return (0); |
249 | 0 | } else { |
250 | | /* |
251 | | * strange we got no data for checking |
252 | | */ |
253 | 0 | if (((ctxt->lastError.code != XML_ERR_UNDECLARED_ENTITY) && |
254 | 0 | (ctxt->lastError.code != XML_WAR_UNDECLARED_ENTITY)) || |
255 | 0 | (ctxt->nbentities <= 10000)) |
256 | 0 | return (0); |
257 | 0 | } |
258 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
259 | 0 | return (1); |
260 | 0 | } |
261 | | |
262 | | /** |
263 | | * xmlParserMaxDepth: |
264 | | * |
265 | | * arbitrary depth limit for the XML documents that we allow to |
266 | | * process. This is not a limitation of the parser but a safety |
267 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
268 | | * parser option. |
269 | | */ |
270 | | unsigned int xmlParserMaxDepth = 256; |
271 | | |
272 | | |
273 | | |
274 | | #define SAX2 1 |
275 | 8.65M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
276 | 14.8M | #define XML_PARSER_BUFFER_SIZE 100 |
277 | 8.81k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
278 | | |
279 | | /** |
280 | | * XML_PARSER_CHUNK_SIZE |
281 | | * |
282 | | * When calling GROW that's the minimal amount of data |
283 | | * the parser expected to have received. It is not a hard |
284 | | * limit but an optimization when reading strings like Names |
285 | | * It is not strictly needed as long as inputs available characters |
286 | | * are followed by 0, which should be provided by the I/O level |
287 | | */ |
288 | 219M | #define XML_PARSER_CHUNK_SIZE 100 |
289 | | |
290 | | /* |
291 | | * List of XML prefixed PI allowed by W3C specs |
292 | | */ |
293 | | |
294 | | static const char* const xmlW3CPIs[] = { |
295 | | "xml-stylesheet", |
296 | | "xml-model", |
297 | | NULL |
298 | | }; |
299 | | |
300 | | |
301 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
302 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
303 | | const xmlChar **str); |
304 | | |
305 | | static xmlParserErrors |
306 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
307 | | xmlSAXHandlerPtr sax, |
308 | | void *user_data, int depth, const xmlChar *URL, |
309 | | const xmlChar *ID, xmlNodePtr *list); |
310 | | |
311 | | static int |
312 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, |
313 | | const char *encoding); |
314 | | #ifdef LIBXML_LEGACY_ENABLED |
315 | | static void |
316 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
317 | | xmlNodePtr lastNode); |
318 | | #endif /* LIBXML_LEGACY_ENABLED */ |
319 | | |
320 | | static xmlParserErrors |
321 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
322 | | const xmlChar *string, void *user_data, xmlNodePtr *lst); |
323 | | |
324 | | static int |
325 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
326 | | |
327 | | /************************************************************************ |
328 | | * * |
329 | | * Some factorized error routines * |
330 | | * * |
331 | | ************************************************************************/ |
332 | | |
333 | | /** |
334 | | * xmlErrAttributeDup: |
335 | | * @ctxt: an XML parser context |
336 | | * @prefix: the attribute prefix |
337 | | * @localname: the attribute localname |
338 | | * |
339 | | * Handle a redefinition of attribute error |
340 | | */ |
341 | | static void |
342 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
343 | | const xmlChar * localname) |
344 | 180k | { |
345 | 180k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
346 | 180k | (ctxt->instate == XML_PARSER_EOF)) |
347 | 0 | return; |
348 | 180k | if (ctxt != NULL) |
349 | 180k | ctxt->errNo = XML_ERR_ATTRIBUTE_REDEFINED; |
350 | | |
351 | 180k | if (prefix == NULL) |
352 | 171k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
353 | 171k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
354 | 171k | (const char *) localname, NULL, NULL, 0, 0, |
355 | 171k | "Attribute %s redefined\n", localname); |
356 | 8.34k | else |
357 | 8.34k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
358 | 8.34k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
359 | 8.34k | (const char *) prefix, (const char *) localname, |
360 | 8.34k | NULL, 0, 0, "Attribute %s:%s redefined\n", prefix, |
361 | 8.34k | localname); |
362 | 180k | if (ctxt != NULL) { |
363 | 180k | ctxt->wellFormed = 0; |
364 | 180k | if (ctxt->recovery == 0) |
365 | 180k | ctxt->disableSAX = 1; |
366 | 180k | } |
367 | 180k | } |
368 | | |
369 | | /** |
370 | | * xmlFatalErr: |
371 | | * @ctxt: an XML parser context |
372 | | * @error: the error number |
373 | | * @extra: extra information string |
374 | | * |
375 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
376 | | */ |
377 | | static void |
378 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info) |
379 | 306k | { |
380 | 306k | const char *errmsg; |
381 | | |
382 | 306k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
383 | 306k | (ctxt->instate == XML_PARSER_EOF)) |
384 | 993 | return; |
385 | 305k | switch (error) { |
386 | 9.59k | case XML_ERR_INVALID_HEX_CHARREF: |
387 | 9.59k | errmsg = "CharRef: invalid hexadecimal value"; |
388 | 9.59k | break; |
389 | 209k | case XML_ERR_INVALID_DEC_CHARREF: |
390 | 209k | errmsg = "CharRef: invalid decimal value"; |
391 | 209k | break; |
392 | 0 | case XML_ERR_INVALID_CHARREF: |
393 | 0 | errmsg = "CharRef: invalid value"; |
394 | 0 | break; |
395 | 1.77k | case XML_ERR_INTERNAL_ERROR: |
396 | 1.77k | errmsg = "internal error"; |
397 | 1.77k | break; |
398 | 0 | case XML_ERR_PEREF_AT_EOF: |
399 | 0 | errmsg = "PEReference at end of document"; |
400 | 0 | break; |
401 | 0 | case XML_ERR_PEREF_IN_PROLOG: |
402 | 0 | errmsg = "PEReference in prolog"; |
403 | 0 | break; |
404 | 0 | case XML_ERR_PEREF_IN_EPILOG: |
405 | 0 | errmsg = "PEReference in epilog"; |
406 | 0 | break; |
407 | 0 | case XML_ERR_PEREF_NO_NAME: |
408 | 0 | errmsg = "PEReference: no name"; |
409 | 0 | break; |
410 | 720 | case XML_ERR_PEREF_SEMICOL_MISSING: |
411 | 720 | errmsg = "PEReference: expecting ';'"; |
412 | 720 | break; |
413 | 0 | case XML_ERR_ENTITY_LOOP: |
414 | 0 | errmsg = "Detected an entity reference loop"; |
415 | 0 | break; |
416 | 0 | case XML_ERR_ENTITY_NOT_STARTED: |
417 | 0 | errmsg = "EntityValue: \" or ' expected"; |
418 | 0 | break; |
419 | 225 | case XML_ERR_ENTITY_PE_INTERNAL: |
420 | 225 | errmsg = "PEReferences forbidden in internal subset"; |
421 | 225 | break; |
422 | 60 | case XML_ERR_ENTITY_NOT_FINISHED: |
423 | 60 | errmsg = "EntityValue: \" or ' expected"; |
424 | 60 | break; |
425 | 942 | case XML_ERR_ATTRIBUTE_NOT_STARTED: |
426 | 942 | errmsg = "AttValue: \" or ' expected"; |
427 | 942 | break; |
428 | 833 | case XML_ERR_LT_IN_ATTRIBUTE: |
429 | 833 | errmsg = "Unescaped '<' not allowed in attributes values"; |
430 | 833 | break; |
431 | 2.29k | case XML_ERR_LITERAL_NOT_STARTED: |
432 | 2.29k | errmsg = "SystemLiteral \" or ' expected"; |
433 | 2.29k | break; |
434 | 964 | case XML_ERR_LITERAL_NOT_FINISHED: |
435 | 964 | errmsg = "Unfinished System or Public ID \" or ' expected"; |
436 | 964 | break; |
437 | 224 | case XML_ERR_MISPLACED_CDATA_END: |
438 | 224 | errmsg = "Sequence ']]>' not allowed in content"; |
439 | 224 | break; |
440 | 1.70k | case XML_ERR_URI_REQUIRED: |
441 | 1.70k | errmsg = "SYSTEM or PUBLIC, the URI is missing"; |
442 | 1.70k | break; |
443 | 586 | case XML_ERR_PUBID_REQUIRED: |
444 | 586 | errmsg = "PUBLIC, the Public Identifier is missing"; |
445 | 586 | break; |
446 | 1.52k | case XML_ERR_HYPHEN_IN_COMMENT: |
447 | 1.52k | errmsg = "Comment must not contain '--' (double-hyphen)"; |
448 | 1.52k | break; |
449 | 1.21k | case XML_ERR_PI_NOT_STARTED: |
450 | 1.21k | errmsg = "xmlParsePI : no target name"; |
451 | 1.21k | break; |
452 | 720 | case XML_ERR_RESERVED_XML_NAME: |
453 | 720 | errmsg = "Invalid PI name"; |
454 | 720 | break; |
455 | 469 | case XML_ERR_NOTATION_NOT_STARTED: |
456 | 469 | errmsg = "NOTATION: Name expected here"; |
457 | 469 | break; |
458 | 1.16k | case XML_ERR_NOTATION_NOT_FINISHED: |
459 | 1.16k | errmsg = "'>' required to close NOTATION declaration"; |
460 | 1.16k | break; |
461 | 1.78k | case XML_ERR_VALUE_REQUIRED: |
462 | 1.78k | errmsg = "Entity value required"; |
463 | 1.78k | break; |
464 | 457 | case XML_ERR_URI_FRAGMENT: |
465 | 457 | errmsg = "Fragment not allowed"; |
466 | 457 | break; |
467 | 458 | case XML_ERR_ATTLIST_NOT_STARTED: |
468 | 458 | errmsg = "'(' required to start ATTLIST enumeration"; |
469 | 458 | break; |
470 | 243 | case XML_ERR_NMTOKEN_REQUIRED: |
471 | 243 | errmsg = "NmToken expected in ATTLIST enumeration"; |
472 | 243 | break; |
473 | 570 | case XML_ERR_ATTLIST_NOT_FINISHED: |
474 | 570 | errmsg = "')' required to finish ATTLIST enumeration"; |
475 | 570 | break; |
476 | 625 | case XML_ERR_MIXED_NOT_STARTED: |
477 | 625 | errmsg = "MixedContentDecl : '|' or ')*' expected"; |
478 | 625 | break; |
479 | 0 | case XML_ERR_PCDATA_REQUIRED: |
480 | 0 | errmsg = "MixedContentDecl : '#PCDATA' expected"; |
481 | 0 | break; |
482 | 801 | case XML_ERR_ELEMCONTENT_NOT_STARTED: |
483 | 801 | errmsg = "ContentDecl : Name or '(' expected"; |
484 | 801 | break; |
485 | 1.80k | case XML_ERR_ELEMCONTENT_NOT_FINISHED: |
486 | 1.80k | errmsg = "ContentDecl : ',' '|' or ')' expected"; |
487 | 1.80k | break; |
488 | 0 | case XML_ERR_PEREF_IN_INT_SUBSET: |
489 | 0 | errmsg = |
490 | 0 | "PEReference: forbidden within markup decl in internal subset"; |
491 | 0 | break; |
492 | 4.51k | case XML_ERR_GT_REQUIRED: |
493 | 4.51k | errmsg = "expected '>'"; |
494 | 4.51k | break; |
495 | 0 | case XML_ERR_CONDSEC_INVALID: |
496 | 0 | errmsg = "XML conditional section '[' expected"; |
497 | 0 | break; |
498 | 0 | case XML_ERR_EXT_SUBSET_NOT_FINISHED: |
499 | 0 | errmsg = "Content error in the external subset"; |
500 | 0 | break; |
501 | 0 | case XML_ERR_CONDSEC_INVALID_KEYWORD: |
502 | 0 | errmsg = |
503 | 0 | "conditional section INCLUDE or IGNORE keyword expected"; |
504 | 0 | break; |
505 | 0 | case XML_ERR_CONDSEC_NOT_FINISHED: |
506 | 0 | errmsg = "XML conditional section not closed"; |
507 | 0 | break; |
508 | 0 | case XML_ERR_XMLDECL_NOT_STARTED: |
509 | 0 | errmsg = "Text declaration '<?xml' required"; |
510 | 0 | break; |
511 | 425 | case XML_ERR_XMLDECL_NOT_FINISHED: |
512 | 425 | errmsg = "parsing XML declaration: '?>' expected"; |
513 | 425 | break; |
514 | 0 | case XML_ERR_EXT_ENTITY_STANDALONE: |
515 | 0 | errmsg = "external parsed entities cannot be standalone"; |
516 | 0 | break; |
517 | 52.4k | case XML_ERR_ENTITYREF_SEMICOL_MISSING: |
518 | 52.4k | errmsg = "EntityRef: expecting ';'"; |
519 | 52.4k | break; |
520 | 2.14k | case XML_ERR_DOCTYPE_NOT_FINISHED: |
521 | 2.14k | errmsg = "DOCTYPE improperly terminated"; |
522 | 2.14k | break; |
523 | 0 | case XML_ERR_LTSLASH_REQUIRED: |
524 | 0 | errmsg = "EndTag: '</' not found"; |
525 | 0 | break; |
526 | 19 | case XML_ERR_EQUAL_REQUIRED: |
527 | 19 | errmsg = "expected '='"; |
528 | 19 | break; |
529 | 197 | case XML_ERR_STRING_NOT_CLOSED: |
530 | 197 | errmsg = "String not closed expecting \" or '"; |
531 | 197 | break; |
532 | 19 | case XML_ERR_STRING_NOT_STARTED: |
533 | 19 | errmsg = "String not started expecting ' or \""; |
534 | 19 | break; |
535 | 12 | case XML_ERR_ENCODING_NAME: |
536 | 12 | errmsg = "Invalid XML encoding name"; |
537 | 12 | break; |
538 | 18 | case XML_ERR_STANDALONE_VALUE: |
539 | 18 | errmsg = "standalone accepts only 'yes' or 'no'"; |
540 | 18 | break; |
541 | 300 | case XML_ERR_DOCUMENT_EMPTY: |
542 | 300 | errmsg = "Document is empty"; |
543 | 300 | break; |
544 | 2.62k | case XML_ERR_DOCUMENT_END: |
545 | 2.62k | errmsg = "Extra content at the end of the document"; |
546 | 2.62k | break; |
547 | 0 | case XML_ERR_NOT_WELL_BALANCED: |
548 | 0 | errmsg = "chunk is not well balanced"; |
549 | 0 | break; |
550 | 0 | case XML_ERR_EXTRA_CONTENT: |
551 | 0 | errmsg = "extra content at the end of well balanced chunk"; |
552 | 0 | break; |
553 | 788 | case XML_ERR_VERSION_MISSING: |
554 | 788 | errmsg = "Malformed declaration expecting version"; |
555 | 788 | break; |
556 | 33 | case XML_ERR_NAME_TOO_LONG: |
557 | 33 | errmsg = "Name too long"; |
558 | 33 | break; |
559 | | #if 0 |
560 | | case: |
561 | | errmsg = ""; |
562 | | break; |
563 | | #endif |
564 | 0 | default: |
565 | 0 | errmsg = "Unregistered error message"; |
566 | 305k | } |
567 | 305k | if (ctxt != NULL) |
568 | 305k | ctxt->errNo = error; |
569 | 305k | if (info == NULL) { |
570 | 303k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
571 | 303k | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s\n", |
572 | 303k | errmsg); |
573 | 303k | } else { |
574 | 1.81k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
575 | 1.81k | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s: %s\n", |
576 | 1.81k | errmsg, info); |
577 | 1.81k | } |
578 | 305k | if (ctxt != NULL) { |
579 | 305k | ctxt->wellFormed = 0; |
580 | 305k | if (ctxt->recovery == 0) |
581 | 305k | ctxt->disableSAX = 1; |
582 | 305k | } |
583 | 305k | } |
584 | | |
585 | | /** |
586 | | * xmlFatalErrMsg: |
587 | | * @ctxt: an XML parser context |
588 | | * @error: the error number |
589 | | * @msg: the error message |
590 | | * |
591 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
592 | | */ |
593 | | static void LIBXML_ATTR_FORMAT(3,0) |
594 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
595 | | const char *msg) |
596 | 232k | { |
597 | 232k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
598 | 232k | (ctxt->instate == XML_PARSER_EOF)) |
599 | 0 | return; |
600 | 232k | if (ctxt != NULL) |
601 | 232k | ctxt->errNo = error; |
602 | 232k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
603 | 232k | XML_ERR_FATAL, NULL, 0, NULL, NULL, NULL, 0, 0, "%s", msg); |
604 | 232k | if (ctxt != NULL) { |
605 | 232k | ctxt->wellFormed = 0; |
606 | 232k | if (ctxt->recovery == 0) |
607 | 232k | ctxt->disableSAX = 1; |
608 | 232k | } |
609 | 232k | } |
610 | | |
611 | | /** |
612 | | * xmlWarningMsg: |
613 | | * @ctxt: an XML parser context |
614 | | * @error: the error number |
615 | | * @msg: the error message |
616 | | * @str1: extra data |
617 | | * @str2: extra data |
618 | | * |
619 | | * Handle a warning. |
620 | | */ |
621 | | static void LIBXML_ATTR_FORMAT(3,0) |
622 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
623 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
624 | 4.41k | { |
625 | 4.41k | xmlStructuredErrorFunc schannel = NULL; |
626 | | |
627 | 4.41k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
628 | 4.41k | (ctxt->instate == XML_PARSER_EOF)) |
629 | 0 | return; |
630 | 4.41k | if ((ctxt != NULL) && (ctxt->sax != NULL) && |
631 | 4.41k | (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
632 | 4.41k | schannel = ctxt->sax->serror; |
633 | 4.41k | if (ctxt != NULL) { |
634 | 4.41k | __xmlRaiseError(schannel, |
635 | 4.41k | (ctxt->sax) ? ctxt->sax->warning : NULL, |
636 | 4.41k | ctxt->userData, |
637 | 4.41k | ctxt, NULL, XML_FROM_PARSER, error, |
638 | 4.41k | XML_ERR_WARNING, NULL, 0, |
639 | 4.41k | (const char *) str1, (const char *) str2, NULL, 0, 0, |
640 | 4.41k | msg, (const char *) str1, (const char *) str2); |
641 | 4.41k | } else { |
642 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
643 | 0 | ctxt, NULL, XML_FROM_PARSER, error, |
644 | 0 | XML_ERR_WARNING, NULL, 0, |
645 | 0 | (const char *) str1, (const char *) str2, NULL, 0, 0, |
646 | 0 | msg, (const char *) str1, (const char *) str2); |
647 | 0 | } |
648 | 4.41k | } |
649 | | |
650 | | /** |
651 | | * xmlValidityError: |
652 | | * @ctxt: an XML parser context |
653 | | * @error: the error number |
654 | | * @msg: the error message |
655 | | * @str1: extra data |
656 | | * |
657 | | * Handle a validity error. |
658 | | */ |
659 | | static void LIBXML_ATTR_FORMAT(3,0) |
660 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
661 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
662 | 2.08k | { |
663 | 2.08k | xmlStructuredErrorFunc schannel = NULL; |
664 | | |
665 | 2.08k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
666 | 2.08k | (ctxt->instate == XML_PARSER_EOF)) |
667 | 0 | return; |
668 | 2.08k | if (ctxt != NULL) { |
669 | 2.08k | ctxt->errNo = error; |
670 | 2.08k | if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
671 | 2.08k | schannel = ctxt->sax->serror; |
672 | 2.08k | } |
673 | 2.08k | if (ctxt != NULL) { |
674 | 2.08k | __xmlRaiseError(schannel, |
675 | 2.08k | ctxt->vctxt.error, ctxt->vctxt.userData, |
676 | 2.08k | ctxt, NULL, XML_FROM_DTD, error, |
677 | 2.08k | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
678 | 2.08k | (const char *) str2, NULL, 0, 0, |
679 | 2.08k | msg, (const char *) str1, (const char *) str2); |
680 | 2.08k | ctxt->valid = 0; |
681 | 2.08k | } else { |
682 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
683 | 0 | ctxt, NULL, XML_FROM_DTD, error, |
684 | 0 | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
685 | 0 | (const char *) str2, NULL, 0, 0, |
686 | 0 | msg, (const char *) str1, (const char *) str2); |
687 | 0 | } |
688 | 2.08k | } |
689 | | |
690 | | /** |
691 | | * xmlFatalErrMsgInt: |
692 | | * @ctxt: an XML parser context |
693 | | * @error: the error number |
694 | | * @msg: the error message |
695 | | * @val: an integer value |
696 | | * |
697 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
698 | | */ |
699 | | static void LIBXML_ATTR_FORMAT(3,0) |
700 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
701 | | const char *msg, int val) |
702 | 247k | { |
703 | 247k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
704 | 247k | (ctxt->instate == XML_PARSER_EOF)) |
705 | 0 | return; |
706 | 247k | if (ctxt != NULL) |
707 | 247k | ctxt->errNo = error; |
708 | 247k | __xmlRaiseError(NULL, NULL, NULL, |
709 | 247k | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
710 | 247k | NULL, 0, NULL, NULL, NULL, val, 0, msg, val); |
711 | 247k | if (ctxt != NULL) { |
712 | 247k | ctxt->wellFormed = 0; |
713 | 247k | if (ctxt->recovery == 0) |
714 | 247k | ctxt->disableSAX = 1; |
715 | 247k | } |
716 | 247k | } |
717 | | |
718 | | /** |
719 | | * xmlFatalErrMsgStrIntStr: |
720 | | * @ctxt: an XML parser context |
721 | | * @error: the error number |
722 | | * @msg: the error message |
723 | | * @str1: an string info |
724 | | * @val: an integer value |
725 | | * @str2: an string info |
726 | | * |
727 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
728 | | */ |
729 | | static void LIBXML_ATTR_FORMAT(3,0) |
730 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
731 | | const char *msg, const xmlChar *str1, int val, |
732 | | const xmlChar *str2) |
733 | 163 | { |
734 | 163 | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
735 | 163 | (ctxt->instate == XML_PARSER_EOF)) |
736 | 0 | return; |
737 | 163 | if (ctxt != NULL) |
738 | 163 | ctxt->errNo = error; |
739 | 163 | __xmlRaiseError(NULL, NULL, NULL, |
740 | 163 | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
741 | 163 | NULL, 0, (const char *) str1, (const char *) str2, |
742 | 163 | NULL, val, 0, msg, str1, val, str2); |
743 | 163 | if (ctxt != NULL) { |
744 | 163 | ctxt->wellFormed = 0; |
745 | 163 | if (ctxt->recovery == 0) |
746 | 163 | ctxt->disableSAX = 1; |
747 | 163 | } |
748 | 163 | } |
749 | | |
750 | | /** |
751 | | * xmlFatalErrMsgStr: |
752 | | * @ctxt: an XML parser context |
753 | | * @error: the error number |
754 | | * @msg: the error message |
755 | | * @val: a string value |
756 | | * |
757 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
758 | | */ |
759 | | static void LIBXML_ATTR_FORMAT(3,0) |
760 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
761 | | const char *msg, const xmlChar * val) |
762 | 28.4k | { |
763 | 28.4k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
764 | 28.4k | (ctxt->instate == XML_PARSER_EOF)) |
765 | 0 | return; |
766 | 28.4k | if (ctxt != NULL) |
767 | 28.4k | ctxt->errNo = error; |
768 | 28.4k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
769 | 28.4k | XML_FROM_PARSER, error, XML_ERR_FATAL, |
770 | 28.4k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
771 | 28.4k | val); |
772 | 28.4k | if (ctxt != NULL) { |
773 | 28.4k | ctxt->wellFormed = 0; |
774 | 28.4k | if (ctxt->recovery == 0) |
775 | 28.4k | ctxt->disableSAX = 1; |
776 | 28.4k | } |
777 | 28.4k | } |
778 | | |
779 | | /** |
780 | | * xmlErrMsgStr: |
781 | | * @ctxt: an XML parser context |
782 | | * @error: the error number |
783 | | * @msg: the error message |
784 | | * @val: a string value |
785 | | * |
786 | | * Handle a non fatal parser error |
787 | | */ |
788 | | static void LIBXML_ATTR_FORMAT(3,0) |
789 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
790 | | const char *msg, const xmlChar * val) |
791 | 1.03k | { |
792 | 1.03k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
793 | 1.03k | (ctxt->instate == XML_PARSER_EOF)) |
794 | 0 | return; |
795 | 1.03k | if (ctxt != NULL) |
796 | 1.03k | ctxt->errNo = error; |
797 | 1.03k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
798 | 1.03k | XML_FROM_PARSER, error, XML_ERR_ERROR, |
799 | 1.03k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
800 | 1.03k | val); |
801 | 1.03k | } |
802 | | |
803 | | /** |
804 | | * xmlNsErr: |
805 | | * @ctxt: an XML parser context |
806 | | * @error: the error number |
807 | | * @msg: the message |
808 | | * @info1: extra information string |
809 | | * @info2: extra information string |
810 | | * |
811 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
812 | | */ |
813 | | static void LIBXML_ATTR_FORMAT(3,0) |
814 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
815 | | const char *msg, |
816 | | const xmlChar * info1, const xmlChar * info2, |
817 | | const xmlChar * info3) |
818 | 69.0k | { |
819 | 69.0k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
820 | 69.0k | (ctxt->instate == XML_PARSER_EOF)) |
821 | 0 | return; |
822 | 69.0k | if (ctxt != NULL) |
823 | 69.0k | ctxt->errNo = error; |
824 | 69.0k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
825 | 69.0k | XML_ERR_ERROR, NULL, 0, (const char *) info1, |
826 | 69.0k | (const char *) info2, (const char *) info3, 0, 0, msg, |
827 | 69.0k | info1, info2, info3); |
828 | 69.0k | if (ctxt != NULL) |
829 | 69.0k | ctxt->nsWellFormed = 0; |
830 | 69.0k | } |
831 | | |
832 | | /** |
833 | | * xmlNsWarn |
834 | | * @ctxt: an XML parser context |
835 | | * @error: the error number |
836 | | * @msg: the message |
837 | | * @info1: extra information string |
838 | | * @info2: extra information string |
839 | | * |
840 | | * Handle a namespace warning error |
841 | | */ |
842 | | static void LIBXML_ATTR_FORMAT(3,0) |
843 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
844 | | const char *msg, |
845 | | const xmlChar * info1, const xmlChar * info2, |
846 | | const xmlChar * info3) |
847 | 10.0k | { |
848 | 10.0k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
849 | 10.0k | (ctxt->instate == XML_PARSER_EOF)) |
850 | 0 | return; |
851 | 10.0k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
852 | 10.0k | XML_ERR_WARNING, NULL, 0, (const char *) info1, |
853 | 10.0k | (const char *) info2, (const char *) info3, 0, 0, msg, |
854 | 10.0k | info1, info2, info3); |
855 | 10.0k | } |
856 | | |
857 | | /************************************************************************ |
858 | | * * |
859 | | * Library wide options * |
860 | | * * |
861 | | ************************************************************************/ |
862 | | |
863 | | /** |
864 | | * xmlHasFeature: |
865 | | * @feature: the feature to be examined |
866 | | * |
867 | | * Examines if the library has been compiled with a given feature. |
868 | | * |
869 | | * Returns a non-zero value if the feature exist, otherwise zero. |
870 | | * Returns zero (0) if the feature does not exist or an unknown |
871 | | * unknown feature is requested, non-zero otherwise. |
872 | | */ |
873 | | int |
874 | | xmlHasFeature(xmlFeature feature) |
875 | 0 | { |
876 | 0 | switch (feature) { |
877 | 0 | case XML_WITH_THREAD: |
878 | 0 | #ifdef LIBXML_THREAD_ENABLED |
879 | 0 | return(1); |
880 | | #else |
881 | | return(0); |
882 | | #endif |
883 | 0 | case XML_WITH_TREE: |
884 | 0 | #ifdef LIBXML_TREE_ENABLED |
885 | 0 | return(1); |
886 | | #else |
887 | | return(0); |
888 | | #endif |
889 | 0 | case XML_WITH_OUTPUT: |
890 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
891 | 0 | return(1); |
892 | | #else |
893 | | return(0); |
894 | | #endif |
895 | 0 | case XML_WITH_PUSH: |
896 | 0 | #ifdef LIBXML_PUSH_ENABLED |
897 | 0 | return(1); |
898 | | #else |
899 | | return(0); |
900 | | #endif |
901 | 0 | case XML_WITH_READER: |
902 | 0 | #ifdef LIBXML_READER_ENABLED |
903 | 0 | return(1); |
904 | | #else |
905 | | return(0); |
906 | | #endif |
907 | 0 | case XML_WITH_PATTERN: |
908 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
909 | 0 | return(1); |
910 | | #else |
911 | | return(0); |
912 | | #endif |
913 | 0 | case XML_WITH_WRITER: |
914 | 0 | #ifdef LIBXML_WRITER_ENABLED |
915 | 0 | return(1); |
916 | | #else |
917 | | return(0); |
918 | | #endif |
919 | 0 | case XML_WITH_SAX1: |
920 | 0 | #ifdef LIBXML_SAX1_ENABLED |
921 | 0 | return(1); |
922 | | #else |
923 | | return(0); |
924 | | #endif |
925 | 0 | case XML_WITH_FTP: |
926 | | #ifdef LIBXML_FTP_ENABLED |
927 | | return(1); |
928 | | #else |
929 | 0 | return(0); |
930 | 0 | #endif |
931 | 0 | case XML_WITH_HTTP: |
932 | 0 | #ifdef LIBXML_HTTP_ENABLED |
933 | 0 | return(1); |
934 | | #else |
935 | | return(0); |
936 | | #endif |
937 | 0 | case XML_WITH_VALID: |
938 | 0 | #ifdef LIBXML_VALID_ENABLED |
939 | 0 | return(1); |
940 | | #else |
941 | | return(0); |
942 | | #endif |
943 | 0 | case XML_WITH_HTML: |
944 | 0 | #ifdef LIBXML_HTML_ENABLED |
945 | 0 | return(1); |
946 | | #else |
947 | | return(0); |
948 | | #endif |
949 | 0 | case XML_WITH_LEGACY: |
950 | | #ifdef LIBXML_LEGACY_ENABLED |
951 | | return(1); |
952 | | #else |
953 | 0 | return(0); |
954 | 0 | #endif |
955 | 0 | case XML_WITH_C14N: |
956 | 0 | #ifdef LIBXML_C14N_ENABLED |
957 | 0 | return(1); |
958 | | #else |
959 | | return(0); |
960 | | #endif |
961 | 0 | case XML_WITH_CATALOG: |
962 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
963 | 0 | return(1); |
964 | | #else |
965 | | return(0); |
966 | | #endif |
967 | 0 | case XML_WITH_XPATH: |
968 | 0 | #ifdef LIBXML_XPATH_ENABLED |
969 | 0 | return(1); |
970 | | #else |
971 | | return(0); |
972 | | #endif |
973 | 0 | case XML_WITH_XPTR: |
974 | 0 | #ifdef LIBXML_XPTR_ENABLED |
975 | 0 | return(1); |
976 | | #else |
977 | | return(0); |
978 | | #endif |
979 | 0 | case XML_WITH_XINCLUDE: |
980 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
981 | 0 | return(1); |
982 | | #else |
983 | | return(0); |
984 | | #endif |
985 | 0 | case XML_WITH_ICONV: |
986 | 0 | #ifdef LIBXML_ICONV_ENABLED |
987 | 0 | return(1); |
988 | | #else |
989 | | return(0); |
990 | | #endif |
991 | 0 | case XML_WITH_ISO8859X: |
992 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
993 | 0 | return(1); |
994 | | #else |
995 | | return(0); |
996 | | #endif |
997 | 0 | case XML_WITH_UNICODE: |
998 | 0 | #ifdef LIBXML_UNICODE_ENABLED |
999 | 0 | return(1); |
1000 | | #else |
1001 | | return(0); |
1002 | | #endif |
1003 | 0 | case XML_WITH_REGEXP: |
1004 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
1005 | 0 | return(1); |
1006 | | #else |
1007 | | return(0); |
1008 | | #endif |
1009 | 0 | case XML_WITH_AUTOMATA: |
1010 | 0 | #ifdef LIBXML_AUTOMATA_ENABLED |
1011 | 0 | return(1); |
1012 | | #else |
1013 | | return(0); |
1014 | | #endif |
1015 | 0 | case XML_WITH_EXPR: |
1016 | | #ifdef LIBXML_EXPR_ENABLED |
1017 | | return(1); |
1018 | | #else |
1019 | 0 | return(0); |
1020 | 0 | #endif |
1021 | 0 | case XML_WITH_SCHEMAS: |
1022 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
1023 | 0 | return(1); |
1024 | | #else |
1025 | | return(0); |
1026 | | #endif |
1027 | 0 | case XML_WITH_SCHEMATRON: |
1028 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
1029 | 0 | return(1); |
1030 | | #else |
1031 | | return(0); |
1032 | | #endif |
1033 | 0 | case XML_WITH_MODULES: |
1034 | | #ifdef LIBXML_MODULES_ENABLED |
1035 | | return(1); |
1036 | | #else |
1037 | 0 | return(0); |
1038 | 0 | #endif |
1039 | 0 | case XML_WITH_DEBUG: |
1040 | 0 | #ifdef LIBXML_DEBUG_ENABLED |
1041 | 0 | return(1); |
1042 | | #else |
1043 | | return(0); |
1044 | | #endif |
1045 | 0 | case XML_WITH_DEBUG_MEM: |
1046 | | #ifdef DEBUG_MEMORY_LOCATION |
1047 | | return(1); |
1048 | | #else |
1049 | 0 | return(0); |
1050 | 0 | #endif |
1051 | 0 | case XML_WITH_DEBUG_RUN: |
1052 | | #ifdef LIBXML_DEBUG_RUNTIME |
1053 | | return(1); |
1054 | | #else |
1055 | 0 | return(0); |
1056 | 0 | #endif |
1057 | 0 | case XML_WITH_ZLIB: |
1058 | | #ifdef LIBXML_ZLIB_ENABLED |
1059 | | return(1); |
1060 | | #else |
1061 | 0 | return(0); |
1062 | 0 | #endif |
1063 | 0 | case XML_WITH_LZMA: |
1064 | | #ifdef LIBXML_LZMA_ENABLED |
1065 | | return(1); |
1066 | | #else |
1067 | 0 | return(0); |
1068 | 0 | #endif |
1069 | 0 | case XML_WITH_ICU: |
1070 | | #ifdef LIBXML_ICU_ENABLED |
1071 | | return(1); |
1072 | | #else |
1073 | 0 | return(0); |
1074 | 0 | #endif |
1075 | 0 | default: |
1076 | 0 | break; |
1077 | 0 | } |
1078 | 0 | return(0); |
1079 | 0 | } |
1080 | | |
1081 | | /************************************************************************ |
1082 | | * * |
1083 | | * SAX2 defaulted attributes handling * |
1084 | | * * |
1085 | | ************************************************************************/ |
1086 | | |
1087 | | /** |
1088 | | * xmlDetectSAX2: |
1089 | | * @ctxt: an XML parser context |
1090 | | * |
1091 | | * Do the SAX2 detection and specific initialization |
1092 | | */ |
1093 | | static void |
1094 | 22.1k | xmlDetectSAX2(xmlParserCtxtPtr ctxt) { |
1095 | 22.1k | xmlSAXHandlerPtr sax; |
1096 | | |
1097 | | /* Avoid unused variable warning if features are disabled. */ |
1098 | 22.1k | (void) sax; |
1099 | | |
1100 | 22.1k | if (ctxt == NULL) return; |
1101 | 22.1k | sax = ctxt->sax; |
1102 | 22.1k | #ifdef LIBXML_SAX1_ENABLED |
1103 | 22.1k | if ((sax) && (sax->initialized == XML_SAX2_MAGIC) && |
1104 | 22.1k | ((sax->startElementNs != NULL) || |
1105 | 22.1k | (sax->endElementNs != NULL) || |
1106 | 22.1k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
1107 | 22.1k | ctxt->sax2 = 1; |
1108 | | #else |
1109 | | ctxt->sax2 = 1; |
1110 | | #endif /* LIBXML_SAX1_ENABLED */ |
1111 | | |
1112 | 22.1k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
1113 | 22.1k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
1114 | 22.1k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
1115 | 22.1k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
1116 | 22.1k | (ctxt->str_xml_ns == NULL)) { |
1117 | 0 | xmlErrMemory(ctxt, NULL); |
1118 | 0 | } |
1119 | 22.1k | } |
1120 | | |
1121 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
1122 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
1123 | | struct _xmlDefAttrs { |
1124 | | int nbAttrs; /* number of defaulted attributes on that element */ |
1125 | | int maxAttrs; /* the size of the array */ |
1126 | | #if __STDC_VERSION__ >= 199901L |
1127 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
1128 | | const xmlChar *values[]; /* array of localname/prefix/values/external */ |
1129 | | #else |
1130 | | const xmlChar *values[5]; |
1131 | | #endif |
1132 | | }; |
1133 | | |
1134 | | /** |
1135 | | * xmlAttrNormalizeSpace: |
1136 | | * @src: the source string |
1137 | | * @dst: the target string |
1138 | | * |
1139 | | * Normalize the space in non CDATA attribute values: |
1140 | | * If the attribute type is not CDATA, then the XML processor MUST further |
1141 | | * process the normalized attribute value by discarding any leading and |
1142 | | * trailing space (#x20) characters, and by replacing sequences of space |
1143 | | * (#x20) characters by a single space (#x20) character. |
1144 | | * Note that the size of dst need to be at least src, and if one doesn't need |
1145 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
1146 | | * passing src as dst is just fine. |
1147 | | * |
1148 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1149 | | * is needed. |
1150 | | */ |
1151 | | static xmlChar * |
1152 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
1153 | 21.1k | { |
1154 | 21.1k | if ((src == NULL) || (dst == NULL)) |
1155 | 0 | return(NULL); |
1156 | | |
1157 | 21.3k | while (*src == 0x20) src++; |
1158 | 196M | while (*src != 0) { |
1159 | 196M | if (*src == 0x20) { |
1160 | 81.5k | while (*src == 0x20) src++; |
1161 | 19.4k | if (*src != 0) |
1162 | 18.5k | *dst++ = 0x20; |
1163 | 196M | } else { |
1164 | 196M | *dst++ = *src++; |
1165 | 196M | } |
1166 | 196M | } |
1167 | 21.1k | *dst = 0; |
1168 | 21.1k | if (dst == src) |
1169 | 20.0k | return(NULL); |
1170 | 1.00k | return(dst); |
1171 | 21.1k | } |
1172 | | |
1173 | | /** |
1174 | | * xmlAttrNormalizeSpace2: |
1175 | | * @src: the source string |
1176 | | * |
1177 | | * Normalize the space in non CDATA attribute values, a slightly more complex |
1178 | | * front end to avoid allocation problems when running on attribute values |
1179 | | * coming from the input. |
1180 | | * |
1181 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1182 | | * is needed. |
1183 | | */ |
1184 | | static const xmlChar * |
1185 | | xmlAttrNormalizeSpace2(xmlParserCtxtPtr ctxt, xmlChar *src, int *len) |
1186 | 2.44k | { |
1187 | 2.44k | int i; |
1188 | 2.44k | int remove_head = 0; |
1189 | 2.44k | int need_realloc = 0; |
1190 | 2.44k | const xmlChar *cur; |
1191 | | |
1192 | 2.44k | if ((ctxt == NULL) || (src == NULL) || (len == NULL)) |
1193 | 0 | return(NULL); |
1194 | 2.44k | i = *len; |
1195 | 2.44k | if (i <= 0) |
1196 | 512 | return(NULL); |
1197 | | |
1198 | 1.93k | cur = src; |
1199 | 2.63k | while (*cur == 0x20) { |
1200 | 697 | cur++; |
1201 | 697 | remove_head++; |
1202 | 697 | } |
1203 | 114k | while (*cur != 0) { |
1204 | 113k | if (*cur == 0x20) { |
1205 | 1.75k | cur++; |
1206 | 1.75k | if ((*cur == 0x20) || (*cur == 0)) { |
1207 | 647 | need_realloc = 1; |
1208 | 647 | break; |
1209 | 647 | } |
1210 | 1.75k | } else |
1211 | 111k | cur++; |
1212 | 113k | } |
1213 | 1.93k | if (need_realloc) { |
1214 | 647 | xmlChar *ret; |
1215 | | |
1216 | 647 | ret = xmlStrndup(src + remove_head, i - remove_head + 1); |
1217 | 647 | if (ret == NULL) { |
1218 | 0 | xmlErrMemory(ctxt, NULL); |
1219 | 0 | return(NULL); |
1220 | 0 | } |
1221 | 647 | xmlAttrNormalizeSpace(ret, ret); |
1222 | 647 | *len = (int) strlen((const char *)ret); |
1223 | 647 | return(ret); |
1224 | 1.29k | } else if (remove_head) { |
1225 | 321 | *len -= remove_head; |
1226 | 321 | memmove(src, src + remove_head, 1 + *len); |
1227 | 321 | return(src); |
1228 | 321 | } |
1229 | 969 | return(NULL); |
1230 | 1.93k | } |
1231 | | |
1232 | | /** |
1233 | | * xmlAddDefAttrs: |
1234 | | * @ctxt: an XML parser context |
1235 | | * @fullname: the element fullname |
1236 | | * @fullattr: the attribute fullname |
1237 | | * @value: the attribute value |
1238 | | * |
1239 | | * Add a defaulted attribute for an element |
1240 | | */ |
1241 | | static void |
1242 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1243 | | const xmlChar *fullname, |
1244 | | const xmlChar *fullattr, |
1245 | 20.5k | const xmlChar *value) { |
1246 | 20.5k | xmlDefAttrsPtr defaults; |
1247 | 20.5k | int len; |
1248 | 20.5k | const xmlChar *name; |
1249 | 20.5k | const xmlChar *prefix; |
1250 | | |
1251 | | /* |
1252 | | * Allows to detect attribute redefinitions |
1253 | | */ |
1254 | 20.5k | if (ctxt->attsSpecial != NULL) { |
1255 | 19.3k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1256 | 3.53k | return; |
1257 | 19.3k | } |
1258 | | |
1259 | 17.0k | if (ctxt->attsDefault == NULL) { |
1260 | 1.18k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1261 | 1.18k | if (ctxt->attsDefault == NULL) |
1262 | 0 | goto mem_error; |
1263 | 1.18k | } |
1264 | | |
1265 | | /* |
1266 | | * split the element name into prefix:localname , the string found |
1267 | | * are within the DTD and then not associated to namespace names. |
1268 | | */ |
1269 | 17.0k | name = xmlSplitQName3(fullname, &len); |
1270 | 17.0k | if (name == NULL) { |
1271 | 7.80k | name = xmlDictLookup(ctxt->dict, fullname, -1); |
1272 | 7.80k | prefix = NULL; |
1273 | 9.23k | } else { |
1274 | 9.23k | name = xmlDictLookup(ctxt->dict, name, -1); |
1275 | 9.23k | prefix = xmlDictLookup(ctxt->dict, fullname, len); |
1276 | 9.23k | } |
1277 | | |
1278 | | /* |
1279 | | * make sure there is some storage |
1280 | | */ |
1281 | 17.0k | defaults = xmlHashLookup2(ctxt->attsDefault, name, prefix); |
1282 | 17.0k | if (defaults == NULL) { |
1283 | 2.50k | defaults = (xmlDefAttrsPtr) xmlMalloc(sizeof(xmlDefAttrs) + |
1284 | 2.50k | (4 * 5) * sizeof(const xmlChar *)); |
1285 | 2.50k | if (defaults == NULL) |
1286 | 0 | goto mem_error; |
1287 | 2.50k | defaults->nbAttrs = 0; |
1288 | 2.50k | defaults->maxAttrs = 4; |
1289 | 2.50k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1290 | 2.50k | defaults, NULL) < 0) { |
1291 | 0 | xmlFree(defaults); |
1292 | 0 | goto mem_error; |
1293 | 0 | } |
1294 | 14.5k | } else if (defaults->nbAttrs >= defaults->maxAttrs) { |
1295 | 2.03k | xmlDefAttrsPtr temp; |
1296 | | |
1297 | 2.03k | temp = (xmlDefAttrsPtr) xmlRealloc(defaults, sizeof(xmlDefAttrs) + |
1298 | 2.03k | (2 * defaults->maxAttrs * 5) * sizeof(const xmlChar *)); |
1299 | 2.03k | if (temp == NULL) |
1300 | 0 | goto mem_error; |
1301 | 2.03k | defaults = temp; |
1302 | 2.03k | defaults->maxAttrs *= 2; |
1303 | 2.03k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1304 | 2.03k | defaults, NULL) < 0) { |
1305 | 0 | xmlFree(defaults); |
1306 | 0 | goto mem_error; |
1307 | 0 | } |
1308 | 2.03k | } |
1309 | | |
1310 | | /* |
1311 | | * Split the element name into prefix:localname , the string found |
1312 | | * are within the DTD and hen not associated to namespace names. |
1313 | | */ |
1314 | 17.0k | name = xmlSplitQName3(fullattr, &len); |
1315 | 17.0k | if (name == NULL) { |
1316 | 7.30k | name = xmlDictLookup(ctxt->dict, fullattr, -1); |
1317 | 7.30k | prefix = NULL; |
1318 | 9.73k | } else { |
1319 | 9.73k | name = xmlDictLookup(ctxt->dict, name, -1); |
1320 | 9.73k | prefix = xmlDictLookup(ctxt->dict, fullattr, len); |
1321 | 9.73k | } |
1322 | | |
1323 | 17.0k | defaults->values[5 * defaults->nbAttrs] = name; |
1324 | 17.0k | defaults->values[5 * defaults->nbAttrs + 1] = prefix; |
1325 | | /* intern the string and precompute the end */ |
1326 | 17.0k | len = xmlStrlen(value); |
1327 | 17.0k | value = xmlDictLookup(ctxt->dict, value, len); |
1328 | 17.0k | defaults->values[5 * defaults->nbAttrs + 2] = value; |
1329 | 17.0k | defaults->values[5 * defaults->nbAttrs + 3] = value + len; |
1330 | 17.0k | if (ctxt->external) |
1331 | 0 | defaults->values[5 * defaults->nbAttrs + 4] = BAD_CAST "external"; |
1332 | 17.0k | else |
1333 | 17.0k | defaults->values[5 * defaults->nbAttrs + 4] = NULL; |
1334 | 17.0k | defaults->nbAttrs++; |
1335 | | |
1336 | 17.0k | return; |
1337 | | |
1338 | 0 | mem_error: |
1339 | 0 | xmlErrMemory(ctxt, NULL); |
1340 | 0 | return; |
1341 | 17.0k | } |
1342 | | |
1343 | | /** |
1344 | | * xmlAddSpecialAttr: |
1345 | | * @ctxt: an XML parser context |
1346 | | * @fullname: the element fullname |
1347 | | * @fullattr: the attribute fullname |
1348 | | * @type: the attribute type |
1349 | | * |
1350 | | * Register this attribute type |
1351 | | */ |
1352 | | static void |
1353 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1354 | | const xmlChar *fullname, |
1355 | | const xmlChar *fullattr, |
1356 | | int type) |
1357 | 21.0k | { |
1358 | 21.0k | if (ctxt->attsSpecial == NULL) { |
1359 | 1.32k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1360 | 1.32k | if (ctxt->attsSpecial == NULL) |
1361 | 0 | goto mem_error; |
1362 | 1.32k | } |
1363 | | |
1364 | 21.0k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1365 | 3.83k | return; |
1366 | | |
1367 | 17.2k | xmlHashAddEntry2(ctxt->attsSpecial, fullname, fullattr, |
1368 | 17.2k | (void *) (ptrdiff_t) type); |
1369 | 17.2k | return; |
1370 | | |
1371 | 0 | mem_error: |
1372 | 0 | xmlErrMemory(ctxt, NULL); |
1373 | 0 | return; |
1374 | 21.0k | } |
1375 | | |
1376 | | /** |
1377 | | * xmlCleanSpecialAttrCallback: |
1378 | | * |
1379 | | * Removes CDATA attributes from the special attribute table |
1380 | | */ |
1381 | | static void |
1382 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1383 | | const xmlChar *fullname, const xmlChar *fullattr, |
1384 | 17.2k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1385 | 17.2k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1386 | | |
1387 | 17.2k | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1388 | 874 | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1389 | 874 | } |
1390 | 17.2k | } |
1391 | | |
1392 | | /** |
1393 | | * xmlCleanSpecialAttr: |
1394 | | * @ctxt: an XML parser context |
1395 | | * |
1396 | | * Trim the list of attributes defined to remove all those of type |
1397 | | * CDATA as they are not special. This call should be done when finishing |
1398 | | * to parse the DTD and before starting to parse the document root. |
1399 | | */ |
1400 | | static void |
1401 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1402 | 4.16k | { |
1403 | 4.16k | if (ctxt->attsSpecial == NULL) |
1404 | 2.85k | return; |
1405 | | |
1406 | 1.31k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1407 | | |
1408 | 1.31k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1409 | 29 | xmlHashFree(ctxt->attsSpecial, NULL); |
1410 | 29 | ctxt->attsSpecial = NULL; |
1411 | 29 | } |
1412 | 1.31k | return; |
1413 | 4.16k | } |
1414 | | |
1415 | | /** |
1416 | | * xmlCheckLanguageID: |
1417 | | * @lang: pointer to the string value |
1418 | | * |
1419 | | * Checks that the value conforms to the LanguageID production: |
1420 | | * |
1421 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1422 | | * the XML Second edition. |
1423 | | * |
1424 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1425 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1426 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1427 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1428 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1429 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1430 | | * |
1431 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1432 | | * |
1433 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1434 | | * langtag = language |
1435 | | * ["-" script] |
1436 | | * ["-" region] |
1437 | | * *("-" variant) |
1438 | | * *("-" extension) |
1439 | | * ["-" privateuse] |
1440 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1441 | | * ["-" extlang] ; sometimes followed by |
1442 | | * ; extended language subtags |
1443 | | * / 4ALPHA ; or reserved for future use |
1444 | | * / 5*8ALPHA ; or registered language subtag |
1445 | | * |
1446 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1447 | | * *2("-" 3ALPHA) ; permanently reserved |
1448 | | * |
1449 | | * script = 4ALPHA ; ISO 15924 code |
1450 | | * |
1451 | | * region = 2ALPHA ; ISO 3166-1 code |
1452 | | * / 3DIGIT ; UN M.49 code |
1453 | | * |
1454 | | * variant = 5*8alphanum ; registered variants |
1455 | | * / (DIGIT 3alphanum) |
1456 | | * |
1457 | | * extension = singleton 1*("-" (2*8alphanum)) |
1458 | | * |
1459 | | * ; Single alphanumerics |
1460 | | * ; "x" reserved for private use |
1461 | | * singleton = DIGIT ; 0 - 9 |
1462 | | * / %x41-57 ; A - W |
1463 | | * / %x59-5A ; Y - Z |
1464 | | * / %x61-77 ; a - w |
1465 | | * / %x79-7A ; y - z |
1466 | | * |
1467 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1468 | | * The parser below doesn't try to cope with extension or privateuse |
1469 | | * that could be added but that's not interoperable anyway |
1470 | | * |
1471 | | * Returns 1 if correct 0 otherwise |
1472 | | **/ |
1473 | | int |
1474 | | xmlCheckLanguageID(const xmlChar * lang) |
1475 | 0 | { |
1476 | 0 | const xmlChar *cur = lang, *nxt; |
1477 | |
|
1478 | 0 | if (cur == NULL) |
1479 | 0 | return (0); |
1480 | 0 | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1481 | 0 | ((cur[0] == 'I') && (cur[1] == '-')) || |
1482 | 0 | ((cur[0] == 'x') && (cur[1] == '-')) || |
1483 | 0 | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1484 | | /* |
1485 | | * Still allow IANA code and user code which were coming |
1486 | | * from the previous version of the XML-1.0 specification |
1487 | | * it's deprecated but we should not fail |
1488 | | */ |
1489 | 0 | cur += 2; |
1490 | 0 | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1491 | 0 | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1492 | 0 | cur++; |
1493 | 0 | return(cur[0] == 0); |
1494 | 0 | } |
1495 | 0 | nxt = cur; |
1496 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1497 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1498 | 0 | nxt++; |
1499 | 0 | if (nxt - cur >= 4) { |
1500 | | /* |
1501 | | * Reserved |
1502 | | */ |
1503 | 0 | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1504 | 0 | return(0); |
1505 | 0 | return(1); |
1506 | 0 | } |
1507 | 0 | if (nxt - cur < 2) |
1508 | 0 | return(0); |
1509 | | /* we got an ISO 639 code */ |
1510 | 0 | if (nxt[0] == 0) |
1511 | 0 | return(1); |
1512 | 0 | if (nxt[0] != '-') |
1513 | 0 | return(0); |
1514 | | |
1515 | 0 | nxt++; |
1516 | 0 | cur = nxt; |
1517 | | /* now we can have extlang or script or region or variant */ |
1518 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1519 | 0 | goto region_m49; |
1520 | | |
1521 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1522 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1523 | 0 | nxt++; |
1524 | 0 | if (nxt - cur == 4) |
1525 | 0 | goto script; |
1526 | 0 | if (nxt - cur == 2) |
1527 | 0 | goto region; |
1528 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1529 | 0 | goto variant; |
1530 | 0 | if (nxt - cur != 3) |
1531 | 0 | return(0); |
1532 | | /* we parsed an extlang */ |
1533 | 0 | if (nxt[0] == 0) |
1534 | 0 | return(1); |
1535 | 0 | if (nxt[0] != '-') |
1536 | 0 | return(0); |
1537 | | |
1538 | 0 | nxt++; |
1539 | 0 | cur = nxt; |
1540 | | /* now we can have script or region or variant */ |
1541 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1542 | 0 | goto region_m49; |
1543 | | |
1544 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1545 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1546 | 0 | nxt++; |
1547 | 0 | if (nxt - cur == 2) |
1548 | 0 | goto region; |
1549 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1550 | 0 | goto variant; |
1551 | 0 | if (nxt - cur != 4) |
1552 | 0 | return(0); |
1553 | | /* we parsed a script */ |
1554 | 0 | script: |
1555 | 0 | if (nxt[0] == 0) |
1556 | 0 | return(1); |
1557 | 0 | if (nxt[0] != '-') |
1558 | 0 | return(0); |
1559 | | |
1560 | 0 | nxt++; |
1561 | 0 | cur = nxt; |
1562 | | /* now we can have region or variant */ |
1563 | 0 | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1564 | 0 | goto region_m49; |
1565 | | |
1566 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1567 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1568 | 0 | nxt++; |
1569 | |
|
1570 | 0 | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1571 | 0 | goto variant; |
1572 | 0 | if (nxt - cur != 2) |
1573 | 0 | return(0); |
1574 | | /* we parsed a region */ |
1575 | 0 | region: |
1576 | 0 | if (nxt[0] == 0) |
1577 | 0 | return(1); |
1578 | 0 | if (nxt[0] != '-') |
1579 | 0 | return(0); |
1580 | | |
1581 | 0 | nxt++; |
1582 | 0 | cur = nxt; |
1583 | | /* now we can just have a variant */ |
1584 | 0 | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1585 | 0 | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1586 | 0 | nxt++; |
1587 | |
|
1588 | 0 | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1589 | 0 | return(0); |
1590 | | |
1591 | | /* we parsed a variant */ |
1592 | 0 | variant: |
1593 | 0 | if (nxt[0] == 0) |
1594 | 0 | return(1); |
1595 | 0 | if (nxt[0] != '-') |
1596 | 0 | return(0); |
1597 | | /* extensions and private use subtags not checked */ |
1598 | 0 | return (1); |
1599 | | |
1600 | 0 | region_m49: |
1601 | 0 | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1602 | 0 | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1603 | 0 | nxt += 3; |
1604 | 0 | goto region; |
1605 | 0 | } |
1606 | 0 | return(0); |
1607 | 0 | } |
1608 | | |
1609 | | /************************************************************************ |
1610 | | * * |
1611 | | * Parser stacks related functions and macros * |
1612 | | * * |
1613 | | ************************************************************************/ |
1614 | | |
1615 | | static xmlEntityPtr xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, |
1616 | | const xmlChar ** str); |
1617 | | |
1618 | | #ifdef SAX2 |
1619 | | /** |
1620 | | * nsPush: |
1621 | | * @ctxt: an XML parser context |
1622 | | * @prefix: the namespace prefix or NULL |
1623 | | * @URL: the namespace name |
1624 | | * |
1625 | | * Pushes a new parser namespace on top of the ns stack |
1626 | | * |
1627 | | * Returns -1 in case of error, -2 if the namespace should be discarded |
1628 | | * and the index in the stack otherwise. |
1629 | | */ |
1630 | | static int |
1631 | | nsPush(xmlParserCtxtPtr ctxt, const xmlChar *prefix, const xmlChar *URL) |
1632 | 18.4k | { |
1633 | 18.4k | if (ctxt->options & XML_PARSE_NSCLEAN) { |
1634 | 18.4k | int i; |
1635 | 25.2k | for (i = ctxt->nsNr - 2;i >= 0;i -= 2) { |
1636 | 19.1k | if (ctxt->nsTab[i] == prefix) { |
1637 | | /* in scope */ |
1638 | 12.3k | if (ctxt->nsTab[i + 1] == URL) |
1639 | 5.14k | return(-2); |
1640 | | /* out of scope keep it */ |
1641 | 7.20k | break; |
1642 | 12.3k | } |
1643 | 19.1k | } |
1644 | 18.4k | } |
1645 | 13.3k | if ((ctxt->nsMax == 0) || (ctxt->nsTab == NULL)) { |
1646 | 3.05k | ctxt->nsMax = 10; |
1647 | 3.05k | ctxt->nsNr = 0; |
1648 | 3.05k | ctxt->nsTab = (const xmlChar **) |
1649 | 3.05k | xmlMalloc(ctxt->nsMax * sizeof(xmlChar *)); |
1650 | 3.05k | if (ctxt->nsTab == NULL) { |
1651 | 0 | xmlErrMemory(ctxt, NULL); |
1652 | 0 | ctxt->nsMax = 0; |
1653 | 0 | return (-1); |
1654 | 0 | } |
1655 | 10.2k | } else if (ctxt->nsNr >= ctxt->nsMax) { |
1656 | 277 | const xmlChar ** tmp; |
1657 | 277 | ctxt->nsMax *= 2; |
1658 | 277 | tmp = (const xmlChar **) xmlRealloc((char *) ctxt->nsTab, |
1659 | 277 | ctxt->nsMax * sizeof(ctxt->nsTab[0])); |
1660 | 277 | if (tmp == NULL) { |
1661 | 0 | xmlErrMemory(ctxt, NULL); |
1662 | 0 | ctxt->nsMax /= 2; |
1663 | 0 | return (-1); |
1664 | 0 | } |
1665 | 277 | ctxt->nsTab = tmp; |
1666 | 277 | } |
1667 | 13.3k | ctxt->nsTab[ctxt->nsNr++] = prefix; |
1668 | 13.3k | ctxt->nsTab[ctxt->nsNr++] = URL; |
1669 | 13.3k | return (ctxt->nsNr); |
1670 | 13.3k | } |
1671 | | /** |
1672 | | * nsPop: |
1673 | | * @ctxt: an XML parser context |
1674 | | * @nr: the number to pop |
1675 | | * |
1676 | | * Pops the top @nr parser prefix/namespace from the ns stack |
1677 | | * |
1678 | | * Returns the number of namespaces removed |
1679 | | */ |
1680 | | static int |
1681 | | nsPop(xmlParserCtxtPtr ctxt, int nr) |
1682 | 1.42k | { |
1683 | 1.42k | int i; |
1684 | | |
1685 | 1.42k | if (ctxt->nsTab == NULL) return(0); |
1686 | 1.42k | if (ctxt->nsNr < nr) { |
1687 | 0 | xmlGenericError(xmlGenericErrorContext, "Pbm popping %d NS\n", nr); |
1688 | 0 | nr = ctxt->nsNr; |
1689 | 0 | } |
1690 | 1.42k | if (ctxt->nsNr <= 0) |
1691 | 0 | return (0); |
1692 | | |
1693 | 6.64k | for (i = 0;i < nr;i++) { |
1694 | 5.22k | ctxt->nsNr--; |
1695 | 5.22k | ctxt->nsTab[ctxt->nsNr] = NULL; |
1696 | 5.22k | } |
1697 | 1.42k | return(nr); |
1698 | 1.42k | } |
1699 | | #endif |
1700 | | |
1701 | | static int |
1702 | 4.27k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1703 | 4.27k | const xmlChar **atts; |
1704 | 4.27k | int *attallocs; |
1705 | 4.27k | int maxatts; |
1706 | | |
1707 | 4.27k | if (ctxt->atts == NULL) { |
1708 | 3.74k | maxatts = 55; /* allow for 10 attrs by default */ |
1709 | 3.74k | atts = (const xmlChar **) |
1710 | 3.74k | xmlMalloc(maxatts * sizeof(xmlChar *)); |
1711 | 3.74k | if (atts == NULL) goto mem_error; |
1712 | 3.74k | ctxt->atts = atts; |
1713 | 3.74k | attallocs = (int *) xmlMalloc((maxatts / 5) * sizeof(int)); |
1714 | 3.74k | if (attallocs == NULL) goto mem_error; |
1715 | 3.74k | ctxt->attallocs = attallocs; |
1716 | 3.74k | ctxt->maxatts = maxatts; |
1717 | 3.74k | } else if (nr + 5 > ctxt->maxatts) { |
1718 | 536 | maxatts = (nr + 5) * 2; |
1719 | 536 | atts = (const xmlChar **) xmlRealloc((void *) ctxt->atts, |
1720 | 536 | maxatts * sizeof(const xmlChar *)); |
1721 | 536 | if (atts == NULL) goto mem_error; |
1722 | 536 | ctxt->atts = atts; |
1723 | 536 | attallocs = (int *) xmlRealloc((void *) ctxt->attallocs, |
1724 | 536 | (maxatts / 5) * sizeof(int)); |
1725 | 536 | if (attallocs == NULL) goto mem_error; |
1726 | 536 | ctxt->attallocs = attallocs; |
1727 | 536 | ctxt->maxatts = maxatts; |
1728 | 536 | } |
1729 | 4.27k | return(ctxt->maxatts); |
1730 | 0 | mem_error: |
1731 | 0 | xmlErrMemory(ctxt, NULL); |
1732 | 0 | return(-1); |
1733 | 4.27k | } |
1734 | | |
1735 | | /** |
1736 | | * inputPush: |
1737 | | * @ctxt: an XML parser context |
1738 | | * @value: the parser input |
1739 | | * |
1740 | | * Pushes a new parser input on top of the input stack |
1741 | | * |
1742 | | * Returns -1 in case of error, the index in the stack otherwise |
1743 | | */ |
1744 | | int |
1745 | | inputPush(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1746 | 18.3k | { |
1747 | 18.3k | if ((ctxt == NULL) || (value == NULL)) |
1748 | 0 | return(-1); |
1749 | 18.3k | if (ctxt->inputNr >= ctxt->inputMax) { |
1750 | 0 | ctxt->inputMax *= 2; |
1751 | 0 | ctxt->inputTab = |
1752 | 0 | (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1753 | 0 | ctxt->inputMax * |
1754 | 0 | sizeof(ctxt->inputTab[0])); |
1755 | 0 | if (ctxt->inputTab == NULL) { |
1756 | 0 | xmlErrMemory(ctxt, NULL); |
1757 | 0 | ctxt->inputMax /= 2; |
1758 | 0 | return (-1); |
1759 | 0 | } |
1760 | 0 | } |
1761 | 18.3k | ctxt->inputTab[ctxt->inputNr] = value; |
1762 | 18.3k | ctxt->input = value; |
1763 | 18.3k | return (ctxt->inputNr++); |
1764 | 18.3k | } |
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 | 54.9k | { |
1776 | 54.9k | xmlParserInputPtr ret; |
1777 | | |
1778 | 54.9k | if (ctxt == NULL) |
1779 | 0 | return(NULL); |
1780 | 54.9k | if (ctxt->inputNr <= 0) |
1781 | 36.6k | return (NULL); |
1782 | 18.3k | ctxt->inputNr--; |
1783 | 18.3k | if (ctxt->inputNr > 0) |
1784 | 0 | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1785 | 18.3k | else |
1786 | 18.3k | ctxt->input = NULL; |
1787 | 18.3k | ret = ctxt->inputTab[ctxt->inputNr]; |
1788 | 18.3k | ctxt->inputTab[ctxt->inputNr] = NULL; |
1789 | 18.3k | return (ret); |
1790 | 54.9k | } |
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 | 0 | { |
1803 | 0 | if (ctxt == NULL) return(0); |
1804 | 0 | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1805 | 0 | xmlNodePtr *tmp; |
1806 | |
|
1807 | 0 | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
1808 | 0 | ctxt->nodeMax * 2 * |
1809 | 0 | sizeof(ctxt->nodeTab[0])); |
1810 | 0 | if (tmp == NULL) { |
1811 | 0 | xmlErrMemory(ctxt, NULL); |
1812 | 0 | return (-1); |
1813 | 0 | } |
1814 | 0 | ctxt->nodeTab = tmp; |
1815 | 0 | ctxt->nodeMax *= 2; |
1816 | 0 | } |
1817 | 0 | if ((((unsigned int) ctxt->nodeNr) > xmlParserMaxDepth) && |
1818 | 0 | ((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 | 0 | ctxt->nodeTab[ctxt->nodeNr] = value; |
1826 | 0 | ctxt->node = value; |
1827 | 0 | return (ctxt->nodeNr++); |
1828 | 0 | } |
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 | 6.34k | { |
1841 | 6.34k | xmlNodePtr ret; |
1842 | | |
1843 | 6.34k | if (ctxt == NULL) return(NULL); |
1844 | 6.34k | if (ctxt->nodeNr <= 0) |
1845 | 6.34k | return (NULL); |
1846 | 0 | ctxt->nodeNr--; |
1847 | 0 | if (ctxt->nodeNr > 0) |
1848 | 0 | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
1849 | 0 | else |
1850 | 0 | ctxt->node = NULL; |
1851 | 0 | ret = ctxt->nodeTab[ctxt->nodeNr]; |
1852 | 0 | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
1853 | 0 | return (ret); |
1854 | 6.34k | } |
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 | 1.47M | { |
1873 | 1.47M | xmlStartTag *tag; |
1874 | | |
1875 | 1.47M | if (ctxt->nameNr >= ctxt->nameMax) { |
1876 | 1.46k | const xmlChar * *tmp; |
1877 | 1.46k | xmlStartTag *tmp2; |
1878 | 1.46k | ctxt->nameMax *= 2; |
1879 | 1.46k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1880 | 1.46k | ctxt->nameMax * |
1881 | 1.46k | sizeof(ctxt->nameTab[0])); |
1882 | 1.46k | if (tmp == NULL) { |
1883 | 0 | ctxt->nameMax /= 2; |
1884 | 0 | goto mem_error; |
1885 | 0 | } |
1886 | 1.46k | ctxt->nameTab = tmp; |
1887 | 1.46k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
1888 | 1.46k | ctxt->nameMax * |
1889 | 1.46k | sizeof(ctxt->pushTab[0])); |
1890 | 1.46k | if (tmp2 == NULL) { |
1891 | 0 | ctxt->nameMax /= 2; |
1892 | 0 | goto mem_error; |
1893 | 0 | } |
1894 | 1.46k | ctxt->pushTab = tmp2; |
1895 | 1.47M | } else if (ctxt->pushTab == NULL) { |
1896 | 9.99k | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
1897 | 9.99k | sizeof(ctxt->pushTab[0])); |
1898 | 9.99k | if (ctxt->pushTab == NULL) |
1899 | 0 | goto mem_error; |
1900 | 9.99k | } |
1901 | 1.47M | ctxt->nameTab[ctxt->nameNr] = value; |
1902 | 1.47M | ctxt->name = value; |
1903 | 1.47M | tag = &ctxt->pushTab[ctxt->nameNr]; |
1904 | 1.47M | tag->prefix = prefix; |
1905 | 1.47M | tag->URI = URI; |
1906 | 1.47M | tag->line = line; |
1907 | 1.47M | tag->nsNr = nsNr; |
1908 | 1.47M | return (ctxt->nameNr++); |
1909 | 0 | mem_error: |
1910 | 0 | xmlErrMemory(ctxt, NULL); |
1911 | 0 | return (-1); |
1912 | 1.47M | } |
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 | 30.7k | { |
1925 | 30.7k | const xmlChar *ret; |
1926 | | |
1927 | 30.7k | if (ctxt->nameNr <= 0) |
1928 | 0 | return (NULL); |
1929 | 30.7k | ctxt->nameNr--; |
1930 | 30.7k | if (ctxt->nameNr > 0) |
1931 | 29.8k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1932 | 870 | else |
1933 | 870 | ctxt->name = NULL; |
1934 | 30.7k | ret = ctxt->nameTab[ctxt->nameNr]; |
1935 | 30.7k | ctxt->nameTab[ctxt->nameNr] = NULL; |
1936 | 30.7k | return (ret); |
1937 | 30.7k | } |
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 | 0 | { |
1983 | 0 | const xmlChar *ret; |
1984 | |
|
1985 | 0 | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
1986 | 0 | return (NULL); |
1987 | 0 | ctxt->nameNr--; |
1988 | 0 | if (ctxt->nameNr > 0) |
1989 | 0 | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1990 | 0 | else |
1991 | 0 | ctxt->name = NULL; |
1992 | 0 | ret = ctxt->nameTab[ctxt->nameNr]; |
1993 | 0 | ctxt->nameTab[ctxt->nameNr] = NULL; |
1994 | 0 | return (ret); |
1995 | 0 | } |
1996 | | |
1997 | 1.73M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
1998 | 1.73M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
1999 | 1.49k | int *tmp; |
2000 | | |
2001 | 1.49k | ctxt->spaceMax *= 2; |
2002 | 1.49k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
2003 | 1.49k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
2004 | 1.49k | if (tmp == NULL) { |
2005 | 0 | xmlErrMemory(ctxt, NULL); |
2006 | 0 | ctxt->spaceMax /=2; |
2007 | 0 | return(-1); |
2008 | 0 | } |
2009 | 1.49k | ctxt->spaceTab = tmp; |
2010 | 1.49k | } |
2011 | 1.73M | ctxt->spaceTab[ctxt->spaceNr] = val; |
2012 | 1.73M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2013 | 1.73M | return(ctxt->spaceNr++); |
2014 | 1.73M | } |
2015 | | |
2016 | 296k | static int spacePop(xmlParserCtxtPtr ctxt) { |
2017 | 296k | int ret; |
2018 | 296k | if (ctxt->spaceNr <= 0) return(0); |
2019 | 296k | ctxt->spaceNr--; |
2020 | 296k | if (ctxt->spaceNr > 0) |
2021 | 296k | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2022 | 0 | else |
2023 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
2024 | 296k | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2025 | 296k | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2026 | 296k | return(ret); |
2027 | 296k | } |
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 | 16.8M | #define RAW (*ctxt->input->cur) |
2065 | 8.23M | #define CUR (*ctxt->input->cur) |
2066 | 120M | #define NXT(val) ctxt->input->cur[(val)] |
2067 | 1.41M | #define CUR_PTR ctxt->input->cur |
2068 | 68.4k | #define BASE_PTR ctxt->input->base |
2069 | | |
2070 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2071 | 475k | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2072 | 237k | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2073 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2074 | 416k | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2075 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2076 | 284k | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2077 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2078 | 198k | ( 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 | 132k | ( 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 | 43.1k | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2083 | 43.1k | ((unsigned char *) s)[ 8 ] == c9 ) |
2084 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2085 | 2.84k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2086 | 2.84k | ((unsigned char *) s)[ 9 ] == c10 ) |
2087 | | |
2088 | 935k | #define SKIP(val) do { \ |
2089 | 935k | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2090 | 935k | if (*ctxt->input->cur == 0) \ |
2091 | 935k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2092 | 935k | } while (0) |
2093 | | |
2094 | 126k | #define SKIPL(val) do { \ |
2095 | 126k | int skipl; \ |
2096 | 25.6M | for(skipl=0; skipl<val; skipl++) { \ |
2097 | 25.5M | if (*(ctxt->input->cur) == '\n') { \ |
2098 | 15.0k | ctxt->input->line++; ctxt->input->col = 1; \ |
2099 | 25.5M | } else ctxt->input->col++; \ |
2100 | 25.5M | ctxt->input->cur++; \ |
2101 | 25.5M | } \ |
2102 | 126k | if (*ctxt->input->cur == 0) \ |
2103 | 126k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2104 | 126k | } while (0) |
2105 | | |
2106 | 4.66M | #define SHRINK if ((ctxt->progressive == 0) && \ |
2107 | 4.66M | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2108 | 4.66M | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2109 | 4.66M | xmlSHRINK (ctxt); |
2110 | | |
2111 | 7.61k | static void xmlSHRINK (xmlParserCtxtPtr ctxt) { |
2112 | 7.61k | xmlParserInputShrink(ctxt->input); |
2113 | 7.61k | if (*ctxt->input->cur == 0) |
2114 | 113 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2115 | 7.61k | } |
2116 | | |
2117 | 176M | #define GROW if ((ctxt->progressive == 0) && \ |
2118 | 176M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2119 | 176M | xmlGROW (ctxt); |
2120 | | |
2121 | 165k | static void xmlGROW (xmlParserCtxtPtr ctxt) { |
2122 | 165k | ptrdiff_t curEnd = ctxt->input->end - ctxt->input->cur; |
2123 | 165k | ptrdiff_t curBase = ctxt->input->cur - ctxt->input->base; |
2124 | | |
2125 | 165k | if (((curEnd > XML_MAX_LOOKUP_LIMIT) || |
2126 | 165k | (curBase > XML_MAX_LOOKUP_LIMIT)) && |
2127 | 165k | ((ctxt->input->buf) && |
2128 | 486 | (ctxt->input->buf->readcallback != xmlInputReadCallbackNop)) && |
2129 | 165k | ((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 | 165k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2135 | 165k | if ((ctxt->input->cur > ctxt->input->end) || |
2136 | 165k | (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 | 165k | if ((ctxt->input->cur != NULL) && (*ctxt->input->cur == 0)) |
2142 | 1.70k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2143 | 165k | } |
2144 | | |
2145 | 4.38M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2146 | | |
2147 | 9.36M | #define NEXT xmlNextChar(ctxt) |
2148 | | |
2149 | 1.76M | #define NEXT1 { \ |
2150 | 1.76M | ctxt->input->col++; \ |
2151 | 1.76M | ctxt->input->cur++; \ |
2152 | 1.76M | if (*ctxt->input->cur == 0) \ |
2153 | 1.76M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2154 | 1.76M | } |
2155 | | |
2156 | 501M | #define NEXTL(l) do { \ |
2157 | 501M | if (*(ctxt->input->cur) == '\n') { \ |
2158 | 34.2k | ctxt->input->line++; ctxt->input->col = 1; \ |
2159 | 501M | } else ctxt->input->col++; \ |
2160 | 501M | ctxt->input->cur += l; \ |
2161 | 501M | } while (0) |
2162 | | |
2163 | 502M | #define CUR_CHAR(l) xmlCurrentChar(ctxt, &l) |
2164 | 39.0M | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2165 | | |
2166 | | #define COPY_BUF(l,b,i,v) \ |
2167 | 332M | if (l == 1) b[i++] = (xmlChar) v; \ |
2168 | 332M | else i += xmlCopyCharMultiByte(&b[i],v) |
2169 | | |
2170 | | #define CUR_CONSUMED \ |
2171 | 4.09M | (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 | 4.38M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2185 | 4.38M | 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 | 4.38M | if (((ctxt->inputNr == 1) && (ctxt->instate != XML_PARSER_DTD)) || |
2192 | 4.38M | (ctxt->instate == XML_PARSER_START)) { |
2193 | 4.02M | const xmlChar *cur; |
2194 | | /* |
2195 | | * if we are in the document content, go really fast |
2196 | | */ |
2197 | 4.02M | cur = ctxt->input->cur; |
2198 | 4.02M | while (IS_BLANK_CH(*cur)) { |
2199 | 1.67M | if (*cur == '\n') { |
2200 | 4.37k | ctxt->input->line++; ctxt->input->col = 1; |
2201 | 1.67M | } else { |
2202 | 1.67M | ctxt->input->col++; |
2203 | 1.67M | } |
2204 | 1.67M | cur++; |
2205 | 1.67M | if (res < INT_MAX) |
2206 | 1.67M | res++; |
2207 | 1.67M | if (*cur == 0) { |
2208 | 735 | ctxt->input->cur = cur; |
2209 | 735 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2210 | 735 | cur = ctxt->input->cur; |
2211 | 735 | } |
2212 | 1.67M | } |
2213 | 4.02M | ctxt->input->cur = cur; |
2214 | 4.02M | } else { |
2215 | 363k | int expandPE = ((ctxt->external != 0) || (ctxt->inputNr != 1)); |
2216 | | |
2217 | 1.72M | while (1) { |
2218 | 1.72M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2219 | 1.36M | NEXT; |
2220 | 1.36M | } else if (CUR == '%') { |
2221 | | /* |
2222 | | * Need to handle support of entities branching here |
2223 | | */ |
2224 | 5.76k | if ((expandPE == 0) || (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2225 | 5.76k | break; |
2226 | 0 | xmlParsePEReference(ctxt); |
2227 | 357k | } else if (CUR == 0) { |
2228 | 776 | if (ctxt->inputNr <= 1) |
2229 | 776 | break; |
2230 | 0 | xmlPopInput(ctxt); |
2231 | 356k | } else { |
2232 | 356k | break; |
2233 | 356k | } |
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 | 1.36M | if (res < INT_MAX) |
2243 | 1.36M | res++; |
2244 | 1.36M | } |
2245 | 363k | } |
2246 | 4.38M | return(res); |
2247 | 4.38M | } |
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 | 0 | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2266 | 0 | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2267 | 0 | if (xmlParserDebugEntities) |
2268 | 0 | xmlGenericError(xmlGenericErrorContext, |
2269 | 0 | "Popping input %d\n", ctxt->inputNr); |
2270 | 0 | if ((ctxt->inputNr > 1) && (ctxt->inSubset == 0) && |
2271 | 0 | (ctxt->instate != XML_PARSER_EOF)) |
2272 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
2273 | 0 | "Unfinished entity outside the DTD"); |
2274 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
2275 | 0 | if (*ctxt->input->cur == 0) |
2276 | 0 | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2277 | 0 | return(CUR); |
2278 | 0 | } |
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 | 0 | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2291 | 0 | int ret; |
2292 | 0 | if (input == NULL) return(-1); |
2293 | | |
2294 | 0 | 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 | 0 | if (((ctxt->inputNr > 40) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2303 | 0 | (ctxt->inputNr > 1024)) { |
2304 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2305 | 0 | while (ctxt->inputNr > 1) |
2306 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
2307 | 0 | return(-1); |
2308 | 0 | } |
2309 | 0 | ret = inputPush(ctxt, input); |
2310 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
2311 | 0 | return(-1); |
2312 | 0 | GROW; |
2313 | 0 | return(ret); |
2314 | 0 | } |
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 | 379k | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2333 | 379k | int val = 0; |
2334 | 379k | int count = 0; |
2335 | | |
2336 | | /* |
2337 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2338 | | */ |
2339 | 379k | if ((RAW == '&') && (NXT(1) == '#') && |
2340 | 379k | (NXT(2) == 'x')) { |
2341 | 24.8k | SKIP(3); |
2342 | 24.8k | GROW; |
2343 | 280k | while (RAW != ';') { /* loop blocked by count */ |
2344 | 264k | if (count++ > 20) { |
2345 | 11.1k | count = 0; |
2346 | 11.1k | GROW; |
2347 | 11.1k | if (ctxt->instate == XML_PARSER_EOF) |
2348 | 0 | return(0); |
2349 | 11.1k | } |
2350 | 264k | if ((RAW >= '0') && (RAW <= '9')) |
2351 | 161k | val = val * 16 + (CUR - '0'); |
2352 | 103k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2353 | 43.5k | val = val * 16 + (CUR - 'a') + 10; |
2354 | 59.8k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2355 | 51.1k | val = val * 16 + (CUR - 'A') + 10; |
2356 | 8.74k | else { |
2357 | 8.74k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2358 | 8.74k | val = 0; |
2359 | 8.74k | break; |
2360 | 8.74k | } |
2361 | 256k | if (val > 0x110000) |
2362 | 143k | val = 0x110000; |
2363 | | |
2364 | 256k | NEXT; |
2365 | 256k | count++; |
2366 | 256k | } |
2367 | 24.8k | if (RAW == ';') { |
2368 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2369 | 16.0k | ctxt->input->col++; |
2370 | 16.0k | ctxt->input->cur++; |
2371 | 16.0k | } |
2372 | 354k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2373 | 354k | SKIP(2); |
2374 | 354k | GROW; |
2375 | 1.11M | while (RAW != ';') { /* loop blocked by count */ |
2376 | 971k | if (count++ > 20) { |
2377 | 13.1k | count = 0; |
2378 | 13.1k | GROW; |
2379 | 13.1k | if (ctxt->instate == XML_PARSER_EOF) |
2380 | 0 | return(0); |
2381 | 13.1k | } |
2382 | 971k | if ((RAW >= '0') && (RAW <= '9')) |
2383 | 762k | val = val * 10 + (CUR - '0'); |
2384 | 209k | else { |
2385 | 209k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2386 | 209k | val = 0; |
2387 | 209k | break; |
2388 | 209k | } |
2389 | 762k | if (val > 0x110000) |
2390 | 5.39k | val = 0x110000; |
2391 | | |
2392 | 762k | NEXT; |
2393 | 762k | count++; |
2394 | 762k | } |
2395 | 354k | if (RAW == ';') { |
2396 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2397 | 145k | ctxt->input->col++; |
2398 | 145k | ctxt->input->cur++; |
2399 | 145k | } |
2400 | 354k | } 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 | 379k | if (val >= 0x110000) { |
2410 | 3.87k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2411 | 3.87k | "xmlParseCharRef: character reference out of bounds\n", |
2412 | 3.87k | val); |
2413 | 375k | } else if (IS_CHAR(val)) { |
2414 | 137k | return(val); |
2415 | 238k | } else { |
2416 | 238k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2417 | 238k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2418 | 238k | val); |
2419 | 238k | } |
2420 | 242k | return(0); |
2421 | 379k | } |
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 | 4.95k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2443 | 4.95k | const xmlChar *ptr; |
2444 | 4.95k | xmlChar cur; |
2445 | 4.95k | int val = 0; |
2446 | | |
2447 | 4.95k | if ((str == NULL) || (*str == NULL)) return(0); |
2448 | 4.95k | ptr = *str; |
2449 | 4.95k | cur = *ptr; |
2450 | 4.95k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2451 | 2.36k | ptr += 3; |
2452 | 2.36k | cur = *ptr; |
2453 | 8.28k | while (cur != ';') { /* Non input consuming loop */ |
2454 | 6.77k | if ((cur >= '0') && (cur <= '9')) |
2455 | 1.82k | val = val * 16 + (cur - '0'); |
2456 | 4.95k | else if ((cur >= 'a') && (cur <= 'f')) |
2457 | 2.47k | val = val * 16 + (cur - 'a') + 10; |
2458 | 2.48k | else if ((cur >= 'A') && (cur <= 'F')) |
2459 | 1.62k | val = val * 16 + (cur - 'A') + 10; |
2460 | 852 | else { |
2461 | 852 | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2462 | 852 | val = 0; |
2463 | 852 | break; |
2464 | 852 | } |
2465 | 5.92k | if (val > 0x110000) |
2466 | 803 | val = 0x110000; |
2467 | | |
2468 | 5.92k | ptr++; |
2469 | 5.92k | cur = *ptr; |
2470 | 5.92k | } |
2471 | 2.36k | if (cur == ';') |
2472 | 1.51k | ptr++; |
2473 | 2.59k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2474 | 2.59k | ptr += 2; |
2475 | 2.59k | cur = *ptr; |
2476 | 8.87k | while (cur != ';') { /* Non input consuming loops */ |
2477 | 7.06k | if ((cur >= '0') && (cur <= '9')) |
2478 | 6.27k | val = val * 10 + (cur - '0'); |
2479 | 788 | else { |
2480 | 788 | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2481 | 788 | val = 0; |
2482 | 788 | break; |
2483 | 788 | } |
2484 | 6.27k | if (val > 0x110000) |
2485 | 1.17k | val = 0x110000; |
2486 | | |
2487 | 6.27k | ptr++; |
2488 | 6.27k | cur = *ptr; |
2489 | 6.27k | } |
2490 | 2.59k | if (cur == ';') |
2491 | 1.80k | ptr++; |
2492 | 2.59k | } else { |
2493 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2494 | 0 | return(0); |
2495 | 0 | } |
2496 | 4.95k | *str = ptr; |
2497 | | |
2498 | | /* |
2499 | | * [ WFC: Legal Character ] |
2500 | | * Characters referred to using character references must match the |
2501 | | * production for Char. |
2502 | | */ |
2503 | 4.95k | if (val >= 0x110000) { |
2504 | 221 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2505 | 221 | "xmlParseStringCharRef: character reference out of bounds\n", |
2506 | 221 | val); |
2507 | 4.73k | } else if (IS_CHAR(val)) { |
2508 | 2.42k | return(val); |
2509 | 2.42k | } else { |
2510 | 2.31k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2511 | 2.31k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2512 | 2.31k | val); |
2513 | 2.31k | } |
2514 | 2.53k | return(0); |
2515 | 4.95k | } |
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 | 10.9k | #define growBuffer(buffer, n) { \ |
2614 | 10.9k | xmlChar *tmp; \ |
2615 | 10.9k | size_t new_size = buffer##_size * 2 + n; \ |
2616 | 10.9k | if (new_size < buffer##_size) goto mem_error; \ |
2617 | 10.9k | tmp = (xmlChar *) xmlRealloc(buffer, new_size); \ |
2618 | 10.9k | if (tmp == NULL) goto mem_error; \ |
2619 | 10.9k | buffer = tmp; \ |
2620 | 10.9k | buffer##_size = new_size; \ |
2621 | 10.9k | } |
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 | 7.25k | int what, xmlChar end, xmlChar end2, xmlChar end3) { |
2645 | 7.25k | xmlChar *buffer = NULL; |
2646 | 7.25k | size_t buffer_size = 0; |
2647 | 7.25k | size_t nbchars = 0; |
2648 | | |
2649 | 7.25k | xmlChar *current = NULL; |
2650 | 7.25k | xmlChar *rep = NULL; |
2651 | 7.25k | const xmlChar *last; |
2652 | 7.25k | xmlEntityPtr ent; |
2653 | 7.25k | int c,l; |
2654 | | |
2655 | 7.25k | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2656 | 0 | return(NULL); |
2657 | 7.25k | last = str + len; |
2658 | | |
2659 | 7.25k | if (((ctxt->depth > 40) && |
2660 | 7.25k | ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2661 | 7.25k | (ctxt->depth > 1024)) { |
2662 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2663 | 0 | return(NULL); |
2664 | 0 | } |
2665 | | |
2666 | | /* |
2667 | | * allocate a translation buffer. |
2668 | | */ |
2669 | 7.25k | buffer_size = XML_PARSER_BIG_BUFFER_SIZE; |
2670 | 7.25k | buffer = (xmlChar *) xmlMallocAtomic(buffer_size); |
2671 | 7.25k | 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 | 7.25k | if (str < last) |
2678 | 5.51k | c = CUR_SCHAR(str, l); |
2679 | 1.74k | else |
2680 | 1.74k | c = 0; |
2681 | 14.6M | while ((c != 0) && (c != end) && /* non input consuming loop */ |
2682 | 14.6M | (c != end2) && (c != end3) && |
2683 | 14.6M | (ctxt->instate != XML_PARSER_EOF)) { |
2684 | | |
2685 | 14.6M | if (c == 0) break; |
2686 | 14.6M | if ((c == '&') && (str[1] == '#')) { |
2687 | 4.95k | int val = xmlParseStringCharRef(ctxt, &str); |
2688 | 4.95k | if (val == 0) |
2689 | 2.53k | goto int_error; |
2690 | 2.42k | COPY_BUF(0,buffer,nbchars,val); |
2691 | 2.42k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2692 | 414 | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2693 | 414 | } |
2694 | 14.6M | } else if ((c == '&') && (what & XML_SUBSTITUTE_REF)) { |
2695 | 0 | if (xmlParserDebugEntities) |
2696 | 0 | xmlGenericError(xmlGenericErrorContext, |
2697 | 0 | "String decoding Entity Reference: %.30s\n", |
2698 | 0 | str); |
2699 | 0 | ent = xmlParseStringEntityRef(ctxt, &str); |
2700 | 0 | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2701 | 0 | if (ent != NULL) |
2702 | 0 | ctxt->nbentities += ent->checked / 2; |
2703 | 0 | if ((ent != NULL) && |
2704 | 0 | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
2705 | 0 | if (ent->content != NULL) { |
2706 | 0 | COPY_BUF(0,buffer,nbchars,ent->content[0]); |
2707 | 0 | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2708 | 0 | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2709 | 0 | } |
2710 | 0 | } 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 | 0 | } else if ((ent != NULL) && (ent->content != NULL)) { |
2716 | 0 | ctxt->depth++; |
2717 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2718 | 0 | 0, 0, 0); |
2719 | 0 | ctxt->depth--; |
2720 | 0 | if (rep == NULL) { |
2721 | 0 | ent->content[0] = 0; |
2722 | 0 | goto int_error; |
2723 | 0 | } |
2724 | | |
2725 | 0 | current = rep; |
2726 | 0 | while (*current != 0) { /* non input consuming loop */ |
2727 | 0 | buffer[nbchars++] = *current++; |
2728 | 0 | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2729 | 0 | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2730 | 0 | goto int_error; |
2731 | 0 | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2732 | 0 | } |
2733 | 0 | } |
2734 | 0 | xmlFree(rep); |
2735 | 0 | rep = NULL; |
2736 | 0 | } else if (ent != NULL) { |
2737 | 0 | int i = xmlStrlen(ent->name); |
2738 | 0 | const xmlChar *cur = ent->name; |
2739 | |
|
2740 | 0 | buffer[nbchars++] = '&'; |
2741 | 0 | if (nbchars + i + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2742 | 0 | growBuffer(buffer, i + XML_PARSER_BUFFER_SIZE); |
2743 | 0 | } |
2744 | 0 | for (;i > 0;i--) |
2745 | 0 | buffer[nbchars++] = *cur++; |
2746 | 0 | buffer[nbchars++] = ';'; |
2747 | 0 | } |
2748 | 14.6M | } else if (c == '%' && (what & XML_SUBSTITUTE_PEREF)) { |
2749 | 0 | if (xmlParserDebugEntities) |
2750 | 0 | xmlGenericError(xmlGenericErrorContext, |
2751 | 0 | "String decoding PE Reference: %.30s\n", str); |
2752 | 0 | ent = xmlParseStringPEReference(ctxt, &str); |
2753 | 0 | xmlParserEntityCheck(ctxt, 0, ent, 0); |
2754 | 0 | if (ent != NULL) |
2755 | 0 | ctxt->nbentities += ent->checked / 2; |
2756 | 0 | if (ent != NULL) { |
2757 | 0 | 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 | 0 | if (((ctxt->options & XML_PARSE_NOENT) != 0) || |
2765 | 0 | ((ctxt->options & XML_PARSE_DTDVALID) != 0) || |
2766 | 0 | (ctxt->validate != 0)) { |
2767 | 0 | xmlLoadEntityContent(ctxt, ent); |
2768 | 0 | } else { |
2769 | 0 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
2770 | 0 | "not validating will not read content for PE entity %s\n", |
2771 | 0 | ent->name, NULL); |
2772 | 0 | } |
2773 | 0 | } |
2774 | 0 | ctxt->depth++; |
2775 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, what, |
2776 | 0 | 0, 0, 0); |
2777 | 0 | ctxt->depth--; |
2778 | 0 | if (rep == NULL) { |
2779 | 0 | if (ent->content != NULL) |
2780 | 0 | ent->content[0] = 0; |
2781 | 0 | goto int_error; |
2782 | 0 | } |
2783 | 0 | current = rep; |
2784 | 0 | while (*current != 0) { /* non input consuming loop */ |
2785 | 0 | buffer[nbchars++] = *current++; |
2786 | 0 | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2787 | 0 | if (xmlParserEntityCheck(ctxt, nbchars, ent, 0)) |
2788 | 0 | goto int_error; |
2789 | 0 | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2790 | 0 | } |
2791 | 0 | } |
2792 | 0 | xmlFree(rep); |
2793 | 0 | rep = NULL; |
2794 | 0 | } |
2795 | 14.6M | } else { |
2796 | 14.6M | COPY_BUF(l,buffer,nbchars,c); |
2797 | 14.6M | str += l; |
2798 | 14.6M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2799 | 1.69k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2800 | 1.69k | } |
2801 | 14.6M | } |
2802 | 14.6M | if (str < last) |
2803 | 14.6M | c = CUR_SCHAR(str, l); |
2804 | 2.98k | else |
2805 | 2.98k | c = 0; |
2806 | 14.6M | } |
2807 | 4.72k | buffer[nbchars] = 0; |
2808 | 4.72k | return(buffer); |
2809 | | |
2810 | 0 | mem_error: |
2811 | 0 | xmlErrMemory(ctxt, NULL); |
2812 | 2.53k | int_error: |
2813 | 2.53k | if (rep != NULL) |
2814 | 0 | xmlFree(rep); |
2815 | 2.53k | if (buffer != NULL) |
2816 | 2.53k | xmlFree(buffer); |
2817 | 2.53k | 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 | 7.25k | xmlChar end, xmlChar end2, xmlChar end3) { |
2841 | 7.25k | if ((ctxt == NULL) || (str == NULL)) return(NULL); |
2842 | 7.25k | return(xmlStringLenDecodeEntities(ctxt, str, xmlStrlen(str), what, |
2843 | 7.25k | end, end2, end3)); |
2844 | 7.25k | } |
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 | 209k | int blank_chars) { |
2866 | 209k | int i, ret; |
2867 | 209k | xmlNodePtr lastChild; |
2868 | | |
2869 | | /* |
2870 | | * Don't spend time trying to differentiate them, the same callback is |
2871 | | * used ! |
2872 | | */ |
2873 | 209k | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2874 | 0 | return(0); |
2875 | | |
2876 | | /* |
2877 | | * Check for xml:space value. |
2878 | | */ |
2879 | 209k | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2880 | 209k | (*(ctxt->space) == -2)) |
2881 | 97.8k | return(0); |
2882 | | |
2883 | | /* |
2884 | | * Check that the string is made of blanks |
2885 | | */ |
2886 | 112k | if (blank_chars == 0) { |
2887 | 106k | for (i = 0;i < len;i++) |
2888 | 99.9k | if (!(IS_BLANK_CH(str[i]))) return(0); |
2889 | 47.9k | } |
2890 | | |
2891 | | /* |
2892 | | * Look if the element is mixed content in the DTD if available |
2893 | | */ |
2894 | 70.5k | 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 | 0 | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefix) { |
2947 | 0 | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2948 | 0 | xmlChar *buffer = NULL; |
2949 | 0 | int len = 0; |
2950 | 0 | int max = XML_MAX_NAMELEN; |
2951 | 0 | xmlChar *ret = NULL; |
2952 | 0 | const xmlChar *cur = name; |
2953 | 0 | int c; |
2954 | |
|
2955 | 0 | if (prefix == NULL) return(NULL); |
2956 | 0 | *prefix = NULL; |
2957 | |
|
2958 | 0 | 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 | 0 | if (cur[0] == ':') |
2969 | 0 | return(xmlStrdup(name)); |
2970 | | |
2971 | 0 | c = *cur++; |
2972 | 0 | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
2973 | 0 | buf[len++] = c; |
2974 | 0 | c = *cur++; |
2975 | 0 | } |
2976 | 0 | 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 | 0 | max = len * 2; |
2982 | |
|
2983 | 0 | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
2984 | 0 | if (buffer == NULL) { |
2985 | 0 | xmlErrMemory(ctxt, NULL); |
2986 | 0 | return(NULL); |
2987 | 0 | } |
2988 | 0 | memcpy(buffer, buf, len); |
2989 | 0 | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
2990 | 0 | if (len + 10 > max) { |
2991 | 0 | xmlChar *tmp; |
2992 | |
|
2993 | 0 | max *= 2; |
2994 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, |
2995 | 0 | max * sizeof(xmlChar)); |
2996 | 0 | if (tmp == NULL) { |
2997 | 0 | xmlFree(buffer); |
2998 | 0 | xmlErrMemory(ctxt, NULL); |
2999 | 0 | return(NULL); |
3000 | 0 | } |
3001 | 0 | buffer = tmp; |
3002 | 0 | } |
3003 | 0 | buffer[len++] = c; |
3004 | 0 | c = *cur++; |
3005 | 0 | } |
3006 | 0 | buffer[len] = 0; |
3007 | 0 | } |
3008 | | |
3009 | 0 | if ((c == ':') && (*cur == 0)) { |
3010 | 0 | if (buffer != NULL) |
3011 | 0 | xmlFree(buffer); |
3012 | 0 | *prefix = NULL; |
3013 | 0 | return(xmlStrdup(name)); |
3014 | 0 | } |
3015 | | |
3016 | 0 | if (buffer == NULL) |
3017 | 0 | ret = xmlStrndup(buf, len); |
3018 | 0 | else { |
3019 | 0 | ret = buffer; |
3020 | 0 | buffer = NULL; |
3021 | 0 | max = XML_MAX_NAMELEN; |
3022 | 0 | } |
3023 | | |
3024 | |
|
3025 | 0 | if (c == ':') { |
3026 | 0 | c = *cur; |
3027 | 0 | *prefix = ret; |
3028 | 0 | if (c == 0) { |
3029 | 0 | return(xmlStrndup(BAD_CAST "", 0)); |
3030 | 0 | } |
3031 | 0 | len = 0; |
3032 | | |
3033 | | /* |
3034 | | * Check that the first character is proper to start |
3035 | | * a new name |
3036 | | */ |
3037 | 0 | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3038 | 0 | ((c >= 0x41) && (c <= 0x5A)) || |
3039 | 0 | (c == '_') || (c == ':'))) { |
3040 | 0 | int l; |
3041 | 0 | int first = CUR_SCHAR(cur, l); |
3042 | |
|
3043 | 0 | if (!IS_LETTER(first) && (first != '_')) { |
3044 | 0 | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3045 | 0 | "Name %s is not XML Namespace compliant\n", |
3046 | 0 | name); |
3047 | 0 | } |
3048 | 0 | } |
3049 | 0 | cur++; |
3050 | |
|
3051 | 0 | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3052 | 0 | buf[len++] = c; |
3053 | 0 | c = *cur++; |
3054 | 0 | } |
3055 | 0 | 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 | 0 | max = len * 2; |
3061 | |
|
3062 | 0 | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3063 | 0 | if (buffer == NULL) { |
3064 | 0 | xmlErrMemory(ctxt, NULL); |
3065 | 0 | return(NULL); |
3066 | 0 | } |
3067 | 0 | memcpy(buffer, buf, len); |
3068 | 0 | while (c != 0) { /* tested bigname2.xml */ |
3069 | 0 | if (len + 10 > max) { |
3070 | 0 | xmlChar *tmp; |
3071 | |
|
3072 | 0 | max *= 2; |
3073 | 0 | tmp = (xmlChar *) xmlRealloc(buffer, |
3074 | 0 | max * sizeof(xmlChar)); |
3075 | 0 | if (tmp == NULL) { |
3076 | 0 | xmlErrMemory(ctxt, NULL); |
3077 | 0 | xmlFree(buffer); |
3078 | 0 | return(NULL); |
3079 | 0 | } |
3080 | 0 | buffer = tmp; |
3081 | 0 | } |
3082 | 0 | buffer[len++] = c; |
3083 | 0 | c = *cur++; |
3084 | 0 | } |
3085 | 0 | buffer[len] = 0; |
3086 | 0 | } |
3087 | | |
3088 | 0 | if (buffer == NULL) |
3089 | 0 | ret = xmlStrndup(buf, len); |
3090 | 0 | else { |
3091 | 0 | ret = buffer; |
3092 | 0 | } |
3093 | 0 | } |
3094 | | |
3095 | 0 | return(ret); |
3096 | 0 | } |
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 | 46.0k | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3131 | 46.0k | 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 | 46.0k | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3137 | 46.0k | (((c >= 'a') && (c <= 'z')) || |
3138 | 45.5k | ((c >= 'A') && (c <= 'Z')) || |
3139 | 45.5k | (c == '_') || (c == ':') || |
3140 | 45.5k | ((c >= 0xC0) && (c <= 0xD6)) || |
3141 | 45.5k | ((c >= 0xD8) && (c <= 0xF6)) || |
3142 | 45.5k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3143 | 45.5k | ((c >= 0x370) && (c <= 0x37D)) || |
3144 | 45.5k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3145 | 45.5k | ((c >= 0x200C) && (c <= 0x200D)) || |
3146 | 45.5k | ((c >= 0x2070) && (c <= 0x218F)) || |
3147 | 45.5k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3148 | 45.5k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3149 | 45.5k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3150 | 45.5k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3151 | 45.5k | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3152 | 43.0k | return(1); |
3153 | 46.0k | } else { |
3154 | 0 | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3155 | 0 | return(1); |
3156 | 0 | } |
3157 | 3.07k | return(0); |
3158 | 46.0k | } |
3159 | | |
3160 | | static int |
3161 | 121M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3162 | 121M | 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 | 121M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3168 | 121M | (((c >= 'a') && (c <= 'z')) || |
3169 | 121M | ((c >= 'A') && (c <= 'Z')) || |
3170 | 121M | ((c >= '0') && (c <= '9')) || /* !start */ |
3171 | 121M | (c == '_') || (c == ':') || |
3172 | 121M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3173 | 121M | ((c >= 0xC0) && (c <= 0xD6)) || |
3174 | 121M | ((c >= 0xD8) && (c <= 0xF6)) || |
3175 | 121M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3176 | 121M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3177 | 121M | ((c >= 0x370) && (c <= 0x37D)) || |
3178 | 121M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3179 | 121M | ((c >= 0x200C) && (c <= 0x200D)) || |
3180 | 121M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3181 | 121M | ((c >= 0x2070) && (c <= 0x218F)) || |
3182 | 121M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3183 | 121M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3184 | 121M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3185 | 121M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3186 | 121M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3187 | 121M | return(1); |
3188 | 121M | } 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 | 19.5k | return(0); |
3197 | 121M | } |
3198 | | |
3199 | | static xmlChar * xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, |
3200 | | int *len, int *alloc, int normalize); |
3201 | | |
3202 | | static const xmlChar * |
3203 | 241k | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3204 | 241k | int len = 0, l; |
3205 | 241k | int c; |
3206 | 241k | int count = 0; |
3207 | 241k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3208 | 241k | XML_MAX_TEXT_LENGTH : |
3209 | 241k | XML_MAX_NAME_LENGTH; |
3210 | | |
3211 | | #ifdef DEBUG |
3212 | | nbParseNameComplex++; |
3213 | | #endif |
3214 | | |
3215 | | /* |
3216 | | * Handler for more complex cases |
3217 | | */ |
3218 | 241k | GROW; |
3219 | 241k | if (ctxt->instate == XML_PARSER_EOF) |
3220 | 0 | return(NULL); |
3221 | 241k | c = CUR_CHAR(l); |
3222 | 241k | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3223 | | /* |
3224 | | * Use the new checks of production [4] [4a] amd [5] of the |
3225 | | * Update 5 of XML-1.0 |
3226 | | */ |
3227 | 241k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3228 | 241k | (!(((c >= 'a') && (c <= 'z')) || |
3229 | 227k | ((c >= 'A') && (c <= 'Z')) || |
3230 | 227k | (c == '_') || (c == ':') || |
3231 | 227k | ((c >= 0xC0) && (c <= 0xD6)) || |
3232 | 227k | ((c >= 0xD8) && (c <= 0xF6)) || |
3233 | 227k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3234 | 227k | ((c >= 0x370) && (c <= 0x37D)) || |
3235 | 227k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3236 | 227k | ((c >= 0x200C) && (c <= 0x200D)) || |
3237 | 227k | ((c >= 0x2070) && (c <= 0x218F)) || |
3238 | 227k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3239 | 227k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3240 | 227k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3241 | 227k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3242 | 227k | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3243 | 184k | return(NULL); |
3244 | 184k | } |
3245 | 56.6k | len += l; |
3246 | 56.6k | NEXTL(l); |
3247 | 56.6k | c = CUR_CHAR(l); |
3248 | 122M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3249 | 122M | (((c >= 'a') && (c <= 'z')) || |
3250 | 122M | ((c >= 'A') && (c <= 'Z')) || |
3251 | 122M | ((c >= '0') && (c <= '9')) || /* !start */ |
3252 | 122M | (c == '_') || (c == ':') || |
3253 | 122M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3254 | 122M | ((c >= 0xC0) && (c <= 0xD6)) || |
3255 | 122M | ((c >= 0xD8) && (c <= 0xF6)) || |
3256 | 122M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3257 | 122M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3258 | 122M | ((c >= 0x370) && (c <= 0x37D)) || |
3259 | 122M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3260 | 122M | ((c >= 0x200C) && (c <= 0x200D)) || |
3261 | 122M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3262 | 122M | ((c >= 0x2070) && (c <= 0x218F)) || |
3263 | 122M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3264 | 122M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3265 | 122M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3266 | 122M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3267 | 122M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3268 | 122M | )) { |
3269 | 122M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3270 | 1.20M | count = 0; |
3271 | 1.20M | GROW; |
3272 | 1.20M | if (ctxt->instate == XML_PARSER_EOF) |
3273 | 0 | return(NULL); |
3274 | 1.20M | } |
3275 | 122M | if (len <= INT_MAX - l) |
3276 | 122M | len += l; |
3277 | 122M | NEXTL(l); |
3278 | 122M | c = CUR_CHAR(l); |
3279 | 122M | } |
3280 | 56.6k | } else { |
3281 | 0 | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3282 | 0 | (!IS_LETTER(c) && (c != '_') && |
3283 | 0 | (c != ':'))) { |
3284 | 0 | return(NULL); |
3285 | 0 | } |
3286 | 0 | len += l; |
3287 | 0 | NEXTL(l); |
3288 | 0 | c = CUR_CHAR(l); |
3289 | |
|
3290 | 0 | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3291 | 0 | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3292 | 0 | (c == '.') || (c == '-') || |
3293 | 0 | (c == '_') || (c == ':') || |
3294 | 0 | (IS_COMBINING(c)) || |
3295 | 0 | (IS_EXTENDER(c)))) { |
3296 | 0 | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3297 | 0 | count = 0; |
3298 | 0 | GROW; |
3299 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
3300 | 0 | return(NULL); |
3301 | 0 | } |
3302 | 0 | if (len <= INT_MAX - l) |
3303 | 0 | len += l; |
3304 | 0 | NEXTL(l); |
3305 | 0 | c = CUR_CHAR(l); |
3306 | 0 | } |
3307 | 0 | } |
3308 | 56.6k | if (len > maxLength) { |
3309 | 15 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3310 | 15 | return(NULL); |
3311 | 15 | } |
3312 | 56.6k | if (ctxt->input->cur - ctxt->input->base < len) { |
3313 | | /* |
3314 | | * There were a couple of bugs where PERefs lead to to a change |
3315 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3316 | | * pointer to xmlDictLookup. |
3317 | | */ |
3318 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3319 | 0 | "unexpected change of input buffer"); |
3320 | 0 | return (NULL); |
3321 | 0 | } |
3322 | 56.6k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3323 | 340 | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len)); |
3324 | 56.3k | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len)); |
3325 | 56.6k | } |
3326 | | |
3327 | | /** |
3328 | | * xmlParseName: |
3329 | | * @ctxt: an XML parser context |
3330 | | * |
3331 | | * parse an XML name. |
3332 | | * |
3333 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3334 | | * CombiningChar | Extender |
3335 | | * |
3336 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3337 | | * |
3338 | | * [6] Names ::= Name (#x20 Name)* |
3339 | | * |
3340 | | * Returns the Name parsed or NULL |
3341 | | */ |
3342 | | |
3343 | | const xmlChar * |
3344 | 402k | xmlParseName(xmlParserCtxtPtr ctxt) { |
3345 | 402k | const xmlChar *in; |
3346 | 402k | const xmlChar *ret; |
3347 | 402k | size_t count = 0; |
3348 | 402k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3349 | 402k | XML_MAX_TEXT_LENGTH : |
3350 | 402k | XML_MAX_NAME_LENGTH; |
3351 | | |
3352 | 402k | GROW; |
3353 | | |
3354 | | #ifdef DEBUG |
3355 | | nbParseName++; |
3356 | | #endif |
3357 | | |
3358 | | /* |
3359 | | * Accelerator for simple ASCII names |
3360 | | */ |
3361 | 402k | in = ctxt->input->cur; |
3362 | 402k | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3363 | 402k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3364 | 402k | (*in == '_') || (*in == ':')) { |
3365 | 178k | in++; |
3366 | 723k | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3367 | 723k | ((*in >= 0x41) && (*in <= 0x5A)) || |
3368 | 723k | ((*in >= 0x30) && (*in <= 0x39)) || |
3369 | 723k | (*in == '_') || (*in == '-') || |
3370 | 723k | (*in == ':') || (*in == '.')) |
3371 | 545k | in++; |
3372 | 178k | if ((*in > 0) && (*in < 0x80)) { |
3373 | 161k | count = in - ctxt->input->cur; |
3374 | 161k | if (count > maxLength) { |
3375 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3376 | 0 | return(NULL); |
3377 | 0 | } |
3378 | 161k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3379 | 161k | ctxt->input->cur = in; |
3380 | 161k | ctxt->input->col += count; |
3381 | 161k | if (ret == NULL) |
3382 | 0 | xmlErrMemory(ctxt, NULL); |
3383 | 161k | return(ret); |
3384 | 161k | } |
3385 | 178k | } |
3386 | | /* accelerator for special cases */ |
3387 | 241k | return(xmlParseNameComplex(ctxt)); |
3388 | 402k | } |
3389 | | |
3390 | | static const xmlChar * |
3391 | 43.0k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3392 | 43.0k | int len = 0, l; |
3393 | 43.0k | int c; |
3394 | 43.0k | int count = 0; |
3395 | 43.0k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3396 | 43.0k | XML_MAX_TEXT_LENGTH : |
3397 | 43.0k | XML_MAX_NAME_LENGTH; |
3398 | 43.0k | size_t startPosition = 0; |
3399 | | |
3400 | | #ifdef DEBUG |
3401 | | nbParseNCNameComplex++; |
3402 | | #endif |
3403 | | |
3404 | | /* |
3405 | | * Handler for more complex cases |
3406 | | */ |
3407 | 43.0k | GROW; |
3408 | 43.0k | startPosition = CUR_PTR - BASE_PTR; |
3409 | 43.0k | c = CUR_CHAR(l); |
3410 | 43.0k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3411 | 43.0k | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3412 | 17.5k | return(NULL); |
3413 | 17.5k | } |
3414 | | |
3415 | 85.4M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3416 | 85.4M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3417 | 85.4M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3418 | 836k | count = 0; |
3419 | 836k | GROW; |
3420 | 836k | if (ctxt->instate == XML_PARSER_EOF) |
3421 | 0 | return(NULL); |
3422 | 836k | } |
3423 | 85.4M | if (len <= INT_MAX - l) |
3424 | 85.4M | len += l; |
3425 | 85.4M | NEXTL(l); |
3426 | 85.4M | c = CUR_CHAR(l); |
3427 | 85.4M | if (c == 0) { |
3428 | 1.27k | count = 0; |
3429 | | /* |
3430 | | * when shrinking to extend the buffer we really need to preserve |
3431 | | * the part of the name we already parsed. Hence rolling back |
3432 | | * by current length. |
3433 | | */ |
3434 | 1.27k | ctxt->input->cur -= l; |
3435 | 1.27k | GROW; |
3436 | 1.27k | if (ctxt->instate == XML_PARSER_EOF) |
3437 | 0 | return(NULL); |
3438 | 1.27k | ctxt->input->cur += l; |
3439 | 1.27k | c = CUR_CHAR(l); |
3440 | 1.27k | } |
3441 | 85.4M | } |
3442 | 25.4k | if (len > maxLength) { |
3443 | 12 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3444 | 12 | return(NULL); |
3445 | 12 | } |
3446 | 25.4k | return(xmlDictLookup(ctxt->dict, (BASE_PTR + startPosition), len)); |
3447 | 25.4k | } |
3448 | | |
3449 | | /** |
3450 | | * xmlParseNCName: |
3451 | | * @ctxt: an XML parser context |
3452 | | * @len: length of the string parsed |
3453 | | * |
3454 | | * parse an XML name. |
3455 | | * |
3456 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3457 | | * CombiningChar | Extender |
3458 | | * |
3459 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3460 | | * |
3461 | | * Returns the Name parsed or NULL |
3462 | | */ |
3463 | | |
3464 | | static const xmlChar * |
3465 | 2.50M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3466 | 2.50M | const xmlChar *in, *e; |
3467 | 2.50M | const xmlChar *ret; |
3468 | 2.50M | size_t count = 0; |
3469 | 2.50M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3470 | 2.50M | XML_MAX_TEXT_LENGTH : |
3471 | 2.50M | XML_MAX_NAME_LENGTH; |
3472 | | |
3473 | | #ifdef DEBUG |
3474 | | nbParseNCName++; |
3475 | | #endif |
3476 | | |
3477 | | /* |
3478 | | * Accelerator for simple ASCII names |
3479 | | */ |
3480 | 2.50M | in = ctxt->input->cur; |
3481 | 2.50M | e = ctxt->input->end; |
3482 | 2.50M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3483 | 2.50M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3484 | 2.50M | (*in == '_')) && (in < e)) { |
3485 | 2.46M | in++; |
3486 | 5.59M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3487 | 5.59M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3488 | 5.59M | ((*in >= 0x30) && (*in <= 0x39)) || |
3489 | 5.59M | (*in == '_') || (*in == '-') || |
3490 | 5.59M | (*in == '.')) && (in < e)) |
3491 | 3.12M | in++; |
3492 | 2.46M | if (in >= e) |
3493 | 709 | goto complex; |
3494 | 2.46M | if ((*in > 0) && (*in < 0x80)) { |
3495 | 2.45M | count = in - ctxt->input->cur; |
3496 | 2.45M | if (count > maxLength) { |
3497 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3498 | 0 | return(NULL); |
3499 | 0 | } |
3500 | 2.45M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3501 | 2.45M | ctxt->input->cur = in; |
3502 | 2.45M | ctxt->input->col += count; |
3503 | 2.45M | if (ret == NULL) { |
3504 | 0 | xmlErrMemory(ctxt, NULL); |
3505 | 0 | } |
3506 | 2.45M | return(ret); |
3507 | 2.45M | } |
3508 | 2.46M | } |
3509 | 43.0k | complex: |
3510 | 43.0k | return(xmlParseNCNameComplex(ctxt)); |
3511 | 2.50M | } |
3512 | | |
3513 | | /** |
3514 | | * xmlParseNameAndCompare: |
3515 | | * @ctxt: an XML parser context |
3516 | | * |
3517 | | * parse an XML name and compares for match |
3518 | | * (specialized for endtag parsing) |
3519 | | * |
3520 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3521 | | * and the name for mismatch |
3522 | | */ |
3523 | | |
3524 | | static const xmlChar * |
3525 | 29.9k | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3526 | 29.9k | register const xmlChar *cmp = other; |
3527 | 29.9k | register const xmlChar *in; |
3528 | 29.9k | const xmlChar *ret; |
3529 | | |
3530 | 29.9k | GROW; |
3531 | 29.9k | if (ctxt->instate == XML_PARSER_EOF) |
3532 | 0 | return(NULL); |
3533 | | |
3534 | 29.9k | in = ctxt->input->cur; |
3535 | 174k | while (*in != 0 && *in == *cmp) { |
3536 | 144k | ++in; |
3537 | 144k | ++cmp; |
3538 | 144k | } |
3539 | 29.9k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3540 | | /* success */ |
3541 | 29.8k | ctxt->input->col += in - ctxt->input->cur; |
3542 | 29.8k | ctxt->input->cur = in; |
3543 | 29.8k | return (const xmlChar*) 1; |
3544 | 29.8k | } |
3545 | | /* failure (or end of input buffer), check with full function */ |
3546 | 151 | ret = xmlParseName (ctxt); |
3547 | | /* strings coming from the dictionary direct compare possible */ |
3548 | 151 | if (ret == other) { |
3549 | 18 | return (const xmlChar*) 1; |
3550 | 18 | } |
3551 | 133 | return ret; |
3552 | 151 | } |
3553 | | |
3554 | | /** |
3555 | | * xmlParseStringName: |
3556 | | * @ctxt: an XML parser context |
3557 | | * @str: a pointer to the string pointer (IN/OUT) |
3558 | | * |
3559 | | * parse an XML name. |
3560 | | * |
3561 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3562 | | * CombiningChar | Extender |
3563 | | * |
3564 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3565 | | * |
3566 | | * [6] Names ::= Name (#x20 Name)* |
3567 | | * |
3568 | | * Returns the Name parsed or NULL. The @str pointer |
3569 | | * is updated to the current location in the string. |
3570 | | */ |
3571 | | |
3572 | | static xmlChar * |
3573 | 4.12k | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3574 | 4.12k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3575 | 4.12k | const xmlChar *cur = *str; |
3576 | 4.12k | int len = 0, l; |
3577 | 4.12k | int c; |
3578 | 4.12k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3579 | 4.12k | XML_MAX_TEXT_LENGTH : |
3580 | 4.12k | XML_MAX_NAME_LENGTH; |
3581 | | |
3582 | | #ifdef DEBUG |
3583 | | nbParseStringName++; |
3584 | | #endif |
3585 | | |
3586 | 4.12k | c = CUR_SCHAR(cur, l); |
3587 | 4.12k | if (!xmlIsNameStartChar(ctxt, c)) { |
3588 | 1.26k | return(NULL); |
3589 | 1.26k | } |
3590 | | |
3591 | 2.85k | COPY_BUF(l,buf,len,c); |
3592 | 2.85k | cur += l; |
3593 | 2.85k | c = CUR_SCHAR(cur, l); |
3594 | 45.6k | while (xmlIsNameChar(ctxt, c)) { |
3595 | 43.5k | COPY_BUF(l,buf,len,c); |
3596 | 43.5k | cur += l; |
3597 | 43.5k | c = CUR_SCHAR(cur, l); |
3598 | 43.5k | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3599 | | /* |
3600 | | * Okay someone managed to make a huge name, so he's ready to pay |
3601 | | * for the processing speed. |
3602 | | */ |
3603 | 693 | xmlChar *buffer; |
3604 | 693 | int max = len * 2; |
3605 | | |
3606 | 693 | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3607 | 693 | if (buffer == NULL) { |
3608 | 0 | xmlErrMemory(ctxt, NULL); |
3609 | 0 | return(NULL); |
3610 | 0 | } |
3611 | 693 | memcpy(buffer, buf, len); |
3612 | 24.2M | while (xmlIsNameChar(ctxt, c)) { |
3613 | 24.2M | if (len + 10 > max) { |
3614 | 2.86k | xmlChar *tmp; |
3615 | | |
3616 | 2.86k | max *= 2; |
3617 | 2.86k | tmp = (xmlChar *) xmlRealloc(buffer, |
3618 | 2.86k | max * sizeof(xmlChar)); |
3619 | 2.86k | if (tmp == NULL) { |
3620 | 0 | xmlErrMemory(ctxt, NULL); |
3621 | 0 | xmlFree(buffer); |
3622 | 0 | return(NULL); |
3623 | 0 | } |
3624 | 2.86k | buffer = tmp; |
3625 | 2.86k | } |
3626 | 24.2M | COPY_BUF(l,buffer,len,c); |
3627 | 24.2M | cur += l; |
3628 | 24.2M | c = CUR_SCHAR(cur, l); |
3629 | 24.2M | if (len > maxLength) { |
3630 | 2 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3631 | 2 | xmlFree(buffer); |
3632 | 2 | return(NULL); |
3633 | 2 | } |
3634 | 24.2M | } |
3635 | 691 | buffer[len] = 0; |
3636 | 691 | *str = cur; |
3637 | 691 | return(buffer); |
3638 | 693 | } |
3639 | 43.5k | } |
3640 | 2.16k | if (len > maxLength) { |
3641 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3642 | 0 | return(NULL); |
3643 | 0 | } |
3644 | 2.16k | *str = cur; |
3645 | 2.16k | return(xmlStrndup(buf, len)); |
3646 | 2.16k | } |
3647 | | |
3648 | | /** |
3649 | | * xmlParseNmtoken: |
3650 | | * @ctxt: an XML parser context |
3651 | | * |
3652 | | * parse an XML Nmtoken. |
3653 | | * |
3654 | | * [7] Nmtoken ::= (NameChar)+ |
3655 | | * |
3656 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3657 | | * |
3658 | | * Returns the Nmtoken parsed or NULL |
3659 | | */ |
3660 | | |
3661 | | xmlChar * |
3662 | 5.63k | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3663 | 5.63k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3664 | 5.63k | int len = 0, l; |
3665 | 5.63k | int c; |
3666 | 5.63k | int count = 0; |
3667 | 5.63k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3668 | 5.63k | XML_MAX_TEXT_LENGTH : |
3669 | 5.63k | XML_MAX_NAME_LENGTH; |
3670 | | |
3671 | | #ifdef DEBUG |
3672 | | nbParseNmToken++; |
3673 | | #endif |
3674 | | |
3675 | 5.63k | GROW; |
3676 | 5.63k | if (ctxt->instate == XML_PARSER_EOF) |
3677 | 0 | return(NULL); |
3678 | 5.63k | c = CUR_CHAR(l); |
3679 | | |
3680 | 38.7k | while (xmlIsNameChar(ctxt, c)) { |
3681 | 33.7k | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3682 | 0 | count = 0; |
3683 | 0 | GROW; |
3684 | 0 | } |
3685 | 33.7k | COPY_BUF(l,buf,len,c); |
3686 | 33.7k | NEXTL(l); |
3687 | 33.7k | c = CUR_CHAR(l); |
3688 | 33.7k | if (c == 0) { |
3689 | 60 | count = 0; |
3690 | 60 | GROW; |
3691 | 60 | if (ctxt->instate == XML_PARSER_EOF) |
3692 | 0 | return(NULL); |
3693 | 60 | c = CUR_CHAR(l); |
3694 | 60 | } |
3695 | 33.7k | if (len >= XML_MAX_NAMELEN) { |
3696 | | /* |
3697 | | * Okay someone managed to make a huge token, so he's ready to pay |
3698 | | * for the processing speed. |
3699 | | */ |
3700 | 625 | xmlChar *buffer; |
3701 | 625 | int max = len * 2; |
3702 | | |
3703 | 625 | buffer = (xmlChar *) xmlMallocAtomic(max * sizeof(xmlChar)); |
3704 | 625 | if (buffer == NULL) { |
3705 | 0 | xmlErrMemory(ctxt, NULL); |
3706 | 0 | return(NULL); |
3707 | 0 | } |
3708 | 625 | memcpy(buffer, buf, len); |
3709 | 11.3M | while (xmlIsNameChar(ctxt, c)) { |
3710 | 11.3M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3711 | 111k | count = 0; |
3712 | 111k | GROW; |
3713 | 111k | if (ctxt->instate == XML_PARSER_EOF) { |
3714 | 0 | xmlFree(buffer); |
3715 | 0 | return(NULL); |
3716 | 0 | } |
3717 | 111k | } |
3718 | 11.3M | if (len + 10 > max) { |
3719 | 853 | xmlChar *tmp; |
3720 | | |
3721 | 853 | max *= 2; |
3722 | 853 | tmp = (xmlChar *) xmlRealloc(buffer, |
3723 | 853 | max * sizeof(xmlChar)); |
3724 | 853 | if (tmp == NULL) { |
3725 | 0 | xmlErrMemory(ctxt, NULL); |
3726 | 0 | xmlFree(buffer); |
3727 | 0 | return(NULL); |
3728 | 0 | } |
3729 | 853 | buffer = tmp; |
3730 | 853 | } |
3731 | 11.3M | COPY_BUF(l,buffer,len,c); |
3732 | 11.3M | NEXTL(l); |
3733 | 11.3M | c = CUR_CHAR(l); |
3734 | 11.3M | if (len > maxLength) { |
3735 | 2 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3736 | 2 | xmlFree(buffer); |
3737 | 2 | return(NULL); |
3738 | 2 | } |
3739 | 11.3M | } |
3740 | 623 | buffer[len] = 0; |
3741 | 623 | return(buffer); |
3742 | 625 | } |
3743 | 33.7k | } |
3744 | 5.00k | if (len == 0) |
3745 | 1.33k | return(NULL); |
3746 | 3.67k | if (len > maxLength) { |
3747 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3748 | 0 | return(NULL); |
3749 | 0 | } |
3750 | 3.67k | return(xmlStrndup(buf, len)); |
3751 | 3.67k | } |
3752 | | |
3753 | | /** |
3754 | | * xmlParseEntityValue: |
3755 | | * @ctxt: an XML parser context |
3756 | | * @orig: if non-NULL store a copy of the original entity value |
3757 | | * |
3758 | | * parse a value for ENTITY declarations |
3759 | | * |
3760 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3761 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3762 | | * |
3763 | | * Returns the EntityValue parsed with reference substituted or NULL |
3764 | | */ |
3765 | | |
3766 | | xmlChar * |
3767 | 10.0k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3768 | 10.0k | xmlChar *buf = NULL; |
3769 | 10.0k | int len = 0; |
3770 | 10.0k | int size = XML_PARSER_BUFFER_SIZE; |
3771 | 10.0k | int c, l; |
3772 | 10.0k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3773 | 10.0k | XML_MAX_HUGE_LENGTH : |
3774 | 10.0k | XML_MAX_TEXT_LENGTH; |
3775 | 10.0k | xmlChar stop; |
3776 | 10.0k | xmlChar *ret = NULL; |
3777 | 10.0k | const xmlChar *cur = NULL; |
3778 | 10.0k | xmlParserInputPtr input; |
3779 | | |
3780 | 10.0k | if (RAW == '"') stop = '"'; |
3781 | 5.01k | else if (RAW == '\'') stop = '\''; |
3782 | 0 | else { |
3783 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_STARTED, NULL); |
3784 | 0 | return(NULL); |
3785 | 0 | } |
3786 | 10.0k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
3787 | 10.0k | if (buf == NULL) { |
3788 | 0 | xmlErrMemory(ctxt, NULL); |
3789 | 0 | return(NULL); |
3790 | 0 | } |
3791 | | |
3792 | | /* |
3793 | | * The content of the entity definition is copied in a buffer. |
3794 | | */ |
3795 | | |
3796 | 10.0k | ctxt->instate = XML_PARSER_ENTITY_VALUE; |
3797 | 10.0k | input = ctxt->input; |
3798 | 10.0k | GROW; |
3799 | 10.0k | if (ctxt->instate == XML_PARSER_EOF) |
3800 | 0 | goto error; |
3801 | 10.0k | NEXT; |
3802 | 10.0k | c = CUR_CHAR(l); |
3803 | | /* |
3804 | | * NOTE: 4.4.5 Included in Literal |
3805 | | * When a parameter entity reference appears in a literal entity |
3806 | | * value, ... a single or double quote character in the replacement |
3807 | | * text is always treated as a normal data character and will not |
3808 | | * terminate the literal. |
3809 | | * In practice it means we stop the loop only when back at parsing |
3810 | | * the initial entity and the quote is found |
3811 | | */ |
3812 | 45.1M | while (((IS_CHAR(c)) && ((c != stop) || /* checked */ |
3813 | 45.1M | (ctxt->input != input))) && (ctxt->instate != XML_PARSER_EOF)) { |
3814 | 45.0M | if (len + 5 >= size) { |
3815 | 3.38k | xmlChar *tmp; |
3816 | | |
3817 | 3.38k | size *= 2; |
3818 | 3.38k | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
3819 | 3.38k | if (tmp == NULL) { |
3820 | 0 | xmlErrMemory(ctxt, NULL); |
3821 | 0 | goto error; |
3822 | 0 | } |
3823 | 3.38k | buf = tmp; |
3824 | 3.38k | } |
3825 | 45.0M | COPY_BUF(l,buf,len,c); |
3826 | 45.0M | NEXTL(l); |
3827 | | |
3828 | 45.0M | GROW; |
3829 | 45.0M | c = CUR_CHAR(l); |
3830 | 45.0M | if (c == 0) { |
3831 | 32 | GROW; |
3832 | 32 | c = CUR_CHAR(l); |
3833 | 32 | } |
3834 | | |
3835 | 45.0M | if (len > maxLength) { |
3836 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
3837 | 0 | "entity value too long\n"); |
3838 | 0 | goto error; |
3839 | 0 | } |
3840 | 45.0M | } |
3841 | 10.0k | buf[len] = 0; |
3842 | 10.0k | if (ctxt->instate == XML_PARSER_EOF) |
3843 | 0 | goto error; |
3844 | 10.0k | if (c != stop) { |
3845 | 60 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3846 | 60 | goto error; |
3847 | 60 | } |
3848 | 9.94k | NEXT; |
3849 | | |
3850 | | /* |
3851 | | * Raise problem w.r.t. '&' and '%' being used in non-entities |
3852 | | * reference constructs. Note Charref will be handled in |
3853 | | * xmlStringDecodeEntities() |
3854 | | */ |
3855 | 9.94k | cur = buf; |
3856 | 8.03M | while (*cur != 0) { /* non input consuming */ |
3857 | 8.02M | if ((*cur == '%') || ((*cur == '&') && (cur[1] != '#'))) { |
3858 | 4.12k | xmlChar *name; |
3859 | 4.12k | xmlChar tmp = *cur; |
3860 | 4.12k | int nameOk = 0; |
3861 | | |
3862 | 4.12k | cur++; |
3863 | 4.12k | name = xmlParseStringName(ctxt, &cur); |
3864 | 4.12k | if (name != NULL) { |
3865 | 2.85k | nameOk = 1; |
3866 | 2.85k | xmlFree(name); |
3867 | 2.85k | } |
3868 | 4.12k | if ((nameOk == 0) || (*cur != ';')) { |
3869 | 2.46k | xmlFatalErrMsgInt(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3870 | 2.46k | "EntityValue: '%c' forbidden except for entities references\n", |
3871 | 2.46k | tmp); |
3872 | 2.46k | goto error; |
3873 | 2.46k | } |
3874 | 1.66k | if ((tmp == '%') && (ctxt->inSubset == 1) && |
3875 | 1.66k | (ctxt->inputNr == 1)) { |
3876 | 225 | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3877 | 225 | goto error; |
3878 | 225 | } |
3879 | 1.43k | if (*cur == 0) |
3880 | 0 | break; |
3881 | 1.43k | } |
3882 | 8.02M | cur++; |
3883 | 8.02M | } |
3884 | | |
3885 | | /* |
3886 | | * Then PEReference entities are substituted. |
3887 | | * |
3888 | | * NOTE: 4.4.7 Bypassed |
3889 | | * When a general entity reference appears in the EntityValue in |
3890 | | * an entity declaration, it is bypassed and left as is. |
3891 | | * so XML_SUBSTITUTE_REF is not set here. |
3892 | | */ |
3893 | 7.25k | ++ctxt->depth; |
3894 | 7.25k | ret = xmlStringDecodeEntities(ctxt, buf, XML_SUBSTITUTE_PEREF, |
3895 | 7.25k | 0, 0, 0); |
3896 | 7.25k | --ctxt->depth; |
3897 | 7.25k | if (orig != NULL) { |
3898 | 7.25k | *orig = buf; |
3899 | 7.25k | buf = NULL; |
3900 | 7.25k | } |
3901 | | |
3902 | 10.0k | error: |
3903 | 10.0k | if (buf != NULL) |
3904 | 2.74k | xmlFree(buf); |
3905 | 10.0k | return(ret); |
3906 | 7.25k | } |
3907 | | |
3908 | | /** |
3909 | | * xmlParseAttValueComplex: |
3910 | | * @ctxt: an XML parser context |
3911 | | * @len: the resulting attribute len |
3912 | | * @normalize: whether to apply the inner normalization |
3913 | | * |
3914 | | * parse a value for an attribute, this is the fallback function |
3915 | | * of xmlParseAttValue() when the attribute parsing requires handling |
3916 | | * of non-ASCII characters, or normalization compaction. |
3917 | | * |
3918 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
3919 | | */ |
3920 | | static xmlChar * |
3921 | 156k | xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) { |
3922 | 156k | xmlChar limit = 0; |
3923 | 156k | xmlChar *buf = NULL; |
3924 | 156k | xmlChar *rep = NULL; |
3925 | 156k | size_t len = 0; |
3926 | 156k | size_t buf_size = 0; |
3927 | 156k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3928 | 156k | XML_MAX_HUGE_LENGTH : |
3929 | 156k | XML_MAX_TEXT_LENGTH; |
3930 | 156k | int c, l, in_space = 0; |
3931 | 156k | xmlChar *current = NULL; |
3932 | 156k | xmlEntityPtr ent; |
3933 | | |
3934 | 156k | if (NXT(0) == '"') { |
3935 | 153k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3936 | 153k | limit = '"'; |
3937 | 153k | NEXT; |
3938 | 153k | } else if (NXT(0) == '\'') { |
3939 | 2.89k | limit = '\''; |
3940 | 2.89k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3941 | 2.89k | NEXT; |
3942 | 2.89k | } else { |
3943 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3944 | 0 | return(NULL); |
3945 | 0 | } |
3946 | | |
3947 | | /* |
3948 | | * allocate a translation buffer. |
3949 | | */ |
3950 | 156k | buf_size = XML_PARSER_BUFFER_SIZE; |
3951 | 156k | buf = (xmlChar *) xmlMallocAtomic(buf_size); |
3952 | 156k | if (buf == NULL) goto mem_error; |
3953 | | |
3954 | | /* |
3955 | | * OK loop until we reach one of the ending char or a size limit. |
3956 | | */ |
3957 | 156k | c = CUR_CHAR(l); |
3958 | 117M | while (((NXT(0) != limit) && /* checked */ |
3959 | 117M | (IS_CHAR(c)) && (c != '<')) && |
3960 | 117M | (ctxt->instate != XML_PARSER_EOF)) { |
3961 | 117M | if (c == '&') { |
3962 | 652k | in_space = 0; |
3963 | 652k | if (NXT(1) == '#') { |
3964 | 378k | int val = xmlParseCharRef(ctxt); |
3965 | | |
3966 | 378k | if (val == '&') { |
3967 | 7.57k | if (ctxt->replaceEntities) { |
3968 | 0 | if (len + 10 > buf_size) { |
3969 | 0 | growBuffer(buf, 10); |
3970 | 0 | } |
3971 | 0 | buf[len++] = '&'; |
3972 | 7.57k | } else { |
3973 | | /* |
3974 | | * The reparsing will be done in xmlStringGetNodeList() |
3975 | | * called by the attribute() function in SAX.c |
3976 | | */ |
3977 | 7.57k | if (len + 10 > buf_size) { |
3978 | 432 | growBuffer(buf, 10); |
3979 | 432 | } |
3980 | 7.57k | buf[len++] = '&'; |
3981 | 7.57k | buf[len++] = '#'; |
3982 | 7.57k | buf[len++] = '3'; |
3983 | 7.57k | buf[len++] = '8'; |
3984 | 7.57k | buf[len++] = ';'; |
3985 | 7.57k | } |
3986 | 371k | } else if (val != 0) { |
3987 | 129k | if (len + 10 > buf_size) { |
3988 | 446 | growBuffer(buf, 10); |
3989 | 446 | } |
3990 | 129k | len += xmlCopyChar(0, &buf[len], val); |
3991 | 129k | } |
3992 | 378k | } else { |
3993 | 273k | ent = xmlParseEntityRef(ctxt); |
3994 | 273k | ctxt->nbentities++; |
3995 | 273k | if (ent != NULL) |
3996 | 44.3k | ctxt->nbentities += ent->owner; |
3997 | 273k | if ((ent != NULL) && |
3998 | 273k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
3999 | 30.2k | if (len + 10 > buf_size) { |
4000 | 672 | growBuffer(buf, 10); |
4001 | 672 | } |
4002 | 30.2k | if ((ctxt->replaceEntities == 0) && |
4003 | 30.2k | (ent->content[0] == '&')) { |
4004 | 18.6k | buf[len++] = '&'; |
4005 | 18.6k | buf[len++] = '#'; |
4006 | 18.6k | buf[len++] = '3'; |
4007 | 18.6k | buf[len++] = '8'; |
4008 | 18.6k | buf[len++] = ';'; |
4009 | 18.6k | } else { |
4010 | 11.5k | buf[len++] = ent->content[0]; |
4011 | 11.5k | } |
4012 | 243k | } else if ((ent != NULL) && |
4013 | 243k | (ctxt->replaceEntities != 0)) { |
4014 | 0 | if (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) { |
4015 | 0 | ++ctxt->depth; |
4016 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4017 | 0 | XML_SUBSTITUTE_REF, |
4018 | 0 | 0, 0, 0); |
4019 | 0 | --ctxt->depth; |
4020 | 0 | if (rep != NULL) { |
4021 | 0 | current = rep; |
4022 | 0 | while (*current != 0) { /* non input consuming */ |
4023 | 0 | if ((*current == 0xD) || (*current == 0xA) || |
4024 | 0 | (*current == 0x9)) { |
4025 | 0 | buf[len++] = 0x20; |
4026 | 0 | current++; |
4027 | 0 | } else |
4028 | 0 | buf[len++] = *current++; |
4029 | 0 | if (len + 10 > buf_size) { |
4030 | 0 | growBuffer(buf, 10); |
4031 | 0 | } |
4032 | 0 | } |
4033 | 0 | xmlFree(rep); |
4034 | 0 | rep = NULL; |
4035 | 0 | } |
4036 | 0 | } else { |
4037 | 0 | if (len + 10 > buf_size) { |
4038 | 0 | growBuffer(buf, 10); |
4039 | 0 | } |
4040 | 0 | if (ent->content != NULL) |
4041 | 0 | buf[len++] = ent->content[0]; |
4042 | 0 | } |
4043 | 243k | } else if (ent != NULL) { |
4044 | 14.1k | int i = xmlStrlen(ent->name); |
4045 | 14.1k | const xmlChar *cur = ent->name; |
4046 | | |
4047 | | /* |
4048 | | * This may look absurd but is needed to detect |
4049 | | * entities problems |
4050 | | */ |
4051 | 14.1k | if ((ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
4052 | 14.1k | (ent->content != NULL) && (ent->checked == 0)) { |
4053 | 0 | unsigned long oldnbent = ctxt->nbentities, diff; |
4054 | |
|
4055 | 0 | ++ctxt->depth; |
4056 | 0 | rep = xmlStringDecodeEntities(ctxt, ent->content, |
4057 | 0 | XML_SUBSTITUTE_REF, 0, 0, 0); |
4058 | 0 | --ctxt->depth; |
4059 | |
|
4060 | 0 | diff = ctxt->nbentities - oldnbent + 1; |
4061 | 0 | if (diff > INT_MAX / 2) |
4062 | 0 | diff = INT_MAX / 2; |
4063 | 0 | ent->checked = diff * 2; |
4064 | 0 | if (rep != NULL) { |
4065 | 0 | if (xmlStrchr(rep, '<')) |
4066 | 0 | ent->checked |= 1; |
4067 | 0 | xmlFree(rep); |
4068 | 0 | rep = NULL; |
4069 | 0 | } else { |
4070 | 0 | ent->content[0] = 0; |
4071 | 0 | } |
4072 | 0 | } |
4073 | | |
4074 | | /* |
4075 | | * Just output the reference |
4076 | | */ |
4077 | 14.1k | buf[len++] = '&'; |
4078 | 14.2k | while (len + i + 10 > buf_size) { |
4079 | 218 | growBuffer(buf, i + 10); |
4080 | 218 | } |
4081 | 14.1k | for (;i > 0;i--) |
4082 | 0 | buf[len++] = *cur++; |
4083 | 14.1k | buf[len++] = ';'; |
4084 | 14.1k | } |
4085 | 273k | } |
4086 | 116M | } else { |
4087 | 116M | if ((c == 0x20) || (c == 0xD) || (c == 0xA) || (c == 0x9)) { |
4088 | 1.40M | if ((len != 0) || (!normalize)) { |
4089 | 1.40M | if ((!normalize) || (!in_space)) { |
4090 | 1.39M | COPY_BUF(l,buf,len,0x20); |
4091 | 1.39M | while (len + 10 > buf_size) { |
4092 | 1.08k | growBuffer(buf, 10); |
4093 | 1.08k | } |
4094 | 1.39M | } |
4095 | 1.40M | in_space = 1; |
4096 | 1.40M | } |
4097 | 115M | } else { |
4098 | 115M | in_space = 0; |
4099 | 115M | COPY_BUF(l,buf,len,c); |
4100 | 115M | if (len + 10 > buf_size) { |
4101 | 16.8k | growBuffer(buf, 10); |
4102 | 16.8k | } |
4103 | 115M | } |
4104 | 116M | NEXTL(l); |
4105 | 116M | } |
4106 | 117M | GROW; |
4107 | 117M | c = CUR_CHAR(l); |
4108 | 117M | if (len > maxLength) { |
4109 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4110 | 0 | "AttValue length too long\n"); |
4111 | 0 | goto mem_error; |
4112 | 0 | } |
4113 | 117M | } |
4114 | 156k | if (ctxt->instate == XML_PARSER_EOF) |
4115 | 0 | goto error; |
4116 | | |
4117 | 156k | if ((in_space) && (normalize)) { |
4118 | 2.61k | while ((len > 0) && (buf[len - 1] == 0x20)) len--; |
4119 | 894 | } |
4120 | 156k | buf[len] = 0; |
4121 | 156k | if (RAW == '<') { |
4122 | 833 | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4123 | 155k | } else if (RAW != limit) { |
4124 | 3.31k | if ((c != 0) && (!IS_CHAR(c))) { |
4125 | 114 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4126 | 114 | "invalid character in attribute value\n"); |
4127 | 3.20k | } else { |
4128 | 3.20k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4129 | 3.20k | "AttValue: ' expected\n"); |
4130 | 3.20k | } |
4131 | 3.31k | } else |
4132 | 152k | NEXT; |
4133 | | |
4134 | 156k | if (attlen != NULL) *attlen = (int) len; |
4135 | 156k | return(buf); |
4136 | | |
4137 | 0 | mem_error: |
4138 | 0 | xmlErrMemory(ctxt, NULL); |
4139 | 0 | error: |
4140 | 0 | if (buf != NULL) |
4141 | 0 | xmlFree(buf); |
4142 | 0 | if (rep != NULL) |
4143 | 0 | xmlFree(rep); |
4144 | 0 | return(NULL); |
4145 | 0 | } |
4146 | | |
4147 | | /** |
4148 | | * xmlParseAttValue: |
4149 | | * @ctxt: an XML parser context |
4150 | | * |
4151 | | * parse a value for an attribute |
4152 | | * Note: the parser won't do substitution of entities here, this |
4153 | | * will be handled later in xmlStringGetNodeList |
4154 | | * |
4155 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4156 | | * "'" ([^<&'] | Reference)* "'" |
4157 | | * |
4158 | | * 3.3.3 Attribute-Value Normalization: |
4159 | | * Before the value of an attribute is passed to the application or |
4160 | | * checked for validity, the XML processor must normalize it as follows: |
4161 | | * - a character reference is processed by appending the referenced |
4162 | | * character to the attribute value |
4163 | | * - an entity reference is processed by recursively processing the |
4164 | | * replacement text of the entity |
4165 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4166 | | * appending #x20 to the normalized value, except that only a single |
4167 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4168 | | * parsed entity or the literal entity value of an internal parsed entity |
4169 | | * - other characters are processed by appending them to the normalized value |
4170 | | * If the declared value is not CDATA, then the XML processor must further |
4171 | | * process the normalized attribute value by discarding any leading and |
4172 | | * trailing space (#x20) characters, and by replacing sequences of space |
4173 | | * (#x20) characters by a single space (#x20) character. |
4174 | | * All attributes for which no declaration has been read should be treated |
4175 | | * by a non-validating parser as if declared CDATA. |
4176 | | * |
4177 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4178 | | */ |
4179 | | |
4180 | | |
4181 | | xmlChar * |
4182 | 22.3k | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4183 | 22.3k | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4184 | 22.3k | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0)); |
4185 | 22.3k | } |
4186 | | |
4187 | | /** |
4188 | | * xmlParseSystemLiteral: |
4189 | | * @ctxt: an XML parser context |
4190 | | * |
4191 | | * parse an XML Literal |
4192 | | * |
4193 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4194 | | * |
4195 | | * Returns the SystemLiteral parsed or NULL |
4196 | | */ |
4197 | | |
4198 | | xmlChar * |
4199 | 4.70k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4200 | 4.70k | xmlChar *buf = NULL; |
4201 | 4.70k | int len = 0; |
4202 | 4.70k | int size = XML_PARSER_BUFFER_SIZE; |
4203 | 4.70k | int cur, l; |
4204 | 4.70k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4205 | 4.70k | XML_MAX_TEXT_LENGTH : |
4206 | 4.70k | XML_MAX_NAME_LENGTH; |
4207 | 4.70k | xmlChar stop; |
4208 | 4.70k | int state = ctxt->instate; |
4209 | 4.70k | int count = 0; |
4210 | | |
4211 | 4.70k | SHRINK; |
4212 | 4.70k | if (RAW == '"') { |
4213 | 1.66k | NEXT; |
4214 | 1.66k | stop = '"'; |
4215 | 3.04k | } else if (RAW == '\'') { |
4216 | 1.34k | NEXT; |
4217 | 1.34k | stop = '\''; |
4218 | 1.70k | } else { |
4219 | 1.70k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4220 | 1.70k | return(NULL); |
4221 | 1.70k | } |
4222 | | |
4223 | 3.00k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4224 | 3.00k | if (buf == NULL) { |
4225 | 0 | xmlErrMemory(ctxt, NULL); |
4226 | 0 | return(NULL); |
4227 | 0 | } |
4228 | 3.00k | ctxt->instate = XML_PARSER_SYSTEM_LITERAL; |
4229 | 3.00k | cur = CUR_CHAR(l); |
4230 | 29.4M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4231 | 29.4M | if (len + 5 >= size) { |
4232 | 1.92k | xmlChar *tmp; |
4233 | | |
4234 | 1.92k | size *= 2; |
4235 | 1.92k | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
4236 | 1.92k | if (tmp == NULL) { |
4237 | 0 | xmlFree(buf); |
4238 | 0 | xmlErrMemory(ctxt, NULL); |
4239 | 0 | ctxt->instate = (xmlParserInputState) state; |
4240 | 0 | return(NULL); |
4241 | 0 | } |
4242 | 1.92k | buf = tmp; |
4243 | 1.92k | } |
4244 | 29.4M | count++; |
4245 | 29.4M | if (count > 50) { |
4246 | 577k | SHRINK; |
4247 | 577k | GROW; |
4248 | 577k | count = 0; |
4249 | 577k | if (ctxt->instate == XML_PARSER_EOF) { |
4250 | 0 | xmlFree(buf); |
4251 | 0 | return(NULL); |
4252 | 0 | } |
4253 | 577k | } |
4254 | 29.4M | COPY_BUF(l,buf,len,cur); |
4255 | 29.4M | NEXTL(l); |
4256 | 29.4M | cur = CUR_CHAR(l); |
4257 | 29.4M | if (cur == 0) { |
4258 | 169 | GROW; |
4259 | 169 | SHRINK; |
4260 | 169 | cur = CUR_CHAR(l); |
4261 | 169 | } |
4262 | 29.4M | if (len > maxLength) { |
4263 | 2 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4264 | 2 | xmlFree(buf); |
4265 | 2 | ctxt->instate = (xmlParserInputState) state; |
4266 | 2 | return(NULL); |
4267 | 2 | } |
4268 | 29.4M | } |
4269 | 3.00k | buf[len] = 0; |
4270 | 3.00k | ctxt->instate = (xmlParserInputState) state; |
4271 | 3.00k | if (!IS_CHAR(cur)) { |
4272 | 244 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4273 | 2.75k | } else { |
4274 | 2.75k | NEXT; |
4275 | 2.75k | } |
4276 | 3.00k | return(buf); |
4277 | 3.00k | } |
4278 | | |
4279 | | /** |
4280 | | * xmlParsePubidLiteral: |
4281 | | * @ctxt: an XML parser context |
4282 | | * |
4283 | | * parse an XML public literal |
4284 | | * |
4285 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4286 | | * |
4287 | | * Returns the PubidLiteral parsed or NULL. |
4288 | | */ |
4289 | | |
4290 | | xmlChar * |
4291 | 3.80k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4292 | 3.80k | xmlChar *buf = NULL; |
4293 | 3.80k | int len = 0; |
4294 | 3.80k | int size = XML_PARSER_BUFFER_SIZE; |
4295 | 3.80k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4296 | 3.80k | XML_MAX_TEXT_LENGTH : |
4297 | 3.80k | XML_MAX_NAME_LENGTH; |
4298 | 3.80k | xmlChar cur; |
4299 | 3.80k | xmlChar stop; |
4300 | 3.80k | int count = 0; |
4301 | 3.80k | xmlParserInputState oldstate = ctxt->instate; |
4302 | | |
4303 | 3.80k | SHRINK; |
4304 | 3.80k | if (RAW == '"') { |
4305 | 2.17k | NEXT; |
4306 | 2.17k | stop = '"'; |
4307 | 2.17k | } else if (RAW == '\'') { |
4308 | 1.04k | NEXT; |
4309 | 1.04k | stop = '\''; |
4310 | 1.04k | } else { |
4311 | 586 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4312 | 586 | return(NULL); |
4313 | 586 | } |
4314 | 3.22k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4315 | 3.22k | if (buf == NULL) { |
4316 | 0 | xmlErrMemory(ctxt, NULL); |
4317 | 0 | return(NULL); |
4318 | 0 | } |
4319 | 3.22k | ctxt->instate = XML_PARSER_PUBLIC_LITERAL; |
4320 | 3.22k | cur = CUR; |
4321 | 60.4k | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop)) { /* checked */ |
4322 | 57.1k | if (len + 1 >= size) { |
4323 | 194 | xmlChar *tmp; |
4324 | | |
4325 | 194 | size *= 2; |
4326 | 194 | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
4327 | 194 | if (tmp == NULL) { |
4328 | 0 | xmlErrMemory(ctxt, NULL); |
4329 | 0 | xmlFree(buf); |
4330 | 0 | return(NULL); |
4331 | 0 | } |
4332 | 194 | buf = tmp; |
4333 | 194 | } |
4334 | 57.1k | buf[len++] = cur; |
4335 | 57.1k | count++; |
4336 | 57.1k | if (count > 50) { |
4337 | 937 | SHRINK; |
4338 | 937 | GROW; |
4339 | 937 | count = 0; |
4340 | 937 | if (ctxt->instate == XML_PARSER_EOF) { |
4341 | 0 | xmlFree(buf); |
4342 | 0 | return(NULL); |
4343 | 0 | } |
4344 | 937 | } |
4345 | 57.1k | NEXT; |
4346 | 57.1k | cur = CUR; |
4347 | 57.1k | if (cur == 0) { |
4348 | 86 | GROW; |
4349 | 86 | SHRINK; |
4350 | 86 | cur = CUR; |
4351 | 86 | } |
4352 | 57.1k | if (len > maxLength) { |
4353 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4354 | 0 | xmlFree(buf); |
4355 | 0 | return(NULL); |
4356 | 0 | } |
4357 | 57.1k | } |
4358 | 3.22k | buf[len] = 0; |
4359 | 3.22k | if (cur != stop) { |
4360 | 720 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4361 | 2.50k | } else { |
4362 | 2.50k | NEXT; |
4363 | 2.50k | } |
4364 | 3.22k | ctxt->instate = oldstate; |
4365 | 3.22k | return(buf); |
4366 | 3.22k | } |
4367 | | |
4368 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata); |
4369 | | |
4370 | | /* |
4371 | | * used for the test in the inner loop of the char data testing |
4372 | | */ |
4373 | | static const unsigned char test_char_data[256] = { |
4374 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4375 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4376 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4377 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4378 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4379 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4380 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4381 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4382 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4383 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4384 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4385 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4386 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4387 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4388 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4389 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4390 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
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 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4398 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4399 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4400 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4401 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4402 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4403 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4404 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4405 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4406 | | }; |
4407 | | |
4408 | | /** |
4409 | | * xmlParseCharData: |
4410 | | * @ctxt: an XML parser context |
4411 | | * @cdata: int indicating whether we are within a CDATA section |
4412 | | * |
4413 | | * parse a CharData section. |
4414 | | * if we are within a CDATA section ']]>' marks an end of section. |
4415 | | * |
4416 | | * The right angle bracket (>) may be represented using the string ">", |
4417 | | * and must, for compatibility, be escaped using ">" or a character |
4418 | | * reference when it appears in the string "]]>" in content, when that |
4419 | | * string is not marking the end of a CDATA section. |
4420 | | * |
4421 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4422 | | */ |
4423 | | |
4424 | | void |
4425 | 356k | xmlParseCharData(xmlParserCtxtPtr ctxt, int cdata) { |
4426 | 356k | const xmlChar *in; |
4427 | 356k | int nbchar = 0; |
4428 | 356k | int line = ctxt->input->line; |
4429 | 356k | int col = ctxt->input->col; |
4430 | 356k | int ccol; |
4431 | | |
4432 | 356k | SHRINK; |
4433 | 356k | GROW; |
4434 | | /* |
4435 | | * Accelerated common case where input don't need to be |
4436 | | * modified before passing it to the handler. |
4437 | | */ |
4438 | 356k | if (!cdata) { |
4439 | 356k | in = ctxt->input->cur; |
4440 | 372k | do { |
4441 | 466k | get_more_space: |
4442 | 478k | while (*in == 0x20) { in++; ctxt->input->col++; } |
4443 | 466k | if (*in == 0xA) { |
4444 | 130k | do { |
4445 | 130k | ctxt->input->line++; ctxt->input->col = 1; |
4446 | 130k | in++; |
4447 | 130k | } while (*in == 0xA); |
4448 | 93.2k | goto get_more_space; |
4449 | 93.2k | } |
4450 | 372k | if (*in == '<') { |
4451 | 74.8k | nbchar = in - ctxt->input->cur; |
4452 | 74.8k | if (nbchar > 0) { |
4453 | 74.8k | const xmlChar *tmp = ctxt->input->cur; |
4454 | 74.8k | ctxt->input->cur = in; |
4455 | | |
4456 | 74.8k | if ((ctxt->sax != NULL) && |
4457 | 74.8k | (ctxt->sax->ignorableWhitespace != |
4458 | 74.8k | ctxt->sax->characters)) { |
4459 | 74.8k | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4460 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4461 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4462 | 0 | tmp, nbchar); |
4463 | 74.8k | } else { |
4464 | 74.8k | if (ctxt->sax->characters != NULL) |
4465 | 74.8k | ctxt->sax->characters(ctxt->userData, |
4466 | 74.8k | tmp, nbchar); |
4467 | 74.8k | if (*ctxt->space == -1) |
4468 | 64.1k | *ctxt->space = -2; |
4469 | 74.8k | } |
4470 | 74.8k | } else if ((ctxt->sax != NULL) && |
4471 | 0 | (ctxt->sax->characters != NULL)) { |
4472 | 0 | ctxt->sax->characters(ctxt->userData, |
4473 | 0 | tmp, nbchar); |
4474 | 0 | } |
4475 | 74.8k | } |
4476 | 74.8k | return; |
4477 | 74.8k | } |
4478 | | |
4479 | 465k | get_more: |
4480 | 465k | ccol = ctxt->input->col; |
4481 | 4.24M | while (test_char_data[*in]) { |
4482 | 3.78M | in++; |
4483 | 3.78M | ccol++; |
4484 | 3.78M | } |
4485 | 465k | ctxt->input->col = ccol; |
4486 | 465k | if (*in == 0xA) { |
4487 | 101k | do { |
4488 | 101k | ctxt->input->line++; ctxt->input->col = 1; |
4489 | 101k | in++; |
4490 | 101k | } while (*in == 0xA); |
4491 | 100k | goto get_more; |
4492 | 100k | } |
4493 | 365k | if (*in == ']') { |
4494 | 67.1k | if ((in[1] == ']') && (in[2] == '>')) { |
4495 | 1 | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4496 | 1 | ctxt->input->cur = in + 1; |
4497 | 1 | return; |
4498 | 1 | } |
4499 | 67.1k | in++; |
4500 | 67.1k | ctxt->input->col++; |
4501 | 67.1k | goto get_more; |
4502 | 67.1k | } |
4503 | 297k | nbchar = in - ctxt->input->cur; |
4504 | 297k | if (nbchar > 0) { |
4505 | 247k | if ((ctxt->sax != NULL) && |
4506 | 247k | (ctxt->sax->ignorableWhitespace != |
4507 | 247k | ctxt->sax->characters) && |
4508 | 247k | (IS_BLANK_CH(*ctxt->input->cur))) { |
4509 | 24.9k | const xmlChar *tmp = ctxt->input->cur; |
4510 | 24.9k | ctxt->input->cur = in; |
4511 | | |
4512 | 24.9k | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4513 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4514 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4515 | 0 | tmp, nbchar); |
4516 | 24.9k | } else { |
4517 | 24.9k | if (ctxt->sax->characters != NULL) |
4518 | 24.9k | ctxt->sax->characters(ctxt->userData, |
4519 | 24.9k | tmp, nbchar); |
4520 | 24.9k | if (*ctxt->space == -1) |
4521 | 13.2k | *ctxt->space = -2; |
4522 | 24.9k | } |
4523 | 24.9k | line = ctxt->input->line; |
4524 | 24.9k | col = ctxt->input->col; |
4525 | 222k | } else if (ctxt->sax != NULL) { |
4526 | 222k | if (ctxt->sax->characters != NULL) |
4527 | 222k | ctxt->sax->characters(ctxt->userData, |
4528 | 222k | ctxt->input->cur, nbchar); |
4529 | 222k | line = ctxt->input->line; |
4530 | 222k | col = ctxt->input->col; |
4531 | 222k | } |
4532 | | /* something really bad happened in the SAX callback */ |
4533 | 247k | if (ctxt->instate != XML_PARSER_CONTENT) |
4534 | 0 | return; |
4535 | 247k | } |
4536 | 297k | ctxt->input->cur = in; |
4537 | 297k | if (*in == 0xD) { |
4538 | 47.1k | in++; |
4539 | 47.1k | if (*in == 0xA) { |
4540 | 16.0k | ctxt->input->cur = in; |
4541 | 16.0k | in++; |
4542 | 16.0k | ctxt->input->line++; ctxt->input->col = 1; |
4543 | 16.0k | continue; /* while */ |
4544 | 16.0k | } |
4545 | 31.0k | in--; |
4546 | 31.0k | } |
4547 | 281k | if (*in == '<') { |
4548 | 238k | return; |
4549 | 238k | } |
4550 | 43.9k | if (*in == '&') { |
4551 | 890 | return; |
4552 | 890 | } |
4553 | 43.0k | SHRINK; |
4554 | 43.0k | GROW; |
4555 | 43.0k | if (ctxt->instate == XML_PARSER_EOF) |
4556 | 0 | return; |
4557 | 43.0k | in = ctxt->input->cur; |
4558 | 59.0k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
4559 | 43.0k | nbchar = 0; |
4560 | 43.0k | } |
4561 | 43.0k | ctxt->input->line = line; |
4562 | 43.0k | ctxt->input->col = col; |
4563 | 43.0k | xmlParseCharDataComplex(ctxt, cdata); |
4564 | 43.0k | } |
4565 | | |
4566 | | /** |
4567 | | * xmlParseCharDataComplex: |
4568 | | * @ctxt: an XML parser context |
4569 | | * @cdata: int indicating whether we are within a CDATA section |
4570 | | * |
4571 | | * parse a CharData section.this is the fallback function |
4572 | | * of xmlParseCharData() when the parsing requires handling |
4573 | | * of non-ASCII characters. |
4574 | | */ |
4575 | | static void |
4576 | 43.0k | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int cdata) { |
4577 | 43.0k | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4578 | 43.0k | int nbchar = 0; |
4579 | 43.0k | int cur, l; |
4580 | 43.0k | int count = 0; |
4581 | | |
4582 | 43.0k | SHRINK; |
4583 | 43.0k | GROW; |
4584 | 43.0k | cur = CUR_CHAR(l); |
4585 | 8.31M | while ((cur != '<') && /* checked */ |
4586 | 8.31M | (cur != '&') && |
4587 | 8.31M | (IS_CHAR(cur))) /* test also done in xmlCurrentChar() */ { |
4588 | 8.27M | if ((cur == ']') && (NXT(1) == ']') && |
4589 | 8.27M | (NXT(2) == '>')) { |
4590 | 223 | if (cdata) break; |
4591 | 223 | else { |
4592 | 223 | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4593 | 223 | } |
4594 | 223 | } |
4595 | 8.27M | COPY_BUF(l,buf,nbchar,cur); |
4596 | | /* move current position before possible calling of ctxt->sax->characters */ |
4597 | 8.27M | NEXTL(l); |
4598 | 8.27M | cur = CUR_CHAR(l); |
4599 | 8.27M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4600 | 68.2k | buf[nbchar] = 0; |
4601 | | |
4602 | | /* |
4603 | | * OK the segment is to be consumed as chars. |
4604 | | */ |
4605 | 68.2k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4606 | 68.0k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4607 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4608 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4609 | 0 | buf, nbchar); |
4610 | 68.0k | } else { |
4611 | 68.0k | if (ctxt->sax->characters != NULL) |
4612 | 68.0k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4613 | 68.0k | if ((ctxt->sax->characters != |
4614 | 68.0k | ctxt->sax->ignorableWhitespace) && |
4615 | 68.0k | (*ctxt->space == -1)) |
4616 | 1.24k | *ctxt->space = -2; |
4617 | 68.0k | } |
4618 | 68.0k | } |
4619 | 68.2k | nbchar = 0; |
4620 | | /* something really bad happened in the SAX callback */ |
4621 | 68.2k | if (ctxt->instate != XML_PARSER_CONTENT) |
4622 | 0 | return; |
4623 | 68.2k | } |
4624 | 8.27M | count++; |
4625 | 8.27M | if (count > 50) { |
4626 | 153k | SHRINK; |
4627 | 153k | GROW; |
4628 | 153k | count = 0; |
4629 | 153k | if (ctxt->instate == XML_PARSER_EOF) |
4630 | 0 | return; |
4631 | 153k | } |
4632 | 8.27M | } |
4633 | 43.0k | if (nbchar != 0) { |
4634 | 42.1k | buf[nbchar] = 0; |
4635 | | /* |
4636 | | * OK the segment is to be consumed as chars. |
4637 | | */ |
4638 | 42.1k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4639 | 42.0k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4640 | 0 | if (ctxt->sax->ignorableWhitespace != NULL) |
4641 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4642 | 42.0k | } else { |
4643 | 42.0k | if (ctxt->sax->characters != NULL) |
4644 | 42.0k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4645 | 42.0k | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4646 | 42.0k | (*ctxt->space == -1)) |
4647 | 33.4k | *ctxt->space = -2; |
4648 | 42.0k | } |
4649 | 42.0k | } |
4650 | 42.1k | } |
4651 | 43.0k | if ((cur != 0) && (!IS_CHAR(cur))) { |
4652 | | /* Generate the error and skip the offending character */ |
4653 | 77 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4654 | 77 | "PCDATA invalid Char value %d\n", |
4655 | 77 | cur); |
4656 | 77 | NEXTL(l); |
4657 | 77 | } |
4658 | 43.0k | } |
4659 | | |
4660 | | /** |
4661 | | * xmlParseExternalID: |
4662 | | * @ctxt: an XML parser context |
4663 | | * @publicID: a xmlChar** receiving PubidLiteral |
4664 | | * @strict: indicate whether we should restrict parsing to only |
4665 | | * production [75], see NOTE below |
4666 | | * |
4667 | | * Parse an External ID or a Public ID |
4668 | | * |
4669 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4670 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
4671 | | * |
4672 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4673 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4674 | | * |
4675 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4676 | | * |
4677 | | * Returns the function returns SystemLiteral and in the second |
4678 | | * case publicID receives PubidLiteral, is strict is off |
4679 | | * it is possible to return NULL and have publicID set. |
4680 | | */ |
4681 | | |
4682 | | xmlChar * |
4683 | 11.8k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
4684 | 11.8k | xmlChar *URI = NULL; |
4685 | | |
4686 | 11.8k | SHRINK; |
4687 | | |
4688 | 11.8k | *publicID = NULL; |
4689 | 11.8k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4690 | 1.34k | SKIP(6); |
4691 | 1.34k | if (SKIP_BLANKS == 0) { |
4692 | 322 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4693 | 322 | "Space required after 'SYSTEM'\n"); |
4694 | 322 | } |
4695 | 1.34k | URI = xmlParseSystemLiteral(ctxt); |
4696 | 1.34k | if (URI == NULL) { |
4697 | 85 | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4698 | 85 | } |
4699 | 10.5k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4700 | 3.80k | SKIP(6); |
4701 | 3.80k | if (SKIP_BLANKS == 0) { |
4702 | 2.12k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4703 | 2.12k | "Space required after 'PUBLIC'\n"); |
4704 | 2.12k | } |
4705 | 3.80k | *publicID = xmlParsePubidLiteral(ctxt); |
4706 | 3.80k | if (*publicID == NULL) { |
4707 | 586 | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4708 | 586 | } |
4709 | 3.80k | if (strict) { |
4710 | | /* |
4711 | | * We don't handle [83] so "S SystemLiteral" is required. |
4712 | | */ |
4713 | 3.08k | if (SKIP_BLANKS == 0) { |
4714 | 2.15k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4715 | 2.15k | "Space required after the Public Identifier\n"); |
4716 | 2.15k | } |
4717 | 3.08k | } else { |
4718 | | /* |
4719 | | * We handle [83] so we return immediately, if |
4720 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4721 | | * system literal was found, but this is harmless since we must |
4722 | | * be at the end of a NotationDecl. |
4723 | | */ |
4724 | 722 | if (SKIP_BLANKS == 0) return(NULL); |
4725 | 531 | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4726 | 531 | } |
4727 | 3.36k | URI = xmlParseSystemLiteral(ctxt); |
4728 | 3.36k | if (URI == NULL) { |
4729 | 1.62k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4730 | 1.62k | } |
4731 | 3.36k | } |
4732 | 11.4k | return(URI); |
4733 | 11.8k | } |
4734 | | |
4735 | | /** |
4736 | | * xmlParseCommentComplex: |
4737 | | * @ctxt: an XML parser context |
4738 | | * @buf: the already parsed part of the buffer |
4739 | | * @len: number of bytes in the buffer |
4740 | | * @size: allocated size of the buffer |
4741 | | * |
4742 | | * Skip an XML (SGML) comment <!-- .... --> |
4743 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4744 | | * must not occur within comments. " |
4745 | | * This is the slow routine in case the accelerator for ascii didn't work |
4746 | | * |
4747 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4748 | | */ |
4749 | | static void |
4750 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4751 | 5.69k | size_t len, size_t size) { |
4752 | 5.69k | int q, ql; |
4753 | 5.69k | int r, rl; |
4754 | 5.69k | int cur, l; |
4755 | 5.69k | size_t count = 0; |
4756 | 5.69k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4757 | 5.69k | XML_MAX_HUGE_LENGTH : |
4758 | 5.69k | XML_MAX_TEXT_LENGTH; |
4759 | 5.69k | int inputid; |
4760 | | |
4761 | 5.69k | inputid = ctxt->input->id; |
4762 | | |
4763 | 5.69k | if (buf == NULL) { |
4764 | 5.69k | len = 0; |
4765 | 5.69k | size = XML_PARSER_BUFFER_SIZE; |
4766 | 5.69k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4767 | 5.69k | if (buf == NULL) { |
4768 | 0 | xmlErrMemory(ctxt, NULL); |
4769 | 0 | return; |
4770 | 0 | } |
4771 | 5.69k | } |
4772 | 5.69k | GROW; /* Assure there's enough input data */ |
4773 | 5.69k | q = CUR_CHAR(ql); |
4774 | 5.69k | if (q == 0) |
4775 | 185 | goto not_terminated; |
4776 | 5.50k | if (!IS_CHAR(q)) { |
4777 | 47 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4778 | 47 | "xmlParseComment: invalid xmlChar value %d\n", |
4779 | 47 | q); |
4780 | 47 | xmlFree (buf); |
4781 | 47 | return; |
4782 | 47 | } |
4783 | 5.46k | NEXTL(ql); |
4784 | 5.46k | r = CUR_CHAR(rl); |
4785 | 5.46k | if (r == 0) |
4786 | 68 | goto not_terminated; |
4787 | 5.39k | if (!IS_CHAR(r)) { |
4788 | 23 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4789 | 23 | "xmlParseComment: invalid xmlChar value %d\n", |
4790 | 23 | q); |
4791 | 23 | xmlFree (buf); |
4792 | 23 | return; |
4793 | 23 | } |
4794 | 5.37k | NEXTL(rl); |
4795 | 5.37k | cur = CUR_CHAR(l); |
4796 | 5.37k | if (cur == 0) |
4797 | 85 | goto not_terminated; |
4798 | 35.6M | while (IS_CHAR(cur) && /* checked */ |
4799 | 35.6M | ((cur != '>') || |
4800 | 35.6M | (r != '-') || (q != '-'))) { |
4801 | 35.6M | if ((r == '-') && (q == '-')) { |
4802 | 1.52k | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
4803 | 1.52k | } |
4804 | 35.6M | if (len + 5 >= size) { |
4805 | 5.05k | xmlChar *new_buf; |
4806 | 5.05k | size_t new_size; |
4807 | | |
4808 | 5.05k | new_size = size * 2; |
4809 | 5.05k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
4810 | 5.05k | if (new_buf == NULL) { |
4811 | 0 | xmlFree (buf); |
4812 | 0 | xmlErrMemory(ctxt, NULL); |
4813 | 0 | return; |
4814 | 0 | } |
4815 | 5.05k | buf = new_buf; |
4816 | 5.05k | size = new_size; |
4817 | 5.05k | } |
4818 | 35.6M | COPY_BUF(ql,buf,len,q); |
4819 | 35.6M | q = r; |
4820 | 35.6M | ql = rl; |
4821 | 35.6M | r = cur; |
4822 | 35.6M | rl = l; |
4823 | | |
4824 | 35.6M | count++; |
4825 | 35.6M | if (count > 50) { |
4826 | 698k | SHRINK; |
4827 | 698k | GROW; |
4828 | 698k | count = 0; |
4829 | 698k | if (ctxt->instate == XML_PARSER_EOF) { |
4830 | 0 | xmlFree(buf); |
4831 | 0 | return; |
4832 | 0 | } |
4833 | 698k | } |
4834 | 35.6M | NEXTL(l); |
4835 | 35.6M | cur = CUR_CHAR(l); |
4836 | 35.6M | if (cur == 0) { |
4837 | 439 | SHRINK; |
4838 | 439 | GROW; |
4839 | 439 | cur = CUR_CHAR(l); |
4840 | 439 | } |
4841 | | |
4842 | 35.6M | if (len > maxLength) { |
4843 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4844 | 0 | "Comment too big found", NULL); |
4845 | 0 | xmlFree (buf); |
4846 | 0 | return; |
4847 | 0 | } |
4848 | 35.6M | } |
4849 | 5.28k | buf[len] = 0; |
4850 | 5.28k | if (cur == 0) { |
4851 | 439 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4852 | 439 | "Comment not terminated \n<!--%.50s\n", buf); |
4853 | 4.84k | } else if (!IS_CHAR(cur)) { |
4854 | 98 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4855 | 98 | "xmlParseComment: invalid xmlChar value %d\n", |
4856 | 98 | cur); |
4857 | 4.74k | } else { |
4858 | 4.74k | if (inputid != ctxt->input->id) { |
4859 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
4860 | 0 | "Comment doesn't start and stop in the same" |
4861 | 0 | " entity\n"); |
4862 | 0 | } |
4863 | 4.74k | NEXT; |
4864 | 4.74k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
4865 | 4.74k | (!ctxt->disableSAX)) |
4866 | 0 | ctxt->sax->comment(ctxt->userData, buf); |
4867 | 4.74k | } |
4868 | 5.28k | xmlFree(buf); |
4869 | 5.28k | return; |
4870 | 338 | not_terminated: |
4871 | 338 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4872 | 338 | "Comment not terminated\n", NULL); |
4873 | 338 | xmlFree(buf); |
4874 | 338 | return; |
4875 | 5.28k | } |
4876 | | |
4877 | | /** |
4878 | | * xmlParseComment: |
4879 | | * @ctxt: an XML parser context |
4880 | | * |
4881 | | * Skip an XML (SGML) comment <!-- .... --> |
4882 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4883 | | * must not occur within comments. " |
4884 | | * |
4885 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4886 | | */ |
4887 | | void |
4888 | 12.0k | xmlParseComment(xmlParserCtxtPtr ctxt) { |
4889 | 12.0k | xmlChar *buf = NULL; |
4890 | 12.0k | size_t size = XML_PARSER_BUFFER_SIZE; |
4891 | 12.0k | size_t len = 0; |
4892 | 12.0k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4893 | 12.0k | XML_MAX_HUGE_LENGTH : |
4894 | 12.0k | XML_MAX_TEXT_LENGTH; |
4895 | 12.0k | xmlParserInputState state; |
4896 | 12.0k | const xmlChar *in; |
4897 | 12.0k | size_t nbchar = 0; |
4898 | 12.0k | int ccol; |
4899 | 12.0k | int inputid; |
4900 | | |
4901 | | /* |
4902 | | * Check that there is a comment right here. |
4903 | | */ |
4904 | 12.0k | if ((RAW != '<') || (NXT(1) != '!') || |
4905 | 12.0k | (NXT(2) != '-') || (NXT(3) != '-')) return; |
4906 | 11.9k | state = ctxt->instate; |
4907 | 11.9k | ctxt->instate = XML_PARSER_COMMENT; |
4908 | 11.9k | inputid = ctxt->input->id; |
4909 | 11.9k | SKIP(4); |
4910 | 11.9k | SHRINK; |
4911 | 11.9k | GROW; |
4912 | | |
4913 | | /* |
4914 | | * Accelerated common case where input don't need to be |
4915 | | * modified before passing it to the handler. |
4916 | | */ |
4917 | 11.9k | in = ctxt->input->cur; |
4918 | 11.9k | do { |
4919 | 11.9k | if (*in == 0xA) { |
4920 | 21.3k | do { |
4921 | 21.3k | ctxt->input->line++; ctxt->input->col = 1; |
4922 | 21.3k | in++; |
4923 | 21.3k | } while (*in == 0xA); |
4924 | 5.40k | } |
4925 | 27.7k | get_more: |
4926 | 27.7k | ccol = ctxt->input->col; |
4927 | 789k | while (((*in > '-') && (*in <= 0x7F)) || |
4928 | 789k | ((*in >= 0x20) && (*in < '-')) || |
4929 | 789k | (*in == 0x09)) { |
4930 | 762k | in++; |
4931 | 762k | ccol++; |
4932 | 762k | } |
4933 | 27.7k | ctxt->input->col = ccol; |
4934 | 27.7k | if (*in == 0xA) { |
4935 | 22.9k | do { |
4936 | 22.9k | ctxt->input->line++; ctxt->input->col = 1; |
4937 | 22.9k | in++; |
4938 | 22.9k | } while (*in == 0xA); |
4939 | 8.42k | goto get_more; |
4940 | 8.42k | } |
4941 | 19.3k | nbchar = in - ctxt->input->cur; |
4942 | | /* |
4943 | | * save current set of data |
4944 | | */ |
4945 | 19.3k | if (nbchar > 0) { |
4946 | 16.2k | if ((ctxt->sax != NULL) && |
4947 | 16.2k | (ctxt->sax->comment != NULL)) { |
4948 | 0 | if (buf == NULL) { |
4949 | 0 | if ((*in == '-') && (in[1] == '-')) |
4950 | 0 | size = nbchar + 1; |
4951 | 0 | else |
4952 | 0 | size = XML_PARSER_BUFFER_SIZE + nbchar; |
4953 | 0 | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
4954 | 0 | if (buf == NULL) { |
4955 | 0 | xmlErrMemory(ctxt, NULL); |
4956 | 0 | ctxt->instate = state; |
4957 | 0 | return; |
4958 | 0 | } |
4959 | 0 | len = 0; |
4960 | 0 | } else if (len + nbchar + 1 >= size) { |
4961 | 0 | xmlChar *new_buf; |
4962 | 0 | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
4963 | 0 | new_buf = (xmlChar *) xmlRealloc(buf, |
4964 | 0 | size * sizeof(xmlChar)); |
4965 | 0 | if (new_buf == NULL) { |
4966 | 0 | xmlFree (buf); |
4967 | 0 | xmlErrMemory(ctxt, NULL); |
4968 | 0 | ctxt->instate = state; |
4969 | 0 | return; |
4970 | 0 | } |
4971 | 0 | buf = new_buf; |
4972 | 0 | } |
4973 | 0 | memcpy(&buf[len], ctxt->input->cur, nbchar); |
4974 | 0 | len += nbchar; |
4975 | 0 | buf[len] = 0; |
4976 | 0 | } |
4977 | 16.2k | } |
4978 | 19.3k | if (len > maxLength) { |
4979 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4980 | 0 | "Comment too big found", NULL); |
4981 | 0 | xmlFree (buf); |
4982 | 0 | return; |
4983 | 0 | } |
4984 | 19.3k | ctxt->input->cur = in; |
4985 | 19.3k | if (*in == 0xA) { |
4986 | 0 | in++; |
4987 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
4988 | 0 | } |
4989 | 19.3k | if (*in == 0xD) { |
4990 | 1.40k | in++; |
4991 | 1.40k | if (*in == 0xA) { |
4992 | 290 | ctxt->input->cur = in; |
4993 | 290 | in++; |
4994 | 290 | ctxt->input->line++; ctxt->input->col = 1; |
4995 | 290 | goto get_more; |
4996 | 290 | } |
4997 | 1.11k | in--; |
4998 | 1.11k | } |
4999 | 19.0k | SHRINK; |
5000 | 19.0k | GROW; |
5001 | 19.0k | if (ctxt->instate == XML_PARSER_EOF) { |
5002 | 0 | xmlFree(buf); |
5003 | 0 | return; |
5004 | 0 | } |
5005 | 19.0k | in = ctxt->input->cur; |
5006 | 19.0k | if (*in == '-') { |
5007 | 13.3k | if (in[1] == '-') { |
5008 | 7.68k | if (in[2] == '>') { |
5009 | 6.30k | if (ctxt->input->id != inputid) { |
5010 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5011 | 0 | "comment doesn't start and stop in the" |
5012 | 0 | " same entity\n"); |
5013 | 0 | } |
5014 | 6.30k | SKIP(3); |
5015 | 6.30k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5016 | 6.30k | (!ctxt->disableSAX)) { |
5017 | 0 | if (buf != NULL) |
5018 | 0 | ctxt->sax->comment(ctxt->userData, buf); |
5019 | 0 | else |
5020 | 0 | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5021 | 0 | } |
5022 | 6.30k | if (buf != NULL) |
5023 | 0 | xmlFree(buf); |
5024 | 6.30k | if (ctxt->instate != XML_PARSER_EOF) |
5025 | 6.30k | ctxt->instate = state; |
5026 | 6.30k | return; |
5027 | 6.30k | } |
5028 | 1.38k | if (buf != NULL) { |
5029 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5030 | 0 | "Double hyphen within comment: " |
5031 | 0 | "<!--%.50s\n", |
5032 | 0 | buf); |
5033 | 0 | } else |
5034 | 1.38k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5035 | 1.38k | "Double hyphen within comment\n", NULL); |
5036 | 1.38k | if (ctxt->instate == XML_PARSER_EOF) { |
5037 | 0 | xmlFree(buf); |
5038 | 0 | return; |
5039 | 0 | } |
5040 | 1.38k | in++; |
5041 | 1.38k | ctxt->input->col++; |
5042 | 1.38k | } |
5043 | 7.02k | in++; |
5044 | 7.02k | ctxt->input->col++; |
5045 | 7.02k | goto get_more; |
5046 | 13.3k | } |
5047 | 19.0k | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5048 | 5.69k | xmlParseCommentComplex(ctxt, buf, len, size); |
5049 | 5.69k | ctxt->instate = state; |
5050 | 5.69k | return; |
5051 | 11.9k | } |
5052 | | |
5053 | | |
5054 | | /** |
5055 | | * xmlParsePITarget: |
5056 | | * @ctxt: an XML parser context |
5057 | | * |
5058 | | * parse the name of a PI |
5059 | | * |
5060 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5061 | | * |
5062 | | * Returns the PITarget name or NULL |
5063 | | */ |
5064 | | |
5065 | | const xmlChar * |
5066 | 21.8k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5067 | 21.8k | const xmlChar *name; |
5068 | | |
5069 | 21.8k | name = xmlParseName(ctxt); |
5070 | 21.8k | if ((name != NULL) && |
5071 | 21.8k | ((name[0] == 'x') || (name[0] == 'X')) && |
5072 | 21.8k | ((name[1] == 'm') || (name[1] == 'M')) && |
5073 | 21.8k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5074 | 2.42k | int i; |
5075 | 2.42k | if ((name[0] == 'x') && (name[1] == 'm') && |
5076 | 2.42k | (name[2] == 'l') && (name[3] == 0)) { |
5077 | 672 | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5078 | 672 | "XML declaration allowed only at the start of the document\n"); |
5079 | 672 | return(name); |
5080 | 1.75k | } else if (name[3] == 0) { |
5081 | 720 | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5082 | 720 | return(name); |
5083 | 720 | } |
5084 | 2.96k | for (i = 0;;i++) { |
5085 | 2.96k | if (xmlW3CPIs[i] == NULL) break; |
5086 | 2.00k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5087 | 68 | return(name); |
5088 | 2.00k | } |
5089 | 965 | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5090 | 965 | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5091 | 965 | NULL, NULL); |
5092 | 965 | } |
5093 | 20.3k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5094 | 872 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5095 | 872 | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5096 | 872 | } |
5097 | 20.3k | return(name); |
5098 | 21.8k | } |
5099 | | |
5100 | | #ifdef LIBXML_CATALOG_ENABLED |
5101 | | /** |
5102 | | * xmlParseCatalogPI: |
5103 | | * @ctxt: an XML parser context |
5104 | | * @catalog: the PI value string |
5105 | | * |
5106 | | * parse an XML Catalog Processing Instruction. |
5107 | | * |
5108 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5109 | | * |
5110 | | * Occurs only if allowed by the user and if happening in the Misc |
5111 | | * part of the document before any doctype information |
5112 | | * This will add the given catalog to the parsing context in order |
5113 | | * to be used if there is a resolution need further down in the document |
5114 | | */ |
5115 | | |
5116 | | static void |
5117 | 2.69k | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5118 | 2.69k | xmlChar *URL = NULL; |
5119 | 2.69k | const xmlChar *tmp, *base; |
5120 | 2.69k | xmlChar marker; |
5121 | | |
5122 | 2.69k | tmp = catalog; |
5123 | 2.69k | while (IS_BLANK_CH(*tmp)) tmp++; |
5124 | 2.69k | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5125 | 509 | goto error; |
5126 | 2.18k | tmp += 7; |
5127 | 2.18k | while (IS_BLANK_CH(*tmp)) tmp++; |
5128 | 2.18k | if (*tmp != '=') { |
5129 | 461 | return; |
5130 | 461 | } |
5131 | 1.72k | tmp++; |
5132 | 1.72k | while (IS_BLANK_CH(*tmp)) tmp++; |
5133 | 1.72k | marker = *tmp; |
5134 | 1.72k | if ((marker != '\'') && (marker != '"')) |
5135 | 351 | goto error; |
5136 | 1.37k | tmp++; |
5137 | 1.37k | base = tmp; |
5138 | 3.16k | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5139 | 1.37k | if (*tmp == 0) |
5140 | 224 | goto error; |
5141 | 1.14k | URL = xmlStrndup(base, tmp - base); |
5142 | 1.14k | tmp++; |
5143 | 1.14k | while (IS_BLANK_CH(*tmp)) tmp++; |
5144 | 1.14k | if (*tmp != 0) |
5145 | 347 | goto error; |
5146 | | |
5147 | 799 | if (URL != NULL) { |
5148 | 799 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5149 | 799 | xmlFree(URL); |
5150 | 799 | } |
5151 | 799 | return; |
5152 | | |
5153 | 1.43k | error: |
5154 | 1.43k | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5155 | 1.43k | "Catalog PI syntax error: %s\n", |
5156 | 1.43k | catalog, NULL); |
5157 | 1.43k | if (URL != NULL) |
5158 | 347 | xmlFree(URL); |
5159 | 1.43k | } |
5160 | | #endif |
5161 | | |
5162 | | /** |
5163 | | * xmlParsePI: |
5164 | | * @ctxt: an XML parser context |
5165 | | * |
5166 | | * parse an XML Processing Instruction. |
5167 | | * |
5168 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5169 | | * |
5170 | | * The processing is transferred to SAX once parsed. |
5171 | | */ |
5172 | | |
5173 | | void |
5174 | 21.8k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5175 | 21.8k | xmlChar *buf = NULL; |
5176 | 21.8k | size_t len = 0; |
5177 | 21.8k | size_t size = XML_PARSER_BUFFER_SIZE; |
5178 | 21.8k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5179 | 21.8k | XML_MAX_HUGE_LENGTH : |
5180 | 21.8k | XML_MAX_TEXT_LENGTH; |
5181 | 21.8k | int cur, l; |
5182 | 21.8k | const xmlChar *target; |
5183 | 21.8k | xmlParserInputState state; |
5184 | 21.8k | int count = 0; |
5185 | | |
5186 | 21.8k | if ((RAW == '<') && (NXT(1) == '?')) { |
5187 | 21.8k | int inputid = ctxt->input->id; |
5188 | 21.8k | state = ctxt->instate; |
5189 | 21.8k | ctxt->instate = XML_PARSER_PI; |
5190 | | /* |
5191 | | * this is a Processing Instruction. |
5192 | | */ |
5193 | 21.8k | SKIP(2); |
5194 | 21.8k | SHRINK; |
5195 | | |
5196 | | /* |
5197 | | * Parse the target name and check for special support like |
5198 | | * namespace. |
5199 | | */ |
5200 | 21.8k | target = xmlParsePITarget(ctxt); |
5201 | 21.8k | if (target != NULL) { |
5202 | 20.6k | if ((RAW == '?') && (NXT(1) == '>')) { |
5203 | 3.83k | if (inputid != ctxt->input->id) { |
5204 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5205 | 0 | "PI declaration doesn't start and stop in" |
5206 | 0 | " the same entity\n"); |
5207 | 0 | } |
5208 | 3.83k | SKIP(2); |
5209 | | |
5210 | | /* |
5211 | | * SAX: PI detected. |
5212 | | */ |
5213 | 3.83k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5214 | 3.83k | (ctxt->sax->processingInstruction != NULL)) |
5215 | 0 | ctxt->sax->processingInstruction(ctxt->userData, |
5216 | 0 | target, NULL); |
5217 | 3.83k | if (ctxt->instate != XML_PARSER_EOF) |
5218 | 3.83k | ctxt->instate = state; |
5219 | 3.83k | return; |
5220 | 3.83k | } |
5221 | 16.7k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
5222 | 16.7k | if (buf == NULL) { |
5223 | 0 | xmlErrMemory(ctxt, NULL); |
5224 | 0 | ctxt->instate = state; |
5225 | 0 | return; |
5226 | 0 | } |
5227 | 16.7k | if (SKIP_BLANKS == 0) { |
5228 | 2.90k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5229 | 2.90k | "ParsePI: PI %s space expected\n", target); |
5230 | 2.90k | } |
5231 | 16.7k | cur = CUR_CHAR(l); |
5232 | 46.6M | while (IS_CHAR(cur) && /* checked */ |
5233 | 46.6M | ((cur != '?') || (NXT(1) != '>'))) { |
5234 | 46.6M | if (len + 5 >= size) { |
5235 | 4.33k | xmlChar *tmp; |
5236 | 4.33k | size_t new_size = size * 2; |
5237 | 4.33k | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5238 | 4.33k | if (tmp == NULL) { |
5239 | 0 | xmlErrMemory(ctxt, NULL); |
5240 | 0 | xmlFree(buf); |
5241 | 0 | ctxt->instate = state; |
5242 | 0 | return; |
5243 | 0 | } |
5244 | 4.33k | buf = tmp; |
5245 | 4.33k | size = new_size; |
5246 | 4.33k | } |
5247 | 46.6M | count++; |
5248 | 46.6M | if (count > 50) { |
5249 | 912k | SHRINK; |
5250 | 912k | GROW; |
5251 | 912k | if (ctxt->instate == XML_PARSER_EOF) { |
5252 | 0 | xmlFree(buf); |
5253 | 0 | return; |
5254 | 0 | } |
5255 | 912k | count = 0; |
5256 | 912k | } |
5257 | 46.6M | COPY_BUF(l,buf,len,cur); |
5258 | 46.6M | NEXTL(l); |
5259 | 46.6M | cur = CUR_CHAR(l); |
5260 | 46.6M | if (cur == 0) { |
5261 | 461 | SHRINK; |
5262 | 461 | GROW; |
5263 | 461 | cur = CUR_CHAR(l); |
5264 | 461 | } |
5265 | 46.6M | if (len > maxLength) { |
5266 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5267 | 0 | "PI %s too big found", target); |
5268 | 0 | xmlFree(buf); |
5269 | 0 | ctxt->instate = state; |
5270 | 0 | return; |
5271 | 0 | } |
5272 | 46.6M | } |
5273 | 16.7k | buf[len] = 0; |
5274 | 16.7k | if (cur != '?') { |
5275 | 1.01k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5276 | 1.01k | "ParsePI: PI %s never end ...\n", target); |
5277 | 15.7k | } else { |
5278 | 15.7k | if (inputid != ctxt->input->id) { |
5279 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5280 | 0 | "PI declaration doesn't start and stop in" |
5281 | 0 | " the same entity\n"); |
5282 | 0 | } |
5283 | 15.7k | SKIP(2); |
5284 | | |
5285 | 15.7k | #ifdef LIBXML_CATALOG_ENABLED |
5286 | 15.7k | if (((state == XML_PARSER_MISC) || |
5287 | 15.7k | (state == XML_PARSER_START)) && |
5288 | 15.7k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5289 | 2.69k | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5290 | 2.69k | if ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5291 | 2.69k | (allow == XML_CATA_ALLOW_ALL)) |
5292 | 2.69k | xmlParseCatalogPI(ctxt, buf); |
5293 | 2.69k | } |
5294 | 15.7k | #endif |
5295 | | |
5296 | | |
5297 | | /* |
5298 | | * SAX: PI detected. |
5299 | | */ |
5300 | 15.7k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5301 | 15.7k | (ctxt->sax->processingInstruction != NULL)) |
5302 | 0 | ctxt->sax->processingInstruction(ctxt->userData, |
5303 | 0 | target, buf); |
5304 | 15.7k | } |
5305 | 16.7k | xmlFree(buf); |
5306 | 16.7k | } else { |
5307 | 1.21k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5308 | 1.21k | } |
5309 | 18.0k | if (ctxt->instate != XML_PARSER_EOF) |
5310 | 18.0k | ctxt->instate = state; |
5311 | 18.0k | } |
5312 | 21.8k | } |
5313 | | |
5314 | | /** |
5315 | | * xmlParseNotationDecl: |
5316 | | * @ctxt: an XML parser context |
5317 | | * |
5318 | | * parse a notation declaration |
5319 | | * |
5320 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5321 | | * |
5322 | | * Hence there is actually 3 choices: |
5323 | | * 'PUBLIC' S PubidLiteral |
5324 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5325 | | * and 'SYSTEM' S SystemLiteral |
5326 | | * |
5327 | | * See the NOTE on xmlParseExternalID(). |
5328 | | */ |
5329 | | |
5330 | | void |
5331 | 2.38k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5332 | 2.38k | const xmlChar *name; |
5333 | 2.38k | xmlChar *Pubid; |
5334 | 2.38k | xmlChar *Systemid; |
5335 | | |
5336 | 2.38k | if (CMP10(CUR_PTR, '<', '!', 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5337 | 2.34k | int inputid = ctxt->input->id; |
5338 | 2.34k | SHRINK; |
5339 | 2.34k | SKIP(10); |
5340 | 2.34k | if (SKIP_BLANKS == 0) { |
5341 | 107 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5342 | 107 | "Space required after '<!NOTATION'\n"); |
5343 | 107 | return; |
5344 | 107 | } |
5345 | | |
5346 | 2.23k | name = xmlParseName(ctxt); |
5347 | 2.23k | if (name == NULL) { |
5348 | 446 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5349 | 446 | return; |
5350 | 446 | } |
5351 | 1.79k | if (xmlStrchr(name, ':') != NULL) { |
5352 | 444 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5353 | 444 | "colons are forbidden from notation names '%s'\n", |
5354 | 444 | name, NULL, NULL); |
5355 | 444 | } |
5356 | 1.79k | if (SKIP_BLANKS == 0) { |
5357 | 777 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5358 | 777 | "Space required after the NOTATION name'\n"); |
5359 | 777 | return; |
5360 | 777 | } |
5361 | | |
5362 | | /* |
5363 | | * Parse the IDs. |
5364 | | */ |
5365 | 1.01k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5366 | 1.01k | SKIP_BLANKS; |
5367 | | |
5368 | 1.01k | if (RAW == '>') { |
5369 | 243 | if (inputid != ctxt->input->id) { |
5370 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5371 | 0 | "Notation declaration doesn't start and stop" |
5372 | 0 | " in the same entity\n"); |
5373 | 0 | } |
5374 | 243 | NEXT; |
5375 | 243 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5376 | 243 | (ctxt->sax->notationDecl != NULL)) |
5377 | 0 | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5378 | 771 | } else { |
5379 | 771 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5380 | 771 | } |
5381 | 1.01k | if (Systemid != NULL) xmlFree(Systemid); |
5382 | 1.01k | if (Pubid != NULL) xmlFree(Pubid); |
5383 | 1.01k | } |
5384 | 2.38k | } |
5385 | | |
5386 | | /** |
5387 | | * xmlParseEntityDecl: |
5388 | | * @ctxt: an XML parser context |
5389 | | * |
5390 | | * parse <!ENTITY declarations |
5391 | | * |
5392 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5393 | | * |
5394 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5395 | | * |
5396 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5397 | | * |
5398 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5399 | | * |
5400 | | * [74] PEDef ::= EntityValue | ExternalID |
5401 | | * |
5402 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5403 | | * |
5404 | | * [ VC: Notation Declared ] |
5405 | | * The Name must match the declared name of a notation. |
5406 | | */ |
5407 | | |
5408 | | void |
5409 | 16.1k | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5410 | 16.1k | const xmlChar *name = NULL; |
5411 | 16.1k | xmlChar *value = NULL; |
5412 | 16.1k | xmlChar *URI = NULL, *literal = NULL; |
5413 | 16.1k | const xmlChar *ndata = NULL; |
5414 | 16.1k | int isParameter = 0; |
5415 | 16.1k | xmlChar *orig = NULL; |
5416 | | |
5417 | | /* GROW; done in the caller */ |
5418 | 16.1k | if (CMP8(CUR_PTR, '<', '!', 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5419 | 16.0k | int inputid = ctxt->input->id; |
5420 | 16.0k | SHRINK; |
5421 | 16.0k | SKIP(8); |
5422 | 16.0k | if (SKIP_BLANKS == 0) { |
5423 | 10.4k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5424 | 10.4k | "Space required after '<!ENTITY'\n"); |
5425 | 10.4k | } |
5426 | | |
5427 | 16.0k | if (RAW == '%') { |
5428 | 2.93k | NEXT; |
5429 | 2.93k | if (SKIP_BLANKS == 0) { |
5430 | 1.99k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5431 | 1.99k | "Space required after '%%'\n"); |
5432 | 1.99k | } |
5433 | 2.93k | isParameter = 1; |
5434 | 2.93k | } |
5435 | | |
5436 | 16.0k | name = xmlParseName(ctxt); |
5437 | 16.0k | if (name == NULL) { |
5438 | 692 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5439 | 692 | "xmlParseEntityDecl: no name\n"); |
5440 | 692 | return; |
5441 | 692 | } |
5442 | 15.3k | if (xmlStrchr(name, ':') != NULL) { |
5443 | 1.47k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5444 | 1.47k | "colons are forbidden from entities names '%s'\n", |
5445 | 1.47k | name, NULL, NULL); |
5446 | 1.47k | } |
5447 | 15.3k | if (SKIP_BLANKS == 0) { |
5448 | 7.81k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5449 | 7.81k | "Space required after the entity name\n"); |
5450 | 7.81k | } |
5451 | | |
5452 | 15.3k | ctxt->instate = XML_PARSER_ENTITY_DECL; |
5453 | | /* |
5454 | | * handle the various case of definitions... |
5455 | | */ |
5456 | 15.3k | if (isParameter) { |
5457 | 2.86k | if ((RAW == '"') || (RAW == '\'')) { |
5458 | 1.18k | value = xmlParseEntityValue(ctxt, &orig); |
5459 | 1.18k | if (value) { |
5460 | 363 | if ((ctxt->sax != NULL) && |
5461 | 363 | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5462 | 0 | ctxt->sax->entityDecl(ctxt->userData, name, |
5463 | 0 | XML_INTERNAL_PARAMETER_ENTITY, |
5464 | 0 | NULL, NULL, value); |
5465 | 363 | } |
5466 | 1.67k | } else { |
5467 | 1.67k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5468 | 1.67k | if ((URI == NULL) && (literal == NULL)) { |
5469 | 186 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5470 | 186 | } |
5471 | 1.67k | if (URI) { |
5472 | 933 | xmlURIPtr uri; |
5473 | | |
5474 | 933 | uri = xmlParseURI((const char *) URI); |
5475 | 933 | if (uri == NULL) { |
5476 | 95 | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5477 | 95 | "Invalid URI: %s\n", URI); |
5478 | | /* |
5479 | | * This really ought to be a well formedness error |
5480 | | * but the XML Core WG decided otherwise c.f. issue |
5481 | | * E26 of the XML erratas. |
5482 | | */ |
5483 | 838 | } else { |
5484 | 838 | if (uri->fragment != NULL) { |
5485 | | /* |
5486 | | * Okay this is foolish to block those but not |
5487 | | * invalid URIs. |
5488 | | */ |
5489 | 34 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5490 | 804 | } else { |
5491 | 804 | if ((ctxt->sax != NULL) && |
5492 | 804 | (!ctxt->disableSAX) && |
5493 | 804 | (ctxt->sax->entityDecl != NULL)) |
5494 | 0 | ctxt->sax->entityDecl(ctxt->userData, name, |
5495 | 0 | XML_EXTERNAL_PARAMETER_ENTITY, |
5496 | 0 | literal, URI, NULL); |
5497 | 804 | } |
5498 | 838 | xmlFreeURI(uri); |
5499 | 838 | } |
5500 | 933 | } |
5501 | 1.67k | } |
5502 | 12.4k | } else { |
5503 | 12.4k | if ((RAW == '"') || (RAW == '\'')) { |
5504 | 8.81k | value = xmlParseEntityValue(ctxt, &orig); |
5505 | 8.81k | if ((ctxt->sax != NULL) && |
5506 | 8.81k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5507 | 0 | ctxt->sax->entityDecl(ctxt->userData, name, |
5508 | 0 | XML_INTERNAL_GENERAL_ENTITY, |
5509 | 0 | NULL, NULL, value); |
5510 | | /* |
5511 | | * For expat compatibility in SAX mode. |
5512 | | */ |
5513 | 8.81k | if ((ctxt->myDoc == NULL) || |
5514 | 8.81k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5515 | 8.81k | if (ctxt->myDoc == NULL) { |
5516 | 894 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5517 | 894 | if (ctxt->myDoc == NULL) { |
5518 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5519 | 0 | return; |
5520 | 0 | } |
5521 | 894 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5522 | 894 | } |
5523 | 8.81k | if (ctxt->myDoc->intSubset == NULL) |
5524 | 894 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5525 | 894 | BAD_CAST "fake", NULL, NULL); |
5526 | | |
5527 | 8.81k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5528 | 8.81k | NULL, NULL, value); |
5529 | 8.81k | } |
5530 | 8.81k | } else { |
5531 | 3.66k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5532 | 3.66k | if ((URI == NULL) && (literal == NULL)) { |
5533 | 1.60k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5534 | 1.60k | } |
5535 | 3.66k | if (URI) { |
5536 | 1.59k | xmlURIPtr uri; |
5537 | | |
5538 | 1.59k | uri = xmlParseURI((const char *)URI); |
5539 | 1.59k | if (uri == NULL) { |
5540 | 944 | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5541 | 944 | "Invalid URI: %s\n", URI); |
5542 | | /* |
5543 | | * This really ought to be a well formedness error |
5544 | | * but the XML Core WG decided otherwise c.f. issue |
5545 | | * E26 of the XML erratas. |
5546 | | */ |
5547 | 944 | } else { |
5548 | 650 | if (uri->fragment != NULL) { |
5549 | | /* |
5550 | | * Okay this is foolish to block those but not |
5551 | | * invalid URIs. |
5552 | | */ |
5553 | 423 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5554 | 423 | } |
5555 | 650 | xmlFreeURI(uri); |
5556 | 650 | } |
5557 | 1.59k | } |
5558 | 3.66k | if ((RAW != '>') && (SKIP_BLANKS == 0)) { |
5559 | 631 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5560 | 631 | "Space required before 'NDATA'\n"); |
5561 | 631 | } |
5562 | 3.66k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5563 | 456 | SKIP(5); |
5564 | 456 | if (SKIP_BLANKS == 0) { |
5565 | 47 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5566 | 47 | "Space required after 'NDATA'\n"); |
5567 | 47 | } |
5568 | 456 | ndata = xmlParseName(ctxt); |
5569 | 456 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5570 | 456 | (ctxt->sax->unparsedEntityDecl != NULL)) |
5571 | 0 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5572 | 0 | literal, URI, ndata); |
5573 | 3.20k | } else { |
5574 | 3.20k | if ((ctxt->sax != NULL) && |
5575 | 3.20k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5576 | 0 | ctxt->sax->entityDecl(ctxt->userData, name, |
5577 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5578 | 0 | literal, URI, NULL); |
5579 | | /* |
5580 | | * For expat compatibility in SAX mode. |
5581 | | * assuming the entity replacement was asked for |
5582 | | */ |
5583 | 3.20k | if ((ctxt->replaceEntities != 0) && |
5584 | 3.20k | ((ctxt->myDoc == NULL) || |
5585 | 0 | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5586 | 0 | if (ctxt->myDoc == NULL) { |
5587 | 0 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5588 | 0 | if (ctxt->myDoc == NULL) { |
5589 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5590 | 0 | return; |
5591 | 0 | } |
5592 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5593 | 0 | } |
5594 | | |
5595 | 0 | if (ctxt->myDoc->intSubset == NULL) |
5596 | 0 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5597 | 0 | BAD_CAST "fake", NULL, NULL); |
5598 | 0 | xmlSAX2EntityDecl(ctxt, name, |
5599 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5600 | 0 | literal, URI, NULL); |
5601 | 0 | } |
5602 | 3.20k | } |
5603 | 3.66k | } |
5604 | 12.4k | } |
5605 | 15.3k | if (ctxt->instate == XML_PARSER_EOF) |
5606 | 0 | goto done; |
5607 | 15.3k | SKIP_BLANKS; |
5608 | 15.3k | if (RAW != '>') { |
5609 | 992 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5610 | 992 | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5611 | 992 | xmlHaltParser(ctxt); |
5612 | 14.3k | } else { |
5613 | 14.3k | if (inputid != ctxt->input->id) { |
5614 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5615 | 0 | "Entity declaration doesn't start and stop in" |
5616 | 0 | " the same entity\n"); |
5617 | 0 | } |
5618 | 14.3k | NEXT; |
5619 | 14.3k | } |
5620 | 15.3k | if (orig != NULL) { |
5621 | | /* |
5622 | | * Ugly mechanism to save the raw entity value. |
5623 | | */ |
5624 | 7.25k | xmlEntityPtr cur = NULL; |
5625 | | |
5626 | 7.25k | if (isParameter) { |
5627 | 732 | if ((ctxt->sax != NULL) && |
5628 | 732 | (ctxt->sax->getParameterEntity != NULL)) |
5629 | 0 | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5630 | 6.52k | } else { |
5631 | 6.52k | if ((ctxt->sax != NULL) && |
5632 | 6.52k | (ctxt->sax->getEntity != NULL)) |
5633 | 6.52k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5634 | 6.52k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5635 | 0 | cur = xmlSAX2GetEntity(ctxt, name); |
5636 | 0 | } |
5637 | 6.52k | } |
5638 | 7.25k | if ((cur != NULL) && (cur->orig == NULL)) { |
5639 | 1 | cur->orig = orig; |
5640 | 1 | orig = NULL; |
5641 | 1 | } |
5642 | 7.25k | } |
5643 | | |
5644 | 15.3k | done: |
5645 | 15.3k | if (value != NULL) xmlFree(value); |
5646 | 15.3k | if (URI != NULL) xmlFree(URI); |
5647 | 15.3k | if (literal != NULL) xmlFree(literal); |
5648 | 15.3k | if (orig != NULL) xmlFree(orig); |
5649 | 15.3k | } |
5650 | 16.1k | } |
5651 | | |
5652 | | /** |
5653 | | * xmlParseDefaultDecl: |
5654 | | * @ctxt: an XML parser context |
5655 | | * @value: Receive a possible fixed default value for the attribute |
5656 | | * |
5657 | | * Parse an attribute default declaration |
5658 | | * |
5659 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5660 | | * |
5661 | | * [ VC: Required Attribute ] |
5662 | | * if the default declaration is the keyword #REQUIRED, then the |
5663 | | * attribute must be specified for all elements of the type in the |
5664 | | * attribute-list declaration. |
5665 | | * |
5666 | | * [ VC: Attribute Default Legal ] |
5667 | | * The declared default value must meet the lexical constraints of |
5668 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5669 | | * |
5670 | | * [ VC: Fixed Attribute Default ] |
5671 | | * if an attribute has a default value declared with the #FIXED |
5672 | | * keyword, instances of that attribute must match the default value. |
5673 | | * |
5674 | | * [ WFC: No < in Attribute Values ] |
5675 | | * handled in xmlParseAttValue() |
5676 | | * |
5677 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5678 | | * or XML_ATTRIBUTE_FIXED. |
5679 | | */ |
5680 | | |
5681 | | int |
5682 | 22.8k | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5683 | 22.8k | int val; |
5684 | 22.8k | xmlChar *ret; |
5685 | | |
5686 | 22.8k | *value = NULL; |
5687 | 22.8k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5688 | 159 | SKIP(9); |
5689 | 159 | return(XML_ATTRIBUTE_REQUIRED); |
5690 | 159 | } |
5691 | 22.6k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5692 | 292 | SKIP(8); |
5693 | 292 | return(XML_ATTRIBUTE_IMPLIED); |
5694 | 292 | } |
5695 | 22.3k | val = XML_ATTRIBUTE_NONE; |
5696 | 22.3k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5697 | 130 | SKIP(6); |
5698 | 130 | val = XML_ATTRIBUTE_FIXED; |
5699 | 130 | if (SKIP_BLANKS == 0) { |
5700 | 52 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5701 | 52 | "Space required after '#FIXED'\n"); |
5702 | 52 | } |
5703 | 130 | } |
5704 | 22.3k | ret = xmlParseAttValue(ctxt); |
5705 | 22.3k | ctxt->instate = XML_PARSER_DTD; |
5706 | 22.3k | if (ret == NULL) { |
5707 | 867 | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5708 | 867 | "Attribute default value declaration error\n"); |
5709 | 867 | } else |
5710 | 21.5k | *value = ret; |
5711 | 22.3k | return(val); |
5712 | 22.6k | } |
5713 | | |
5714 | | /** |
5715 | | * xmlParseNotationType: |
5716 | | * @ctxt: an XML parser context |
5717 | | * |
5718 | | * parse an Notation attribute type. |
5719 | | * |
5720 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5721 | | * |
5722 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5723 | | * |
5724 | | * [ VC: Notation Attributes ] |
5725 | | * Values of this type must match one of the notation names included |
5726 | | * in the declaration; all notation names in the declaration must be declared. |
5727 | | * |
5728 | | * Returns: the notation attribute tree built while parsing |
5729 | | */ |
5730 | | |
5731 | | xmlEnumerationPtr |
5732 | 900 | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5733 | 900 | const xmlChar *name; |
5734 | 900 | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5735 | | |
5736 | 900 | if (RAW != '(') { |
5737 | 23 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5738 | 23 | return(NULL); |
5739 | 23 | } |
5740 | 877 | SHRINK; |
5741 | 2.26k | do { |
5742 | 2.26k | NEXT; |
5743 | 2.26k | SKIP_BLANKS; |
5744 | 2.26k | name = xmlParseName(ctxt); |
5745 | 2.26k | if (name == NULL) { |
5746 | 274 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5747 | 274 | "Name expected in NOTATION declaration\n"); |
5748 | 274 | xmlFreeEnumeration(ret); |
5749 | 274 | return(NULL); |
5750 | 274 | } |
5751 | 1.98k | tmp = ret; |
5752 | 5.92k | while (tmp != NULL) { |
5753 | 4.75k | if (xmlStrEqual(name, tmp->name)) { |
5754 | 813 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5755 | 813 | "standalone: attribute notation value token %s duplicated\n", |
5756 | 813 | name, NULL); |
5757 | 813 | if (!xmlDictOwns(ctxt->dict, name)) |
5758 | 0 | xmlFree((xmlChar *) name); |
5759 | 813 | break; |
5760 | 813 | } |
5761 | 3.93k | tmp = tmp->next; |
5762 | 3.93k | } |
5763 | 1.98k | if (tmp == NULL) { |
5764 | 1.17k | cur = xmlCreateEnumeration(name); |
5765 | 1.17k | if (cur == NULL) { |
5766 | 0 | xmlFreeEnumeration(ret); |
5767 | 0 | return(NULL); |
5768 | 0 | } |
5769 | 1.17k | if (last == NULL) ret = last = cur; |
5770 | 564 | else { |
5771 | 564 | last->next = cur; |
5772 | 564 | last = cur; |
5773 | 564 | } |
5774 | 1.17k | } |
5775 | 1.98k | SKIP_BLANKS; |
5776 | 1.98k | } while (RAW == '|'); |
5777 | 603 | if (RAW != ')') { |
5778 | 397 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5779 | 397 | xmlFreeEnumeration(ret); |
5780 | 397 | return(NULL); |
5781 | 397 | } |
5782 | 206 | NEXT; |
5783 | 206 | return(ret); |
5784 | 603 | } |
5785 | | |
5786 | | /** |
5787 | | * xmlParseEnumerationType: |
5788 | | * @ctxt: an XML parser context |
5789 | | * |
5790 | | * parse an Enumeration attribute type. |
5791 | | * |
5792 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5793 | | * |
5794 | | * [ VC: Enumeration ] |
5795 | | * Values of this type must match one of the Nmtoken tokens in |
5796 | | * the declaration |
5797 | | * |
5798 | | * Returns: the enumeration attribute tree built while parsing |
5799 | | */ |
5800 | | |
5801 | | xmlEnumerationPtr |
5802 | 1.84k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
5803 | 1.84k | xmlChar *name; |
5804 | 1.84k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5805 | | |
5806 | 1.84k | if (RAW != '(') { |
5807 | 458 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5808 | 458 | return(NULL); |
5809 | 458 | } |
5810 | 1.38k | SHRINK; |
5811 | 3.58k | do { |
5812 | 3.58k | NEXT; |
5813 | 3.58k | SKIP_BLANKS; |
5814 | 3.58k | name = xmlParseNmtoken(ctxt); |
5815 | 3.58k | if (name == NULL) { |
5816 | 243 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5817 | 243 | return(ret); |
5818 | 243 | } |
5819 | 3.33k | tmp = ret; |
5820 | 9.53k | while (tmp != NULL) { |
5821 | 7.46k | if (xmlStrEqual(name, tmp->name)) { |
5822 | 1.26k | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5823 | 1.26k | "standalone: attribute enumeration value token %s duplicated\n", |
5824 | 1.26k | name, NULL); |
5825 | 1.26k | if (!xmlDictOwns(ctxt->dict, name)) |
5826 | 1.26k | xmlFree(name); |
5827 | 1.26k | break; |
5828 | 1.26k | } |
5829 | 6.20k | tmp = tmp->next; |
5830 | 6.20k | } |
5831 | 3.33k | if (tmp == NULL) { |
5832 | 2.06k | cur = xmlCreateEnumeration(name); |
5833 | 2.06k | if (!xmlDictOwns(ctxt->dict, name)) |
5834 | 2.06k | xmlFree(name); |
5835 | 2.06k | if (cur == NULL) { |
5836 | 0 | xmlFreeEnumeration(ret); |
5837 | 0 | return(NULL); |
5838 | 0 | } |
5839 | 2.06k | if (last == NULL) ret = last = cur; |
5840 | 729 | else { |
5841 | 729 | last->next = cur; |
5842 | 729 | last = cur; |
5843 | 729 | } |
5844 | 2.06k | } |
5845 | 3.33k | SKIP_BLANKS; |
5846 | 3.33k | } while (RAW == '|'); |
5847 | 1.14k | if (RAW != ')') { |
5848 | 570 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5849 | 570 | return(ret); |
5850 | 570 | } |
5851 | 574 | NEXT; |
5852 | 574 | return(ret); |
5853 | 1.14k | } |
5854 | | |
5855 | | /** |
5856 | | * xmlParseEnumeratedType: |
5857 | | * @ctxt: an XML parser context |
5858 | | * @tree: the enumeration tree built while parsing |
5859 | | * |
5860 | | * parse an Enumerated attribute type. |
5861 | | * |
5862 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5863 | | * |
5864 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5865 | | * |
5866 | | * |
5867 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
5868 | | */ |
5869 | | |
5870 | | int |
5871 | 2.95k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5872 | 2.95k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5873 | 1.10k | SKIP(8); |
5874 | 1.10k | if (SKIP_BLANKS == 0) { |
5875 | 207 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5876 | 207 | "Space required after 'NOTATION'\n"); |
5877 | 207 | return(0); |
5878 | 207 | } |
5879 | 900 | *tree = xmlParseNotationType(ctxt); |
5880 | 900 | if (*tree == NULL) return(0); |
5881 | 206 | return(XML_ATTRIBUTE_NOTATION); |
5882 | 900 | } |
5883 | 1.84k | *tree = xmlParseEnumerationType(ctxt); |
5884 | 1.84k | if (*tree == NULL) return(0); |
5885 | 1.34k | return(XML_ATTRIBUTE_ENUMERATION); |
5886 | 1.84k | } |
5887 | | |
5888 | | /** |
5889 | | * xmlParseAttributeType: |
5890 | | * @ctxt: an XML parser context |
5891 | | * @tree: the enumeration tree built while parsing |
5892 | | * |
5893 | | * parse the Attribute list def for an element |
5894 | | * |
5895 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
5896 | | * |
5897 | | * [55] StringType ::= 'CDATA' |
5898 | | * |
5899 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
5900 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
5901 | | * |
5902 | | * Validity constraints for attribute values syntax are checked in |
5903 | | * xmlValidateAttributeValue() |
5904 | | * |
5905 | | * [ VC: ID ] |
5906 | | * Values of type ID must match the Name production. A name must not |
5907 | | * appear more than once in an XML document as a value of this type; |
5908 | | * i.e., ID values must uniquely identify the elements which bear them. |
5909 | | * |
5910 | | * [ VC: One ID per Element Type ] |
5911 | | * No element type may have more than one ID attribute specified. |
5912 | | * |
5913 | | * [ VC: ID Attribute Default ] |
5914 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
5915 | | * |
5916 | | * [ VC: IDREF ] |
5917 | | * Values of type IDREF must match the Name production, and values |
5918 | | * of type IDREFS must match Names; each IDREF Name must match the value |
5919 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
5920 | | * values must match the value of some ID attribute. |
5921 | | * |
5922 | | * [ VC: Entity Name ] |
5923 | | * Values of type ENTITY must match the Name production, values |
5924 | | * of type ENTITIES must match Names; each Entity Name must match the |
5925 | | * name of an unparsed entity declared in the DTD. |
5926 | | * |
5927 | | * [ VC: Name Token ] |
5928 | | * Values of type NMTOKEN must match the Nmtoken production; values |
5929 | | * of type NMTOKENS must match Nmtokens. |
5930 | | * |
5931 | | * Returns the attribute type |
5932 | | */ |
5933 | | int |
5934 | 25.5k | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5935 | 25.5k | SHRINK; |
5936 | 25.5k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
5937 | 1.09k | SKIP(5); |
5938 | 1.09k | return(XML_ATTRIBUTE_CDATA); |
5939 | 24.4k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
5940 | 200 | SKIP(6); |
5941 | 200 | return(XML_ATTRIBUTE_IDREFS); |
5942 | 24.2k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
5943 | 99 | SKIP(5); |
5944 | 99 | return(XML_ATTRIBUTE_IDREF); |
5945 | 24.1k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
5946 | 20.7k | SKIP(2); |
5947 | 20.7k | return(XML_ATTRIBUTE_ID); |
5948 | 20.7k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5949 | 11 | SKIP(6); |
5950 | 11 | return(XML_ATTRIBUTE_ENTITY); |
5951 | 3.45k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
5952 | 204 | SKIP(8); |
5953 | 204 | return(XML_ATTRIBUTE_ENTITIES); |
5954 | 3.25k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
5955 | 93 | SKIP(8); |
5956 | 93 | return(XML_ATTRIBUTE_NMTOKENS); |
5957 | 3.15k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
5958 | 205 | SKIP(7); |
5959 | 205 | return(XML_ATTRIBUTE_NMTOKEN); |
5960 | 205 | } |
5961 | 2.95k | return(xmlParseEnumeratedType(ctxt, tree)); |
5962 | 25.5k | } |
5963 | | |
5964 | | /** |
5965 | | * xmlParseAttributeListDecl: |
5966 | | * @ctxt: an XML parser context |
5967 | | * |
5968 | | * : parse the Attribute list def for an element |
5969 | | * |
5970 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
5971 | | * |
5972 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
5973 | | * |
5974 | | */ |
5975 | | void |
5976 | 8.53k | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
5977 | 8.53k | const xmlChar *elemName; |
5978 | 8.53k | const xmlChar *attrName; |
5979 | 8.53k | xmlEnumerationPtr tree; |
5980 | | |
5981 | 8.53k | if (CMP9(CUR_PTR, '<', '!', 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
5982 | 8.46k | int inputid = ctxt->input->id; |
5983 | | |
5984 | 8.46k | SKIP(9); |
5985 | 8.46k | if (SKIP_BLANKS == 0) { |
5986 | 4.24k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5987 | 4.24k | "Space required after '<!ATTLIST'\n"); |
5988 | 4.24k | } |
5989 | 8.46k | elemName = xmlParseName(ctxt); |
5990 | 8.46k | if (elemName == NULL) { |
5991 | 118 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5992 | 118 | "ATTLIST: no name for Element\n"); |
5993 | 118 | return; |
5994 | 118 | } |
5995 | 8.34k | SKIP_BLANKS; |
5996 | 8.34k | GROW; |
5997 | 29.4k | while ((RAW != '>') && (ctxt->instate != XML_PARSER_EOF)) { |
5998 | 26.8k | int type; |
5999 | 26.8k | int def; |
6000 | 26.8k | xmlChar *defaultValue = NULL; |
6001 | | |
6002 | 26.8k | GROW; |
6003 | 26.8k | tree = NULL; |
6004 | 26.8k | attrName = xmlParseName(ctxt); |
6005 | 26.8k | if (attrName == NULL) { |
6006 | 878 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6007 | 878 | "ATTLIST: no name for Attribute\n"); |
6008 | 878 | break; |
6009 | 878 | } |
6010 | 26.0k | GROW; |
6011 | 26.0k | if (SKIP_BLANKS == 0) { |
6012 | 438 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6013 | 438 | "Space required after the attribute name\n"); |
6014 | 438 | break; |
6015 | 438 | } |
6016 | | |
6017 | 25.5k | type = xmlParseAttributeType(ctxt, &tree); |
6018 | 25.5k | if (type <= 0) { |
6019 | 1.40k | break; |
6020 | 1.40k | } |
6021 | | |
6022 | 24.1k | GROW; |
6023 | 24.1k | if (SKIP_BLANKS == 0) { |
6024 | 1.33k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6025 | 1.33k | "Space required after the attribute type\n"); |
6026 | 1.33k | if (tree != NULL) |
6027 | 772 | xmlFreeEnumeration(tree); |
6028 | 1.33k | break; |
6029 | 1.33k | } |
6030 | | |
6031 | 22.8k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6032 | 22.8k | if (def <= 0) { |
6033 | 0 | if (defaultValue != NULL) |
6034 | 0 | xmlFree(defaultValue); |
6035 | 0 | if (tree != NULL) |
6036 | 0 | xmlFreeEnumeration(tree); |
6037 | 0 | break; |
6038 | 0 | } |
6039 | 22.8k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6040 | 20.4k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6041 | | |
6042 | 22.8k | GROW; |
6043 | 22.8k | if (RAW != '>') { |
6044 | 20.3k | if (SKIP_BLANKS == 0) { |
6045 | 1.76k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6046 | 1.76k | "Space required after the attribute default value\n"); |
6047 | 1.76k | if (defaultValue != NULL) |
6048 | 946 | xmlFree(defaultValue); |
6049 | 1.76k | if (tree != NULL) |
6050 | 484 | xmlFreeEnumeration(tree); |
6051 | 1.76k | break; |
6052 | 1.76k | } |
6053 | 20.3k | } |
6054 | 21.0k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6055 | 21.0k | (ctxt->sax->attributeDecl != NULL)) |
6056 | 0 | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6057 | 0 | type, def, defaultValue, tree); |
6058 | 21.0k | else if (tree != NULL) |
6059 | 290 | xmlFreeEnumeration(tree); |
6060 | | |
6061 | 21.0k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6062 | 21.0k | (def != XML_ATTRIBUTE_IMPLIED) && |
6063 | 21.0k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6064 | 20.5k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6065 | 20.5k | } |
6066 | 21.0k | if (ctxt->sax2) { |
6067 | 21.0k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6068 | 21.0k | } |
6069 | 21.0k | if (defaultValue != NULL) |
6070 | 20.5k | xmlFree(defaultValue); |
6071 | 21.0k | GROW; |
6072 | 21.0k | } |
6073 | 8.34k | if (RAW == '>') { |
6074 | 2.57k | if (inputid != ctxt->input->id) { |
6075 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6076 | 0 | "Attribute list declaration doesn't start and" |
6077 | 0 | " stop in the same entity\n"); |
6078 | 0 | } |
6079 | 2.57k | NEXT; |
6080 | 2.57k | } |
6081 | 8.34k | } |
6082 | 8.53k | } |
6083 | | |
6084 | | /** |
6085 | | * xmlParseElementMixedContentDecl: |
6086 | | * @ctxt: an XML parser context |
6087 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6088 | | * |
6089 | | * parse the declaration for a Mixed Element content |
6090 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6091 | | * |
6092 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6093 | | * '(' S? '#PCDATA' S? ')' |
6094 | | * |
6095 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6096 | | * |
6097 | | * [ VC: No Duplicate Types ] |
6098 | | * The same name must not appear more than once in a single |
6099 | | * mixed-content declaration. |
6100 | | * |
6101 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6102 | | */ |
6103 | | xmlElementContentPtr |
6104 | 1.87k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6105 | 1.87k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6106 | 1.87k | const xmlChar *elem = NULL; |
6107 | | |
6108 | 1.87k | GROW; |
6109 | 1.87k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6110 | 1.87k | SKIP(7); |
6111 | 1.87k | SKIP_BLANKS; |
6112 | 1.87k | SHRINK; |
6113 | 1.87k | if (RAW == ')') { |
6114 | 555 | if (ctxt->input->id != inputchk) { |
6115 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6116 | 0 | "Element content declaration doesn't start and" |
6117 | 0 | " stop in the same entity\n"); |
6118 | 0 | } |
6119 | 555 | NEXT; |
6120 | 555 | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6121 | 555 | if (ret == NULL) |
6122 | 0 | return(NULL); |
6123 | 555 | if (RAW == '*') { |
6124 | 96 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6125 | 96 | NEXT; |
6126 | 96 | } |
6127 | 555 | return(ret); |
6128 | 555 | } |
6129 | 1.32k | if ((RAW == '(') || (RAW == '|')) { |
6130 | 842 | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6131 | 842 | if (ret == NULL) return(NULL); |
6132 | 842 | } |
6133 | 2.43k | while ((RAW == '|') && (ctxt->instate != XML_PARSER_EOF)) { |
6134 | 1.52k | NEXT; |
6135 | 1.52k | if (elem == NULL) { |
6136 | 840 | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6137 | 840 | if (ret == NULL) { |
6138 | 0 | xmlFreeDocElementContent(ctxt->myDoc, cur); |
6139 | 0 | return(NULL); |
6140 | 0 | } |
6141 | 840 | ret->c1 = cur; |
6142 | 840 | if (cur != NULL) |
6143 | 840 | cur->parent = ret; |
6144 | 840 | cur = ret; |
6145 | 840 | } else { |
6146 | 682 | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6147 | 682 | if (n == NULL) { |
6148 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6149 | 0 | return(NULL); |
6150 | 0 | } |
6151 | 682 | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6152 | 682 | if (n->c1 != NULL) |
6153 | 682 | n->c1->parent = n; |
6154 | 682 | cur->c2 = n; |
6155 | 682 | if (n != NULL) |
6156 | 682 | n->parent = cur; |
6157 | 682 | cur = n; |
6158 | 682 | } |
6159 | 1.52k | SKIP_BLANKS; |
6160 | 1.52k | elem = xmlParseName(ctxt); |
6161 | 1.52k | if (elem == NULL) { |
6162 | 409 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6163 | 409 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6164 | 409 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6165 | 409 | return(NULL); |
6166 | 409 | } |
6167 | 1.11k | SKIP_BLANKS; |
6168 | 1.11k | GROW; |
6169 | 1.11k | } |
6170 | 912 | if ((RAW == ')') && (NXT(1) == '*')) { |
6171 | 287 | if (elem != NULL) { |
6172 | 287 | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6173 | 287 | XML_ELEMENT_CONTENT_ELEMENT); |
6174 | 287 | if (cur->c2 != NULL) |
6175 | 287 | cur->c2->parent = cur; |
6176 | 287 | } |
6177 | 287 | if (ret != NULL) |
6178 | 287 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6179 | 287 | if (ctxt->input->id != inputchk) { |
6180 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6181 | 0 | "Element content declaration doesn't start and" |
6182 | 0 | " stop in the same entity\n"); |
6183 | 0 | } |
6184 | 287 | SKIP(2); |
6185 | 625 | } else { |
6186 | 625 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6187 | 625 | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6188 | 625 | return(NULL); |
6189 | 625 | } |
6190 | | |
6191 | 912 | } else { |
6192 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6193 | 0 | } |
6194 | 287 | return(ret); |
6195 | 1.87k | } |
6196 | | |
6197 | | /** |
6198 | | * xmlParseElementChildrenContentDeclPriv: |
6199 | | * @ctxt: an XML parser context |
6200 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6201 | | * @depth: the level of recursion |
6202 | | * |
6203 | | * parse the declaration for a Mixed Element content |
6204 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6205 | | * |
6206 | | * |
6207 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6208 | | * |
6209 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6210 | | * |
6211 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6212 | | * |
6213 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6214 | | * |
6215 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6216 | | * TODO Parameter-entity replacement text must be properly nested |
6217 | | * with parenthesized groups. That is to say, if either of the |
6218 | | * opening or closing parentheses in a choice, seq, or Mixed |
6219 | | * construct is contained in the replacement text for a parameter |
6220 | | * entity, both must be contained in the same replacement text. For |
6221 | | * interoperability, if a parameter-entity reference appears in a |
6222 | | * choice, seq, or Mixed construct, its replacement text should not |
6223 | | * be empty, and neither the first nor last non-blank character of |
6224 | | * the replacement text should be a connector (| or ,). |
6225 | | * |
6226 | | * Returns the tree of xmlElementContentPtr describing the element |
6227 | | * hierarchy. |
6228 | | */ |
6229 | | static xmlElementContentPtr |
6230 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6231 | 54.2k | int depth) { |
6232 | 54.2k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6233 | 54.2k | const xmlChar *elem; |
6234 | 54.2k | xmlChar type = 0; |
6235 | | |
6236 | 54.2k | if (((depth > 128) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
6237 | 54.2k | (depth > 2048)) { |
6238 | 2 | xmlFatalErrMsgInt(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, |
6239 | 2 | "xmlParseElementChildrenContentDecl : depth %d too deep, use XML_PARSE_HUGE\n", |
6240 | 2 | depth); |
6241 | 2 | return(NULL); |
6242 | 2 | } |
6243 | 54.2k | SKIP_BLANKS; |
6244 | 54.2k | GROW; |
6245 | 54.2k | if (RAW == '(') { |
6246 | 43.7k | int inputid = ctxt->input->id; |
6247 | | |
6248 | | /* Recurse on first child */ |
6249 | 43.7k | NEXT; |
6250 | 43.7k | SKIP_BLANKS; |
6251 | 43.7k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6252 | 43.7k | depth + 1); |
6253 | 43.7k | if (cur == NULL) |
6254 | 40.1k | return(NULL); |
6255 | 3.53k | SKIP_BLANKS; |
6256 | 3.53k | GROW; |
6257 | 10.5k | } else { |
6258 | 10.5k | elem = xmlParseName(ctxt); |
6259 | 10.5k | if (elem == NULL) { |
6260 | 266 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6261 | 266 | return(NULL); |
6262 | 266 | } |
6263 | 10.2k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6264 | 10.2k | if (cur == NULL) { |
6265 | 0 | xmlErrMemory(ctxt, NULL); |
6266 | 0 | return(NULL); |
6267 | 0 | } |
6268 | 10.2k | GROW; |
6269 | 10.2k | if (RAW == '?') { |
6270 | 3.77k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6271 | 3.77k | NEXT; |
6272 | 6.51k | } else if (RAW == '*') { |
6273 | 283 | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6274 | 283 | NEXT; |
6275 | 6.23k | } else if (RAW == '+') { |
6276 | 352 | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6277 | 352 | NEXT; |
6278 | 5.88k | } else { |
6279 | 5.88k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6280 | 5.88k | } |
6281 | 10.2k | GROW; |
6282 | 10.2k | } |
6283 | 13.8k | SKIP_BLANKS; |
6284 | 13.8k | SHRINK; |
6285 | 21.7k | while ((RAW != ')') && (ctxt->instate != XML_PARSER_EOF)) { |
6286 | | /* |
6287 | | * Each loop we parse one separator and one element. |
6288 | | */ |
6289 | 14.5k | if (RAW == ',') { |
6290 | 2.11k | if (type == 0) type = CUR; |
6291 | | |
6292 | | /* |
6293 | | * Detect "Name | Name , Name" error |
6294 | | */ |
6295 | 1.45k | else if (type != CUR) { |
6296 | 1 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6297 | 1 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6298 | 1 | type); |
6299 | 1 | if ((last != NULL) && (last != ret)) |
6300 | 1 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6301 | 1 | if (ret != NULL) |
6302 | 1 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6303 | 1 | return(NULL); |
6304 | 1 | } |
6305 | 2.11k | NEXT; |
6306 | | |
6307 | 2.11k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6308 | 2.11k | if (op == NULL) { |
6309 | 0 | if ((last != NULL) && (last != ret)) |
6310 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6311 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6312 | 0 | return(NULL); |
6313 | 0 | } |
6314 | 2.11k | if (last == NULL) { |
6315 | 663 | op->c1 = ret; |
6316 | 663 | if (ret != NULL) |
6317 | 663 | ret->parent = op; |
6318 | 663 | ret = cur = op; |
6319 | 1.45k | } else { |
6320 | 1.45k | cur->c2 = op; |
6321 | 1.45k | if (op != NULL) |
6322 | 1.45k | op->parent = cur; |
6323 | 1.45k | op->c1 = last; |
6324 | 1.45k | if (last != NULL) |
6325 | 1.45k | last->parent = op; |
6326 | 1.45k | cur =op; |
6327 | 1.45k | last = NULL; |
6328 | 1.45k | } |
6329 | 12.3k | } else if (RAW == '|') { |
6330 | 10.5k | if (type == 0) type = CUR; |
6331 | | |
6332 | | /* |
6333 | | * Detect "Name , Name | Name" error |
6334 | | */ |
6335 | 3.31k | else if (type != CUR) { |
6336 | 2 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6337 | 2 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6338 | 2 | type); |
6339 | 2 | if ((last != NULL) && (last != ret)) |
6340 | 2 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6341 | 2 | if (ret != NULL) |
6342 | 2 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6343 | 2 | return(NULL); |
6344 | 2 | } |
6345 | 10.5k | NEXT; |
6346 | | |
6347 | 10.5k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6348 | 10.5k | if (op == NULL) { |
6349 | 0 | if ((last != NULL) && (last != ret)) |
6350 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6351 | 0 | if (ret != NULL) |
6352 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6353 | 0 | return(NULL); |
6354 | 0 | } |
6355 | 10.5k | if (last == NULL) { |
6356 | 7.28k | op->c1 = ret; |
6357 | 7.28k | if (ret != NULL) |
6358 | 7.28k | ret->parent = op; |
6359 | 7.28k | ret = cur = op; |
6360 | 7.28k | } else { |
6361 | 3.30k | cur->c2 = op; |
6362 | 3.30k | if (op != NULL) |
6363 | 3.30k | op->parent = cur; |
6364 | 3.30k | op->c1 = last; |
6365 | 3.30k | if (last != NULL) |
6366 | 3.30k | last->parent = op; |
6367 | 3.30k | cur =op; |
6368 | 3.30k | last = NULL; |
6369 | 3.30k | } |
6370 | 10.5k | } else { |
6371 | 1.80k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6372 | 1.80k | if ((last != NULL) && (last != ret)) |
6373 | 669 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6374 | 1.80k | if (ret != NULL) |
6375 | 1.80k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6376 | 1.80k | return(NULL); |
6377 | 1.80k | } |
6378 | 12.7k | GROW; |
6379 | 12.7k | SKIP_BLANKS; |
6380 | 12.7k | GROW; |
6381 | 12.7k | if (RAW == '(') { |
6382 | 7.65k | int inputid = ctxt->input->id; |
6383 | | /* Recurse on second child */ |
6384 | 7.65k | NEXT; |
6385 | 7.65k | SKIP_BLANKS; |
6386 | 7.65k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6387 | 7.65k | depth + 1); |
6388 | 7.65k | if (last == NULL) { |
6389 | 4.29k | if (ret != NULL) |
6390 | 4.29k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6391 | 4.29k | return(NULL); |
6392 | 4.29k | } |
6393 | 3.35k | SKIP_BLANKS; |
6394 | 5.05k | } else { |
6395 | 5.05k | elem = xmlParseName(ctxt); |
6396 | 5.05k | if (elem == NULL) { |
6397 | 535 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6398 | 535 | if (ret != NULL) |
6399 | 535 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6400 | 535 | return(NULL); |
6401 | 535 | } |
6402 | 4.52k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6403 | 4.52k | if (last == NULL) { |
6404 | 0 | if (ret != NULL) |
6405 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6406 | 0 | return(NULL); |
6407 | 0 | } |
6408 | 4.52k | if (RAW == '?') { |
6409 | 693 | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6410 | 693 | NEXT; |
6411 | 3.82k | } else if (RAW == '*') { |
6412 | 461 | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6413 | 461 | NEXT; |
6414 | 3.36k | } else if (RAW == '+') { |
6415 | 209 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6416 | 209 | NEXT; |
6417 | 3.15k | } else { |
6418 | 3.15k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6419 | 3.15k | } |
6420 | 4.52k | } |
6421 | 7.87k | SKIP_BLANKS; |
6422 | 7.87k | GROW; |
6423 | 7.87k | } |
6424 | 7.19k | if ((cur != NULL) && (last != NULL)) { |
6425 | 2.44k | cur->c2 = last; |
6426 | 2.44k | if (last != NULL) |
6427 | 2.44k | last->parent = cur; |
6428 | 2.44k | } |
6429 | 7.19k | if (ctxt->input->id != inputchk) { |
6430 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6431 | 0 | "Element content declaration doesn't start and stop in" |
6432 | 0 | " the same entity\n"); |
6433 | 0 | } |
6434 | 7.19k | NEXT; |
6435 | 7.19k | if (RAW == '?') { |
6436 | 688 | if (ret != NULL) { |
6437 | 688 | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6438 | 688 | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6439 | 316 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6440 | 372 | else |
6441 | 372 | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6442 | 688 | } |
6443 | 688 | NEXT; |
6444 | 6.50k | } else if (RAW == '*') { |
6445 | 731 | if (ret != NULL) { |
6446 | 731 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6447 | 731 | cur = ret; |
6448 | | /* |
6449 | | * Some normalization: |
6450 | | * (a | b* | c?)* == (a | b | c)* |
6451 | | */ |
6452 | 1.92k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6453 | 1.19k | if ((cur->c1 != NULL) && |
6454 | 1.19k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6455 | 1.19k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6456 | 526 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6457 | 1.19k | if ((cur->c2 != NULL) && |
6458 | 1.19k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6459 | 1.19k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6460 | 423 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6461 | 1.19k | cur = cur->c2; |
6462 | 1.19k | } |
6463 | 731 | } |
6464 | 731 | NEXT; |
6465 | 5.77k | } else if (RAW == '+') { |
6466 | 4.38k | if (ret != NULL) { |
6467 | 4.38k | int found = 0; |
6468 | | |
6469 | 4.38k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6470 | 4.38k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6471 | 2.79k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6472 | 1.59k | else |
6473 | 1.59k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6474 | | /* |
6475 | | * Some normalization: |
6476 | | * (a | b*)+ == (a | b)* |
6477 | | * (a | b?)+ == (a | b)* |
6478 | | */ |
6479 | 11.5k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6480 | 7.14k | if ((cur->c1 != NULL) && |
6481 | 7.14k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6482 | 7.14k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6483 | 1.26k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6484 | 1.26k | found = 1; |
6485 | 1.26k | } |
6486 | 7.14k | if ((cur->c2 != NULL) && |
6487 | 7.14k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6488 | 7.14k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6489 | 957 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6490 | 957 | found = 1; |
6491 | 957 | } |
6492 | 7.14k | cur = cur->c2; |
6493 | 7.14k | } |
6494 | 4.38k | if (found) |
6495 | 1.20k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6496 | 4.38k | } |
6497 | 4.38k | NEXT; |
6498 | 4.38k | } |
6499 | 7.19k | return(ret); |
6500 | 13.8k | } |
6501 | | |
6502 | | /** |
6503 | | * xmlParseElementChildrenContentDecl: |
6504 | | * @ctxt: an XML parser context |
6505 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6506 | | * |
6507 | | * parse the declaration for a Mixed Element content |
6508 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6509 | | * |
6510 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6511 | | * |
6512 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6513 | | * |
6514 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6515 | | * |
6516 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6517 | | * |
6518 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6519 | | * TODO Parameter-entity replacement text must be properly nested |
6520 | | * with parenthesized groups. That is to say, if either of the |
6521 | | * opening or closing parentheses in a choice, seq, or Mixed |
6522 | | * construct is contained in the replacement text for a parameter |
6523 | | * entity, both must be contained in the same replacement text. For |
6524 | | * interoperability, if a parameter-entity reference appears in a |
6525 | | * choice, seq, or Mixed construct, its replacement text should not |
6526 | | * be empty, and neither the first nor last non-blank character of |
6527 | | * the replacement text should be a connector (| or ,). |
6528 | | * |
6529 | | * Returns the tree of xmlElementContentPtr describing the element |
6530 | | * hierarchy. |
6531 | | */ |
6532 | | xmlElementContentPtr |
6533 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6534 | | /* stub left for API/ABI compat */ |
6535 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6536 | 0 | } |
6537 | | |
6538 | | /** |
6539 | | * xmlParseElementContentDecl: |
6540 | | * @ctxt: an XML parser context |
6541 | | * @name: the name of the element being defined. |
6542 | | * @result: the Element Content pointer will be stored here if any |
6543 | | * |
6544 | | * parse the declaration for an Element content either Mixed or Children, |
6545 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6546 | | * |
6547 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6548 | | * |
6549 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6550 | | */ |
6551 | | |
6552 | | int |
6553 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6554 | 4.79k | xmlElementContentPtr *result) { |
6555 | | |
6556 | 4.79k | xmlElementContentPtr tree = NULL; |
6557 | 4.79k | int inputid = ctxt->input->id; |
6558 | 4.79k | int res; |
6559 | | |
6560 | 4.79k | *result = NULL; |
6561 | | |
6562 | 4.79k | if (RAW != '(') { |
6563 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6564 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6565 | 0 | return(-1); |
6566 | 0 | } |
6567 | 4.79k | NEXT; |
6568 | 4.79k | GROW; |
6569 | 4.79k | if (ctxt->instate == XML_PARSER_EOF) |
6570 | 0 | return(-1); |
6571 | 4.79k | SKIP_BLANKS; |
6572 | 4.79k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6573 | 1.87k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6574 | 1.87k | res = XML_ELEMENT_TYPE_MIXED; |
6575 | 2.91k | } else { |
6576 | 2.91k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6577 | 2.91k | res = XML_ELEMENT_TYPE_ELEMENT; |
6578 | 2.91k | } |
6579 | 4.79k | SKIP_BLANKS; |
6580 | 4.79k | *result = tree; |
6581 | 4.79k | return(res); |
6582 | 4.79k | } |
6583 | | |
6584 | | /** |
6585 | | * xmlParseElementDecl: |
6586 | | * @ctxt: an XML parser context |
6587 | | * |
6588 | | * parse an Element declaration. |
6589 | | * |
6590 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6591 | | * |
6592 | | * [ VC: Unique Element Type Declaration ] |
6593 | | * No element type may be declared more than once |
6594 | | * |
6595 | | * Returns the type of the element, or -1 in case of error |
6596 | | */ |
6597 | | int |
6598 | 6.10k | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6599 | 6.10k | const xmlChar *name; |
6600 | 6.10k | int ret = -1; |
6601 | 6.10k | xmlElementContentPtr content = NULL; |
6602 | | |
6603 | | /* GROW; done in the caller */ |
6604 | 6.10k | if (CMP9(CUR_PTR, '<', '!', 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6605 | 6.04k | int inputid = ctxt->input->id; |
6606 | | |
6607 | 6.04k | SKIP(9); |
6608 | 6.04k | if (SKIP_BLANKS == 0) { |
6609 | 124 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6610 | 124 | "Space required after 'ELEMENT'\n"); |
6611 | 124 | return(-1); |
6612 | 124 | } |
6613 | 5.92k | name = xmlParseName(ctxt); |
6614 | 5.92k | if (name == NULL) { |
6615 | 346 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6616 | 346 | "xmlParseElementDecl: no name for Element\n"); |
6617 | 346 | return(-1); |
6618 | 346 | } |
6619 | 5.57k | if (SKIP_BLANKS == 0) { |
6620 | 4.61k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6621 | 4.61k | "Space required after the element name\n"); |
6622 | 4.61k | } |
6623 | 5.57k | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6624 | 11 | SKIP(5); |
6625 | | /* |
6626 | | * Element must always be empty. |
6627 | | */ |
6628 | 11 | ret = XML_ELEMENT_TYPE_EMPTY; |
6629 | 5.56k | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6630 | 5.56k | (NXT(2) == 'Y')) { |
6631 | 67 | SKIP(3); |
6632 | | /* |
6633 | | * Element is a generic container. |
6634 | | */ |
6635 | 67 | ret = XML_ELEMENT_TYPE_ANY; |
6636 | 5.49k | } else if (RAW == '(') { |
6637 | 4.79k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6638 | 4.79k | } else { |
6639 | | /* |
6640 | | * [ WFC: PEs in Internal Subset ] error handling. |
6641 | | */ |
6642 | 709 | if ((RAW == '%') && (ctxt->external == 0) && |
6643 | 709 | (ctxt->inputNr == 1)) { |
6644 | 67 | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_IN_INT_SUBSET, |
6645 | 67 | "PEReference: forbidden within markup decl in internal subset\n"); |
6646 | 642 | } else { |
6647 | 642 | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6648 | 642 | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6649 | 642 | } |
6650 | 709 | return(-1); |
6651 | 709 | } |
6652 | | |
6653 | 4.86k | SKIP_BLANKS; |
6654 | | |
6655 | 4.86k | if (RAW != '>') { |
6656 | 4.33k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6657 | 4.33k | if (content != NULL) { |
6658 | 703 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6659 | 703 | } |
6660 | 4.33k | } else { |
6661 | 538 | if (inputid != ctxt->input->id) { |
6662 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6663 | 0 | "Element declaration doesn't start and stop in" |
6664 | 0 | " the same entity\n"); |
6665 | 0 | } |
6666 | | |
6667 | 538 | NEXT; |
6668 | 538 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6669 | 538 | (ctxt->sax->elementDecl != NULL)) { |
6670 | 0 | if (content != NULL) |
6671 | 0 | content->parent = NULL; |
6672 | 0 | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6673 | 0 | content); |
6674 | 0 | if ((content != NULL) && (content->parent == NULL)) { |
6675 | | /* |
6676 | | * this is a trick: if xmlAddElementDecl is called, |
6677 | | * instead of copying the full tree it is plugged directly |
6678 | | * if called from the parser. Avoid duplicating the |
6679 | | * interfaces or change the API/ABI |
6680 | | */ |
6681 | 0 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6682 | 0 | } |
6683 | 538 | } else if (content != NULL) { |
6684 | 446 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6685 | 446 | } |
6686 | 538 | } |
6687 | 4.86k | } |
6688 | 4.92k | return(ret); |
6689 | 6.10k | } |
6690 | | |
6691 | | /** |
6692 | | * xmlParseConditionalSections |
6693 | | * @ctxt: an XML parser context |
6694 | | * |
6695 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6696 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6697 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6698 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6699 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6700 | | */ |
6701 | | |
6702 | | static void |
6703 | 0 | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6704 | 0 | int *inputIds = NULL; |
6705 | 0 | size_t inputIdsSize = 0; |
6706 | 0 | size_t depth = 0; |
6707 | |
|
6708 | 0 | while (ctxt->instate != XML_PARSER_EOF) { |
6709 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6710 | 0 | int id = ctxt->input->id; |
6711 | |
|
6712 | 0 | SKIP(3); |
6713 | 0 | SKIP_BLANKS; |
6714 | |
|
6715 | 0 | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6716 | 0 | SKIP(7); |
6717 | 0 | SKIP_BLANKS; |
6718 | 0 | if (RAW != '[') { |
6719 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6720 | 0 | xmlHaltParser(ctxt); |
6721 | 0 | goto error; |
6722 | 0 | } |
6723 | 0 | if (ctxt->input->id != id) { |
6724 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6725 | 0 | "All markup of the conditional section is" |
6726 | 0 | " not in the same entity\n"); |
6727 | 0 | } |
6728 | 0 | NEXT; |
6729 | |
|
6730 | 0 | if (inputIdsSize <= depth) { |
6731 | 0 | int *tmp; |
6732 | |
|
6733 | 0 | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
6734 | 0 | tmp = (int *) xmlRealloc(inputIds, |
6735 | 0 | inputIdsSize * sizeof(int)); |
6736 | 0 | if (tmp == NULL) { |
6737 | 0 | xmlErrMemory(ctxt, NULL); |
6738 | 0 | goto error; |
6739 | 0 | } |
6740 | 0 | inputIds = tmp; |
6741 | 0 | } |
6742 | 0 | inputIds[depth] = id; |
6743 | 0 | depth++; |
6744 | 0 | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6745 | 0 | int state; |
6746 | 0 | xmlParserInputState instate; |
6747 | 0 | size_t ignoreDepth = 0; |
6748 | |
|
6749 | 0 | SKIP(6); |
6750 | 0 | SKIP_BLANKS; |
6751 | 0 | if (RAW != '[') { |
6752 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6753 | 0 | xmlHaltParser(ctxt); |
6754 | 0 | goto error; |
6755 | 0 | } |
6756 | 0 | if (ctxt->input->id != id) { |
6757 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6758 | 0 | "All markup of the conditional section is" |
6759 | 0 | " not in the same entity\n"); |
6760 | 0 | } |
6761 | 0 | NEXT; |
6762 | | |
6763 | | /* |
6764 | | * Parse up to the end of the conditional section but disable |
6765 | | * SAX event generating DTD building in the meantime |
6766 | | */ |
6767 | 0 | state = ctxt->disableSAX; |
6768 | 0 | instate = ctxt->instate; |
6769 | 0 | if (ctxt->recovery == 0) ctxt->disableSAX = 1; |
6770 | 0 | ctxt->instate = XML_PARSER_IGNORE; |
6771 | |
|
6772 | 0 | while (RAW != 0) { |
6773 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6774 | 0 | SKIP(3); |
6775 | 0 | ignoreDepth++; |
6776 | | /* Check for integer overflow */ |
6777 | 0 | if (ignoreDepth == 0) { |
6778 | 0 | xmlErrMemory(ctxt, NULL); |
6779 | 0 | goto error; |
6780 | 0 | } |
6781 | 0 | } else if ((RAW == ']') && (NXT(1) == ']') && |
6782 | 0 | (NXT(2) == '>')) { |
6783 | 0 | if (ignoreDepth == 0) |
6784 | 0 | break; |
6785 | 0 | SKIP(3); |
6786 | 0 | ignoreDepth--; |
6787 | 0 | } else { |
6788 | 0 | NEXT; |
6789 | 0 | } |
6790 | 0 | } |
6791 | | |
6792 | 0 | ctxt->disableSAX = state; |
6793 | 0 | ctxt->instate = instate; |
6794 | |
|
6795 | 0 | if (RAW == 0) { |
6796 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6797 | 0 | goto error; |
6798 | 0 | } |
6799 | 0 | if (ctxt->input->id != id) { |
6800 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6801 | 0 | "All markup of the conditional section is" |
6802 | 0 | " not in the same entity\n"); |
6803 | 0 | } |
6804 | 0 | SKIP(3); |
6805 | 0 | } else { |
6806 | 0 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6807 | 0 | xmlHaltParser(ctxt); |
6808 | 0 | goto error; |
6809 | 0 | } |
6810 | 0 | } else if ((depth > 0) && |
6811 | 0 | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6812 | 0 | depth--; |
6813 | 0 | if (ctxt->input->id != inputIds[depth]) { |
6814 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6815 | 0 | "All markup of the conditional section is not" |
6816 | 0 | " in the same entity\n"); |
6817 | 0 | } |
6818 | 0 | SKIP(3); |
6819 | 0 | } else { |
6820 | 0 | int id = ctxt->input->id; |
6821 | 0 | unsigned long cons = CUR_CONSUMED; |
6822 | |
|
6823 | 0 | xmlParseMarkupDecl(ctxt); |
6824 | |
|
6825 | 0 | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
6826 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6827 | 0 | xmlHaltParser(ctxt); |
6828 | 0 | goto error; |
6829 | 0 | } |
6830 | 0 | } |
6831 | | |
6832 | 0 | if (depth == 0) |
6833 | 0 | break; |
6834 | | |
6835 | 0 | SKIP_BLANKS; |
6836 | 0 | GROW; |
6837 | 0 | } |
6838 | | |
6839 | 0 | error: |
6840 | 0 | xmlFree(inputIds); |
6841 | 0 | } |
6842 | | |
6843 | | /** |
6844 | | * xmlParseMarkupDecl: |
6845 | | * @ctxt: an XML parser context |
6846 | | * |
6847 | | * parse Markup declarations |
6848 | | * |
6849 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6850 | | * NotationDecl | PI | Comment |
6851 | | * |
6852 | | * [ VC: Proper Declaration/PE Nesting ] |
6853 | | * Parameter-entity replacement text must be properly nested with |
6854 | | * markup declarations. That is to say, if either the first character |
6855 | | * or the last character of a markup declaration (markupdecl above) is |
6856 | | * contained in the replacement text for a parameter-entity reference, |
6857 | | * both must be contained in the same replacement text. |
6858 | | * |
6859 | | * [ WFC: PEs in Internal Subset ] |
6860 | | * In the internal DTD subset, parameter-entity references can occur |
6861 | | * only where markup declarations can occur, not within markup declarations. |
6862 | | * (This does not apply to references that occur in external parameter |
6863 | | * entities or to the external subset.) |
6864 | | */ |
6865 | | void |
6866 | 43.4k | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
6867 | 43.4k | GROW; |
6868 | 43.4k | if (CUR == '<') { |
6869 | 40.0k | if (NXT(1) == '!') { |
6870 | 34.8k | switch (NXT(2)) { |
6871 | 22.2k | case 'E': |
6872 | 22.2k | if (NXT(3) == 'L') |
6873 | 6.10k | xmlParseElementDecl(ctxt); |
6874 | 16.1k | else if (NXT(3) == 'N') |
6875 | 16.1k | xmlParseEntityDecl(ctxt); |
6876 | 22.2k | break; |
6877 | 8.53k | case 'A': |
6878 | 8.53k | xmlParseAttributeListDecl(ctxt); |
6879 | 8.53k | break; |
6880 | 2.38k | case 'N': |
6881 | 2.38k | xmlParseNotationDecl(ctxt); |
6882 | 2.38k | break; |
6883 | 1.62k | case '-': |
6884 | 1.62k | xmlParseComment(ctxt); |
6885 | 1.62k | break; |
6886 | 63 | default: |
6887 | | /* there is an error but it will be detected later */ |
6888 | 63 | break; |
6889 | 34.8k | } |
6890 | 34.8k | } else if (NXT(1) == '?') { |
6891 | 5.04k | xmlParsePI(ctxt); |
6892 | 5.04k | } |
6893 | 40.0k | } |
6894 | | |
6895 | | /* |
6896 | | * detect requirement to exit there and act accordingly |
6897 | | * and avoid having instate overridden later on |
6898 | | */ |
6899 | 43.4k | if (ctxt->instate == XML_PARSER_EOF) |
6900 | 992 | return; |
6901 | | |
6902 | 42.4k | ctxt->instate = XML_PARSER_DTD; |
6903 | 42.4k | } |
6904 | | |
6905 | | /** |
6906 | | * xmlParseTextDecl: |
6907 | | * @ctxt: an XML parser context |
6908 | | * |
6909 | | * parse an XML declaration header for external entities |
6910 | | * |
6911 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
6912 | | */ |
6913 | | |
6914 | | void |
6915 | 0 | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
6916 | 0 | xmlChar *version; |
6917 | 0 | const xmlChar *encoding; |
6918 | 0 | int oldstate; |
6919 | | |
6920 | | /* |
6921 | | * We know that '<?xml' is here. |
6922 | | */ |
6923 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
6924 | 0 | SKIP(5); |
6925 | 0 | } else { |
6926 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
6927 | 0 | return; |
6928 | 0 | } |
6929 | | |
6930 | | /* Avoid expansion of parameter entities when skipping blanks. */ |
6931 | 0 | oldstate = ctxt->instate; |
6932 | 0 | ctxt->instate = XML_PARSER_START; |
6933 | |
|
6934 | 0 | if (SKIP_BLANKS == 0) { |
6935 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6936 | 0 | "Space needed after '<?xml'\n"); |
6937 | 0 | } |
6938 | | |
6939 | | /* |
6940 | | * We may have the VersionInfo here. |
6941 | | */ |
6942 | 0 | version = xmlParseVersionInfo(ctxt); |
6943 | 0 | if (version == NULL) |
6944 | 0 | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
6945 | 0 | else { |
6946 | 0 | if (SKIP_BLANKS == 0) { |
6947 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6948 | 0 | "Space needed here\n"); |
6949 | 0 | } |
6950 | 0 | } |
6951 | 0 | ctxt->input->version = version; |
6952 | | |
6953 | | /* |
6954 | | * We must have the encoding declaration |
6955 | | */ |
6956 | 0 | encoding = xmlParseEncodingDecl(ctxt); |
6957 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
6958 | | /* |
6959 | | * The XML REC instructs us to stop parsing right here |
6960 | | */ |
6961 | 0 | ctxt->instate = oldstate; |
6962 | 0 | return; |
6963 | 0 | } |
6964 | 0 | if ((encoding == NULL) && (ctxt->errNo == XML_ERR_OK)) { |
6965 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_MISSING_ENCODING, |
6966 | 0 | "Missing encoding in text declaration\n"); |
6967 | 0 | } |
6968 | |
|
6969 | 0 | SKIP_BLANKS; |
6970 | 0 | if ((RAW == '?') && (NXT(1) == '>')) { |
6971 | 0 | SKIP(2); |
6972 | 0 | } else if (RAW == '>') { |
6973 | | /* Deprecated old WD ... */ |
6974 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
6975 | 0 | NEXT; |
6976 | 0 | } else { |
6977 | 0 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
6978 | 0 | MOVETO_ENDTAG(CUR_PTR); |
6979 | 0 | NEXT; |
6980 | 0 | } |
6981 | |
|
6982 | 0 | ctxt->instate = oldstate; |
6983 | 0 | } |
6984 | | |
6985 | | /** |
6986 | | * xmlParseExternalSubset: |
6987 | | * @ctxt: an XML parser context |
6988 | | * @ExternalID: the external identifier |
6989 | | * @SystemID: the system identifier (or URL) |
6990 | | * |
6991 | | * parse Markup declarations from an external subset |
6992 | | * |
6993 | | * [30] extSubset ::= textDecl? extSubsetDecl |
6994 | | * |
6995 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
6996 | | */ |
6997 | | void |
6998 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
6999 | 0 | const xmlChar *SystemID) { |
7000 | 0 | xmlDetectSAX2(ctxt); |
7001 | 0 | GROW; |
7002 | |
|
7003 | 0 | if ((ctxt->encoding == NULL) && |
7004 | 0 | (ctxt->input->end - ctxt->input->cur >= 4)) { |
7005 | 0 | xmlChar start[4]; |
7006 | 0 | xmlCharEncoding enc; |
7007 | |
|
7008 | 0 | start[0] = RAW; |
7009 | 0 | start[1] = NXT(1); |
7010 | 0 | start[2] = NXT(2); |
7011 | 0 | start[3] = NXT(3); |
7012 | 0 | enc = xmlDetectCharEncoding(start, 4); |
7013 | 0 | if (enc != XML_CHAR_ENCODING_NONE) |
7014 | 0 | xmlSwitchEncoding(ctxt, enc); |
7015 | 0 | } |
7016 | |
|
7017 | 0 | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7018 | 0 | xmlParseTextDecl(ctxt); |
7019 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7020 | | /* |
7021 | | * The XML REC instructs us to stop parsing right here |
7022 | | */ |
7023 | 0 | xmlHaltParser(ctxt); |
7024 | 0 | return; |
7025 | 0 | } |
7026 | 0 | } |
7027 | 0 | if (ctxt->myDoc == NULL) { |
7028 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7029 | 0 | if (ctxt->myDoc == NULL) { |
7030 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
7031 | 0 | return; |
7032 | 0 | } |
7033 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7034 | 0 | } |
7035 | 0 | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL)) |
7036 | 0 | xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID); |
7037 | |
|
7038 | 0 | ctxt->instate = XML_PARSER_DTD; |
7039 | 0 | ctxt->external = 1; |
7040 | 0 | SKIP_BLANKS; |
7041 | 0 | while (((RAW == '<') && (NXT(1) == '?')) || |
7042 | 0 | ((RAW == '<') && (NXT(1) == '!')) || |
7043 | 0 | (RAW == '%')) { |
7044 | 0 | int id = ctxt->input->id; |
7045 | 0 | unsigned long cons = CUR_CONSUMED; |
7046 | |
|
7047 | 0 | GROW; |
7048 | 0 | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7049 | 0 | xmlParseConditionalSections(ctxt); |
7050 | 0 | } else |
7051 | 0 | xmlParseMarkupDecl(ctxt); |
7052 | 0 | SKIP_BLANKS; |
7053 | |
|
7054 | 0 | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
7055 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7056 | 0 | break; |
7057 | 0 | } |
7058 | 0 | } |
7059 | |
|
7060 | 0 | if (RAW != 0) { |
7061 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7062 | 0 | } |
7063 | |
|
7064 | 0 | } |
7065 | | |
7066 | | /** |
7067 | | * xmlParseReference: |
7068 | | * @ctxt: an XML parser context |
7069 | | * |
7070 | | * parse and handle entity references in content, depending on the SAX |
7071 | | * interface, this may end-up in a call to character() if this is a |
7072 | | * CharRef, a predefined entity, if there is no reference() callback. |
7073 | | * or if the parser was asked to switch to that mode. |
7074 | | * |
7075 | | * [67] Reference ::= EntityRef | CharRef |
7076 | | */ |
7077 | | void |
7078 | 2.91k | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7079 | 2.91k | xmlEntityPtr ent; |
7080 | 2.91k | xmlChar *val; |
7081 | 2.91k | int was_checked; |
7082 | 2.91k | xmlNodePtr list = NULL; |
7083 | 2.91k | xmlParserErrors ret = XML_ERR_OK; |
7084 | | |
7085 | | |
7086 | 2.91k | if (RAW != '&') |
7087 | 0 | return; |
7088 | | |
7089 | | /* |
7090 | | * Simple case of a CharRef |
7091 | | */ |
7092 | 2.91k | if (NXT(1) == '#') { |
7093 | 485 | int i = 0; |
7094 | 485 | xmlChar out[16]; |
7095 | 485 | int hex = NXT(2); |
7096 | 485 | int value = xmlParseCharRef(ctxt); |
7097 | | |
7098 | 485 | if (value == 0) |
7099 | 136 | return; |
7100 | 349 | if (ctxt->charset != XML_CHAR_ENCODING_UTF8) { |
7101 | | /* |
7102 | | * So we are using non-UTF-8 buffers |
7103 | | * Check that the char fit on 8bits, if not |
7104 | | * generate a CharRef. |
7105 | | */ |
7106 | 0 | if (value <= 0xFF) { |
7107 | 0 | out[0] = value; |
7108 | 0 | out[1] = 0; |
7109 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7110 | 0 | (!ctxt->disableSAX)) |
7111 | 0 | ctxt->sax->characters(ctxt->userData, out, 1); |
7112 | 0 | } else { |
7113 | 0 | if ((hex == 'x') || (hex == 'X')) |
7114 | 0 | snprintf((char *)out, sizeof(out), "#x%X", value); |
7115 | 0 | else |
7116 | 0 | snprintf((char *)out, sizeof(out), "#%d", value); |
7117 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7118 | 0 | (!ctxt->disableSAX)) |
7119 | 0 | ctxt->sax->reference(ctxt->userData, out); |
7120 | 0 | } |
7121 | 349 | } else { |
7122 | | /* |
7123 | | * Just encode the value in UTF-8 |
7124 | | */ |
7125 | 349 | COPY_BUF(0 ,out, i, value); |
7126 | 349 | out[i] = 0; |
7127 | 349 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7128 | 349 | (!ctxt->disableSAX)) |
7129 | 349 | ctxt->sax->characters(ctxt->userData, out, i); |
7130 | 349 | } |
7131 | 349 | return; |
7132 | 485 | } |
7133 | | |
7134 | | /* |
7135 | | * We are seeing an entity reference |
7136 | | */ |
7137 | 2.43k | ent = xmlParseEntityRef(ctxt); |
7138 | 2.43k | if (ent == NULL) return; |
7139 | 2.35k | if (!ctxt->wellFormed) |
7140 | 1 | return; |
7141 | 2.35k | was_checked = ent->checked; |
7142 | | |
7143 | | /* special case of predefined entities */ |
7144 | 2.35k | if ((ent->name == NULL) || |
7145 | 2.35k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7146 | 2.35k | val = ent->content; |
7147 | 2.35k | if (val == NULL) return; |
7148 | | /* |
7149 | | * inline the entity. |
7150 | | */ |
7151 | 933 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7152 | 933 | (!ctxt->disableSAX)) |
7153 | 933 | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7154 | 933 | return; |
7155 | 2.35k | } |
7156 | | |
7157 | | /* |
7158 | | * The first reference to the entity trigger a parsing phase |
7159 | | * where the ent->children is filled with the result from |
7160 | | * the parsing. |
7161 | | * Note: external parsed entities will not be loaded, it is not |
7162 | | * required for a non-validating parser, unless the parsing option |
7163 | | * of validating, or substituting entities were given. Doing so is |
7164 | | * far more secure as the parser will only process data coming from |
7165 | | * the document entity by default. |
7166 | | */ |
7167 | 0 | if (((ent->checked == 0) || |
7168 | 0 | ((ent->children == NULL) && (ctxt->options & XML_PARSE_NOENT))) && |
7169 | 0 | ((ent->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY) || |
7170 | 0 | (ctxt->options & (XML_PARSE_NOENT | XML_PARSE_DTDVALID)))) { |
7171 | 0 | unsigned long oldnbent = ctxt->nbentities, diff; |
7172 | | |
7173 | | /* |
7174 | | * This is a bit hackish but this seems the best |
7175 | | * way to make sure both SAX and DOM entity support |
7176 | | * behaves okay. |
7177 | | */ |
7178 | 0 | void *user_data; |
7179 | 0 | if (ctxt->userData == ctxt) |
7180 | 0 | user_data = NULL; |
7181 | 0 | else |
7182 | 0 | user_data = ctxt->userData; |
7183 | | |
7184 | | /* |
7185 | | * Check that this entity is well formed |
7186 | | * 4.3.2: An internal general parsed entity is well-formed |
7187 | | * if its replacement text matches the production labeled |
7188 | | * content. |
7189 | | */ |
7190 | 0 | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7191 | 0 | ctxt->depth++; |
7192 | 0 | ret = xmlParseBalancedChunkMemoryInternal(ctxt, ent->content, |
7193 | 0 | user_data, &list); |
7194 | 0 | ctxt->depth--; |
7195 | |
|
7196 | 0 | } else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7197 | 0 | ctxt->depth++; |
7198 | 0 | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, ctxt->sax, |
7199 | 0 | user_data, ctxt->depth, ent->URI, |
7200 | 0 | ent->ExternalID, &list); |
7201 | 0 | ctxt->depth--; |
7202 | 0 | } else { |
7203 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7204 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7205 | 0 | "invalid entity type found\n", NULL); |
7206 | 0 | } |
7207 | | |
7208 | | /* |
7209 | | * Store the number of entities needing parsing for this entity |
7210 | | * content and do checkings |
7211 | | */ |
7212 | 0 | diff = ctxt->nbentities - oldnbent + 1; |
7213 | 0 | if (diff > INT_MAX / 2) |
7214 | 0 | diff = INT_MAX / 2; |
7215 | 0 | ent->checked = diff * 2; |
7216 | 0 | if ((ent->content != NULL) && (xmlStrchr(ent->content, '<'))) |
7217 | 0 | ent->checked |= 1; |
7218 | 0 | if (ret == XML_ERR_ENTITY_LOOP) { |
7219 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7220 | 0 | xmlHaltParser(ctxt); |
7221 | 0 | xmlFreeNodeList(list); |
7222 | 0 | return; |
7223 | 0 | } |
7224 | 0 | if (xmlParserEntityCheck(ctxt, 0, ent, 0)) { |
7225 | 0 | xmlFreeNodeList(list); |
7226 | 0 | return; |
7227 | 0 | } |
7228 | | |
7229 | 0 | if ((ret == XML_ERR_OK) && (list != NULL)) { |
7230 | 0 | if (((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7231 | 0 | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY))&& |
7232 | 0 | (ent->children == NULL)) { |
7233 | 0 | ent->children = list; |
7234 | | /* |
7235 | | * Prune it directly in the generated document |
7236 | | * except for single text nodes. |
7237 | | */ |
7238 | 0 | if ((ctxt->replaceEntities == 0) || |
7239 | 0 | (ctxt->parseMode == XML_PARSE_READER) || |
7240 | 0 | ((list->type == XML_TEXT_NODE) && |
7241 | 0 | (list->next == NULL))) { |
7242 | 0 | ent->owner = 1; |
7243 | 0 | while (list != NULL) { |
7244 | 0 | list->parent = (xmlNodePtr) ent; |
7245 | 0 | if (list->doc != ent->doc) |
7246 | 0 | xmlSetTreeDoc(list, ent->doc); |
7247 | 0 | if (list->next == NULL) |
7248 | 0 | ent->last = list; |
7249 | 0 | list = list->next; |
7250 | 0 | } |
7251 | 0 | list = NULL; |
7252 | 0 | } else { |
7253 | 0 | ent->owner = 0; |
7254 | 0 | while (list != NULL) { |
7255 | 0 | list->parent = (xmlNodePtr) ctxt->node; |
7256 | 0 | list->doc = ctxt->myDoc; |
7257 | 0 | if (list->next == NULL) |
7258 | 0 | ent->last = list; |
7259 | 0 | list = list->next; |
7260 | 0 | } |
7261 | 0 | list = ent->children; |
7262 | | #ifdef LIBXML_LEGACY_ENABLED |
7263 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7264 | | xmlAddEntityReference(ent, list, NULL); |
7265 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7266 | 0 | } |
7267 | 0 | } else { |
7268 | 0 | xmlFreeNodeList(list); |
7269 | 0 | list = NULL; |
7270 | 0 | } |
7271 | 0 | } else if ((ret != XML_ERR_OK) && |
7272 | 0 | (ret != XML_WAR_UNDECLARED_ENTITY)) { |
7273 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7274 | 0 | "Entity '%s' failed to parse\n", ent->name); |
7275 | 0 | if (ent->content != NULL) |
7276 | 0 | ent->content[0] = 0; |
7277 | 0 | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7278 | 0 | } else if (list != NULL) { |
7279 | 0 | xmlFreeNodeList(list); |
7280 | 0 | list = NULL; |
7281 | 0 | } |
7282 | 0 | if (ent->checked == 0) |
7283 | 0 | ent->checked = 2; |
7284 | | |
7285 | | /* Prevent entity from being parsed and expanded twice (Bug 760367). */ |
7286 | 0 | was_checked = 0; |
7287 | 0 | } else if (ent->checked != 1) { |
7288 | 0 | ctxt->nbentities += ent->checked / 2; |
7289 | 0 | } |
7290 | | |
7291 | | /* |
7292 | | * Now that the entity content has been gathered |
7293 | | * provide it to the application, this can take different forms based |
7294 | | * on the parsing modes. |
7295 | | */ |
7296 | 0 | if (ent->children == NULL) { |
7297 | | /* |
7298 | | * Probably running in SAX mode and the callbacks don't |
7299 | | * build the entity content. So unless we already went |
7300 | | * though parsing for first checking go though the entity |
7301 | | * content to generate callbacks associated to the entity |
7302 | | */ |
7303 | 0 | if (was_checked != 0) { |
7304 | 0 | void *user_data; |
7305 | | /* |
7306 | | * This is a bit hackish but this seems the best |
7307 | | * way to make sure both SAX and DOM entity support |
7308 | | * behaves okay. |
7309 | | */ |
7310 | 0 | if (ctxt->userData == ctxt) |
7311 | 0 | user_data = NULL; |
7312 | 0 | else |
7313 | 0 | user_data = ctxt->userData; |
7314 | |
|
7315 | 0 | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7316 | 0 | ctxt->depth++; |
7317 | 0 | ret = xmlParseBalancedChunkMemoryInternal(ctxt, |
7318 | 0 | ent->content, user_data, NULL); |
7319 | 0 | ctxt->depth--; |
7320 | 0 | } else if (ent->etype == |
7321 | 0 | XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7322 | 0 | ctxt->depth++; |
7323 | 0 | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, |
7324 | 0 | ctxt->sax, user_data, ctxt->depth, |
7325 | 0 | ent->URI, ent->ExternalID, NULL); |
7326 | 0 | ctxt->depth--; |
7327 | 0 | } else { |
7328 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7329 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7330 | 0 | "invalid entity type found\n", NULL); |
7331 | 0 | } |
7332 | 0 | if (ret == XML_ERR_ENTITY_LOOP) { |
7333 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7334 | 0 | return; |
7335 | 0 | } |
7336 | 0 | } |
7337 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7338 | 0 | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7339 | | /* |
7340 | | * Entity reference callback comes second, it's somewhat |
7341 | | * superfluous but a compatibility to historical behaviour |
7342 | | */ |
7343 | 0 | ctxt->sax->reference(ctxt->userData, ent->name); |
7344 | 0 | } |
7345 | 0 | return; |
7346 | 0 | } |
7347 | | |
7348 | | /* |
7349 | | * If we didn't get any children for the entity being built |
7350 | | */ |
7351 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7352 | 0 | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7353 | | /* |
7354 | | * Create a node. |
7355 | | */ |
7356 | 0 | ctxt->sax->reference(ctxt->userData, ent->name); |
7357 | 0 | return; |
7358 | 0 | } |
7359 | | |
7360 | 0 | if ((ctxt->replaceEntities) || (ent->children == NULL)) { |
7361 | | /* |
7362 | | * There is a problem on the handling of _private for entities |
7363 | | * (bug 155816): Should we copy the content of the field from |
7364 | | * the entity (possibly overwriting some value set by the user |
7365 | | * when a copy is created), should we leave it alone, or should |
7366 | | * we try to take care of different situations? The problem |
7367 | | * is exacerbated by the usage of this field by the xmlReader. |
7368 | | * To fix this bug, we look at _private on the created node |
7369 | | * and, if it's NULL, we copy in whatever was in the entity. |
7370 | | * If it's not NULL we leave it alone. This is somewhat of a |
7371 | | * hack - maybe we should have further tests to determine |
7372 | | * what to do. |
7373 | | */ |
7374 | 0 | if ((ctxt->node != NULL) && (ent->children != NULL)) { |
7375 | | /* |
7376 | | * Seems we are generating the DOM content, do |
7377 | | * a simple tree copy for all references except the first |
7378 | | * In the first occurrence list contains the replacement. |
7379 | | */ |
7380 | 0 | if (((list == NULL) && (ent->owner == 0)) || |
7381 | 0 | (ctxt->parseMode == XML_PARSE_READER)) { |
7382 | 0 | xmlNodePtr nw = NULL, cur, firstChild = NULL; |
7383 | | |
7384 | | /* |
7385 | | * We are copying here, make sure there is no abuse |
7386 | | */ |
7387 | 0 | ctxt->sizeentcopy += ent->length + 5; |
7388 | 0 | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7389 | 0 | return; |
7390 | | |
7391 | | /* |
7392 | | * when operating on a reader, the entities definitions |
7393 | | * are always owning the entities subtree. |
7394 | | if (ctxt->parseMode == XML_PARSE_READER) |
7395 | | ent->owner = 1; |
7396 | | */ |
7397 | | |
7398 | 0 | cur = ent->children; |
7399 | 0 | while (cur != NULL) { |
7400 | 0 | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7401 | 0 | if (nw != NULL) { |
7402 | 0 | if (nw->_private == NULL) |
7403 | 0 | nw->_private = cur->_private; |
7404 | 0 | if (firstChild == NULL){ |
7405 | 0 | firstChild = nw; |
7406 | 0 | } |
7407 | 0 | nw = xmlAddChild(ctxt->node, nw); |
7408 | 0 | } |
7409 | 0 | if (cur == ent->last) { |
7410 | | /* |
7411 | | * needed to detect some strange empty |
7412 | | * node cases in the reader tests |
7413 | | */ |
7414 | 0 | if ((ctxt->parseMode == XML_PARSE_READER) && |
7415 | 0 | (nw != NULL) && |
7416 | 0 | (nw->type == XML_ELEMENT_NODE) && |
7417 | 0 | (nw->children == NULL)) |
7418 | 0 | nw->extra = 1; |
7419 | |
|
7420 | 0 | break; |
7421 | 0 | } |
7422 | 0 | cur = cur->next; |
7423 | 0 | } |
7424 | | #ifdef LIBXML_LEGACY_ENABLED |
7425 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7426 | | xmlAddEntityReference(ent, firstChild, nw); |
7427 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7428 | 0 | } else if ((list == NULL) || (ctxt->inputNr > 0)) { |
7429 | 0 | xmlNodePtr nw = NULL, cur, next, last, |
7430 | 0 | firstChild = NULL; |
7431 | | |
7432 | | /* |
7433 | | * We are copying here, make sure there is no abuse |
7434 | | */ |
7435 | 0 | ctxt->sizeentcopy += ent->length + 5; |
7436 | 0 | if (xmlParserEntityCheck(ctxt, 0, ent, ctxt->sizeentcopy)) |
7437 | 0 | return; |
7438 | | |
7439 | | /* |
7440 | | * Copy the entity child list and make it the new |
7441 | | * entity child list. The goal is to make sure any |
7442 | | * ID or REF referenced will be the one from the |
7443 | | * document content and not the entity copy. |
7444 | | */ |
7445 | 0 | cur = ent->children; |
7446 | 0 | ent->children = NULL; |
7447 | 0 | last = ent->last; |
7448 | 0 | ent->last = NULL; |
7449 | 0 | while (cur != NULL) { |
7450 | 0 | next = cur->next; |
7451 | 0 | cur->next = NULL; |
7452 | 0 | cur->parent = NULL; |
7453 | 0 | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7454 | 0 | if (nw != NULL) { |
7455 | 0 | if (nw->_private == NULL) |
7456 | 0 | nw->_private = cur->_private; |
7457 | 0 | if (firstChild == NULL){ |
7458 | 0 | firstChild = cur; |
7459 | 0 | } |
7460 | 0 | xmlAddChild((xmlNodePtr) ent, nw); |
7461 | 0 | xmlAddChild(ctxt->node, cur); |
7462 | 0 | } |
7463 | 0 | if (cur == last) |
7464 | 0 | break; |
7465 | 0 | cur = next; |
7466 | 0 | } |
7467 | 0 | if (ent->owner == 0) |
7468 | 0 | ent->owner = 1; |
7469 | | #ifdef LIBXML_LEGACY_ENABLED |
7470 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7471 | | xmlAddEntityReference(ent, firstChild, nw); |
7472 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7473 | 0 | } else { |
7474 | 0 | const xmlChar *nbktext; |
7475 | | |
7476 | | /* |
7477 | | * the name change is to avoid coalescing of the |
7478 | | * node with a possible previous text one which |
7479 | | * would make ent->children a dangling pointer |
7480 | | */ |
7481 | 0 | nbktext = xmlDictLookup(ctxt->dict, BAD_CAST "nbktext", |
7482 | 0 | -1); |
7483 | 0 | if (ent->children->type == XML_TEXT_NODE) |
7484 | 0 | ent->children->name = nbktext; |
7485 | 0 | if ((ent->last != ent->children) && |
7486 | 0 | (ent->last->type == XML_TEXT_NODE)) |
7487 | 0 | ent->last->name = nbktext; |
7488 | 0 | xmlAddChildList(ctxt->node, ent->children); |
7489 | 0 | } |
7490 | | |
7491 | | /* |
7492 | | * This is to avoid a nasty side effect, see |
7493 | | * characters() in SAX.c |
7494 | | */ |
7495 | 0 | ctxt->nodemem = 0; |
7496 | 0 | ctxt->nodelen = 0; |
7497 | 0 | return; |
7498 | 0 | } |
7499 | 0 | } |
7500 | 0 | } |
7501 | | |
7502 | | /** |
7503 | | * xmlParseEntityRef: |
7504 | | * @ctxt: an XML parser context |
7505 | | * |
7506 | | * parse ENTITY references declarations |
7507 | | * |
7508 | | * [68] EntityRef ::= '&' Name ';' |
7509 | | * |
7510 | | * [ WFC: Entity Declared ] |
7511 | | * In a document without any DTD, a document with only an internal DTD |
7512 | | * subset which contains no parameter entity references, or a document |
7513 | | * with "standalone='yes'", the Name given in the entity reference |
7514 | | * must match that in an entity declaration, except that well-formed |
7515 | | * documents need not declare any of the following entities: amp, lt, |
7516 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7517 | | * any reference to it. Similarly, the declaration of a general entity |
7518 | | * must precede any reference to it which appears in a default value in an |
7519 | | * attribute-list declaration. Note that if entities are declared in the |
7520 | | * external subset or in external parameter entities, a non-validating |
7521 | | * processor is not obligated to read and process their declarations; |
7522 | | * for such documents, the rule that an entity must be declared is a |
7523 | | * well-formedness constraint only if standalone='yes'. |
7524 | | * |
7525 | | * [ WFC: Parsed Entity ] |
7526 | | * An entity reference must not contain the name of an unparsed entity |
7527 | | * |
7528 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7529 | | */ |
7530 | | xmlEntityPtr |
7531 | 276k | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7532 | 276k | const xmlChar *name; |
7533 | 276k | xmlEntityPtr ent = NULL; |
7534 | | |
7535 | 276k | GROW; |
7536 | 276k | if (ctxt->instate == XML_PARSER_EOF) |
7537 | 0 | return(NULL); |
7538 | | |
7539 | 276k | if (RAW != '&') |
7540 | 0 | return(NULL); |
7541 | 276k | NEXT; |
7542 | 276k | name = xmlParseName(ctxt); |
7543 | 276k | if (name == NULL) { |
7544 | 176k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7545 | 176k | "xmlParseEntityRef: no name\n"); |
7546 | 176k | return(NULL); |
7547 | 176k | } |
7548 | 99.1k | if (RAW != ';') { |
7549 | 52.4k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7550 | 52.4k | return(NULL); |
7551 | 52.4k | } |
7552 | 46.6k | NEXT; |
7553 | | |
7554 | | /* |
7555 | | * Predefined entities override any extra definition |
7556 | | */ |
7557 | 46.6k | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7558 | 46.6k | ent = xmlGetPredefinedEntity(name); |
7559 | 46.6k | if (ent != NULL) |
7560 | 31.1k | return(ent); |
7561 | 46.6k | } |
7562 | | |
7563 | | /* |
7564 | | * Increase the number of entity references parsed |
7565 | | */ |
7566 | 15.5k | ctxt->nbentities++; |
7567 | | |
7568 | | /* |
7569 | | * Ask first SAX for entity resolution, otherwise try the |
7570 | | * entities which may have stored in the parser context. |
7571 | | */ |
7572 | 15.5k | if (ctxt->sax != NULL) { |
7573 | 15.5k | if (ctxt->sax->getEntity != NULL) |
7574 | 15.5k | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7575 | 15.5k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7576 | 15.5k | (ctxt->options & XML_PARSE_OLDSAX)) |
7577 | 0 | ent = xmlGetPredefinedEntity(name); |
7578 | 15.5k | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7579 | 15.5k | (ctxt->userData==ctxt)) { |
7580 | 0 | ent = xmlSAX2GetEntity(ctxt, name); |
7581 | 0 | } |
7582 | 15.5k | } |
7583 | 15.5k | if (ctxt->instate == XML_PARSER_EOF) |
7584 | 0 | return(NULL); |
7585 | | /* |
7586 | | * [ WFC: Entity Declared ] |
7587 | | * In a document without any DTD, a document with only an |
7588 | | * internal DTD subset which contains no parameter entity |
7589 | | * references, or a document with "standalone='yes'", the |
7590 | | * Name given in the entity reference must match that in an |
7591 | | * entity declaration, except that well-formed documents |
7592 | | * need not declare any of the following entities: amp, lt, |
7593 | | * gt, apos, quot. |
7594 | | * The declaration of a parameter entity must precede any |
7595 | | * reference to it. |
7596 | | * Similarly, the declaration of a general entity must |
7597 | | * precede any reference to it which appears in a default |
7598 | | * value in an attribute-list declaration. Note that if |
7599 | | * entities are declared in the external subset or in |
7600 | | * external parameter entities, a non-validating processor |
7601 | | * is not obligated to read and process their declarations; |
7602 | | * for such documents, the rule that an entity must be |
7603 | | * declared is a well-formedness constraint only if |
7604 | | * standalone='yes'. |
7605 | | */ |
7606 | 15.5k | if (ent == NULL) { |
7607 | 0 | if ((ctxt->standalone == 1) || |
7608 | 0 | ((ctxt->hasExternalSubset == 0) && |
7609 | 0 | (ctxt->hasPErefs == 0))) { |
7610 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7611 | 0 | "Entity '%s' not defined\n", name); |
7612 | 0 | } else { |
7613 | 0 | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7614 | 0 | "Entity '%s' not defined\n", name); |
7615 | 0 | if ((ctxt->inSubset == 0) && |
7616 | 0 | (ctxt->sax != NULL) && |
7617 | 0 | (ctxt->sax->reference != NULL)) { |
7618 | 0 | ctxt->sax->reference(ctxt->userData, name); |
7619 | 0 | } |
7620 | 0 | } |
7621 | 0 | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7622 | 0 | ctxt->valid = 0; |
7623 | 0 | } |
7624 | | |
7625 | | /* |
7626 | | * [ WFC: Parsed Entity ] |
7627 | | * An entity reference must not contain the name of an |
7628 | | * unparsed entity |
7629 | | */ |
7630 | 15.5k | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7631 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7632 | 0 | "Entity reference to unparsed entity %s\n", name); |
7633 | 0 | } |
7634 | | |
7635 | | /* |
7636 | | * [ WFC: No External Entity References ] |
7637 | | * Attribute values cannot contain direct or indirect |
7638 | | * entity references to external entities. |
7639 | | */ |
7640 | 15.5k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7641 | 15.5k | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7642 | 14.1k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7643 | 14.1k | "Attribute references external entity '%s'\n", name); |
7644 | 14.1k | } |
7645 | | /* |
7646 | | * [ WFC: No < in Attribute Values ] |
7647 | | * The replacement text of any entity referred to directly or |
7648 | | * indirectly in an attribute value (other than "<") must |
7649 | | * not contain a <. |
7650 | | */ |
7651 | 1.41k | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7652 | 1.41k | (ent != NULL) && |
7653 | 1.41k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
7654 | 0 | if (((ent->checked & 1) || (ent->checked == 0)) && |
7655 | 0 | (ent->content != NULL) && (xmlStrchr(ent->content, '<'))) { |
7656 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7657 | 0 | "'<' in entity '%s' is not allowed in attributes values\n", name); |
7658 | 0 | } |
7659 | 0 | } |
7660 | | |
7661 | | /* |
7662 | | * Internal check, no parameter entities here ... |
7663 | | */ |
7664 | 1.41k | else { |
7665 | 1.41k | switch (ent->etype) { |
7666 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7667 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7668 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7669 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7670 | 0 | name); |
7671 | 0 | break; |
7672 | 1.41k | default: |
7673 | 1.41k | break; |
7674 | 1.41k | } |
7675 | 1.41k | } |
7676 | | |
7677 | | /* |
7678 | | * [ WFC: No Recursion ] |
7679 | | * A parsed entity must not contain a recursive reference |
7680 | | * to itself, either directly or indirectly. |
7681 | | * Done somewhere else |
7682 | | */ |
7683 | 15.5k | return(ent); |
7684 | 15.5k | } |
7685 | | |
7686 | | /** |
7687 | | * xmlParseStringEntityRef: |
7688 | | * @ctxt: an XML parser context |
7689 | | * @str: a pointer to an index in the string |
7690 | | * |
7691 | | * parse ENTITY references declarations, but this version parses it from |
7692 | | * a string value. |
7693 | | * |
7694 | | * [68] EntityRef ::= '&' Name ';' |
7695 | | * |
7696 | | * [ WFC: Entity Declared ] |
7697 | | * In a document without any DTD, a document with only an internal DTD |
7698 | | * subset which contains no parameter entity references, or a document |
7699 | | * with "standalone='yes'", the Name given in the entity reference |
7700 | | * must match that in an entity declaration, except that well-formed |
7701 | | * documents need not declare any of the following entities: amp, lt, |
7702 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7703 | | * any reference to it. Similarly, the declaration of a general entity |
7704 | | * must precede any reference to it which appears in a default value in an |
7705 | | * attribute-list declaration. Note that if entities are declared in the |
7706 | | * external subset or in external parameter entities, a non-validating |
7707 | | * processor is not obligated to read and process their declarations; |
7708 | | * for such documents, the rule that an entity must be declared is a |
7709 | | * well-formedness constraint only if standalone='yes'. |
7710 | | * |
7711 | | * [ WFC: Parsed Entity ] |
7712 | | * An entity reference must not contain the name of an unparsed entity |
7713 | | * |
7714 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7715 | | * is updated to the current location in the string. |
7716 | | */ |
7717 | | static xmlEntityPtr |
7718 | 0 | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7719 | 0 | xmlChar *name; |
7720 | 0 | const xmlChar *ptr; |
7721 | 0 | xmlChar cur; |
7722 | 0 | xmlEntityPtr ent = NULL; |
7723 | |
|
7724 | 0 | if ((str == NULL) || (*str == NULL)) |
7725 | 0 | return(NULL); |
7726 | 0 | ptr = *str; |
7727 | 0 | cur = *ptr; |
7728 | 0 | if (cur != '&') |
7729 | 0 | return(NULL); |
7730 | | |
7731 | 0 | ptr++; |
7732 | 0 | name = xmlParseStringName(ctxt, &ptr); |
7733 | 0 | if (name == NULL) { |
7734 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7735 | 0 | "xmlParseStringEntityRef: no name\n"); |
7736 | 0 | *str = ptr; |
7737 | 0 | return(NULL); |
7738 | 0 | } |
7739 | 0 | if (*ptr != ';') { |
7740 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7741 | 0 | xmlFree(name); |
7742 | 0 | *str = ptr; |
7743 | 0 | return(NULL); |
7744 | 0 | } |
7745 | 0 | ptr++; |
7746 | | |
7747 | | |
7748 | | /* |
7749 | | * Predefined entities override any extra definition |
7750 | | */ |
7751 | 0 | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7752 | 0 | ent = xmlGetPredefinedEntity(name); |
7753 | 0 | if (ent != NULL) { |
7754 | 0 | xmlFree(name); |
7755 | 0 | *str = ptr; |
7756 | 0 | return(ent); |
7757 | 0 | } |
7758 | 0 | } |
7759 | | |
7760 | | /* |
7761 | | * Increase the number of entity references parsed |
7762 | | */ |
7763 | 0 | ctxt->nbentities++; |
7764 | | |
7765 | | /* |
7766 | | * Ask first SAX for entity resolution, otherwise try the |
7767 | | * entities which may have stored in the parser context. |
7768 | | */ |
7769 | 0 | if (ctxt->sax != NULL) { |
7770 | 0 | if (ctxt->sax->getEntity != NULL) |
7771 | 0 | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7772 | 0 | if ((ent == NULL) && (ctxt->options & XML_PARSE_OLDSAX)) |
7773 | 0 | ent = xmlGetPredefinedEntity(name); |
7774 | 0 | if ((ent == NULL) && (ctxt->userData==ctxt)) { |
7775 | 0 | ent = xmlSAX2GetEntity(ctxt, name); |
7776 | 0 | } |
7777 | 0 | } |
7778 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
7779 | 0 | xmlFree(name); |
7780 | 0 | return(NULL); |
7781 | 0 | } |
7782 | | |
7783 | | /* |
7784 | | * [ WFC: Entity Declared ] |
7785 | | * In a document without any DTD, a document with only an |
7786 | | * internal DTD subset which contains no parameter entity |
7787 | | * references, or a document with "standalone='yes'", the |
7788 | | * Name given in the entity reference must match that in an |
7789 | | * entity declaration, except that well-formed documents |
7790 | | * need not declare any of the following entities: amp, lt, |
7791 | | * gt, apos, quot. |
7792 | | * The declaration of a parameter entity must precede any |
7793 | | * reference to it. |
7794 | | * Similarly, the declaration of a general entity must |
7795 | | * precede any reference to it which appears in a default |
7796 | | * value in an attribute-list declaration. Note that if |
7797 | | * entities are declared in the external subset or in |
7798 | | * external parameter entities, a non-validating processor |
7799 | | * is not obligated to read and process their declarations; |
7800 | | * for such documents, the rule that an entity must be |
7801 | | * declared is a well-formedness constraint only if |
7802 | | * standalone='yes'. |
7803 | | */ |
7804 | 0 | if (ent == NULL) { |
7805 | 0 | if ((ctxt->standalone == 1) || |
7806 | 0 | ((ctxt->hasExternalSubset == 0) && |
7807 | 0 | (ctxt->hasPErefs == 0))) { |
7808 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7809 | 0 | "Entity '%s' not defined\n", name); |
7810 | 0 | } else { |
7811 | 0 | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7812 | 0 | "Entity '%s' not defined\n", |
7813 | 0 | name); |
7814 | 0 | } |
7815 | 0 | xmlParserEntityCheck(ctxt, 0, ent, 0); |
7816 | | /* TODO ? check regressions ctxt->valid = 0; */ |
7817 | 0 | } |
7818 | | |
7819 | | /* |
7820 | | * [ WFC: Parsed Entity ] |
7821 | | * An entity reference must not contain the name of an |
7822 | | * unparsed entity |
7823 | | */ |
7824 | 0 | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7825 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7826 | 0 | "Entity reference to unparsed entity %s\n", name); |
7827 | 0 | } |
7828 | | |
7829 | | /* |
7830 | | * [ WFC: No External Entity References ] |
7831 | | * Attribute values cannot contain direct or indirect |
7832 | | * entity references to external entities. |
7833 | | */ |
7834 | 0 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7835 | 0 | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7836 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7837 | 0 | "Attribute references external entity '%s'\n", name); |
7838 | 0 | } |
7839 | | /* |
7840 | | * [ WFC: No < in Attribute Values ] |
7841 | | * The replacement text of any entity referred to directly or |
7842 | | * indirectly in an attribute value (other than "<") must |
7843 | | * not contain a <. |
7844 | | */ |
7845 | 0 | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7846 | 0 | (ent != NULL) && (ent->content != NULL) && |
7847 | 0 | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
7848 | 0 | (xmlStrchr(ent->content, '<'))) { |
7849 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7850 | 0 | "'<' in entity '%s' is not allowed in attributes values\n", |
7851 | 0 | name); |
7852 | 0 | } |
7853 | | |
7854 | | /* |
7855 | | * Internal check, no parameter entities here ... |
7856 | | */ |
7857 | 0 | else { |
7858 | 0 | switch (ent->etype) { |
7859 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7860 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7861 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7862 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7863 | 0 | name); |
7864 | 0 | break; |
7865 | 0 | default: |
7866 | 0 | break; |
7867 | 0 | } |
7868 | 0 | } |
7869 | | |
7870 | | /* |
7871 | | * [ WFC: No Recursion ] |
7872 | | * A parsed entity must not contain a recursive reference |
7873 | | * to itself, either directly or indirectly. |
7874 | | * Done somewhere else |
7875 | | */ |
7876 | | |
7877 | 0 | xmlFree(name); |
7878 | 0 | *str = ptr; |
7879 | 0 | return(ent); |
7880 | 0 | } |
7881 | | |
7882 | | /** |
7883 | | * xmlParsePEReference: |
7884 | | * @ctxt: an XML parser context |
7885 | | * |
7886 | | * parse PEReference declarations |
7887 | | * The entity content is handled directly by pushing it's content as |
7888 | | * a new input stream. |
7889 | | * |
7890 | | * [69] PEReference ::= '%' Name ';' |
7891 | | * |
7892 | | * [ WFC: No Recursion ] |
7893 | | * A parsed entity must not contain a recursive |
7894 | | * reference to itself, either directly or indirectly. |
7895 | | * |
7896 | | * [ WFC: Entity Declared ] |
7897 | | * In a document without any DTD, a document with only an internal DTD |
7898 | | * subset which contains no parameter entity references, or a document |
7899 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7900 | | * entity must precede any reference to it... |
7901 | | * |
7902 | | * [ VC: Entity Declared ] |
7903 | | * In a document with an external subset or external parameter entities |
7904 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7905 | | * must precede any reference to it... |
7906 | | * |
7907 | | * [ WFC: In DTD ] |
7908 | | * Parameter-entity references may only appear in the DTD. |
7909 | | * NOTE: misleading but this is handled. |
7910 | | */ |
7911 | | void |
7912 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7913 | 43.4k | { |
7914 | 43.4k | const xmlChar *name; |
7915 | 43.4k | xmlEntityPtr entity = NULL; |
7916 | 43.4k | xmlParserInputPtr input; |
7917 | | |
7918 | 43.4k | if (RAW != '%') |
7919 | 40.7k | return; |
7920 | 2.66k | NEXT; |
7921 | 2.66k | name = xmlParseName(ctxt); |
7922 | 2.66k | if (name == NULL) { |
7923 | 1.49k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7924 | 1.49k | return; |
7925 | 1.49k | } |
7926 | 1.16k | if (xmlParserDebugEntities) |
7927 | 0 | xmlGenericError(xmlGenericErrorContext, |
7928 | 0 | "PEReference: %s\n", name); |
7929 | 1.16k | if (RAW != ';') { |
7930 | 720 | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7931 | 720 | return; |
7932 | 720 | } |
7933 | | |
7934 | 448 | NEXT; |
7935 | | |
7936 | | /* |
7937 | | * Increase the number of entity references parsed |
7938 | | */ |
7939 | 448 | ctxt->nbentities++; |
7940 | | |
7941 | | /* |
7942 | | * Request the entity from SAX |
7943 | | */ |
7944 | 448 | if ((ctxt->sax != NULL) && |
7945 | 448 | (ctxt->sax->getParameterEntity != NULL)) |
7946 | 0 | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7947 | 448 | if (ctxt->instate == XML_PARSER_EOF) |
7948 | 0 | return; |
7949 | 448 | if (entity == NULL) { |
7950 | | /* |
7951 | | * [ WFC: Entity Declared ] |
7952 | | * In a document without any DTD, a document with only an |
7953 | | * internal DTD subset which contains no parameter entity |
7954 | | * references, or a document with "standalone='yes'", ... |
7955 | | * ... The declaration of a parameter entity must precede |
7956 | | * any reference to it... |
7957 | | */ |
7958 | 448 | if ((ctxt->standalone == 1) || |
7959 | 448 | ((ctxt->hasExternalSubset == 0) && |
7960 | 448 | (ctxt->hasPErefs == 0))) { |
7961 | 14 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7962 | 14 | "PEReference: %%%s; not found\n", |
7963 | 14 | name); |
7964 | 434 | } else { |
7965 | | /* |
7966 | | * [ VC: Entity Declared ] |
7967 | | * In a document with an external subset or external |
7968 | | * parameter entities with "standalone='no'", ... |
7969 | | * ... The declaration of a parameter entity must |
7970 | | * precede any reference to it... |
7971 | | */ |
7972 | 434 | if ((ctxt->validate) && (ctxt->vctxt.error != NULL)) { |
7973 | 0 | xmlValidityError(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7974 | 0 | "PEReference: %%%s; not found\n", |
7975 | 0 | name, NULL); |
7976 | 0 | } else |
7977 | 434 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7978 | 434 | "PEReference: %%%s; not found\n", |
7979 | 434 | name, NULL); |
7980 | 434 | ctxt->valid = 0; |
7981 | 434 | } |
7982 | 448 | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
7983 | 448 | } else { |
7984 | | /* |
7985 | | * Internal checking in case the entity quest barfed |
7986 | | */ |
7987 | 0 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7988 | 0 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7989 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7990 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7991 | 0 | name, NULL); |
7992 | 0 | } else { |
7993 | 0 | xmlChar start[4]; |
7994 | 0 | xmlCharEncoding enc; |
7995 | |
|
7996 | 0 | if (xmlParserEntityCheck(ctxt, 0, entity, 0)) |
7997 | 0 | return; |
7998 | | |
7999 | 0 | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
8000 | 0 | ((ctxt->options & XML_PARSE_NOENT) == 0) && |
8001 | 0 | ((ctxt->options & XML_PARSE_DTDVALID) == 0) && |
8002 | 0 | ((ctxt->options & XML_PARSE_DTDLOAD) == 0) && |
8003 | 0 | ((ctxt->options & XML_PARSE_DTDATTR) == 0) && |
8004 | 0 | (ctxt->replaceEntities == 0) && |
8005 | 0 | (ctxt->validate == 0)) |
8006 | 0 | return; |
8007 | | |
8008 | 0 | input = xmlNewEntityInputStream(ctxt, entity); |
8009 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
8010 | 0 | xmlFreeInputStream(input); |
8011 | 0 | return; |
8012 | 0 | } |
8013 | | |
8014 | 0 | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
8015 | | /* |
8016 | | * Get the 4 first bytes and decode the charset |
8017 | | * if enc != XML_CHAR_ENCODING_NONE |
8018 | | * plug some encoding conversion routines. |
8019 | | * Note that, since we may have some non-UTF8 |
8020 | | * encoding (like UTF16, bug 135229), the 'length' |
8021 | | * is not known, but we can calculate based upon |
8022 | | * the amount of data in the buffer. |
8023 | | */ |
8024 | 0 | GROW |
8025 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
8026 | 0 | return; |
8027 | 0 | if ((ctxt->input->end - ctxt->input->cur)>=4) { |
8028 | 0 | start[0] = RAW; |
8029 | 0 | start[1] = NXT(1); |
8030 | 0 | start[2] = NXT(2); |
8031 | 0 | start[3] = NXT(3); |
8032 | 0 | enc = xmlDetectCharEncoding(start, 4); |
8033 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
8034 | 0 | xmlSwitchEncoding(ctxt, enc); |
8035 | 0 | } |
8036 | 0 | } |
8037 | |
|
8038 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
8039 | 0 | (IS_BLANK_CH(NXT(5)))) { |
8040 | 0 | xmlParseTextDecl(ctxt); |
8041 | 0 | } |
8042 | 0 | } |
8043 | 0 | } |
8044 | 0 | } |
8045 | 448 | ctxt->hasPErefs = 1; |
8046 | 448 | } |
8047 | | |
8048 | | /** |
8049 | | * xmlLoadEntityContent: |
8050 | | * @ctxt: an XML parser context |
8051 | | * @entity: an unloaded system entity |
8052 | | * |
8053 | | * Load the original content of the given system entity from the |
8054 | | * ExternalID/SystemID given. This is to be used for Included in Literal |
8055 | | * http://www.w3.org/TR/REC-xml/#inliteral processing of entities references |
8056 | | * |
8057 | | * Returns 0 in case of success and -1 in case of failure |
8058 | | */ |
8059 | | static int |
8060 | 0 | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
8061 | 0 | xmlParserInputPtr input; |
8062 | 0 | xmlBufferPtr buf; |
8063 | 0 | int l, c; |
8064 | 0 | int count = 0; |
8065 | |
|
8066 | 0 | if ((ctxt == NULL) || (entity == NULL) || |
8067 | 0 | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
8068 | 0 | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
8069 | 0 | (entity->content != NULL)) { |
8070 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8071 | 0 | "xmlLoadEntityContent parameter error"); |
8072 | 0 | return(-1); |
8073 | 0 | } |
8074 | | |
8075 | 0 | if (xmlParserDebugEntities) |
8076 | 0 | xmlGenericError(xmlGenericErrorContext, |
8077 | 0 | "Reading %s entity content input\n", entity->name); |
8078 | |
|
8079 | 0 | buf = xmlBufferCreate(); |
8080 | 0 | if (buf == NULL) { |
8081 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8082 | 0 | "xmlLoadEntityContent parameter error"); |
8083 | 0 | return(-1); |
8084 | 0 | } |
8085 | 0 | xmlBufferSetAllocationScheme(buf, XML_BUFFER_ALLOC_DOUBLEIT); |
8086 | |
|
8087 | 0 | input = xmlNewEntityInputStream(ctxt, entity); |
8088 | 0 | if (input == NULL) { |
8089 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8090 | 0 | "xmlLoadEntityContent input error"); |
8091 | 0 | xmlBufferFree(buf); |
8092 | 0 | return(-1); |
8093 | 0 | } |
8094 | | |
8095 | | /* |
8096 | | * Push the entity as the current input, read char by char |
8097 | | * saving to the buffer until the end of the entity or an error |
8098 | | */ |
8099 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
8100 | 0 | xmlBufferFree(buf); |
8101 | 0 | xmlFreeInputStream(input); |
8102 | 0 | return(-1); |
8103 | 0 | } |
8104 | | |
8105 | 0 | GROW; |
8106 | 0 | c = CUR_CHAR(l); |
8107 | 0 | while ((ctxt->input == input) && (ctxt->input->cur < ctxt->input->end) && |
8108 | 0 | (IS_CHAR(c))) { |
8109 | 0 | xmlBufferAdd(buf, ctxt->input->cur, l); |
8110 | 0 | if (count++ > XML_PARSER_CHUNK_SIZE) { |
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 | } |
8118 | 0 | NEXTL(l); |
8119 | 0 | c = CUR_CHAR(l); |
8120 | 0 | if (c == 0) { |
8121 | 0 | count = 0; |
8122 | 0 | GROW; |
8123 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
8124 | 0 | xmlBufferFree(buf); |
8125 | 0 | return(-1); |
8126 | 0 | } |
8127 | 0 | c = CUR_CHAR(l); |
8128 | 0 | } |
8129 | 0 | } |
8130 | | |
8131 | 0 | if ((ctxt->input == input) && (ctxt->input->cur >= ctxt->input->end)) { |
8132 | 0 | xmlPopInput(ctxt); |
8133 | 0 | } else if (!IS_CHAR(c)) { |
8134 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8135 | 0 | "xmlLoadEntityContent: invalid char value %d\n", |
8136 | 0 | c); |
8137 | 0 | xmlBufferFree(buf); |
8138 | 0 | return(-1); |
8139 | 0 | } |
8140 | 0 | entity->content = buf->content; |
8141 | 0 | buf->content = NULL; |
8142 | 0 | xmlBufferFree(buf); |
8143 | |
|
8144 | 0 | return(0); |
8145 | 0 | } |
8146 | | |
8147 | | /** |
8148 | | * xmlParseStringPEReference: |
8149 | | * @ctxt: an XML parser context |
8150 | | * @str: a pointer to an index in the string |
8151 | | * |
8152 | | * parse PEReference declarations |
8153 | | * |
8154 | | * [69] PEReference ::= '%' Name ';' |
8155 | | * |
8156 | | * [ WFC: No Recursion ] |
8157 | | * A parsed entity must not contain a recursive |
8158 | | * reference to itself, either directly or indirectly. |
8159 | | * |
8160 | | * [ WFC: Entity Declared ] |
8161 | | * In a document without any DTD, a document with only an internal DTD |
8162 | | * subset which contains no parameter entity references, or a document |
8163 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8164 | | * entity must precede any reference to it... |
8165 | | * |
8166 | | * [ VC: Entity Declared ] |
8167 | | * In a document with an external subset or external parameter entities |
8168 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8169 | | * must precede any reference to it... |
8170 | | * |
8171 | | * [ WFC: In DTD ] |
8172 | | * Parameter-entity references may only appear in the DTD. |
8173 | | * NOTE: misleading but this is handled. |
8174 | | * |
8175 | | * Returns the string of the entity content. |
8176 | | * str is updated to the current value of the index |
8177 | | */ |
8178 | | static xmlEntityPtr |
8179 | 0 | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8180 | 0 | const xmlChar *ptr; |
8181 | 0 | xmlChar cur; |
8182 | 0 | xmlChar *name; |
8183 | 0 | xmlEntityPtr entity = NULL; |
8184 | |
|
8185 | 0 | if ((str == NULL) || (*str == NULL)) return(NULL); |
8186 | 0 | ptr = *str; |
8187 | 0 | cur = *ptr; |
8188 | 0 | if (cur != '%') |
8189 | 0 | return(NULL); |
8190 | 0 | ptr++; |
8191 | 0 | name = xmlParseStringName(ctxt, &ptr); |
8192 | 0 | if (name == NULL) { |
8193 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8194 | 0 | "xmlParseStringPEReference: no name\n"); |
8195 | 0 | *str = ptr; |
8196 | 0 | return(NULL); |
8197 | 0 | } |
8198 | 0 | cur = *ptr; |
8199 | 0 | if (cur != ';') { |
8200 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8201 | 0 | xmlFree(name); |
8202 | 0 | *str = ptr; |
8203 | 0 | return(NULL); |
8204 | 0 | } |
8205 | 0 | ptr++; |
8206 | | |
8207 | | /* |
8208 | | * Increase the number of entity references parsed |
8209 | | */ |
8210 | 0 | ctxt->nbentities++; |
8211 | | |
8212 | | /* |
8213 | | * Request the entity from SAX |
8214 | | */ |
8215 | 0 | if ((ctxt->sax != NULL) && |
8216 | 0 | (ctxt->sax->getParameterEntity != NULL)) |
8217 | 0 | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8218 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
8219 | 0 | xmlFree(name); |
8220 | 0 | *str = ptr; |
8221 | 0 | return(NULL); |
8222 | 0 | } |
8223 | 0 | if (entity == NULL) { |
8224 | | /* |
8225 | | * [ WFC: Entity Declared ] |
8226 | | * In a document without any DTD, a document with only an |
8227 | | * internal DTD subset which contains no parameter entity |
8228 | | * references, or a document with "standalone='yes'", ... |
8229 | | * ... The declaration of a parameter entity must precede |
8230 | | * any reference to it... |
8231 | | */ |
8232 | 0 | if ((ctxt->standalone == 1) || |
8233 | 0 | ((ctxt->hasExternalSubset == 0) && (ctxt->hasPErefs == 0))) { |
8234 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8235 | 0 | "PEReference: %%%s; not found\n", name); |
8236 | 0 | } else { |
8237 | | /* |
8238 | | * [ VC: Entity Declared ] |
8239 | | * In a document with an external subset or external |
8240 | | * parameter entities with "standalone='no'", ... |
8241 | | * ... The declaration of a parameter entity must |
8242 | | * precede any reference to it... |
8243 | | */ |
8244 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8245 | 0 | "PEReference: %%%s; not found\n", |
8246 | 0 | name, NULL); |
8247 | 0 | ctxt->valid = 0; |
8248 | 0 | } |
8249 | 0 | xmlParserEntityCheck(ctxt, 0, NULL, 0); |
8250 | 0 | } else { |
8251 | | /* |
8252 | | * Internal checking in case the entity quest barfed |
8253 | | */ |
8254 | 0 | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8255 | 0 | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8256 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8257 | 0 | "%%%s; is not a parameter entity\n", |
8258 | 0 | name, NULL); |
8259 | 0 | } |
8260 | 0 | } |
8261 | 0 | ctxt->hasPErefs = 1; |
8262 | 0 | xmlFree(name); |
8263 | 0 | *str = ptr; |
8264 | 0 | return(entity); |
8265 | 0 | } |
8266 | | |
8267 | | /** |
8268 | | * xmlParseDocTypeDecl: |
8269 | | * @ctxt: an XML parser context |
8270 | | * |
8271 | | * parse a DOCTYPE declaration |
8272 | | * |
8273 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8274 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8275 | | * |
8276 | | * [ VC: Root Element Type ] |
8277 | | * The Name in the document type declaration must match the element |
8278 | | * type of the root element. |
8279 | | */ |
8280 | | |
8281 | | void |
8282 | 5.50k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8283 | 5.50k | const xmlChar *name = NULL; |
8284 | 5.50k | xmlChar *ExternalID = NULL; |
8285 | 5.50k | xmlChar *URI = NULL; |
8286 | | |
8287 | | /* |
8288 | | * We know that '<!DOCTYPE' has been detected. |
8289 | | */ |
8290 | 5.50k | SKIP(9); |
8291 | | |
8292 | 5.50k | SKIP_BLANKS; |
8293 | | |
8294 | | /* |
8295 | | * Parse the DOCTYPE name. |
8296 | | */ |
8297 | 5.50k | name = xmlParseName(ctxt); |
8298 | 5.50k | if (name == NULL) { |
8299 | 52 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8300 | 52 | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8301 | 52 | } |
8302 | 5.50k | ctxt->intSubName = name; |
8303 | | |
8304 | 5.50k | SKIP_BLANKS; |
8305 | | |
8306 | | /* |
8307 | | * Check for SystemID and ExternalID |
8308 | | */ |
8309 | 5.50k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8310 | | |
8311 | 5.50k | if ((URI != NULL) || (ExternalID != NULL)) { |
8312 | 299 | ctxt->hasExternalSubset = 1; |
8313 | 299 | } |
8314 | 5.50k | ctxt->extSubURI = URI; |
8315 | 5.50k | ctxt->extSubSystem = ExternalID; |
8316 | | |
8317 | 5.50k | SKIP_BLANKS; |
8318 | | |
8319 | | /* |
8320 | | * Create and update the internal subset. |
8321 | | */ |
8322 | 5.50k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8323 | 5.50k | (!ctxt->disableSAX)) |
8324 | 0 | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8325 | 5.50k | if (ctxt->instate == XML_PARSER_EOF) |
8326 | 0 | return; |
8327 | | |
8328 | | /* |
8329 | | * Is there any internal subset declarations ? |
8330 | | * they are handled separately in xmlParseInternalSubset() |
8331 | | */ |
8332 | 5.50k | if (RAW == '[') |
8333 | 4.84k | return; |
8334 | | |
8335 | | /* |
8336 | | * We should be at the end of the DOCTYPE declaration. |
8337 | | */ |
8338 | 656 | if (RAW != '>') { |
8339 | 423 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8340 | 423 | } |
8341 | 656 | NEXT; |
8342 | 656 | } |
8343 | | |
8344 | | /** |
8345 | | * xmlParseInternalSubset: |
8346 | | * @ctxt: an XML parser context |
8347 | | * |
8348 | | * parse the internal subset declaration |
8349 | | * |
8350 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8351 | | */ |
8352 | | |
8353 | | static void |
8354 | 4.50k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8355 | | /* |
8356 | | * Is there any DTD definition ? |
8357 | | */ |
8358 | 4.50k | if (RAW == '[') { |
8359 | 4.50k | int baseInputNr = ctxt->inputNr; |
8360 | 4.50k | ctxt->instate = XML_PARSER_DTD; |
8361 | 4.50k | NEXT; |
8362 | | /* |
8363 | | * Parse the succession of Markup declarations and |
8364 | | * PEReferences. |
8365 | | * Subsequence (markupdecl | PEReference | S)* |
8366 | | */ |
8367 | 46.2k | while (((RAW != ']') || (ctxt->inputNr > baseInputNr)) && |
8368 | 46.2k | (ctxt->instate != XML_PARSER_EOF)) { |
8369 | 43.4k | int id = ctxt->input->id; |
8370 | 43.4k | unsigned long cons = CUR_CONSUMED; |
8371 | | |
8372 | 43.4k | SKIP_BLANKS; |
8373 | 43.4k | xmlParseMarkupDecl(ctxt); |
8374 | 43.4k | xmlParsePEReference(ctxt); |
8375 | | |
8376 | | /* |
8377 | | * Conditional sections are allowed from external entities included |
8378 | | * by PE References in the internal subset. |
8379 | | */ |
8380 | 43.4k | if ((ctxt->inputNr > 1) && (ctxt->input->filename != NULL) && |
8381 | 43.4k | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8382 | 0 | xmlParseConditionalSections(ctxt); |
8383 | 0 | } |
8384 | | |
8385 | 43.4k | if ((id == ctxt->input->id) && (cons == CUR_CONSUMED)) { |
8386 | 1.72k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8387 | 1.72k | "xmlParseInternalSubset: error detected in Markup declaration\n"); |
8388 | 1.72k | if (ctxt->inputNr > baseInputNr) |
8389 | 0 | xmlPopInput(ctxt); |
8390 | 1.72k | else |
8391 | 1.72k | break; |
8392 | 1.72k | } |
8393 | 43.4k | } |
8394 | 4.50k | if (RAW == ']') { |
8395 | 1.78k | NEXT; |
8396 | 1.78k | SKIP_BLANKS; |
8397 | 1.78k | } |
8398 | 4.50k | } |
8399 | | |
8400 | | /* |
8401 | | * We should be at the end of the DOCTYPE declaration. |
8402 | | */ |
8403 | 4.50k | if (RAW != '>') { |
8404 | 2.71k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8405 | 2.71k | return; |
8406 | 2.71k | } |
8407 | 1.79k | NEXT; |
8408 | 1.79k | } |
8409 | | |
8410 | | #ifdef LIBXML_SAX1_ENABLED |
8411 | | /** |
8412 | | * xmlParseAttribute: |
8413 | | * @ctxt: an XML parser context |
8414 | | * @value: a xmlChar ** used to store the value of the attribute |
8415 | | * |
8416 | | * parse an attribute |
8417 | | * |
8418 | | * [41] Attribute ::= Name Eq AttValue |
8419 | | * |
8420 | | * [ WFC: No External Entity References ] |
8421 | | * Attribute values cannot contain direct or indirect entity references |
8422 | | * to external entities. |
8423 | | * |
8424 | | * [ WFC: No < in Attribute Values ] |
8425 | | * The replacement text of any entity referred to directly or indirectly in |
8426 | | * an attribute value (other than "<") must not contain a <. |
8427 | | * |
8428 | | * [ VC: Attribute Value Type ] |
8429 | | * The attribute must have been declared; the value must be of the type |
8430 | | * declared for it. |
8431 | | * |
8432 | | * [25] Eq ::= S? '=' S? |
8433 | | * |
8434 | | * With namespace: |
8435 | | * |
8436 | | * [NS 11] Attribute ::= QName Eq AttValue |
8437 | | * |
8438 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8439 | | * definition. |
8440 | | * |
8441 | | * Returns the attribute name, and the value in *value. |
8442 | | */ |
8443 | | |
8444 | | const xmlChar * |
8445 | 0 | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8446 | 0 | const xmlChar *name; |
8447 | 0 | xmlChar *val; |
8448 | |
|
8449 | 0 | *value = NULL; |
8450 | 0 | GROW; |
8451 | 0 | name = xmlParseName(ctxt); |
8452 | 0 | if (name == NULL) { |
8453 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8454 | 0 | "error parsing attribute name\n"); |
8455 | 0 | return(NULL); |
8456 | 0 | } |
8457 | | |
8458 | | /* |
8459 | | * read the value |
8460 | | */ |
8461 | 0 | SKIP_BLANKS; |
8462 | 0 | if (RAW == '=') { |
8463 | 0 | NEXT; |
8464 | 0 | SKIP_BLANKS; |
8465 | 0 | val = xmlParseAttValue(ctxt); |
8466 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
8467 | 0 | } else { |
8468 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8469 | 0 | "Specification mandates value for attribute %s\n", name); |
8470 | 0 | return(NULL); |
8471 | 0 | } |
8472 | | |
8473 | | /* |
8474 | | * Check that xml:lang conforms to the specification |
8475 | | * No more registered as an error, just generate a warning now |
8476 | | * since this was deprecated in XML second edition |
8477 | | */ |
8478 | 0 | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8479 | 0 | if (!xmlCheckLanguageID(val)) { |
8480 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8481 | 0 | "Malformed value for xml:lang : %s\n", |
8482 | 0 | val, NULL); |
8483 | 0 | } |
8484 | 0 | } |
8485 | | |
8486 | | /* |
8487 | | * Check that xml:space conforms to the specification |
8488 | | */ |
8489 | 0 | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8490 | 0 | if (xmlStrEqual(val, BAD_CAST "default")) |
8491 | 0 | *(ctxt->space) = 0; |
8492 | 0 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8493 | 0 | *(ctxt->space) = 1; |
8494 | 0 | else { |
8495 | 0 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8496 | 0 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8497 | 0 | val, NULL); |
8498 | 0 | } |
8499 | 0 | } |
8500 | |
|
8501 | 0 | *value = val; |
8502 | 0 | return(name); |
8503 | 0 | } |
8504 | | |
8505 | | /** |
8506 | | * xmlParseStartTag: |
8507 | | * @ctxt: an XML parser context |
8508 | | * |
8509 | | * parse a start of tag either for rule element or |
8510 | | * EmptyElement. In both case we don't parse the tag closing chars. |
8511 | | * |
8512 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8513 | | * |
8514 | | * [ WFC: Unique Att Spec ] |
8515 | | * No attribute name may appear more than once in the same start-tag or |
8516 | | * empty-element tag. |
8517 | | * |
8518 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8519 | | * |
8520 | | * [ WFC: Unique Att Spec ] |
8521 | | * No attribute name may appear more than once in the same start-tag or |
8522 | | * empty-element tag. |
8523 | | * |
8524 | | * With namespace: |
8525 | | * |
8526 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8527 | | * |
8528 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8529 | | * |
8530 | | * Returns the element name parsed |
8531 | | */ |
8532 | | |
8533 | | const xmlChar * |
8534 | 0 | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8535 | 0 | const xmlChar *name; |
8536 | 0 | const xmlChar *attname; |
8537 | 0 | xmlChar *attvalue; |
8538 | 0 | const xmlChar **atts = ctxt->atts; |
8539 | 0 | int nbatts = 0; |
8540 | 0 | int maxatts = ctxt->maxatts; |
8541 | 0 | int i; |
8542 | |
|
8543 | 0 | if (RAW != '<') return(NULL); |
8544 | 0 | NEXT1; |
8545 | |
|
8546 | 0 | name = xmlParseName(ctxt); |
8547 | 0 | if (name == NULL) { |
8548 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8549 | 0 | "xmlParseStartTag: invalid element name\n"); |
8550 | 0 | return(NULL); |
8551 | 0 | } |
8552 | | |
8553 | | /* |
8554 | | * Now parse the attributes, it ends up with the ending |
8555 | | * |
8556 | | * (S Attribute)* S? |
8557 | | */ |
8558 | 0 | SKIP_BLANKS; |
8559 | 0 | GROW; |
8560 | |
|
8561 | 0 | while (((RAW != '>') && |
8562 | 0 | ((RAW != '/') || (NXT(1) != '>')) && |
8563 | 0 | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
8564 | 0 | int id = ctxt->input->id; |
8565 | 0 | unsigned long cons = CUR_CONSUMED; |
8566 | |
|
8567 | 0 | attname = xmlParseAttribute(ctxt, &attvalue); |
8568 | 0 | if ((attname != NULL) && (attvalue != NULL)) { |
8569 | | /* |
8570 | | * [ WFC: Unique Att Spec ] |
8571 | | * No attribute name may appear more than once in the same |
8572 | | * start-tag or empty-element tag. |
8573 | | */ |
8574 | 0 | for (i = 0; i < nbatts;i += 2) { |
8575 | 0 | if (xmlStrEqual(atts[i], attname)) { |
8576 | 0 | xmlErrAttributeDup(ctxt, NULL, attname); |
8577 | 0 | xmlFree(attvalue); |
8578 | 0 | goto failed; |
8579 | 0 | } |
8580 | 0 | } |
8581 | | /* |
8582 | | * Add the pair to atts |
8583 | | */ |
8584 | 0 | if (atts == NULL) { |
8585 | 0 | maxatts = 22; /* allow for 10 attrs by default */ |
8586 | 0 | atts = (const xmlChar **) |
8587 | 0 | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8588 | 0 | if (atts == NULL) { |
8589 | 0 | xmlErrMemory(ctxt, NULL); |
8590 | 0 | if (attvalue != NULL) |
8591 | 0 | xmlFree(attvalue); |
8592 | 0 | goto failed; |
8593 | 0 | } |
8594 | 0 | ctxt->atts = atts; |
8595 | 0 | ctxt->maxatts = maxatts; |
8596 | 0 | } else if (nbatts + 4 > maxatts) { |
8597 | 0 | const xmlChar **n; |
8598 | |
|
8599 | 0 | maxatts *= 2; |
8600 | 0 | n = (const xmlChar **) xmlRealloc((void *) atts, |
8601 | 0 | maxatts * sizeof(const xmlChar *)); |
8602 | 0 | if (n == NULL) { |
8603 | 0 | xmlErrMemory(ctxt, NULL); |
8604 | 0 | if (attvalue != NULL) |
8605 | 0 | xmlFree(attvalue); |
8606 | 0 | goto failed; |
8607 | 0 | } |
8608 | 0 | atts = n; |
8609 | 0 | ctxt->atts = atts; |
8610 | 0 | ctxt->maxatts = maxatts; |
8611 | 0 | } |
8612 | 0 | atts[nbatts++] = attname; |
8613 | 0 | atts[nbatts++] = attvalue; |
8614 | 0 | atts[nbatts] = NULL; |
8615 | 0 | atts[nbatts + 1] = NULL; |
8616 | 0 | } else { |
8617 | 0 | if (attvalue != NULL) |
8618 | 0 | xmlFree(attvalue); |
8619 | 0 | } |
8620 | | |
8621 | 0 | failed: |
8622 | |
|
8623 | 0 | GROW |
8624 | 0 | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8625 | 0 | break; |
8626 | 0 | if (SKIP_BLANKS == 0) { |
8627 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8628 | 0 | "attributes construct error\n"); |
8629 | 0 | } |
8630 | 0 | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
8631 | 0 | (attname == NULL) && (attvalue == NULL)) { |
8632 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
8633 | 0 | "xmlParseStartTag: problem parsing attributes\n"); |
8634 | 0 | break; |
8635 | 0 | } |
8636 | 0 | SHRINK; |
8637 | 0 | GROW; |
8638 | 0 | } |
8639 | | |
8640 | | /* |
8641 | | * SAX: Start of Element ! |
8642 | | */ |
8643 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8644 | 0 | (!ctxt->disableSAX)) { |
8645 | 0 | if (nbatts > 0) |
8646 | 0 | ctxt->sax->startElement(ctxt->userData, name, atts); |
8647 | 0 | else |
8648 | 0 | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8649 | 0 | } |
8650 | |
|
8651 | 0 | if (atts != NULL) { |
8652 | | /* Free only the content strings */ |
8653 | 0 | for (i = 1;i < nbatts;i+=2) |
8654 | 0 | if (atts[i] != NULL) |
8655 | 0 | xmlFree((xmlChar *) atts[i]); |
8656 | 0 | } |
8657 | 0 | return(name); |
8658 | 0 | } |
8659 | | |
8660 | | /** |
8661 | | * xmlParseEndTag1: |
8662 | | * @ctxt: an XML parser context |
8663 | | * @line: line of the start tag |
8664 | | * @nsNr: number of namespaces on the start tag |
8665 | | * |
8666 | | * parse an end of tag |
8667 | | * |
8668 | | * [42] ETag ::= '</' Name S? '>' |
8669 | | * |
8670 | | * With namespace |
8671 | | * |
8672 | | * [NS 9] ETag ::= '</' QName S? '>' |
8673 | | */ |
8674 | | |
8675 | | static void |
8676 | 0 | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8677 | 0 | const xmlChar *name; |
8678 | |
|
8679 | 0 | GROW; |
8680 | 0 | if ((RAW != '<') || (NXT(1) != '/')) { |
8681 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8682 | 0 | "xmlParseEndTag: '</' not found\n"); |
8683 | 0 | return; |
8684 | 0 | } |
8685 | 0 | SKIP(2); |
8686 | |
|
8687 | 0 | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8688 | | |
8689 | | /* |
8690 | | * We should definitely be at the ending "S? '>'" part |
8691 | | */ |
8692 | 0 | GROW; |
8693 | 0 | SKIP_BLANKS; |
8694 | 0 | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8695 | 0 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8696 | 0 | } else |
8697 | 0 | NEXT1; |
8698 | | |
8699 | | /* |
8700 | | * [ WFC: Element Type Match ] |
8701 | | * The Name in an element's end-tag must match the element type in the |
8702 | | * start-tag. |
8703 | | * |
8704 | | */ |
8705 | 0 | if (name != (xmlChar*)1) { |
8706 | 0 | if (name == NULL) name = BAD_CAST "unparsable"; |
8707 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8708 | 0 | "Opening and ending tag mismatch: %s line %d and %s\n", |
8709 | 0 | ctxt->name, line, name); |
8710 | 0 | } |
8711 | | |
8712 | | /* |
8713 | | * SAX: End of Tag |
8714 | | */ |
8715 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8716 | 0 | (!ctxt->disableSAX)) |
8717 | 0 | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8718 | |
|
8719 | 0 | namePop(ctxt); |
8720 | 0 | spacePop(ctxt); |
8721 | 0 | return; |
8722 | 0 | } |
8723 | | |
8724 | | /** |
8725 | | * xmlParseEndTag: |
8726 | | * @ctxt: an XML parser context |
8727 | | * |
8728 | | * parse an end of tag |
8729 | | * |
8730 | | * [42] ETag ::= '</' Name S? '>' |
8731 | | * |
8732 | | * With namespace |
8733 | | * |
8734 | | * [NS 9] ETag ::= '</' QName S? '>' |
8735 | | */ |
8736 | | |
8737 | | void |
8738 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8739 | 0 | xmlParseEndTag1(ctxt, 0); |
8740 | 0 | } |
8741 | | #endif /* LIBXML_SAX1_ENABLED */ |
8742 | | |
8743 | | /************************************************************************ |
8744 | | * * |
8745 | | * SAX 2 specific operations * |
8746 | | * * |
8747 | | ************************************************************************/ |
8748 | | |
8749 | | /* |
8750 | | * xmlGetNamespace: |
8751 | | * @ctxt: an XML parser context |
8752 | | * @prefix: the prefix to lookup |
8753 | | * |
8754 | | * Lookup the namespace name for the @prefix (which ca be NULL) |
8755 | | * The prefix must come from the @ctxt->dict dictionary |
8756 | | * |
8757 | | * Returns the namespace name or NULL if not bound |
8758 | | */ |
8759 | | static const xmlChar * |
8760 | 1.76M | xmlGetNamespace(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
8761 | 1.76M | int i; |
8762 | | |
8763 | 1.76M | if (prefix == ctxt->str_xml) return(ctxt->str_xml_ns); |
8764 | 2.46M | for (i = ctxt->nsNr - 2;i >= 0;i-=2) |
8765 | 722k | if (ctxt->nsTab[i] == prefix) { |
8766 | 22.7k | if ((prefix == NULL) && (*ctxt->nsTab[i + 1] == 0)) |
8767 | 2.40k | return(NULL); |
8768 | 20.3k | return(ctxt->nsTab[i + 1]); |
8769 | 22.7k | } |
8770 | 1.74M | return(NULL); |
8771 | 1.76M | } |
8772 | | |
8773 | | /** |
8774 | | * xmlParseQName: |
8775 | | * @ctxt: an XML parser context |
8776 | | * @prefix: pointer to store the prefix part |
8777 | | * |
8778 | | * parse an XML Namespace QName |
8779 | | * |
8780 | | * [6] QName ::= (Prefix ':')? LocalPart |
8781 | | * [7] Prefix ::= NCName |
8782 | | * [8] LocalPart ::= NCName |
8783 | | * |
8784 | | * Returns the Name parsed or NULL |
8785 | | */ |
8786 | | |
8787 | | static const xmlChar * |
8788 | 2.48M | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8789 | 2.48M | const xmlChar *l, *p; |
8790 | | |
8791 | 2.48M | GROW; |
8792 | | |
8793 | 2.48M | l = xmlParseNCName(ctxt); |
8794 | 2.48M | if (l == NULL) { |
8795 | 15.5k | if (CUR == ':') { |
8796 | 14.5k | l = xmlParseName(ctxt); |
8797 | 14.5k | if (l != NULL) { |
8798 | 14.5k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8799 | 14.5k | "Failed to parse QName '%s'\n", l, NULL, NULL); |
8800 | 14.5k | *prefix = NULL; |
8801 | 14.5k | return(l); |
8802 | 14.5k | } |
8803 | 14.5k | } |
8804 | 995 | return(NULL); |
8805 | 15.5k | } |
8806 | 2.46M | if (CUR == ':') { |
8807 | 21.2k | NEXT; |
8808 | 21.2k | p = l; |
8809 | 21.2k | l = xmlParseNCName(ctxt); |
8810 | 21.2k | if (l == NULL) { |
8811 | 2.05k | xmlChar *tmp; |
8812 | | |
8813 | 2.05k | if (ctxt->instate == XML_PARSER_EOF) |
8814 | 0 | return(NULL); |
8815 | 2.05k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8816 | 2.05k | "Failed to parse QName '%s:'\n", p, NULL, NULL); |
8817 | 2.05k | l = xmlParseNmtoken(ctxt); |
8818 | 2.05k | if (l == NULL) { |
8819 | 1.09k | if (ctxt->instate == XML_PARSER_EOF) |
8820 | 0 | return(NULL); |
8821 | 1.09k | tmp = xmlBuildQName(BAD_CAST "", p, NULL, 0); |
8822 | 1.09k | } else { |
8823 | 961 | tmp = xmlBuildQName(l, p, NULL, 0); |
8824 | 961 | xmlFree((char *)l); |
8825 | 961 | } |
8826 | 2.05k | p = xmlDictLookup(ctxt->dict, tmp, -1); |
8827 | 2.05k | if (tmp != NULL) xmlFree(tmp); |
8828 | 2.05k | *prefix = NULL; |
8829 | 2.05k | return(p); |
8830 | 2.05k | } |
8831 | 19.2k | if (CUR == ':') { |
8832 | 2.55k | xmlChar *tmp; |
8833 | | |
8834 | 2.55k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8835 | 2.55k | "Failed to parse QName '%s:%s:'\n", p, l, NULL); |
8836 | 2.55k | NEXT; |
8837 | 2.55k | tmp = (xmlChar *) xmlParseName(ctxt); |
8838 | 2.55k | if (tmp != NULL) { |
8839 | 1.55k | tmp = xmlBuildQName(tmp, l, NULL, 0); |
8840 | 1.55k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8841 | 1.55k | if (tmp != NULL) xmlFree(tmp); |
8842 | 1.55k | *prefix = p; |
8843 | 1.55k | return(l); |
8844 | 1.55k | } |
8845 | 1.00k | if (ctxt->instate == XML_PARSER_EOF) |
8846 | 0 | return(NULL); |
8847 | 1.00k | tmp = xmlBuildQName(BAD_CAST "", l, NULL, 0); |
8848 | 1.00k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8849 | 1.00k | if (tmp != NULL) xmlFree(tmp); |
8850 | 1.00k | *prefix = p; |
8851 | 1.00k | return(l); |
8852 | 1.00k | } |
8853 | 16.6k | *prefix = p; |
8854 | 16.6k | } else |
8855 | 2.44M | *prefix = NULL; |
8856 | 2.46M | return(l); |
8857 | 2.46M | } |
8858 | | |
8859 | | /** |
8860 | | * xmlParseQNameAndCompare: |
8861 | | * @ctxt: an XML parser context |
8862 | | * @name: the localname |
8863 | | * @prefix: the prefix, if any. |
8864 | | * |
8865 | | * parse an XML name and compares for match |
8866 | | * (specialized for endtag parsing) |
8867 | | * |
8868 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8869 | | * and the name for mismatch |
8870 | | */ |
8871 | | |
8872 | | static const xmlChar * |
8873 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8874 | 742 | xmlChar const *prefix) { |
8875 | 742 | const xmlChar *cmp; |
8876 | 742 | const xmlChar *in; |
8877 | 742 | const xmlChar *ret; |
8878 | 742 | const xmlChar *prefix2; |
8879 | | |
8880 | 742 | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8881 | | |
8882 | 742 | GROW; |
8883 | 742 | in = ctxt->input->cur; |
8884 | | |
8885 | 742 | cmp = prefix; |
8886 | 1.74k | while (*in != 0 && *in == *cmp) { |
8887 | 1.00k | ++in; |
8888 | 1.00k | ++cmp; |
8889 | 1.00k | } |
8890 | 742 | if ((*cmp == 0) && (*in == ':')) { |
8891 | 718 | in++; |
8892 | 718 | cmp = name; |
8893 | 1.81k | while (*in != 0 && *in == *cmp) { |
8894 | 1.09k | ++in; |
8895 | 1.09k | ++cmp; |
8896 | 1.09k | } |
8897 | 718 | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8898 | | /* success */ |
8899 | 703 | ctxt->input->col += in - ctxt->input->cur; |
8900 | 703 | ctxt->input->cur = in; |
8901 | 703 | return((const xmlChar*) 1); |
8902 | 703 | } |
8903 | 718 | } |
8904 | | /* |
8905 | | * all strings coms from the dictionary, equality can be done directly |
8906 | | */ |
8907 | 39 | ret = xmlParseQName (ctxt, &prefix2); |
8908 | 39 | if ((ret == name) && (prefix == prefix2)) |
8909 | 9 | return((const xmlChar*) 1); |
8910 | 30 | return ret; |
8911 | 39 | } |
8912 | | |
8913 | | /** |
8914 | | * xmlParseAttValueInternal: |
8915 | | * @ctxt: an XML parser context |
8916 | | * @len: attribute len result |
8917 | | * @alloc: whether the attribute was reallocated as a new string |
8918 | | * @normalize: if 1 then further non-CDATA normalization must be done |
8919 | | * |
8920 | | * parse a value for an attribute. |
8921 | | * NOTE: if no normalization is needed, the routine will return pointers |
8922 | | * directly from the data buffer. |
8923 | | * |
8924 | | * 3.3.3 Attribute-Value Normalization: |
8925 | | * Before the value of an attribute is passed to the application or |
8926 | | * checked for validity, the XML processor must normalize it as follows: |
8927 | | * - a character reference is processed by appending the referenced |
8928 | | * character to the attribute value |
8929 | | * - an entity reference is processed by recursively processing the |
8930 | | * replacement text of the entity |
8931 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
8932 | | * appending #x20 to the normalized value, except that only a single |
8933 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
8934 | | * parsed entity or the literal entity value of an internal parsed entity |
8935 | | * - other characters are processed by appending them to the normalized value |
8936 | | * If the declared value is not CDATA, then the XML processor must further |
8937 | | * process the normalized attribute value by discarding any leading and |
8938 | | * trailing space (#x20) characters, and by replacing sequences of space |
8939 | | * (#x20) characters by a single space (#x20) character. |
8940 | | * All attributes for which no declaration has been read should be treated |
8941 | | * by a non-validating parser as if declared CDATA. |
8942 | | * |
8943 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
8944 | | * caller if it was copied, this can be detected by val[*len] == 0. |
8945 | | */ |
8946 | | |
8947 | | #define GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) \ |
8948 | 2.04k | const xmlChar *oldbase = ctxt->input->base;\ |
8949 | 2.04k | GROW;\ |
8950 | 2.04k | if (ctxt->instate == XML_PARSER_EOF)\ |
8951 | 2.04k | return(NULL);\ |
8952 | 2.04k | if (oldbase != ctxt->input->base) {\ |
8953 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase;\ |
8954 | 0 | start = start + delta;\ |
8955 | 0 | in = in + delta;\ |
8956 | 0 | }\ |
8957 | 2.04k | end = ctxt->input->end; |
8958 | | |
8959 | | static xmlChar * |
8960 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *len, int *alloc, |
8961 | | int normalize) |
8962 | 762k | { |
8963 | 762k | xmlChar limit = 0; |
8964 | 762k | const xmlChar *in = NULL, *start, *end, *last; |
8965 | 762k | xmlChar *ret = NULL; |
8966 | 762k | int line, col; |
8967 | 762k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
8968 | 762k | XML_MAX_HUGE_LENGTH : |
8969 | 762k | XML_MAX_TEXT_LENGTH; |
8970 | | |
8971 | 762k | GROW; |
8972 | 762k | in = (xmlChar *) CUR_PTR; |
8973 | 762k | line = ctxt->input->line; |
8974 | 762k | col = ctxt->input->col; |
8975 | 762k | if (*in != '"' && *in != '\'') { |
8976 | 942 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
8977 | 942 | return (NULL); |
8978 | 942 | } |
8979 | 761k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
8980 | | |
8981 | | /* |
8982 | | * try to handle in this routine the most common case where no |
8983 | | * allocation of a new string is required and where content is |
8984 | | * pure ASCII. |
8985 | | */ |
8986 | 761k | limit = *in++; |
8987 | 761k | col++; |
8988 | 761k | end = ctxt->input->end; |
8989 | 761k | start = in; |
8990 | 761k | if (in >= end) { |
8991 | 134 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
8992 | 134 | } |
8993 | 761k | if (normalize) { |
8994 | | /* |
8995 | | * Skip any leading spaces |
8996 | | */ |
8997 | 8.44k | while ((in < end) && (*in != limit) && |
8998 | 8.44k | ((*in == 0x20) || (*in == 0x9) || |
8999 | 7.09k | (*in == 0xA) || (*in == 0xD))) { |
9000 | 2.83k | if (*in == 0xA) { |
9001 | 553 | line++; col = 1; |
9002 | 2.27k | } else { |
9003 | 2.27k | col++; |
9004 | 2.27k | } |
9005 | 2.83k | in++; |
9006 | 2.83k | start = in; |
9007 | 2.83k | if (in >= end) { |
9008 | 22 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9009 | 22 | if ((in - start) > maxLength) { |
9010 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9011 | 0 | "AttValue length too long\n"); |
9012 | 0 | return(NULL); |
9013 | 0 | } |
9014 | 22 | } |
9015 | 2.83k | } |
9016 | 112k | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9017 | 112k | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9018 | 107k | col++; |
9019 | 107k | if ((*in++ == 0x20) && (*in == 0x20)) break; |
9020 | 107k | if (in >= end) { |
9021 | 49 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9022 | 49 | if ((in - start) > maxLength) { |
9023 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9024 | 0 | "AttValue length too long\n"); |
9025 | 0 | return(NULL); |
9026 | 0 | } |
9027 | 49 | } |
9028 | 107k | } |
9029 | 5.61k | last = in; |
9030 | | /* |
9031 | | * skip the trailing blanks |
9032 | | */ |
9033 | 6.42k | while ((last[-1] == 0x20) && (last > start)) last--; |
9034 | 13.7k | while ((in < end) && (*in != limit) && |
9035 | 13.7k | ((*in == 0x20) || (*in == 0x9) || |
9036 | 10.4k | (*in == 0xA) || (*in == 0xD))) { |
9037 | 8.09k | if (*in == 0xA) { |
9038 | 206 | line++, col = 1; |
9039 | 7.88k | } else { |
9040 | 7.88k | col++; |
9041 | 7.88k | } |
9042 | 8.09k | in++; |
9043 | 8.09k | if (in >= end) { |
9044 | 34 | const xmlChar *oldbase = ctxt->input->base; |
9045 | 34 | GROW; |
9046 | 34 | if (ctxt->instate == XML_PARSER_EOF) |
9047 | 0 | return(NULL); |
9048 | 34 | if (oldbase != ctxt->input->base) { |
9049 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase; |
9050 | 0 | start = start + delta; |
9051 | 0 | in = in + delta; |
9052 | 0 | last = last + delta; |
9053 | 0 | } |
9054 | 34 | end = ctxt->input->end; |
9055 | 34 | if ((in - start) > maxLength) { |
9056 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9057 | 0 | "AttValue length too long\n"); |
9058 | 0 | return(NULL); |
9059 | 0 | } |
9060 | 34 | } |
9061 | 8.09k | } |
9062 | 5.61k | if ((in - start) > maxLength) { |
9063 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9064 | 0 | "AttValue length too long\n"); |
9065 | 0 | return(NULL); |
9066 | 0 | } |
9067 | 5.61k | if (*in != limit) goto need_complex; |
9068 | 756k | } else { |
9069 | 3.35M | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9070 | 3.35M | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9071 | 2.59M | in++; |
9072 | 2.59M | col++; |
9073 | 2.59M | if (in >= end) { |
9074 | 1.83k | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9075 | 1.83k | if ((in - start) > maxLength) { |
9076 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9077 | 0 | "AttValue length too long\n"); |
9078 | 0 | return(NULL); |
9079 | 0 | } |
9080 | 1.83k | } |
9081 | 2.59M | } |
9082 | 756k | last = in; |
9083 | 756k | if ((in - start) > maxLength) { |
9084 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9085 | 0 | "AttValue length too long\n"); |
9086 | 0 | return(NULL); |
9087 | 0 | } |
9088 | 756k | if (*in != limit) goto need_complex; |
9089 | 756k | } |
9090 | 605k | in++; |
9091 | 605k | col++; |
9092 | 605k | if (len != NULL) { |
9093 | 585k | *len = last - start; |
9094 | 585k | ret = (xmlChar *) start; |
9095 | 585k | } else { |
9096 | 19.7k | if (alloc) *alloc = 1; |
9097 | 19.7k | ret = xmlStrndup(start, last - start); |
9098 | 19.7k | } |
9099 | 605k | CUR_PTR = in; |
9100 | 605k | ctxt->input->line = line; |
9101 | 605k | ctxt->input->col = col; |
9102 | 605k | if (alloc) *alloc = 0; |
9103 | 605k | return ret; |
9104 | 156k | need_complex: |
9105 | 156k | if (alloc) *alloc = 1; |
9106 | 156k | return xmlParseAttValueComplex(ctxt, len, normalize); |
9107 | 761k | } |
9108 | | |
9109 | | /** |
9110 | | * xmlParseAttribute2: |
9111 | | * @ctxt: an XML parser context |
9112 | | * @pref: the element prefix |
9113 | | * @elem: the element name |
9114 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
9115 | | * @value: a xmlChar ** used to store the value of the attribute |
9116 | | * @len: an int * to save the length of the attribute |
9117 | | * @alloc: an int * to indicate if the attribute was allocated |
9118 | | * |
9119 | | * parse an attribute in the new SAX2 framework. |
9120 | | * |
9121 | | * Returns the attribute name, and the value in *value, . |
9122 | | */ |
9123 | | |
9124 | | static const xmlChar * |
9125 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
9126 | | const xmlChar * pref, const xmlChar * elem, |
9127 | | const xmlChar ** prefix, xmlChar ** value, |
9128 | | int *len, int *alloc) |
9129 | 741k | { |
9130 | 741k | const xmlChar *name; |
9131 | 741k | xmlChar *val, *internal_val = NULL; |
9132 | 741k | int normalize = 0; |
9133 | | |
9134 | 741k | *value = NULL; |
9135 | 741k | GROW; |
9136 | 741k | name = xmlParseQName(ctxt, prefix); |
9137 | 741k | if (name == NULL) { |
9138 | 623 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9139 | 623 | "error parsing attribute name\n"); |
9140 | 623 | return (NULL); |
9141 | 623 | } |
9142 | | |
9143 | | /* |
9144 | | * get the type if needed |
9145 | | */ |
9146 | 741k | if (ctxt->attsSpecial != NULL) { |
9147 | 153k | int type; |
9148 | | |
9149 | 153k | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
9150 | 153k | pref, elem, *prefix, name); |
9151 | 153k | if (type != 0) |
9152 | 5.65k | normalize = 1; |
9153 | 153k | } |
9154 | | |
9155 | | /* |
9156 | | * read the value |
9157 | | */ |
9158 | 741k | SKIP_BLANKS; |
9159 | 741k | if (RAW == '=') { |
9160 | 740k | NEXT; |
9161 | 740k | SKIP_BLANKS; |
9162 | 740k | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize); |
9163 | 740k | if (normalize) { |
9164 | | /* |
9165 | | * Sometimes a second normalisation pass for spaces is needed |
9166 | | * but that only happens if charrefs or entities references |
9167 | | * have been used in the attribute value, i.e. the attribute |
9168 | | * value have been extracted in an allocated string already. |
9169 | | */ |
9170 | 5.63k | if (*alloc) { |
9171 | 2.44k | const xmlChar *val2; |
9172 | | |
9173 | 2.44k | val2 = xmlAttrNormalizeSpace2(ctxt, val, len); |
9174 | 2.44k | if ((val2 != NULL) && (val2 != val)) { |
9175 | 647 | xmlFree(val); |
9176 | 647 | val = (xmlChar *) val2; |
9177 | 647 | } |
9178 | 2.44k | } |
9179 | 5.63k | } |
9180 | 740k | ctxt->instate = XML_PARSER_CONTENT; |
9181 | 740k | } else { |
9182 | 478 | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
9183 | 478 | "Specification mandates value for attribute %s\n", |
9184 | 478 | name); |
9185 | 478 | return (NULL); |
9186 | 478 | } |
9187 | | |
9188 | 740k | if (*prefix == ctxt->str_xml) { |
9189 | | /* |
9190 | | * Check that xml:lang conforms to the specification |
9191 | | * No more registered as an error, just generate a warning now |
9192 | | * since this was deprecated in XML second edition |
9193 | | */ |
9194 | 1.40k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
9195 | 0 | internal_val = xmlStrndup(val, *len); |
9196 | 0 | if (!xmlCheckLanguageID(internal_val)) { |
9197 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
9198 | 0 | "Malformed value for xml:lang : %s\n", |
9199 | 0 | internal_val, NULL); |
9200 | 0 | } |
9201 | 0 | } |
9202 | | |
9203 | | /* |
9204 | | * Check that xml:space conforms to the specification |
9205 | | */ |
9206 | 1.40k | if (xmlStrEqual(name, BAD_CAST "space")) { |
9207 | 529 | internal_val = xmlStrndup(val, *len); |
9208 | 529 | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
9209 | 37 | *(ctxt->space) = 0; |
9210 | 492 | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
9211 | 328 | *(ctxt->space) = 1; |
9212 | 164 | else { |
9213 | 164 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
9214 | 164 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
9215 | 164 | internal_val, NULL); |
9216 | 164 | } |
9217 | 529 | } |
9218 | 1.40k | if (internal_val) { |
9219 | 528 | xmlFree(internal_val); |
9220 | 528 | } |
9221 | 1.40k | } |
9222 | | |
9223 | 740k | *value = val; |
9224 | 740k | return (name); |
9225 | 741k | } |
9226 | | /** |
9227 | | * xmlParseStartTag2: |
9228 | | * @ctxt: an XML parser context |
9229 | | * |
9230 | | * parse a start of tag either for rule element or |
9231 | | * EmptyElement. In both case we don't parse the tag closing chars. |
9232 | | * This routine is called when running SAX2 parsing |
9233 | | * |
9234 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
9235 | | * |
9236 | | * [ WFC: Unique Att Spec ] |
9237 | | * No attribute name may appear more than once in the same start-tag or |
9238 | | * empty-element tag. |
9239 | | * |
9240 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
9241 | | * |
9242 | | * [ WFC: Unique Att Spec ] |
9243 | | * No attribute name may appear more than once in the same start-tag or |
9244 | | * empty-element tag. |
9245 | | * |
9246 | | * With namespace: |
9247 | | * |
9248 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
9249 | | * |
9250 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
9251 | | * |
9252 | | * Returns the element name parsed |
9253 | | */ |
9254 | | |
9255 | | static const xmlChar * |
9256 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
9257 | 1.73M | const xmlChar **URI, int *tlen) { |
9258 | 1.73M | const xmlChar *localname; |
9259 | 1.73M | const xmlChar *prefix; |
9260 | 1.73M | const xmlChar *attname; |
9261 | 1.73M | const xmlChar *aprefix; |
9262 | 1.73M | const xmlChar *nsname; |
9263 | 1.73M | xmlChar *attvalue; |
9264 | 1.73M | const xmlChar **atts = ctxt->atts; |
9265 | 1.73M | int maxatts = ctxt->maxatts; |
9266 | 1.73M | int nratts, nbatts, nbdef, inputid; |
9267 | 1.73M | int i, j, nbNs, attval; |
9268 | 1.73M | unsigned long cur; |
9269 | 1.73M | int nsNr = ctxt->nsNr; |
9270 | | |
9271 | 1.73M | if (RAW != '<') return(NULL); |
9272 | 1.73M | NEXT1; |
9273 | | |
9274 | | /* |
9275 | | * NOTE: it is crucial with the SAX2 API to never call SHRINK beyond that |
9276 | | * point since the attribute values may be stored as pointers to |
9277 | | * the buffer and calling SHRINK would destroy them ! |
9278 | | * The Shrinking is only possible once the full set of attribute |
9279 | | * callbacks have been done. |
9280 | | */ |
9281 | 1.73M | SHRINK; |
9282 | 1.73M | cur = ctxt->input->cur - ctxt->input->base; |
9283 | 1.73M | inputid = ctxt->input->id; |
9284 | 1.73M | nbatts = 0; |
9285 | 1.73M | nratts = 0; |
9286 | 1.73M | nbdef = 0; |
9287 | 1.73M | nbNs = 0; |
9288 | 1.73M | attval = 0; |
9289 | | /* Forget any namespaces added during an earlier parse of this element. */ |
9290 | 1.73M | ctxt->nsNr = nsNr; |
9291 | | |
9292 | 1.73M | localname = xmlParseQName(ctxt, &prefix); |
9293 | 1.73M | if (localname == NULL) { |
9294 | 364 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9295 | 364 | "StartTag: invalid element name\n"); |
9296 | 364 | return(NULL); |
9297 | 364 | } |
9298 | 1.73M | *tlen = ctxt->input->cur - ctxt->input->base - cur; |
9299 | | |
9300 | | /* |
9301 | | * Now parse the attributes, it ends up with the ending |
9302 | | * |
9303 | | * (S Attribute)* S? |
9304 | | */ |
9305 | 1.73M | SKIP_BLANKS; |
9306 | 1.73M | GROW; |
9307 | | |
9308 | 2.38M | while (((RAW != '>') && |
9309 | 2.38M | ((RAW != '/') || (NXT(1) != '>')) && |
9310 | 2.38M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
9311 | 741k | int id = ctxt->input->id; |
9312 | 741k | unsigned long cons = CUR_CONSUMED; |
9313 | 741k | int len = -1, alloc = 0; |
9314 | | |
9315 | 741k | attname = xmlParseAttribute2(ctxt, prefix, localname, |
9316 | 741k | &aprefix, &attvalue, &len, &alloc); |
9317 | 741k | if ((attname == NULL) || (attvalue == NULL)) |
9318 | 1.17k | goto next_attr; |
9319 | 740k | if (len < 0) len = xmlStrlen(attvalue); |
9320 | | |
9321 | 740k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9322 | 41.2k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9323 | 41.2k | xmlURIPtr uri; |
9324 | | |
9325 | 41.2k | if (URL == NULL) { |
9326 | 0 | xmlErrMemory(ctxt, "dictionary allocation failure"); |
9327 | 0 | if ((attvalue != NULL) && (alloc != 0)) |
9328 | 0 | xmlFree(attvalue); |
9329 | 0 | localname = NULL; |
9330 | 0 | goto done; |
9331 | 0 | } |
9332 | 41.2k | if (*URL != 0) { |
9333 | 40.0k | uri = xmlParseURI((const char *) URL); |
9334 | 40.0k | if (uri == NULL) { |
9335 | 25.4k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9336 | 25.4k | "xmlns: '%s' is not a valid URI\n", |
9337 | 25.4k | URL, NULL, NULL); |
9338 | 25.4k | } else { |
9339 | 14.5k | if (uri->scheme == NULL) { |
9340 | 10.0k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9341 | 10.0k | "xmlns: URI %s is not absolute\n", |
9342 | 10.0k | URL, NULL, NULL); |
9343 | 10.0k | } |
9344 | 14.5k | xmlFreeURI(uri); |
9345 | 14.5k | } |
9346 | 40.0k | if (URL == ctxt->str_xml_ns) { |
9347 | 196 | if (attname != ctxt->str_xml) { |
9348 | 196 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9349 | 196 | "xml namespace URI cannot be the default namespace\n", |
9350 | 196 | NULL, NULL, NULL); |
9351 | 196 | } |
9352 | 196 | goto next_attr; |
9353 | 196 | } |
9354 | 39.8k | if ((len == 29) && |
9355 | 39.8k | (xmlStrEqual(URL, |
9356 | 496 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9357 | 196 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9358 | 196 | "reuse of the xmlns namespace name is forbidden\n", |
9359 | 196 | NULL, NULL, NULL); |
9360 | 196 | goto next_attr; |
9361 | 196 | } |
9362 | 39.8k | } |
9363 | | /* |
9364 | | * check that it's not a defined namespace |
9365 | | */ |
9366 | 44.3k | for (j = 1;j <= nbNs;j++) |
9367 | 33.8k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9368 | 30.3k | break; |
9369 | 40.8k | if (j <= nbNs) |
9370 | 30.3k | xmlErrAttributeDup(ctxt, NULL, attname); |
9371 | 10.5k | else |
9372 | 10.5k | if (nsPush(ctxt, NULL, URL) > 0) nbNs++; |
9373 | | |
9374 | 699k | } else if (aprefix == ctxt->str_xmlns) { |
9375 | 10.4k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9376 | 10.4k | xmlURIPtr uri; |
9377 | | |
9378 | 10.4k | if (attname == ctxt->str_xml) { |
9379 | 233 | if (URL != ctxt->str_xml_ns) { |
9380 | 167 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9381 | 167 | "xml namespace prefix mapped to wrong URI\n", |
9382 | 167 | NULL, NULL, NULL); |
9383 | 167 | } |
9384 | | /* |
9385 | | * Do not keep a namespace definition node |
9386 | | */ |
9387 | 233 | goto next_attr; |
9388 | 233 | } |
9389 | 10.1k | if (URL == ctxt->str_xml_ns) { |
9390 | 228 | if (attname != ctxt->str_xml) { |
9391 | 228 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9392 | 228 | "xml namespace URI mapped to wrong prefix\n", |
9393 | 228 | NULL, NULL, NULL); |
9394 | 228 | } |
9395 | 228 | goto next_attr; |
9396 | 228 | } |
9397 | 9.93k | if (attname == ctxt->str_xmlns) { |
9398 | 459 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9399 | 459 | "redefinition of the xmlns prefix is forbidden\n", |
9400 | 459 | NULL, NULL, NULL); |
9401 | 459 | goto next_attr; |
9402 | 459 | } |
9403 | 9.48k | if ((len == 29) && |
9404 | 9.48k | (xmlStrEqual(URL, |
9405 | 2.23k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9406 | 1.78k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9407 | 1.78k | "reuse of the xmlns namespace name is forbidden\n", |
9408 | 1.78k | NULL, NULL, NULL); |
9409 | 1.78k | goto next_attr; |
9410 | 1.78k | } |
9411 | 7.69k | if ((URL == NULL) || (URL[0] == 0)) { |
9412 | 473 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9413 | 473 | "xmlns:%s: Empty XML namespace is not allowed\n", |
9414 | 473 | attname, NULL, NULL); |
9415 | 473 | goto next_attr; |
9416 | 7.22k | } else { |
9417 | 7.22k | uri = xmlParseURI((const char *) URL); |
9418 | 7.22k | if (uri == NULL) { |
9419 | 3.93k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9420 | 3.93k | "xmlns:%s: '%s' is not a valid URI\n", |
9421 | 3.93k | attname, URL, NULL); |
9422 | 3.93k | } else { |
9423 | 3.28k | if ((ctxt->pedantic) && (uri->scheme == NULL)) { |
9424 | 0 | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9425 | 0 | "xmlns:%s: URI %s is not absolute\n", |
9426 | 0 | attname, URL, NULL); |
9427 | 0 | } |
9428 | 3.28k | xmlFreeURI(uri); |
9429 | 3.28k | } |
9430 | 7.22k | } |
9431 | | |
9432 | | /* |
9433 | | * check that it's not a defined namespace |
9434 | | */ |
9435 | 16.0k | for (j = 1;j <= nbNs;j++) |
9436 | 13.9k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9437 | 5.09k | break; |
9438 | 7.22k | if (j <= nbNs) |
9439 | 5.09k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9440 | 2.13k | else |
9441 | 2.13k | if (nsPush(ctxt, attname, URL) > 0) nbNs++; |
9442 | | |
9443 | 688k | } else { |
9444 | | /* |
9445 | | * Add the pair to atts |
9446 | | */ |
9447 | 688k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9448 | 4.08k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9449 | 0 | goto next_attr; |
9450 | 0 | } |
9451 | 4.08k | maxatts = ctxt->maxatts; |
9452 | 4.08k | atts = ctxt->atts; |
9453 | 4.08k | } |
9454 | 688k | ctxt->attallocs[nratts++] = alloc; |
9455 | 688k | atts[nbatts++] = attname; |
9456 | 688k | atts[nbatts++] = aprefix; |
9457 | | /* |
9458 | | * The namespace URI field is used temporarily to point at the |
9459 | | * base of the current input buffer for non-alloced attributes. |
9460 | | * When the input buffer is reallocated, all the pointers become |
9461 | | * invalid, but they can be reconstructed later. |
9462 | | */ |
9463 | 688k | if (alloc) |
9464 | 135k | atts[nbatts++] = NULL; |
9465 | 553k | else |
9466 | 553k | atts[nbatts++] = ctxt->input->base; |
9467 | 688k | atts[nbatts++] = attvalue; |
9468 | 688k | attvalue += len; |
9469 | 688k | atts[nbatts++] = attvalue; |
9470 | | /* |
9471 | | * tag if some deallocation is needed |
9472 | | */ |
9473 | 688k | if (alloc != 0) attval = 1; |
9474 | 688k | attvalue = NULL; /* moved into atts */ |
9475 | 688k | } |
9476 | | |
9477 | 741k | next_attr: |
9478 | 741k | if ((attvalue != NULL) && (alloc != 0)) { |
9479 | 19.3k | xmlFree(attvalue); |
9480 | 19.3k | attvalue = NULL; |
9481 | 19.3k | } |
9482 | | |
9483 | 741k | GROW |
9484 | 741k | if (ctxt->instate == XML_PARSER_EOF) |
9485 | 0 | break; |
9486 | 741k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9487 | 90.5k | break; |
9488 | 651k | if (SKIP_BLANKS == 0) { |
9489 | 4.73k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9490 | 4.73k | "attributes construct error\n"); |
9491 | 4.73k | break; |
9492 | 4.73k | } |
9493 | 646k | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id) && |
9494 | 646k | (attname == NULL) && (attvalue == NULL)) { |
9495 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9496 | 0 | "xmlParseStartTag: problem parsing attributes\n"); |
9497 | 0 | break; |
9498 | 0 | } |
9499 | 646k | GROW; |
9500 | 646k | } |
9501 | | |
9502 | 1.73M | if (ctxt->input->id != inputid) { |
9503 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9504 | 0 | "Unexpected change of input\n"); |
9505 | 0 | localname = NULL; |
9506 | 0 | goto done; |
9507 | 0 | } |
9508 | | |
9509 | | /* Reconstruct attribute value pointers. */ |
9510 | 2.42M | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9511 | 688k | if (atts[i+2] != NULL) { |
9512 | | /* |
9513 | | * Arithmetic on dangling pointers is technically undefined |
9514 | | * behavior, but well... |
9515 | | */ |
9516 | 553k | ptrdiff_t offset = ctxt->input->base - atts[i+2]; |
9517 | 553k | atts[i+2] = NULL; /* Reset repurposed namespace URI */ |
9518 | 553k | atts[i+3] += offset; /* value */ |
9519 | 553k | atts[i+4] += offset; /* valuend */ |
9520 | 553k | } |
9521 | 688k | } |
9522 | | |
9523 | | /* |
9524 | | * The attributes defaulting |
9525 | | */ |
9526 | 1.73M | if (ctxt->attsDefault != NULL) { |
9527 | 9.01k | xmlDefAttrsPtr defaults; |
9528 | | |
9529 | 9.01k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9530 | 9.01k | if (defaults != NULL) { |
9531 | 32.5k | for (i = 0;i < defaults->nbAttrs;i++) { |
9532 | 25.1k | attname = defaults->values[5 * i]; |
9533 | 25.1k | aprefix = defaults->values[5 * i + 1]; |
9534 | | |
9535 | | /* |
9536 | | * special work for namespaces defaulted defs |
9537 | | */ |
9538 | 25.1k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9539 | | /* |
9540 | | * check that it's not a defined namespace |
9541 | | */ |
9542 | 4.53k | for (j = 1;j <= nbNs;j++) |
9543 | 2.54k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9544 | 1.59k | break; |
9545 | 3.58k | if (j <= nbNs) continue; |
9546 | | |
9547 | 1.98k | nsname = xmlGetNamespace(ctxt, NULL); |
9548 | 1.98k | if (nsname != defaults->values[5 * i + 2]) { |
9549 | 1.51k | if (nsPush(ctxt, NULL, |
9550 | 1.51k | defaults->values[5 * i + 2]) > 0) |
9551 | 1.17k | nbNs++; |
9552 | 1.51k | } |
9553 | 21.5k | } else if (aprefix == ctxt->str_xmlns) { |
9554 | | /* |
9555 | | * check that it's not a defined namespace |
9556 | | */ |
9557 | 7.39k | for (j = 1;j <= nbNs;j++) |
9558 | 2.24k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9559 | 726 | break; |
9560 | 5.87k | if (j <= nbNs) continue; |
9561 | | |
9562 | 5.14k | nsname = xmlGetNamespace(ctxt, attname); |
9563 | 5.14k | if (nsname != defaults->values[2]) { |
9564 | 4.32k | if (nsPush(ctxt, attname, |
9565 | 4.32k | defaults->values[5 * i + 2]) > 0) |
9566 | 1.64k | nbNs++; |
9567 | 4.32k | } |
9568 | 15.6k | } else { |
9569 | | /* |
9570 | | * check that it's not a defined attribute |
9571 | | */ |
9572 | 485k | for (j = 0;j < nbatts;j+=5) { |
9573 | 470k | if ((attname == atts[j]) && (aprefix == atts[j+1])) |
9574 | 357 | break; |
9575 | 470k | } |
9576 | 15.6k | if (j < nbatts) continue; |
9577 | | |
9578 | 15.2k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9579 | 193 | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9580 | 0 | localname = NULL; |
9581 | 0 | goto done; |
9582 | 0 | } |
9583 | 193 | maxatts = ctxt->maxatts; |
9584 | 193 | atts = ctxt->atts; |
9585 | 193 | } |
9586 | 15.2k | atts[nbatts++] = attname; |
9587 | 15.2k | atts[nbatts++] = aprefix; |
9588 | 15.2k | if (aprefix == NULL) |
9589 | 6.31k | atts[nbatts++] = NULL; |
9590 | 8.98k | else |
9591 | 8.98k | atts[nbatts++] = xmlGetNamespace(ctxt, aprefix); |
9592 | 15.2k | atts[nbatts++] = defaults->values[5 * i + 2]; |
9593 | 15.2k | atts[nbatts++] = defaults->values[5 * i + 3]; |
9594 | 15.2k | if ((ctxt->standalone == 1) && |
9595 | 15.2k | (defaults->values[5 * i + 4] != NULL)) { |
9596 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9597 | 0 | "standalone: attribute %s on %s defaulted from external subset\n", |
9598 | 0 | attname, localname); |
9599 | 0 | } |
9600 | 15.2k | nbdef++; |
9601 | 15.2k | } |
9602 | 25.1k | } |
9603 | 7.47k | } |
9604 | 9.01k | } |
9605 | | |
9606 | | /* |
9607 | | * The attributes checkings |
9608 | | */ |
9609 | 2.44M | for (i = 0; i < nbatts;i += 5) { |
9610 | | /* |
9611 | | * The default namespace does not apply to attribute names. |
9612 | | */ |
9613 | 704k | if (atts[i + 1] != NULL) { |
9614 | 14.3k | nsname = xmlGetNamespace(ctxt, atts[i + 1]); |
9615 | 14.3k | if (nsname == NULL) { |
9616 | 10.7k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9617 | 10.7k | "Namespace prefix %s for %s on %s is not defined\n", |
9618 | 10.7k | atts[i + 1], atts[i], localname); |
9619 | 10.7k | } |
9620 | 14.3k | atts[i + 2] = nsname; |
9621 | 14.3k | } else |
9622 | 689k | nsname = NULL; |
9623 | | /* |
9624 | | * [ WFC: Unique Att Spec ] |
9625 | | * No attribute name may appear more than once in the same |
9626 | | * start-tag or empty-element tag. |
9627 | | * As extended by the Namespace in XML REC. |
9628 | | */ |
9629 | 2.79M | for (j = 0; j < i;j += 5) { |
9630 | 2.23M | if (atts[i] == atts[j]) { |
9631 | 200k | if (atts[i+1] == atts[j+1]) { |
9632 | 144k | xmlErrAttributeDup(ctxt, atts[i+1], atts[i]); |
9633 | 144k | break; |
9634 | 144k | } |
9635 | 55.5k | if ((nsname != NULL) && (atts[j + 2] == nsname)) { |
9636 | 402 | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9637 | 402 | "Namespaced Attribute %s in '%s' redefined\n", |
9638 | 402 | atts[i], nsname, NULL); |
9639 | 402 | break; |
9640 | 402 | } |
9641 | 55.5k | } |
9642 | 2.23M | } |
9643 | 704k | } |
9644 | | |
9645 | 1.73M | nsname = xmlGetNamespace(ctxt, prefix); |
9646 | 1.73M | if ((prefix != NULL) && (nsname == NULL)) { |
9647 | 3.01k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9648 | 3.01k | "Namespace prefix %s on %s is not defined\n", |
9649 | 3.01k | prefix, localname, NULL); |
9650 | 3.01k | } |
9651 | 1.73M | *pref = prefix; |
9652 | 1.73M | *URI = nsname; |
9653 | | |
9654 | | /* |
9655 | | * SAX: Start of Element ! |
9656 | | */ |
9657 | 1.73M | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9658 | 1.73M | (!ctxt->disableSAX)) { |
9659 | 1.73M | if (nbNs > 0) |
9660 | 8.83k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9661 | 8.83k | nsname, nbNs, &ctxt->nsTab[ctxt->nsNr - 2 * nbNs], |
9662 | 8.83k | nbatts / 5, nbdef, atts); |
9663 | 1.72M | else |
9664 | 1.72M | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9665 | 1.72M | nsname, 0, NULL, nbatts / 5, nbdef, atts); |
9666 | 1.73M | } |
9667 | | |
9668 | 1.73M | done: |
9669 | | /* |
9670 | | * Free up attribute allocated strings if needed |
9671 | | */ |
9672 | 1.73M | if (attval != 0) { |
9673 | 147k | for (i = 3,j = 0; j < nratts;i += 5,j++) |
9674 | 145k | if ((ctxt->attallocs[j] != 0) && (atts[i] != NULL)) |
9675 | 135k | xmlFree((xmlChar *) atts[i]); |
9676 | 1.71k | } |
9677 | | |
9678 | 1.73M | return(localname); |
9679 | 1.73M | } |
9680 | | |
9681 | | /** |
9682 | | * xmlParseEndTag2: |
9683 | | * @ctxt: an XML parser context |
9684 | | * @line: line of the start tag |
9685 | | * @nsNr: number of namespaces on the start tag |
9686 | | * |
9687 | | * parse an end of tag |
9688 | | * |
9689 | | * [42] ETag ::= '</' Name S? '>' |
9690 | | * |
9691 | | * With namespace |
9692 | | * |
9693 | | * [NS 9] ETag ::= '</' QName S? '>' |
9694 | | */ |
9695 | | |
9696 | | static void |
9697 | 30.7k | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9698 | 30.7k | const xmlChar *name; |
9699 | | |
9700 | 30.7k | GROW; |
9701 | 30.7k | if ((RAW != '<') || (NXT(1) != '/')) { |
9702 | 0 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9703 | 0 | return; |
9704 | 0 | } |
9705 | 30.7k | SKIP(2); |
9706 | | |
9707 | 30.7k | if (tag->prefix == NULL) |
9708 | 29.9k | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9709 | 742 | else |
9710 | 742 | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9711 | | |
9712 | | /* |
9713 | | * We should definitely be at the ending "S? '>'" part |
9714 | | */ |
9715 | 30.7k | GROW; |
9716 | 30.7k | if (ctxt->instate == XML_PARSER_EOF) |
9717 | 0 | return; |
9718 | 30.7k | SKIP_BLANKS; |
9719 | 30.7k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9720 | 186 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9721 | 186 | } else |
9722 | 30.5k | NEXT1; |
9723 | | |
9724 | | /* |
9725 | | * [ WFC: Element Type Match ] |
9726 | | * The Name in an element's end-tag must match the element type in the |
9727 | | * start-tag. |
9728 | | * |
9729 | | */ |
9730 | 30.7k | if (name != (xmlChar*)1) { |
9731 | 163 | if (name == NULL) name = BAD_CAST "unparsable"; |
9732 | 163 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9733 | 163 | "Opening and ending tag mismatch: %s line %d and %s\n", |
9734 | 163 | ctxt->name, tag->line, name); |
9735 | 163 | } |
9736 | | |
9737 | | /* |
9738 | | * SAX: End of Tag |
9739 | | */ |
9740 | 30.7k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9741 | 30.7k | (!ctxt->disableSAX)) |
9742 | 30.4k | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9743 | 30.4k | tag->URI); |
9744 | | |
9745 | 30.7k | spacePop(ctxt); |
9746 | 30.7k | if (tag->nsNr != 0) |
9747 | 196 | nsPop(ctxt, tag->nsNr); |
9748 | 30.7k | } |
9749 | | |
9750 | | /** |
9751 | | * xmlParseCDSect: |
9752 | | * @ctxt: an XML parser context |
9753 | | * |
9754 | | * Parse escaped pure raw content. |
9755 | | * |
9756 | | * [18] CDSect ::= CDStart CData CDEnd |
9757 | | * |
9758 | | * [19] CDStart ::= '<![CDATA[' |
9759 | | * |
9760 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9761 | | * |
9762 | | * [21] CDEnd ::= ']]>' |
9763 | | */ |
9764 | | void |
9765 | 0 | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9766 | 0 | xmlChar *buf = NULL; |
9767 | 0 | int len = 0; |
9768 | 0 | int size = XML_PARSER_BUFFER_SIZE; |
9769 | 0 | int r, rl; |
9770 | 0 | int s, sl; |
9771 | 0 | int cur, l; |
9772 | 0 | int count = 0; |
9773 | 0 | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9774 | 0 | XML_MAX_HUGE_LENGTH : |
9775 | 0 | XML_MAX_TEXT_LENGTH; |
9776 | | |
9777 | | /* Check 2.6.0 was NXT(0) not RAW */ |
9778 | 0 | if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9779 | 0 | SKIP(9); |
9780 | 0 | } else |
9781 | 0 | return; |
9782 | | |
9783 | 0 | ctxt->instate = XML_PARSER_CDATA_SECTION; |
9784 | 0 | r = CUR_CHAR(rl); |
9785 | 0 | if (!IS_CHAR(r)) { |
9786 | 0 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9787 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
9788 | 0 | return; |
9789 | 0 | } |
9790 | 0 | NEXTL(rl); |
9791 | 0 | s = CUR_CHAR(sl); |
9792 | 0 | if (!IS_CHAR(s)) { |
9793 | 0 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9794 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
9795 | 0 | return; |
9796 | 0 | } |
9797 | 0 | NEXTL(sl); |
9798 | 0 | cur = CUR_CHAR(l); |
9799 | 0 | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
9800 | 0 | if (buf == NULL) { |
9801 | 0 | xmlErrMemory(ctxt, NULL); |
9802 | 0 | return; |
9803 | 0 | } |
9804 | 0 | while (IS_CHAR(cur) && |
9805 | 0 | ((r != ']') || (s != ']') || (cur != '>'))) { |
9806 | 0 | if (len + 5 >= size) { |
9807 | 0 | xmlChar *tmp; |
9808 | |
|
9809 | 0 | tmp = (xmlChar *) xmlRealloc(buf, size * 2 * sizeof(xmlChar)); |
9810 | 0 | if (tmp == NULL) { |
9811 | 0 | xmlFree(buf); |
9812 | 0 | xmlErrMemory(ctxt, NULL); |
9813 | 0 | return; |
9814 | 0 | } |
9815 | 0 | buf = tmp; |
9816 | 0 | size *= 2; |
9817 | 0 | } |
9818 | 0 | COPY_BUF(rl,buf,len,r); |
9819 | 0 | r = s; |
9820 | 0 | rl = sl; |
9821 | 0 | s = cur; |
9822 | 0 | sl = l; |
9823 | 0 | count++; |
9824 | 0 | if (count > 50) { |
9825 | 0 | SHRINK; |
9826 | 0 | GROW; |
9827 | 0 | if (ctxt->instate == XML_PARSER_EOF) { |
9828 | 0 | xmlFree(buf); |
9829 | 0 | return; |
9830 | 0 | } |
9831 | 0 | count = 0; |
9832 | 0 | } |
9833 | 0 | NEXTL(l); |
9834 | 0 | cur = CUR_CHAR(l); |
9835 | 0 | if (len > maxLength) { |
9836 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9837 | 0 | "CData section too big found\n"); |
9838 | 0 | xmlFree(buf); |
9839 | 0 | return; |
9840 | 0 | } |
9841 | 0 | } |
9842 | 0 | buf[len] = 0; |
9843 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
9844 | 0 | if (cur != '>') { |
9845 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9846 | 0 | "CData section not finished\n%.50s\n", buf); |
9847 | 0 | xmlFree(buf); |
9848 | 0 | return; |
9849 | 0 | } |
9850 | 0 | NEXTL(l); |
9851 | | |
9852 | | /* |
9853 | | * OK the buffer is to be consumed as cdata. |
9854 | | */ |
9855 | 0 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9856 | 0 | if (ctxt->sax->cdataBlock != NULL) |
9857 | 0 | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9858 | 0 | else if (ctxt->sax->characters != NULL) |
9859 | 0 | ctxt->sax->characters(ctxt->userData, buf, len); |
9860 | 0 | } |
9861 | 0 | xmlFree(buf); |
9862 | 0 | } |
9863 | | |
9864 | | /** |
9865 | | * xmlParseContentInternal: |
9866 | | * @ctxt: an XML parser context |
9867 | | * |
9868 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9869 | | * unexpected EOF to the caller. |
9870 | | */ |
9871 | | |
9872 | | static void |
9873 | 0 | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9874 | 0 | int nameNr = ctxt->nameNr; |
9875 | |
|
9876 | 0 | GROW; |
9877 | 0 | while ((RAW != 0) && |
9878 | 0 | (ctxt->instate != XML_PARSER_EOF)) { |
9879 | 0 | int id = ctxt->input->id; |
9880 | 0 | unsigned long cons = CUR_CONSUMED; |
9881 | 0 | const xmlChar *cur = ctxt->input->cur; |
9882 | | |
9883 | | /* |
9884 | | * First case : a Processing Instruction. |
9885 | | */ |
9886 | 0 | if ((*cur == '<') && (cur[1] == '?')) { |
9887 | 0 | xmlParsePI(ctxt); |
9888 | 0 | } |
9889 | | |
9890 | | /* |
9891 | | * Second case : a CDSection |
9892 | | */ |
9893 | | /* 2.6.0 test was *cur not RAW */ |
9894 | 0 | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9895 | 0 | xmlParseCDSect(ctxt); |
9896 | 0 | } |
9897 | | |
9898 | | /* |
9899 | | * Third case : a comment |
9900 | | */ |
9901 | 0 | else if ((*cur == '<') && (NXT(1) == '!') && |
9902 | 0 | (NXT(2) == '-') && (NXT(3) == '-')) { |
9903 | 0 | xmlParseComment(ctxt); |
9904 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
9905 | 0 | } |
9906 | | |
9907 | | /* |
9908 | | * Fourth case : a sub-element. |
9909 | | */ |
9910 | 0 | else if (*cur == '<') { |
9911 | 0 | if (NXT(1) == '/') { |
9912 | 0 | if (ctxt->nameNr <= nameNr) |
9913 | 0 | break; |
9914 | 0 | xmlParseElementEnd(ctxt); |
9915 | 0 | } else { |
9916 | 0 | xmlParseElementStart(ctxt); |
9917 | 0 | } |
9918 | 0 | } |
9919 | | |
9920 | | /* |
9921 | | * Fifth case : a reference. If if has not been resolved, |
9922 | | * parsing returns it's Name, create the node |
9923 | | */ |
9924 | | |
9925 | 0 | else if (*cur == '&') { |
9926 | 0 | xmlParseReference(ctxt); |
9927 | 0 | } |
9928 | | |
9929 | | /* |
9930 | | * Last case, text. Note that References are handled directly. |
9931 | | */ |
9932 | 0 | else { |
9933 | 0 | xmlParseCharData(ctxt, 0); |
9934 | 0 | } |
9935 | | |
9936 | 0 | GROW; |
9937 | 0 | SHRINK; |
9938 | |
|
9939 | 0 | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
9940 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9941 | 0 | "detected an error in element content\n"); |
9942 | 0 | xmlHaltParser(ctxt); |
9943 | 0 | break; |
9944 | 0 | } |
9945 | 0 | } |
9946 | 0 | } |
9947 | | |
9948 | | /** |
9949 | | * xmlParseContent: |
9950 | | * @ctxt: an XML parser context |
9951 | | * |
9952 | | * Parse a content sequence. Stops at EOF or '</'. |
9953 | | * |
9954 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
9955 | | */ |
9956 | | |
9957 | | void |
9958 | 0 | xmlParseContent(xmlParserCtxtPtr ctxt) { |
9959 | 0 | int nameNr = ctxt->nameNr; |
9960 | |
|
9961 | 0 | xmlParseContentInternal(ctxt); |
9962 | |
|
9963 | 0 | if ((ctxt->instate != XML_PARSER_EOF) && (ctxt->nameNr > nameNr)) { |
9964 | 0 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9965 | 0 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9966 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9967 | 0 | "Premature end of data in tag %s line %d\n", |
9968 | 0 | name, line, NULL); |
9969 | 0 | } |
9970 | 0 | } |
9971 | | |
9972 | | /** |
9973 | | * xmlParseElement: |
9974 | | * @ctxt: an XML parser context |
9975 | | * |
9976 | | * parse an XML element |
9977 | | * |
9978 | | * [39] element ::= EmptyElemTag | STag content ETag |
9979 | | * |
9980 | | * [ WFC: Element Type Match ] |
9981 | | * The Name in an element's end-tag must match the element type in the |
9982 | | * start-tag. |
9983 | | * |
9984 | | */ |
9985 | | |
9986 | | void |
9987 | 0 | xmlParseElement(xmlParserCtxtPtr ctxt) { |
9988 | 0 | if (xmlParseElementStart(ctxt) != 0) |
9989 | 0 | return; |
9990 | | |
9991 | 0 | xmlParseContentInternal(ctxt); |
9992 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
9993 | 0 | return; |
9994 | | |
9995 | 0 | if (CUR == 0) { |
9996 | 0 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9997 | 0 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9998 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9999 | 0 | "Premature end of data in tag %s line %d\n", |
10000 | 0 | name, line, NULL); |
10001 | 0 | return; |
10002 | 0 | } |
10003 | | |
10004 | 0 | xmlParseElementEnd(ctxt); |
10005 | 0 | } |
10006 | | |
10007 | | /** |
10008 | | * xmlParseElementStart: |
10009 | | * @ctxt: an XML parser context |
10010 | | * |
10011 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
10012 | | * opening tag was parsed, 1 if an empty element was parsed. |
10013 | | */ |
10014 | | static int |
10015 | 0 | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
10016 | 0 | const xmlChar *name; |
10017 | 0 | const xmlChar *prefix = NULL; |
10018 | 0 | const xmlChar *URI = NULL; |
10019 | 0 | xmlParserNodeInfo node_info; |
10020 | 0 | int line, tlen = 0; |
10021 | 0 | xmlNodePtr ret; |
10022 | 0 | int nsNr = ctxt->nsNr; |
10023 | |
|
10024 | 0 | if (((unsigned int) ctxt->nameNr > xmlParserMaxDepth) && |
10025 | 0 | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
10026 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
10027 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
10028 | 0 | xmlParserMaxDepth); |
10029 | 0 | xmlHaltParser(ctxt); |
10030 | 0 | return(-1); |
10031 | 0 | } |
10032 | | |
10033 | | /* Capture start position */ |
10034 | 0 | if (ctxt->record_info) { |
10035 | 0 | node_info.begin_pos = ctxt->input->consumed + |
10036 | 0 | (CUR_PTR - ctxt->input->base); |
10037 | 0 | node_info.begin_line = ctxt->input->line; |
10038 | 0 | } |
10039 | |
|
10040 | 0 | if (ctxt->spaceNr == 0) |
10041 | 0 | spacePush(ctxt, -1); |
10042 | 0 | else if (*ctxt->space == -2) |
10043 | 0 | spacePush(ctxt, -1); |
10044 | 0 | else |
10045 | 0 | spacePush(ctxt, *ctxt->space); |
10046 | |
|
10047 | 0 | line = ctxt->input->line; |
10048 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10049 | 0 | if (ctxt->sax2) |
10050 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10051 | 0 | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
10052 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10053 | 0 | else |
10054 | 0 | name = xmlParseStartTag(ctxt); |
10055 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10056 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10057 | 0 | return(-1); |
10058 | 0 | if (name == NULL) { |
10059 | 0 | spacePop(ctxt); |
10060 | 0 | return(-1); |
10061 | 0 | } |
10062 | 0 | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
10063 | 0 | ret = ctxt->node; |
10064 | |
|
10065 | 0 | #ifdef LIBXML_VALID_ENABLED |
10066 | | /* |
10067 | | * [ VC: Root Element Type ] |
10068 | | * The Name in the document type declaration must match the element |
10069 | | * type of the root element. |
10070 | | */ |
10071 | 0 | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10072 | 0 | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10073 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10074 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
10075 | | |
10076 | | /* |
10077 | | * Check for an Empty Element. |
10078 | | */ |
10079 | 0 | if ((RAW == '/') && (NXT(1) == '>')) { |
10080 | 0 | SKIP(2); |
10081 | 0 | if (ctxt->sax2) { |
10082 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
10083 | 0 | (!ctxt->disableSAX)) |
10084 | 0 | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
10085 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10086 | 0 | } else { |
10087 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
10088 | 0 | (!ctxt->disableSAX)) |
10089 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
10090 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10091 | 0 | } |
10092 | 0 | namePop(ctxt); |
10093 | 0 | spacePop(ctxt); |
10094 | 0 | if (nsNr != ctxt->nsNr) |
10095 | 0 | nsPop(ctxt, ctxt->nsNr - nsNr); |
10096 | 0 | if ( ret != NULL && ctxt->record_info ) { |
10097 | 0 | node_info.end_pos = ctxt->input->consumed + |
10098 | 0 | (CUR_PTR - ctxt->input->base); |
10099 | 0 | node_info.end_line = ctxt->input->line; |
10100 | 0 | node_info.node = ret; |
10101 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10102 | 0 | } |
10103 | 0 | return(1); |
10104 | 0 | } |
10105 | 0 | if (RAW == '>') { |
10106 | 0 | NEXT1; |
10107 | 0 | } else { |
10108 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
10109 | 0 | "Couldn't find end of Start Tag %s line %d\n", |
10110 | 0 | name, line, NULL); |
10111 | | |
10112 | | /* |
10113 | | * end of parsing of this node. |
10114 | | */ |
10115 | 0 | nodePop(ctxt); |
10116 | 0 | namePop(ctxt); |
10117 | 0 | spacePop(ctxt); |
10118 | 0 | if (nsNr != ctxt->nsNr) |
10119 | 0 | nsPop(ctxt, ctxt->nsNr - nsNr); |
10120 | | |
10121 | | /* |
10122 | | * Capture end position and add node |
10123 | | */ |
10124 | 0 | if ( ret != NULL && ctxt->record_info ) { |
10125 | 0 | node_info.end_pos = ctxt->input->consumed + |
10126 | 0 | (CUR_PTR - ctxt->input->base); |
10127 | 0 | node_info.end_line = ctxt->input->line; |
10128 | 0 | node_info.node = ret; |
10129 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10130 | 0 | } |
10131 | 0 | return(-1); |
10132 | 0 | } |
10133 | | |
10134 | 0 | return(0); |
10135 | 0 | } |
10136 | | |
10137 | | /** |
10138 | | * xmlParseElementEnd: |
10139 | | * @ctxt: an XML parser context |
10140 | | * |
10141 | | * Parse the end of an XML element. |
10142 | | */ |
10143 | | static void |
10144 | 0 | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
10145 | 0 | xmlParserNodeInfo node_info; |
10146 | 0 | xmlNodePtr ret = ctxt->node; |
10147 | |
|
10148 | 0 | if (ctxt->nameNr <= 0) |
10149 | 0 | return; |
10150 | | |
10151 | | /* |
10152 | | * parse the end of tag: '</' should be here. |
10153 | | */ |
10154 | 0 | if (ctxt->sax2) { |
10155 | 0 | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
10156 | 0 | namePop(ctxt); |
10157 | 0 | } |
10158 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10159 | 0 | else |
10160 | 0 | xmlParseEndTag1(ctxt, 0); |
10161 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10162 | | |
10163 | | /* |
10164 | | * Capture end position and add node |
10165 | | */ |
10166 | 0 | if ( ret != NULL && ctxt->record_info ) { |
10167 | 0 | node_info.end_pos = ctxt->input->consumed + |
10168 | 0 | (CUR_PTR - ctxt->input->base); |
10169 | 0 | node_info.end_line = ctxt->input->line; |
10170 | 0 | node_info.node = ret; |
10171 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10172 | 0 | } |
10173 | 0 | } |
10174 | | |
10175 | | /** |
10176 | | * xmlParseVersionNum: |
10177 | | * @ctxt: an XML parser context |
10178 | | * |
10179 | | * parse the XML version value. |
10180 | | * |
10181 | | * [26] VersionNum ::= '1.' [0-9]+ |
10182 | | * |
10183 | | * In practice allow [0-9].[0-9]+ at that level |
10184 | | * |
10185 | | * Returns the string giving the XML version number, or NULL |
10186 | | */ |
10187 | | xmlChar * |
10188 | 2.97k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10189 | 2.97k | xmlChar *buf = NULL; |
10190 | 2.97k | int len = 0; |
10191 | 2.97k | int size = 10; |
10192 | 2.97k | xmlChar cur; |
10193 | | |
10194 | 2.97k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
10195 | 2.97k | if (buf == NULL) { |
10196 | 0 | xmlErrMemory(ctxt, NULL); |
10197 | 0 | return(NULL); |
10198 | 0 | } |
10199 | 2.97k | cur = CUR; |
10200 | 2.97k | if (!((cur >= '0') && (cur <= '9'))) { |
10201 | 17 | xmlFree(buf); |
10202 | 17 | return(NULL); |
10203 | 17 | } |
10204 | 2.95k | buf[len++] = cur; |
10205 | 2.95k | NEXT; |
10206 | 2.95k | cur=CUR; |
10207 | 2.95k | if (cur != '.') { |
10208 | 12 | xmlFree(buf); |
10209 | 12 | return(NULL); |
10210 | 12 | } |
10211 | 2.94k | buf[len++] = cur; |
10212 | 2.94k | NEXT; |
10213 | 2.94k | cur=CUR; |
10214 | 2.77M | while ((cur >= '0') && (cur <= '9')) { |
10215 | 2.76M | if (len + 1 >= size) { |
10216 | 325 | xmlChar *tmp; |
10217 | | |
10218 | 325 | size *= 2; |
10219 | 325 | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
10220 | 325 | if (tmp == NULL) { |
10221 | 0 | xmlFree(buf); |
10222 | 0 | xmlErrMemory(ctxt, NULL); |
10223 | 0 | return(NULL); |
10224 | 0 | } |
10225 | 325 | buf = tmp; |
10226 | 325 | } |
10227 | 2.76M | buf[len++] = cur; |
10228 | 2.76M | NEXT; |
10229 | 2.76M | cur=CUR; |
10230 | 2.76M | } |
10231 | 2.94k | buf[len] = 0; |
10232 | 2.94k | return(buf); |
10233 | 2.94k | } |
10234 | | |
10235 | | /** |
10236 | | * xmlParseVersionInfo: |
10237 | | * @ctxt: an XML parser context |
10238 | | * |
10239 | | * parse the XML version. |
10240 | | * |
10241 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10242 | | * |
10243 | | * [25] Eq ::= S? '=' S? |
10244 | | * |
10245 | | * Returns the version string, e.g. "1.0" |
10246 | | */ |
10247 | | |
10248 | | xmlChar * |
10249 | 3.73k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10250 | 3.73k | xmlChar *version = NULL; |
10251 | | |
10252 | 3.73k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10253 | 2.99k | SKIP(7); |
10254 | 2.99k | SKIP_BLANKS; |
10255 | 2.99k | if (RAW != '=') { |
10256 | 11 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10257 | 11 | return(NULL); |
10258 | 11 | } |
10259 | 2.98k | NEXT; |
10260 | 2.98k | SKIP_BLANKS; |
10261 | 2.98k | if (RAW == '"') { |
10262 | 2.96k | NEXT; |
10263 | 2.96k | version = xmlParseVersionNum(ctxt); |
10264 | 2.96k | if (RAW != '"') { |
10265 | 63 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10266 | 63 | } else |
10267 | 2.90k | NEXT; |
10268 | 2.96k | } else if (RAW == '\''){ |
10269 | 8 | NEXT; |
10270 | 8 | version = xmlParseVersionNum(ctxt); |
10271 | 8 | if (RAW != '\'') { |
10272 | 6 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10273 | 6 | } else |
10274 | 2 | NEXT; |
10275 | 12 | } else { |
10276 | 12 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10277 | 12 | } |
10278 | 2.98k | } |
10279 | 3.72k | return(version); |
10280 | 3.73k | } |
10281 | | |
10282 | | /** |
10283 | | * xmlParseEncName: |
10284 | | * @ctxt: an XML parser context |
10285 | | * |
10286 | | * parse the XML encoding name |
10287 | | * |
10288 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10289 | | * |
10290 | | * Returns the encoding name value or NULL |
10291 | | */ |
10292 | | xmlChar * |
10293 | 3.50k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10294 | 3.50k | xmlChar *buf = NULL; |
10295 | 3.50k | int len = 0; |
10296 | 3.50k | int size = 10; |
10297 | 3.50k | xmlChar cur; |
10298 | | |
10299 | 3.50k | cur = CUR; |
10300 | 3.50k | if (((cur >= 'a') && (cur <= 'z')) || |
10301 | 3.50k | ((cur >= 'A') && (cur <= 'Z'))) { |
10302 | 3.48k | buf = (xmlChar *) xmlMallocAtomic(size * sizeof(xmlChar)); |
10303 | 3.48k | if (buf == NULL) { |
10304 | 0 | xmlErrMemory(ctxt, NULL); |
10305 | 0 | return(NULL); |
10306 | 0 | } |
10307 | | |
10308 | 3.48k | buf[len++] = cur; |
10309 | 3.48k | NEXT; |
10310 | 3.48k | cur = CUR; |
10311 | 1.09M | while (((cur >= 'a') && (cur <= 'z')) || |
10312 | 1.09M | ((cur >= 'A') && (cur <= 'Z')) || |
10313 | 1.09M | ((cur >= '0') && (cur <= '9')) || |
10314 | 1.09M | (cur == '.') || (cur == '_') || |
10315 | 1.09M | (cur == '-')) { |
10316 | 1.08M | if (len + 1 >= size) { |
10317 | 408 | xmlChar *tmp; |
10318 | | |
10319 | 408 | size *= 2; |
10320 | 408 | tmp = (xmlChar *) xmlRealloc(buf, size * sizeof(xmlChar)); |
10321 | 408 | if (tmp == NULL) { |
10322 | 0 | xmlErrMemory(ctxt, NULL); |
10323 | 0 | xmlFree(buf); |
10324 | 0 | return(NULL); |
10325 | 0 | } |
10326 | 408 | buf = tmp; |
10327 | 408 | } |
10328 | 1.08M | buf[len++] = cur; |
10329 | 1.08M | NEXT; |
10330 | 1.08M | cur = CUR; |
10331 | 1.08M | if (cur == 0) { |
10332 | 71 | SHRINK; |
10333 | 71 | GROW; |
10334 | 71 | cur = CUR; |
10335 | 71 | } |
10336 | 1.08M | } |
10337 | 3.48k | buf[len] = 0; |
10338 | 3.48k | } else { |
10339 | 12 | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10340 | 12 | } |
10341 | 3.50k | return(buf); |
10342 | 3.50k | } |
10343 | | |
10344 | | /** |
10345 | | * xmlParseEncodingDecl: |
10346 | | * @ctxt: an XML parser context |
10347 | | * |
10348 | | * parse the XML encoding declaration |
10349 | | * |
10350 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10351 | | * |
10352 | | * this setups the conversion filters. |
10353 | | * |
10354 | | * Returns the encoding value or NULL |
10355 | | */ |
10356 | | |
10357 | | const xmlChar * |
10358 | 3.73k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10359 | 3.73k | xmlChar *encoding = NULL; |
10360 | | |
10361 | 3.73k | SKIP_BLANKS; |
10362 | 3.73k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g')) { |
10363 | 3.50k | SKIP(8); |
10364 | 3.50k | SKIP_BLANKS; |
10365 | 3.50k | if (RAW != '=') { |
10366 | 5 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10367 | 5 | return(NULL); |
10368 | 5 | } |
10369 | 3.50k | NEXT; |
10370 | 3.50k | SKIP_BLANKS; |
10371 | 3.50k | if (RAW == '"') { |
10372 | 3.48k | NEXT; |
10373 | 3.48k | encoding = xmlParseEncName(ctxt); |
10374 | 3.48k | if (RAW != '"') { |
10375 | 97 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10376 | 97 | xmlFree((xmlChar *) encoding); |
10377 | 97 | return(NULL); |
10378 | 97 | } else |
10379 | 3.39k | NEXT; |
10380 | 3.48k | } else if (RAW == '\''){ |
10381 | 12 | NEXT; |
10382 | 12 | encoding = xmlParseEncName(ctxt); |
10383 | 12 | if (RAW != '\'') { |
10384 | 8 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10385 | 8 | xmlFree((xmlChar *) encoding); |
10386 | 8 | return(NULL); |
10387 | 8 | } else |
10388 | 4 | NEXT; |
10389 | 12 | } else { |
10390 | 3 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10391 | 3 | } |
10392 | | |
10393 | | /* |
10394 | | * Non standard parsing, allowing the user to ignore encoding |
10395 | | */ |
10396 | 3.39k | if (ctxt->options & XML_PARSE_IGNORE_ENC) { |
10397 | 0 | xmlFree((xmlChar *) encoding); |
10398 | 0 | return(NULL); |
10399 | 0 | } |
10400 | | |
10401 | | /* |
10402 | | * UTF-16 encoding switch has already taken place at this stage, |
10403 | | * more over the little-endian/big-endian selection is already done |
10404 | | */ |
10405 | 3.39k | if ((encoding != NULL) && |
10406 | 3.39k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-16")) || |
10407 | 3.39k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF16")))) { |
10408 | | /* |
10409 | | * If no encoding was passed to the parser, that we are |
10410 | | * using UTF-16 and no decoder is present i.e. the |
10411 | | * document is apparently UTF-8 compatible, then raise an |
10412 | | * encoding mismatch fatal error |
10413 | | */ |
10414 | 3 | if ((ctxt->encoding == NULL) && |
10415 | 3 | (ctxt->input->buf != NULL) && |
10416 | 3 | (ctxt->input->buf->encoder == NULL)) { |
10417 | 2 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_ENCODING, |
10418 | 2 | "Document labelled UTF-16 but has UTF-8 content\n"); |
10419 | 2 | } |
10420 | 3 | if (ctxt->encoding != NULL) |
10421 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10422 | 3 | ctxt->encoding = encoding; |
10423 | 3 | } |
10424 | | /* |
10425 | | * UTF-8 encoding is handled natively |
10426 | | */ |
10427 | 3.39k | else if ((encoding != NULL) && |
10428 | 3.39k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-8")) || |
10429 | 3.39k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF8")))) { |
10430 | 35 | if (ctxt->encoding != NULL) |
10431 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
10432 | 35 | ctxt->encoding = encoding; |
10433 | 35 | } |
10434 | 3.36k | else if (encoding != NULL) { |
10435 | 3.35k | xmlCharEncodingHandlerPtr handler; |
10436 | | |
10437 | 3.35k | if (ctxt->input->encoding != NULL) |
10438 | 0 | xmlFree((xmlChar *) ctxt->input->encoding); |
10439 | 3.35k | ctxt->input->encoding = encoding; |
10440 | | |
10441 | 3.35k | handler = xmlFindCharEncodingHandler((const char *) encoding); |
10442 | 3.35k | if (handler != NULL) { |
10443 | 2.93k | if (xmlSwitchToEncoding(ctxt, handler) < 0) { |
10444 | | /* failed to convert */ |
10445 | 5 | ctxt->errNo = XML_ERR_UNSUPPORTED_ENCODING; |
10446 | 5 | return(NULL); |
10447 | 5 | } |
10448 | 2.93k | } else { |
10449 | 424 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
10450 | 424 | "Unsupported encoding %s\n", encoding); |
10451 | 424 | return(NULL); |
10452 | 424 | } |
10453 | 3.35k | } |
10454 | 3.39k | } |
10455 | 3.19k | return(encoding); |
10456 | 3.73k | } |
10457 | | |
10458 | | /** |
10459 | | * xmlParseSDDecl: |
10460 | | * @ctxt: an XML parser context |
10461 | | * |
10462 | | * parse the XML standalone declaration |
10463 | | * |
10464 | | * [32] SDDecl ::= S 'standalone' Eq |
10465 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10466 | | * |
10467 | | * [ VC: Standalone Document Declaration ] |
10468 | | * TODO The standalone document declaration must have the value "no" |
10469 | | * if any external markup declarations contain declarations of: |
10470 | | * - attributes with default values, if elements to which these |
10471 | | * attributes apply appear in the document without specifications |
10472 | | * of values for these attributes, or |
10473 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10474 | | * to those entities appear in the document, or |
10475 | | * - attributes with values subject to normalization, where the |
10476 | | * attribute appears in the document with a value which will change |
10477 | | * as a result of normalization, or |
10478 | | * - element types with element content, if white space occurs directly |
10479 | | * within any instance of those types. |
10480 | | * |
10481 | | * Returns: |
10482 | | * 1 if standalone="yes" |
10483 | | * 0 if standalone="no" |
10484 | | * -2 if standalone attribute is missing or invalid |
10485 | | * (A standalone value of -2 means that the XML declaration was found, |
10486 | | * but no value was specified for the standalone attribute). |
10487 | | */ |
10488 | | |
10489 | | int |
10490 | 460 | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10491 | 460 | int standalone = -2; |
10492 | | |
10493 | 460 | SKIP_BLANKS; |
10494 | 460 | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10495 | 33 | SKIP(10); |
10496 | 33 | SKIP_BLANKS; |
10497 | 33 | if (RAW != '=') { |
10498 | 3 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10499 | 3 | return(standalone); |
10500 | 3 | } |
10501 | 30 | NEXT; |
10502 | 30 | SKIP_BLANKS; |
10503 | 30 | if (RAW == '\''){ |
10504 | 14 | NEXT; |
10505 | 14 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10506 | 2 | standalone = 0; |
10507 | 2 | SKIP(2); |
10508 | 12 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10509 | 12 | (NXT(2) == 's')) { |
10510 | 2 | standalone = 1; |
10511 | 2 | SKIP(3); |
10512 | 10 | } else { |
10513 | 10 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10514 | 10 | } |
10515 | 14 | if (RAW != '\'') { |
10516 | 13 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10517 | 13 | } else |
10518 | 1 | NEXT; |
10519 | 16 | } else if (RAW == '"'){ |
10520 | 12 | NEXT; |
10521 | 12 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10522 | 2 | standalone = 0; |
10523 | 2 | SKIP(2); |
10524 | 10 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10525 | 10 | (NXT(2) == 's')) { |
10526 | 2 | standalone = 1; |
10527 | 2 | SKIP(3); |
10528 | 8 | } else { |
10529 | 8 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10530 | 8 | } |
10531 | 12 | if (RAW != '"') { |
10532 | 10 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10533 | 10 | } else |
10534 | 2 | NEXT; |
10535 | 12 | } else { |
10536 | 4 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10537 | 4 | } |
10538 | 30 | } |
10539 | 457 | return(standalone); |
10540 | 460 | } |
10541 | | |
10542 | | /** |
10543 | | * xmlParseXMLDecl: |
10544 | | * @ctxt: an XML parser context |
10545 | | * |
10546 | | * parse an XML declaration header |
10547 | | * |
10548 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10549 | | */ |
10550 | | |
10551 | | void |
10552 | 3.73k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10553 | 3.73k | xmlChar *version; |
10554 | | |
10555 | | /* |
10556 | | * This value for standalone indicates that the document has an |
10557 | | * XML declaration but it does not have a standalone attribute. |
10558 | | * It will be overwritten later if a standalone attribute is found. |
10559 | | */ |
10560 | 3.73k | ctxt->input->standalone = -2; |
10561 | | |
10562 | | /* |
10563 | | * We know that '<?xml' is here. |
10564 | | */ |
10565 | 3.73k | SKIP(5); |
10566 | | |
10567 | 3.73k | if (!IS_BLANK_CH(RAW)) { |
10568 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10569 | 0 | "Blank needed after '<?xml'\n"); |
10570 | 0 | } |
10571 | 3.73k | SKIP_BLANKS; |
10572 | | |
10573 | | /* |
10574 | | * We must have the VersionInfo here. |
10575 | | */ |
10576 | 3.73k | version = xmlParseVersionInfo(ctxt); |
10577 | 3.73k | if (version == NULL) { |
10578 | 788 | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10579 | 2.94k | } else { |
10580 | 2.94k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10581 | | /* |
10582 | | * Changed here for XML-1.0 5th edition |
10583 | | */ |
10584 | 1.44k | if (ctxt->options & XML_PARSE_OLD10) { |
10585 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10586 | 0 | "Unsupported version '%s'\n", |
10587 | 0 | version); |
10588 | 1.44k | } else { |
10589 | 1.44k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10590 | 1.41k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10591 | 1.41k | "Unsupported version '%s'\n", |
10592 | 1.41k | version, NULL); |
10593 | 1.41k | } else { |
10594 | 25 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10595 | 25 | "Unsupported version '%s'\n", |
10596 | 25 | version); |
10597 | 25 | } |
10598 | 1.44k | } |
10599 | 1.44k | } |
10600 | 2.94k | if (ctxt->version != NULL) |
10601 | 0 | xmlFree((void *) ctxt->version); |
10602 | 2.94k | ctxt->version = version; |
10603 | 2.94k | } |
10604 | | |
10605 | | /* |
10606 | | * We may have the encoding declaration |
10607 | | */ |
10608 | 3.73k | if (!IS_BLANK_CH(RAW)) { |
10609 | 828 | if ((RAW == '?') && (NXT(1) == '>')) { |
10610 | 4 | SKIP(2); |
10611 | 4 | return; |
10612 | 4 | } |
10613 | 824 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10614 | 824 | } |
10615 | 3.73k | xmlParseEncodingDecl(ctxt); |
10616 | 3.73k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10617 | 3.73k | (ctxt->instate == XML_PARSER_EOF)) { |
10618 | | /* |
10619 | | * The XML REC instructs us to stop parsing right here |
10620 | | */ |
10621 | 429 | return; |
10622 | 429 | } |
10623 | | |
10624 | | /* |
10625 | | * We may have the standalone status. |
10626 | | */ |
10627 | 3.30k | if ((ctxt->input->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10628 | 2.92k | if ((RAW == '?') && (NXT(1) == '>')) { |
10629 | 2.84k | SKIP(2); |
10630 | 2.84k | return; |
10631 | 2.84k | } |
10632 | 79 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10633 | 79 | } |
10634 | | |
10635 | | /* |
10636 | | * We can grow the input buffer freely at that point |
10637 | | */ |
10638 | 460 | GROW; |
10639 | | |
10640 | 460 | SKIP_BLANKS; |
10641 | 460 | ctxt->input->standalone = xmlParseSDDecl(ctxt); |
10642 | | |
10643 | 460 | SKIP_BLANKS; |
10644 | 460 | if ((RAW == '?') && (NXT(1) == '>')) { |
10645 | 35 | SKIP(2); |
10646 | 425 | } else if (RAW == '>') { |
10647 | | /* Deprecated old WD ... */ |
10648 | 5 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10649 | 5 | NEXT; |
10650 | 420 | } else { |
10651 | 420 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10652 | 420 | MOVETO_ENDTAG(CUR_PTR); |
10653 | 420 | NEXT; |
10654 | 420 | } |
10655 | 460 | } |
10656 | | |
10657 | | /** |
10658 | | * xmlParseMisc: |
10659 | | * @ctxt: an XML parser context |
10660 | | * |
10661 | | * parse an XML Misc* optional field. |
10662 | | * |
10663 | | * [27] Misc ::= Comment | PI | S |
10664 | | */ |
10665 | | |
10666 | | void |
10667 | 0 | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10668 | 0 | while (ctxt->instate != XML_PARSER_EOF) { |
10669 | 0 | SKIP_BLANKS; |
10670 | 0 | GROW; |
10671 | 0 | if ((RAW == '<') && (NXT(1) == '?')) { |
10672 | 0 | xmlParsePI(ctxt); |
10673 | 0 | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10674 | 0 | xmlParseComment(ctxt); |
10675 | 0 | } else { |
10676 | 0 | break; |
10677 | 0 | } |
10678 | 0 | } |
10679 | 0 | } |
10680 | | |
10681 | | /** |
10682 | | * xmlParseDocument: |
10683 | | * @ctxt: an XML parser context |
10684 | | * |
10685 | | * parse an XML document (and build a tree if using the standard SAX |
10686 | | * interface). |
10687 | | * |
10688 | | * [1] document ::= prolog element Misc* |
10689 | | * |
10690 | | * [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? |
10691 | | * |
10692 | | * Returns 0, -1 in case of error. the parser context is augmented |
10693 | | * as a result of the parsing. |
10694 | | */ |
10695 | | |
10696 | | int |
10697 | 0 | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10698 | 0 | xmlChar start[4]; |
10699 | 0 | xmlCharEncoding enc; |
10700 | |
|
10701 | 0 | xmlInitParser(); |
10702 | |
|
10703 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10704 | 0 | return(-1); |
10705 | | |
10706 | 0 | GROW; |
10707 | | |
10708 | | /* |
10709 | | * SAX: detecting the level. |
10710 | | */ |
10711 | 0 | xmlDetectSAX2(ctxt); |
10712 | | |
10713 | | /* |
10714 | | * SAX: beginning of the document processing. |
10715 | | */ |
10716 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10717 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10718 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10719 | 0 | return(-1); |
10720 | | |
10721 | 0 | if ((ctxt->encoding == NULL) && |
10722 | 0 | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
10723 | | /* |
10724 | | * Get the 4 first bytes and decode the charset |
10725 | | * if enc != XML_CHAR_ENCODING_NONE |
10726 | | * plug some encoding conversion routines. |
10727 | | */ |
10728 | 0 | start[0] = RAW; |
10729 | 0 | start[1] = NXT(1); |
10730 | 0 | start[2] = NXT(2); |
10731 | 0 | start[3] = NXT(3); |
10732 | 0 | enc = xmlDetectCharEncoding(&start[0], 4); |
10733 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
10734 | 0 | xmlSwitchEncoding(ctxt, enc); |
10735 | 0 | } |
10736 | 0 | } |
10737 | | |
10738 | |
|
10739 | 0 | if (CUR == 0) { |
10740 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10741 | 0 | return(-1); |
10742 | 0 | } |
10743 | | |
10744 | | /* |
10745 | | * Check for the XMLDecl in the Prolog. |
10746 | | * do not GROW here to avoid the detected encoder to decode more |
10747 | | * than just the first line, unless the amount of data is really |
10748 | | * too small to hold "<?xml version="1.0" encoding="foo" |
10749 | | */ |
10750 | 0 | if ((ctxt->input->end - ctxt->input->cur) < 35) { |
10751 | 0 | GROW; |
10752 | 0 | } |
10753 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10754 | | |
10755 | | /* |
10756 | | * Note that we will switch encoding on the fly. |
10757 | | */ |
10758 | 0 | xmlParseXMLDecl(ctxt); |
10759 | 0 | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10760 | 0 | (ctxt->instate == XML_PARSER_EOF)) { |
10761 | | /* |
10762 | | * The XML REC instructs us to stop parsing right here |
10763 | | */ |
10764 | 0 | return(-1); |
10765 | 0 | } |
10766 | 0 | ctxt->standalone = ctxt->input->standalone; |
10767 | 0 | SKIP_BLANKS; |
10768 | 0 | } else { |
10769 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10770 | 0 | } |
10771 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10772 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10773 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10774 | 0 | return(-1); |
10775 | 0 | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10776 | 0 | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10777 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10778 | 0 | } |
10779 | | |
10780 | | /* |
10781 | | * The Misc part of the Prolog |
10782 | | */ |
10783 | 0 | xmlParseMisc(ctxt); |
10784 | | |
10785 | | /* |
10786 | | * Then possibly doc type declaration(s) and more Misc |
10787 | | * (doctypedecl Misc*)? |
10788 | | */ |
10789 | 0 | GROW; |
10790 | 0 | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10791 | |
|
10792 | 0 | ctxt->inSubset = 1; |
10793 | 0 | xmlParseDocTypeDecl(ctxt); |
10794 | 0 | if (RAW == '[') { |
10795 | 0 | ctxt->instate = XML_PARSER_DTD; |
10796 | 0 | xmlParseInternalSubset(ctxt); |
10797 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10798 | 0 | return(-1); |
10799 | 0 | } |
10800 | | |
10801 | | /* |
10802 | | * Create and update the external subset. |
10803 | | */ |
10804 | 0 | ctxt->inSubset = 2; |
10805 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10806 | 0 | (!ctxt->disableSAX)) |
10807 | 0 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10808 | 0 | ctxt->extSubSystem, ctxt->extSubURI); |
10809 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10810 | 0 | return(-1); |
10811 | 0 | ctxt->inSubset = 0; |
10812 | |
|
10813 | 0 | xmlCleanSpecialAttr(ctxt); |
10814 | |
|
10815 | 0 | ctxt->instate = XML_PARSER_PROLOG; |
10816 | 0 | xmlParseMisc(ctxt); |
10817 | 0 | } |
10818 | | |
10819 | | /* |
10820 | | * Time to start parsing the tree itself |
10821 | | */ |
10822 | 0 | GROW; |
10823 | 0 | if (RAW != '<') { |
10824 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10825 | 0 | "Start tag expected, '<' not found\n"); |
10826 | 0 | } else { |
10827 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
10828 | 0 | xmlParseElement(ctxt); |
10829 | 0 | ctxt->instate = XML_PARSER_EPILOG; |
10830 | | |
10831 | | |
10832 | | /* |
10833 | | * The Misc part at the end |
10834 | | */ |
10835 | 0 | xmlParseMisc(ctxt); |
10836 | |
|
10837 | 0 | if (RAW != 0) { |
10838 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10839 | 0 | } |
10840 | 0 | ctxt->instate = XML_PARSER_EOF; |
10841 | 0 | } |
10842 | | |
10843 | | /* |
10844 | | * SAX: end of the document processing. |
10845 | | */ |
10846 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10847 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10848 | | |
10849 | | /* |
10850 | | * Remove locally kept entity definitions if the tree was not built |
10851 | | */ |
10852 | 0 | if ((ctxt->myDoc != NULL) && |
10853 | 0 | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
10854 | 0 | xmlFreeDoc(ctxt->myDoc); |
10855 | 0 | ctxt->myDoc = NULL; |
10856 | 0 | } |
10857 | |
|
10858 | 0 | if ((ctxt->wellFormed) && (ctxt->myDoc != NULL)) { |
10859 | 0 | ctxt->myDoc->properties |= XML_DOC_WELLFORMED; |
10860 | 0 | if (ctxt->valid) |
10861 | 0 | ctxt->myDoc->properties |= XML_DOC_DTDVALID; |
10862 | 0 | if (ctxt->nsWellFormed) |
10863 | 0 | ctxt->myDoc->properties |= XML_DOC_NSVALID; |
10864 | 0 | if (ctxt->options & XML_PARSE_OLD10) |
10865 | 0 | ctxt->myDoc->properties |= XML_DOC_OLD10; |
10866 | 0 | } |
10867 | 0 | if (! ctxt->wellFormed) { |
10868 | 0 | ctxt->valid = 0; |
10869 | 0 | return(-1); |
10870 | 0 | } |
10871 | 0 | return(0); |
10872 | 0 | } |
10873 | | |
10874 | | /** |
10875 | | * xmlParseExtParsedEnt: |
10876 | | * @ctxt: an XML parser context |
10877 | | * |
10878 | | * parse a general parsed entity |
10879 | | * An external general parsed entity is well-formed if it matches the |
10880 | | * production labeled extParsedEnt. |
10881 | | * |
10882 | | * [78] extParsedEnt ::= TextDecl? content |
10883 | | * |
10884 | | * Returns 0, -1 in case of error. the parser context is augmented |
10885 | | * as a result of the parsing. |
10886 | | */ |
10887 | | |
10888 | | int |
10889 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
10890 | 0 | xmlChar start[4]; |
10891 | 0 | xmlCharEncoding enc; |
10892 | |
|
10893 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10894 | 0 | return(-1); |
10895 | | |
10896 | 0 | xmlDetectSAX2(ctxt); |
10897 | |
|
10898 | 0 | GROW; |
10899 | | |
10900 | | /* |
10901 | | * SAX: beginning of the document processing. |
10902 | | */ |
10903 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10904 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10905 | | |
10906 | | /* |
10907 | | * Get the 4 first bytes and decode the charset |
10908 | | * if enc != XML_CHAR_ENCODING_NONE |
10909 | | * plug some encoding conversion routines. |
10910 | | */ |
10911 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
10912 | 0 | start[0] = RAW; |
10913 | 0 | start[1] = NXT(1); |
10914 | 0 | start[2] = NXT(2); |
10915 | 0 | start[3] = NXT(3); |
10916 | 0 | enc = xmlDetectCharEncoding(start, 4); |
10917 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
10918 | 0 | xmlSwitchEncoding(ctxt, enc); |
10919 | 0 | } |
10920 | 0 | } |
10921 | | |
10922 | |
|
10923 | 0 | if (CUR == 0) { |
10924 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10925 | 0 | } |
10926 | | |
10927 | | /* |
10928 | | * Check for the XMLDecl in the Prolog. |
10929 | | */ |
10930 | 0 | GROW; |
10931 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10932 | | |
10933 | | /* |
10934 | | * Note that we will switch encoding on the fly. |
10935 | | */ |
10936 | 0 | xmlParseXMLDecl(ctxt); |
10937 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
10938 | | /* |
10939 | | * The XML REC instructs us to stop parsing right here |
10940 | | */ |
10941 | 0 | return(-1); |
10942 | 0 | } |
10943 | 0 | SKIP_BLANKS; |
10944 | 0 | } else { |
10945 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10946 | 0 | } |
10947 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10948 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10949 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10950 | 0 | return(-1); |
10951 | | |
10952 | | /* |
10953 | | * Doing validity checking on chunk doesn't make sense |
10954 | | */ |
10955 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
10956 | 0 | ctxt->validate = 0; |
10957 | 0 | ctxt->loadsubset = 0; |
10958 | 0 | ctxt->depth = 0; |
10959 | |
|
10960 | 0 | xmlParseContent(ctxt); |
10961 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
10962 | 0 | return(-1); |
10963 | | |
10964 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
10965 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10966 | 0 | } else if (RAW != 0) { |
10967 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
10968 | 0 | } |
10969 | | |
10970 | | /* |
10971 | | * SAX: end of the document processing. |
10972 | | */ |
10973 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10974 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10975 | |
|
10976 | 0 | if (! ctxt->wellFormed) return(-1); |
10977 | 0 | return(0); |
10978 | 0 | } |
10979 | | |
10980 | | #ifdef LIBXML_PUSH_ENABLED |
10981 | | /************************************************************************ |
10982 | | * * |
10983 | | * Progressive parsing interfaces * |
10984 | | * * |
10985 | | ************************************************************************/ |
10986 | | |
10987 | | /** |
10988 | | * xmlParseLookupSequence: |
10989 | | * @ctxt: an XML parser context |
10990 | | * @first: the first char to lookup |
10991 | | * @next: the next char to lookup or zero |
10992 | | * @third: the next char to lookup or zero |
10993 | | * |
10994 | | * Try to find if a sequence (first, next, third) or just (first next) or |
10995 | | * (first) is available in the input stream. |
10996 | | * This function has a side effect of (possibly) incrementing ctxt->checkIndex |
10997 | | * to avoid rescanning sequences of bytes, it DOES change the state of the |
10998 | | * parser, do not use liberally. |
10999 | | * |
11000 | | * Returns the index to the current parsing point if the full sequence |
11001 | | * is available, -1 otherwise. |
11002 | | */ |
11003 | | static int |
11004 | | xmlParseLookupSequence(xmlParserCtxtPtr ctxt, xmlChar first, |
11005 | 214k | xmlChar next, xmlChar third) { |
11006 | 214k | int base, len; |
11007 | 214k | xmlParserInputPtr in; |
11008 | 214k | const xmlChar *buf; |
11009 | | |
11010 | 214k | in = ctxt->input; |
11011 | 214k | if (in == NULL) return(-1); |
11012 | 214k | base = in->cur - in->base; |
11013 | 214k | if (base < 0) return(-1); |
11014 | 214k | if (ctxt->checkIndex > base) |
11015 | 64.7k | base = ctxt->checkIndex; |
11016 | 214k | if (in->buf == NULL) { |
11017 | 0 | buf = in->base; |
11018 | 0 | len = in->length; |
11019 | 214k | } else { |
11020 | 214k | buf = xmlBufContent(in->buf->buffer); |
11021 | 214k | len = xmlBufUse(in->buf->buffer); |
11022 | 214k | } |
11023 | | /* take into account the sequence length */ |
11024 | 214k | if (third) len -= 2; |
11025 | 78.1k | else if (next) len --; |
11026 | 2.22G | for (;base < len;base++) { |
11027 | 2.22G | if (buf[base] == first) { |
11028 | 1.40M | if (third != 0) { |
11029 | 1.38M | if ((buf[base + 1] != next) || |
11030 | 1.38M | (buf[base + 2] != third)) continue; |
11031 | 1.38M | } else if (next != 0) { |
11032 | 27.9k | if (buf[base + 1] != next) continue; |
11033 | 27.9k | } |
11034 | 144k | ctxt->checkIndex = 0; |
11035 | | #ifdef DEBUG_PUSH |
11036 | | if (next == 0) |
11037 | | xmlGenericError(xmlGenericErrorContext, |
11038 | | "PP: lookup '%c' found at %d\n", |
11039 | | first, base); |
11040 | | else if (third == 0) |
11041 | | xmlGenericError(xmlGenericErrorContext, |
11042 | | "PP: lookup '%c%c' found at %d\n", |
11043 | | first, next, base); |
11044 | | else |
11045 | | xmlGenericError(xmlGenericErrorContext, |
11046 | | "PP: lookup '%c%c%c' found at %d\n", |
11047 | | first, next, third, base); |
11048 | | #endif |
11049 | 144k | return(base - (in->cur - in->base)); |
11050 | 1.40M | } |
11051 | 2.22G | } |
11052 | 70.0k | ctxt->checkIndex = base; |
11053 | | #ifdef DEBUG_PUSH |
11054 | | if (next == 0) |
11055 | | xmlGenericError(xmlGenericErrorContext, |
11056 | | "PP: lookup '%c' failed\n", first); |
11057 | | else if (third == 0) |
11058 | | xmlGenericError(xmlGenericErrorContext, |
11059 | | "PP: lookup '%c%c' failed\n", first, next); |
11060 | | else |
11061 | | xmlGenericError(xmlGenericErrorContext, |
11062 | | "PP: lookup '%c%c%c' failed\n", first, next, third); |
11063 | | #endif |
11064 | 70.0k | return(-1); |
11065 | 214k | } |
11066 | | |
11067 | | /** |
11068 | | * xmlParseGetLasts: |
11069 | | * @ctxt: an XML parser context |
11070 | | * @lastlt: pointer to store the last '<' from the input |
11071 | | * @lastgt: pointer to store the last '>' from the input |
11072 | | * |
11073 | | * Lookup the last < and > in the current chunk |
11074 | | */ |
11075 | | static void |
11076 | | xmlParseGetLasts(xmlParserCtxtPtr ctxt, const xmlChar **lastlt, |
11077 | 115k | const xmlChar **lastgt) { |
11078 | 115k | const xmlChar *tmp; |
11079 | | |
11080 | 115k | if ((ctxt == NULL) || (lastlt == NULL) || (lastgt == NULL)) { |
11081 | 0 | xmlGenericError(xmlGenericErrorContext, |
11082 | 0 | "Internal error: xmlParseGetLasts\n"); |
11083 | 0 | return; |
11084 | 0 | } |
11085 | 115k | if ((ctxt->progressive != 0) && (ctxt->inputNr == 1)) { |
11086 | 30.9k | tmp = ctxt->input->end; |
11087 | 30.9k | tmp--; |
11088 | 2.76G | while ((tmp >= ctxt->input->base) && (*tmp != '<')) tmp--; |
11089 | 30.9k | if (tmp < ctxt->input->base) { |
11090 | 2.03k | *lastlt = NULL; |
11091 | 2.03k | *lastgt = NULL; |
11092 | 28.9k | } else { |
11093 | 28.9k | *lastlt = tmp; |
11094 | 28.9k | tmp++; |
11095 | 1.51G | while ((tmp < ctxt->input->end) && (*tmp != '>')) { |
11096 | 1.51G | if (*tmp == '\'') { |
11097 | 9.07k | tmp++; |
11098 | 515M | while ((tmp < ctxt->input->end) && (*tmp != '\'')) tmp++; |
11099 | 9.07k | if (tmp < ctxt->input->end) tmp++; |
11100 | 1.51G | } else if (*tmp == '"') { |
11101 | 437k | tmp++; |
11102 | 516M | while ((tmp < ctxt->input->end) && (*tmp != '"')) tmp++; |
11103 | 437k | if (tmp < ctxt->input->end) tmp++; |
11104 | 437k | } else |
11105 | 1.51G | tmp++; |
11106 | 1.51G | } |
11107 | 28.9k | if (tmp < ctxt->input->end) |
11108 | 8.70k | *lastgt = tmp; |
11109 | 20.2k | else { |
11110 | 20.2k | tmp = *lastlt; |
11111 | 20.2k | tmp--; |
11112 | 1.15G | while ((tmp >= ctxt->input->base) && (*tmp != '>')) tmp--; |
11113 | 20.2k | if (tmp >= ctxt->input->base) |
11114 | 14.6k | *lastgt = tmp; |
11115 | 5.63k | else |
11116 | 5.63k | *lastgt = NULL; |
11117 | 20.2k | } |
11118 | 28.9k | } |
11119 | 84.2k | } else { |
11120 | 84.2k | *lastlt = NULL; |
11121 | 84.2k | *lastgt = NULL; |
11122 | 84.2k | } |
11123 | 115k | } |
11124 | | /** |
11125 | | * xmlCheckCdataPush: |
11126 | | * @cur: pointer to the block of characters |
11127 | | * @len: length of the block in bytes |
11128 | | * @complete: 1 if complete CDATA block is passed in, 0 if partial block |
11129 | | * |
11130 | | * Check that the block of characters is okay as SCdata content [20] |
11131 | | * |
11132 | | * Returns the number of bytes to pass if okay, a negative index where an |
11133 | | * UTF-8 error occurred otherwise |
11134 | | */ |
11135 | | static int |
11136 | 126k | xmlCheckCdataPush(const xmlChar *utf, int len, int complete) { |
11137 | 126k | int ix; |
11138 | 126k | unsigned char c; |
11139 | 126k | int codepoint; |
11140 | | |
11141 | 126k | if ((utf == NULL) || (len <= 0)) |
11142 | 26.5k | return(0); |
11143 | | |
11144 | 13.8M | for (ix = 0; ix < len;) { /* string is 0-terminated */ |
11145 | 13.7M | c = utf[ix]; |
11146 | 13.7M | if ((c & 0x80) == 0x00) { /* 1-byte code, starts with 10 */ |
11147 | 2.12M | if (c >= 0x20) |
11148 | 2.06M | ix++; |
11149 | 65.5k | else if ((c == 0xA) || (c == 0xD) || (c == 0x9)) |
11150 | 65.1k | ix++; |
11151 | 427 | else |
11152 | 427 | return(-ix); |
11153 | 11.6M | } else if ((c & 0xe0) == 0xc0) {/* 2-byte code, starts with 110 */ |
11154 | 1.25M | if (ix + 2 > len) return(complete ? -ix : ix); |
11155 | 1.25M | if ((utf[ix+1] & 0xc0 ) != 0x80) |
11156 | 212 | return(-ix); |
11157 | 1.25M | codepoint = (utf[ix] & 0x1f) << 6; |
11158 | 1.25M | codepoint |= utf[ix+1] & 0x3f; |
11159 | 1.25M | if (!xmlIsCharQ(codepoint)) |
11160 | 197 | return(-ix); |
11161 | 1.25M | ix += 2; |
11162 | 10.3M | } else if ((c & 0xf0) == 0xe0) {/* 3-byte code, starts with 1110 */ |
11163 | 10.3M | if (ix + 3 > len) return(complete ? -ix : ix); |
11164 | 10.3M | if (((utf[ix+1] & 0xc0) != 0x80) || |
11165 | 10.3M | ((utf[ix+2] & 0xc0) != 0x80)) |
11166 | 553 | return(-ix); |
11167 | 10.3M | codepoint = (utf[ix] & 0xf) << 12; |
11168 | 10.3M | codepoint |= (utf[ix+1] & 0x3f) << 6; |
11169 | 10.3M | codepoint |= utf[ix+2] & 0x3f; |
11170 | 10.3M | if (!xmlIsCharQ(codepoint)) |
11171 | 598 | return(-ix); |
11172 | 10.3M | ix += 3; |
11173 | 10.3M | } else if ((c & 0xf8) == 0xf0) {/* 4-byte code, starts with 11110 */ |
11174 | 2.31k | if (ix + 4 > len) return(complete ? -ix : ix); |
11175 | 2.29k | if (((utf[ix+1] & 0xc0) != 0x80) || |
11176 | 2.29k | ((utf[ix+2] & 0xc0) != 0x80) || |
11177 | 2.29k | ((utf[ix+3] & 0xc0) != 0x80)) |
11178 | 584 | return(-ix); |
11179 | 1.71k | codepoint = (utf[ix] & 0x7) << 18; |
11180 | 1.71k | codepoint |= (utf[ix+1] & 0x3f) << 12; |
11181 | 1.71k | codepoint |= (utf[ix+2] & 0x3f) << 6; |
11182 | 1.71k | codepoint |= utf[ix+3] & 0x3f; |
11183 | 1.71k | if (!xmlIsCharQ(codepoint)) |
11184 | 599 | return(-ix); |
11185 | 1.11k | ix += 4; |
11186 | 1.11k | } else /* unknown encoding */ |
11187 | 212 | return(-ix); |
11188 | 13.7M | } |
11189 | 95.9k | return(ix); |
11190 | 99.9k | } |
11191 | | |
11192 | | /** |
11193 | | * xmlParseTryOrFinish: |
11194 | | * @ctxt: an XML parser context |
11195 | | * @terminate: last chunk indicator |
11196 | | * |
11197 | | * Try to progress on parsing |
11198 | | * |
11199 | | * Returns zero if no parsing was possible |
11200 | | */ |
11201 | | static int |
11202 | 104k | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11203 | 104k | int ret = 0; |
11204 | 104k | int avail, tlen; |
11205 | 104k | xmlChar cur, next; |
11206 | 104k | const xmlChar *lastlt, *lastgt; |
11207 | | |
11208 | 104k | if (ctxt->input == NULL) |
11209 | 0 | return(0); |
11210 | | |
11211 | | #ifdef DEBUG_PUSH |
11212 | | switch (ctxt->instate) { |
11213 | | case XML_PARSER_EOF: |
11214 | | xmlGenericError(xmlGenericErrorContext, |
11215 | | "PP: try EOF\n"); break; |
11216 | | case XML_PARSER_START: |
11217 | | xmlGenericError(xmlGenericErrorContext, |
11218 | | "PP: try START\n"); break; |
11219 | | case XML_PARSER_MISC: |
11220 | | xmlGenericError(xmlGenericErrorContext, |
11221 | | "PP: try MISC\n");break; |
11222 | | case XML_PARSER_COMMENT: |
11223 | | xmlGenericError(xmlGenericErrorContext, |
11224 | | "PP: try COMMENT\n");break; |
11225 | | case XML_PARSER_PROLOG: |
11226 | | xmlGenericError(xmlGenericErrorContext, |
11227 | | "PP: try PROLOG\n");break; |
11228 | | case XML_PARSER_START_TAG: |
11229 | | xmlGenericError(xmlGenericErrorContext, |
11230 | | "PP: try START_TAG\n");break; |
11231 | | case XML_PARSER_CONTENT: |
11232 | | xmlGenericError(xmlGenericErrorContext, |
11233 | | "PP: try CONTENT\n");break; |
11234 | | case XML_PARSER_CDATA_SECTION: |
11235 | | xmlGenericError(xmlGenericErrorContext, |
11236 | | "PP: try CDATA_SECTION\n");break; |
11237 | | case XML_PARSER_END_TAG: |
11238 | | xmlGenericError(xmlGenericErrorContext, |
11239 | | "PP: try END_TAG\n");break; |
11240 | | case XML_PARSER_ENTITY_DECL: |
11241 | | xmlGenericError(xmlGenericErrorContext, |
11242 | | "PP: try ENTITY_DECL\n");break; |
11243 | | case XML_PARSER_ENTITY_VALUE: |
11244 | | xmlGenericError(xmlGenericErrorContext, |
11245 | | "PP: try ENTITY_VALUE\n");break; |
11246 | | case XML_PARSER_ATTRIBUTE_VALUE: |
11247 | | xmlGenericError(xmlGenericErrorContext, |
11248 | | "PP: try ATTRIBUTE_VALUE\n");break; |
11249 | | case XML_PARSER_DTD: |
11250 | | xmlGenericError(xmlGenericErrorContext, |
11251 | | "PP: try DTD\n");break; |
11252 | | case XML_PARSER_EPILOG: |
11253 | | xmlGenericError(xmlGenericErrorContext, |
11254 | | "PP: try EPILOG\n");break; |
11255 | | case XML_PARSER_PI: |
11256 | | xmlGenericError(xmlGenericErrorContext, |
11257 | | "PP: try PI\n");break; |
11258 | | case XML_PARSER_IGNORE: |
11259 | | xmlGenericError(xmlGenericErrorContext, |
11260 | | "PP: try IGNORE\n");break; |
11261 | | } |
11262 | | #endif |
11263 | | |
11264 | 104k | if ((ctxt->input != NULL) && |
11265 | 104k | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11266 | 6.49k | xmlSHRINK(ctxt); |
11267 | 6.49k | ctxt->checkIndex = 0; |
11268 | 6.49k | } |
11269 | 104k | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11270 | | |
11271 | 4.30M | while (ctxt->instate != XML_PARSER_EOF) { |
11272 | 4.30M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
11273 | 12.5k | return(0); |
11274 | | |
11275 | 4.29M | if (ctxt->input == NULL) break; |
11276 | 4.29M | if (ctxt->input->buf == NULL) |
11277 | 0 | avail = ctxt->input->length - |
11278 | 0 | (ctxt->input->cur - ctxt->input->base); |
11279 | 4.29M | else { |
11280 | | /* |
11281 | | * If we are operating on converted input, try to flush |
11282 | | * remaining chars to avoid them stalling in the non-converted |
11283 | | * buffer. But do not do this in document start where |
11284 | | * encoding="..." may not have been read and we work on a |
11285 | | * guessed encoding. |
11286 | | */ |
11287 | 4.29M | if ((ctxt->instate != XML_PARSER_START) && |
11288 | 4.29M | (ctxt->input->buf->raw != NULL) && |
11289 | 4.29M | (xmlBufIsEmpty(ctxt->input->buf->raw) == 0)) { |
11290 | 18.9k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
11291 | 18.9k | ctxt->input); |
11292 | 18.9k | size_t current = ctxt->input->cur - ctxt->input->base; |
11293 | | |
11294 | 18.9k | xmlParserInputBufferPush(ctxt->input->buf, 0, ""); |
11295 | 18.9k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
11296 | 18.9k | base, current); |
11297 | 18.9k | } |
11298 | 4.29M | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11299 | 4.29M | (ctxt->input->cur - ctxt->input->base); |
11300 | 4.29M | } |
11301 | 4.29M | if (avail < 1) |
11302 | 3.83k | goto done; |
11303 | 4.28M | switch (ctxt->instate) { |
11304 | 0 | case XML_PARSER_EOF: |
11305 | | /* |
11306 | | * Document parsing is done ! |
11307 | | */ |
11308 | 0 | goto done; |
11309 | 79.9k | case XML_PARSER_START: |
11310 | 79.9k | if (ctxt->charset == XML_CHAR_ENCODING_NONE) { |
11311 | 0 | xmlChar start[4]; |
11312 | 0 | xmlCharEncoding enc; |
11313 | | |
11314 | | /* |
11315 | | * Very first chars read from the document flow. |
11316 | | */ |
11317 | 0 | if (avail < 4) |
11318 | 0 | goto done; |
11319 | | |
11320 | | /* |
11321 | | * Get the 4 first bytes and decode the charset |
11322 | | * if enc != XML_CHAR_ENCODING_NONE |
11323 | | * plug some encoding conversion routines, |
11324 | | * else xmlSwitchEncoding will set to (default) |
11325 | | * UTF8. |
11326 | | */ |
11327 | 0 | start[0] = RAW; |
11328 | 0 | start[1] = NXT(1); |
11329 | 0 | start[2] = NXT(2); |
11330 | 0 | start[3] = NXT(3); |
11331 | 0 | enc = xmlDetectCharEncoding(start, 4); |
11332 | 0 | xmlSwitchEncoding(ctxt, enc); |
11333 | 0 | break; |
11334 | 0 | } |
11335 | | |
11336 | 79.9k | if (avail < 2) |
11337 | 11 | goto done; |
11338 | 79.9k | cur = ctxt->input->cur[0]; |
11339 | 79.9k | next = ctxt->input->cur[1]; |
11340 | 79.9k | if (cur == 0) { |
11341 | 46 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11342 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11343 | 0 | &xmlDefaultSAXLocator); |
11344 | 46 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11345 | 46 | xmlHaltParser(ctxt); |
11346 | | #ifdef DEBUG_PUSH |
11347 | | xmlGenericError(xmlGenericErrorContext, |
11348 | | "PP: entering EOF\n"); |
11349 | | #endif |
11350 | 46 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11351 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11352 | 46 | goto done; |
11353 | 46 | } |
11354 | 79.8k | if ((cur == '<') && (next == '?')) { |
11355 | | /* PI or XML decl */ |
11356 | 66.5k | if (avail < 5) return(ret); |
11357 | 66.5k | if ((!terminate) && |
11358 | 66.5k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) |
11359 | 61.6k | return(ret); |
11360 | 4.81k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11361 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11362 | 0 | &xmlDefaultSAXLocator); |
11363 | 4.81k | if ((ctxt->input->cur[2] == 'x') && |
11364 | 4.81k | (ctxt->input->cur[3] == 'm') && |
11365 | 4.81k | (ctxt->input->cur[4] == 'l') && |
11366 | 4.81k | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11367 | 3.73k | ret += 5; |
11368 | | #ifdef DEBUG_PUSH |
11369 | | xmlGenericError(xmlGenericErrorContext, |
11370 | | "PP: Parsing XML Decl\n"); |
11371 | | #endif |
11372 | 3.73k | xmlParseXMLDecl(ctxt); |
11373 | 3.73k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11374 | | /* |
11375 | | * The XML REC instructs us to stop parsing right |
11376 | | * here |
11377 | | */ |
11378 | 429 | xmlHaltParser(ctxt); |
11379 | 429 | return(0); |
11380 | 429 | } |
11381 | 3.30k | ctxt->standalone = ctxt->input->standalone; |
11382 | 3.30k | if ((ctxt->encoding == NULL) && |
11383 | 3.30k | (ctxt->input->encoding != NULL)) |
11384 | 2.92k | ctxt->encoding = xmlStrdup(ctxt->input->encoding); |
11385 | 3.30k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11386 | 3.30k | (!ctxt->disableSAX)) |
11387 | 2.86k | ctxt->sax->startDocument(ctxt->userData); |
11388 | 3.30k | ctxt->instate = XML_PARSER_MISC; |
11389 | | #ifdef DEBUG_PUSH |
11390 | | xmlGenericError(xmlGenericErrorContext, |
11391 | | "PP: entering MISC\n"); |
11392 | | #endif |
11393 | 3.30k | } else { |
11394 | 1.07k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11395 | 1.07k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11396 | 1.07k | (!ctxt->disableSAX)) |
11397 | 1.07k | ctxt->sax->startDocument(ctxt->userData); |
11398 | 1.07k | ctxt->instate = XML_PARSER_MISC; |
11399 | | #ifdef DEBUG_PUSH |
11400 | | xmlGenericError(xmlGenericErrorContext, |
11401 | | "PP: entering MISC\n"); |
11402 | | #endif |
11403 | 1.07k | } |
11404 | 13.3k | } else { |
11405 | 13.3k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11406 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11407 | 0 | &xmlDefaultSAXLocator); |
11408 | 13.3k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11409 | 13.3k | if (ctxt->version == NULL) { |
11410 | 0 | xmlErrMemory(ctxt, NULL); |
11411 | 0 | break; |
11412 | 0 | } |
11413 | 13.3k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11414 | 13.3k | (!ctxt->disableSAX)) |
11415 | 13.3k | ctxt->sax->startDocument(ctxt->userData); |
11416 | 13.3k | ctxt->instate = XML_PARSER_MISC; |
11417 | | #ifdef DEBUG_PUSH |
11418 | | xmlGenericError(xmlGenericErrorContext, |
11419 | | "PP: entering MISC\n"); |
11420 | | #endif |
11421 | 13.3k | } |
11422 | 17.7k | break; |
11423 | 1.74M | case XML_PARSER_START_TAG: { |
11424 | 1.74M | const xmlChar *name; |
11425 | 1.74M | const xmlChar *prefix = NULL; |
11426 | 1.74M | const xmlChar *URI = NULL; |
11427 | 1.74M | int line = ctxt->input->line; |
11428 | 1.74M | int nsNr = ctxt->nsNr; |
11429 | | |
11430 | 1.74M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11431 | 0 | goto done; |
11432 | 1.74M | cur = ctxt->input->cur[0]; |
11433 | 1.74M | if (cur != '<') { |
11434 | 254 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11435 | 254 | xmlHaltParser(ctxt); |
11436 | 254 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11437 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11438 | 254 | goto done; |
11439 | 254 | } |
11440 | 1.74M | if (!terminate) { |
11441 | 1.67M | if (ctxt->progressive) { |
11442 | | /* > can be found unescaped in attribute values */ |
11443 | 1.67M | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11444 | 3.37k | goto done; |
11445 | 1.67M | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11446 | 0 | goto done; |
11447 | 0 | } |
11448 | 1.67M | } |
11449 | 1.73M | if (ctxt->spaceNr == 0) |
11450 | 0 | spacePush(ctxt, -1); |
11451 | 1.73M | else if (*ctxt->space == -2) |
11452 | 198k | spacePush(ctxt, -1); |
11453 | 1.54M | else |
11454 | 1.54M | spacePush(ctxt, *ctxt->space); |
11455 | 1.73M | #ifdef LIBXML_SAX1_ENABLED |
11456 | 1.73M | if (ctxt->sax2) |
11457 | 1.73M | #endif /* LIBXML_SAX1_ENABLED */ |
11458 | 1.73M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
11459 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11460 | 0 | else |
11461 | 0 | name = xmlParseStartTag(ctxt); |
11462 | 1.73M | #endif /* LIBXML_SAX1_ENABLED */ |
11463 | 1.73M | if (ctxt->instate == XML_PARSER_EOF) |
11464 | 0 | goto done; |
11465 | 1.73M | if (name == NULL) { |
11466 | 364 | spacePop(ctxt); |
11467 | 364 | xmlHaltParser(ctxt); |
11468 | 364 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11469 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11470 | 364 | goto done; |
11471 | 364 | } |
11472 | 1.73M | #ifdef LIBXML_VALID_ENABLED |
11473 | | /* |
11474 | | * [ VC: Root Element Type ] |
11475 | | * The Name in the document type declaration must match |
11476 | | * the element type of the root element. |
11477 | | */ |
11478 | 1.73M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11479 | 1.73M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11480 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11481 | 1.73M | #endif /* LIBXML_VALID_ENABLED */ |
11482 | | |
11483 | | /* |
11484 | | * Check for an Empty Element. |
11485 | | */ |
11486 | 1.73M | if ((RAW == '/') && (NXT(1) == '>')) { |
11487 | 258k | SKIP(2); |
11488 | | |
11489 | 258k | if (ctxt->sax2) { |
11490 | 258k | if ((ctxt->sax != NULL) && |
11491 | 258k | (ctxt->sax->endElementNs != NULL) && |
11492 | 258k | (!ctxt->disableSAX)) |
11493 | 258k | ctxt->sax->endElementNs(ctxt->userData, name, |
11494 | 258k | prefix, URI); |
11495 | 258k | if (ctxt->nsNr - nsNr > 0) |
11496 | 1.22k | nsPop(ctxt, ctxt->nsNr - nsNr); |
11497 | 258k | #ifdef LIBXML_SAX1_ENABLED |
11498 | 258k | } else { |
11499 | 0 | if ((ctxt->sax != NULL) && |
11500 | 0 | (ctxt->sax->endElement != NULL) && |
11501 | 0 | (!ctxt->disableSAX)) |
11502 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
11503 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
11504 | 0 | } |
11505 | 258k | if (ctxt->instate == XML_PARSER_EOF) |
11506 | 0 | goto done; |
11507 | 258k | spacePop(ctxt); |
11508 | 258k | if (ctxt->nameNr == 0) { |
11509 | 112 | ctxt->instate = XML_PARSER_EPILOG; |
11510 | 258k | } else { |
11511 | 258k | ctxt->instate = XML_PARSER_CONTENT; |
11512 | 258k | } |
11513 | 258k | ctxt->progressive = 1; |
11514 | 258k | break; |
11515 | 258k | } |
11516 | 1.47M | if (RAW == '>') { |
11517 | 1.47M | NEXT; |
11518 | 1.47M | } else { |
11519 | 6.34k | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11520 | 6.34k | "Couldn't find end of Start Tag %s\n", |
11521 | 6.34k | name); |
11522 | 6.34k | nodePop(ctxt); |
11523 | 6.34k | spacePop(ctxt); |
11524 | 6.34k | } |
11525 | 1.47M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
11526 | | |
11527 | 1.47M | ctxt->instate = XML_PARSER_CONTENT; |
11528 | 1.47M | ctxt->progressive = 1; |
11529 | 1.47M | break; |
11530 | 1.73M | } |
11531 | 2.25M | case XML_PARSER_CONTENT: { |
11532 | 2.25M | int id; |
11533 | 2.25M | unsigned long cons; |
11534 | 2.25M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11535 | 911 | goto done; |
11536 | 2.25M | cur = ctxt->input->cur[0]; |
11537 | 2.25M | next = ctxt->input->cur[1]; |
11538 | | |
11539 | 2.25M | id = ctxt->input->id; |
11540 | 2.25M | cons = CUR_CONSUMED; |
11541 | 2.25M | if ((cur == '<') && (next == '/')) { |
11542 | 30.7k | ctxt->instate = XML_PARSER_END_TAG; |
11543 | 30.7k | break; |
11544 | 2.22M | } else if ((cur == '<') && (next == '?')) { |
11545 | 1.61k | if ((!terminate) && |
11546 | 1.61k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11547 | 328 | ctxt->progressive = XML_PARSER_PI; |
11548 | 328 | goto done; |
11549 | 328 | } |
11550 | 1.28k | xmlParsePI(ctxt); |
11551 | 1.28k | ctxt->instate = XML_PARSER_CONTENT; |
11552 | 1.28k | ctxt->progressive = 1; |
11553 | 2.22M | } else if ((cur == '<') && (next != '!')) { |
11554 | 1.72M | ctxt->instate = XML_PARSER_START_TAG; |
11555 | 1.72M | break; |
11556 | 1.72M | } else if ((cur == '<') && (next == '!') && |
11557 | 494k | (ctxt->input->cur[2] == '-') && |
11558 | 494k | (ctxt->input->cur[3] == '-')) { |
11559 | 7.28k | int term; |
11560 | | |
11561 | 7.28k | if (avail < 4) |
11562 | 0 | goto done; |
11563 | 7.28k | ctxt->input->cur += 4; |
11564 | 7.28k | term = xmlParseLookupSequence(ctxt, '-', '-', '>'); |
11565 | 7.28k | ctxt->input->cur -= 4; |
11566 | 7.28k | if ((!terminate) && (term < 0)) { |
11567 | 402 | ctxt->progressive = XML_PARSER_COMMENT; |
11568 | 402 | goto done; |
11569 | 402 | } |
11570 | 6.88k | xmlParseComment(ctxt); |
11571 | 6.88k | ctxt->instate = XML_PARSER_CONTENT; |
11572 | 6.88k | ctxt->progressive = 1; |
11573 | 486k | } else if ((cur == '<') && (ctxt->input->cur[1] == '!') && |
11574 | 486k | (ctxt->input->cur[2] == '[') && |
11575 | 486k | (ctxt->input->cur[3] == 'C') && |
11576 | 486k | (ctxt->input->cur[4] == 'D') && |
11577 | 486k | (ctxt->input->cur[5] == 'A') && |
11578 | 486k | (ctxt->input->cur[6] == 'T') && |
11579 | 486k | (ctxt->input->cur[7] == 'A') && |
11580 | 486k | (ctxt->input->cur[8] == '[')) { |
11581 | 122k | SKIP(9); |
11582 | 122k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11583 | 122k | break; |
11584 | 364k | } else if ((cur == '<') && (next == '!') && |
11585 | 364k | (avail < 9)) { |
11586 | 1.74k | goto done; |
11587 | 362k | } else if (cur == '&') { |
11588 | 3.71k | if ((!terminate) && |
11589 | 3.71k | (xmlParseLookupSequence(ctxt, ';', 0, 0) < 0)) |
11590 | 795 | goto done; |
11591 | 2.91k | xmlParseReference(ctxt); |
11592 | 358k | } else { |
11593 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11594 | | /* |
11595 | | * Goal of the following test is: |
11596 | | * - minimize calls to the SAX 'character' callback |
11597 | | * when they are mergeable |
11598 | | * - handle an problem for isBlank when we only parse |
11599 | | * a sequence of blank chars and the next one is |
11600 | | * not available to check against '<' presence. |
11601 | | * - tries to homogenize the differences in SAX |
11602 | | * callbacks between the push and pull versions |
11603 | | * of the parser. |
11604 | | */ |
11605 | 358k | if ((ctxt->inputNr == 1) && |
11606 | 358k | (avail < XML_PARSER_BIG_BUFFER_SIZE)) { |
11607 | 33.9k | if (!terminate) { |
11608 | 24.3k | if (ctxt->progressive) { |
11609 | 24.3k | if ((lastlt == NULL) || |
11610 | 24.3k | (ctxt->input->cur > lastlt)) |
11611 | 1.87k | goto done; |
11612 | 24.3k | } else if (xmlParseLookupSequence(ctxt, |
11613 | 0 | '<', 0, 0) < 0) { |
11614 | 0 | goto done; |
11615 | 0 | } |
11616 | 24.3k | } |
11617 | 33.9k | } |
11618 | 356k | ctxt->checkIndex = 0; |
11619 | 356k | xmlParseCharData(ctxt, 0); |
11620 | 356k | } |
11621 | 367k | if ((cons == CUR_CONSUMED) && (id == ctxt->input->id)) { |
11622 | 54 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
11623 | 54 | "detected an error in element content\n"); |
11624 | 54 | xmlHaltParser(ctxt); |
11625 | 54 | break; |
11626 | 54 | } |
11627 | 367k | break; |
11628 | 367k | } |
11629 | 367k | case XML_PARSER_END_TAG: |
11630 | 31.1k | if (avail < 2) |
11631 | 0 | goto done; |
11632 | 31.1k | if (!terminate) { |
11633 | 24.5k | if (ctxt->progressive) { |
11634 | | /* > can be found unescaped in attribute values */ |
11635 | 24.5k | if ((lastgt == NULL) || (ctxt->input->cur >= lastgt)) |
11636 | 450 | goto done; |
11637 | 24.5k | } else if (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0) { |
11638 | 0 | goto done; |
11639 | 0 | } |
11640 | 24.5k | } |
11641 | 30.7k | if (ctxt->sax2) { |
11642 | 30.7k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11643 | 30.7k | nameNsPop(ctxt); |
11644 | 30.7k | } |
11645 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11646 | 0 | else |
11647 | 0 | xmlParseEndTag1(ctxt, 0); |
11648 | 30.7k | #endif /* LIBXML_SAX1_ENABLED */ |
11649 | 30.7k | if (ctxt->instate == XML_PARSER_EOF) { |
11650 | | /* Nothing */ |
11651 | 30.7k | } else if (ctxt->nameNr == 0) { |
11652 | 870 | ctxt->instate = XML_PARSER_EPILOG; |
11653 | 29.8k | } else { |
11654 | 29.8k | ctxt->instate = XML_PARSER_CONTENT; |
11655 | 29.8k | } |
11656 | 30.7k | break; |
11657 | 127k | case XML_PARSER_CDATA_SECTION: { |
11658 | | /* |
11659 | | * The Push mode need to have the SAX callback for |
11660 | | * cdataBlock merge back contiguous callbacks. |
11661 | | */ |
11662 | 127k | int base; |
11663 | | |
11664 | 127k | base = xmlParseLookupSequence(ctxt, ']', ']', '>'); |
11665 | 127k | if (base < 0) { |
11666 | 4.60k | if (avail >= XML_PARSER_BIG_BUFFER_SIZE + 2) { |
11667 | 4.12k | int tmp; |
11668 | | |
11669 | 4.12k | tmp = xmlCheckCdataPush(ctxt->input->cur, |
11670 | 4.12k | XML_PARSER_BIG_BUFFER_SIZE, 0); |
11671 | 4.12k | if (tmp < 0) { |
11672 | 12 | tmp = -tmp; |
11673 | 12 | ctxt->input->cur += tmp; |
11674 | 12 | goto encoding_error; |
11675 | 12 | } |
11676 | 4.11k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
11677 | 4.11k | if (ctxt->sax->cdataBlock != NULL) |
11678 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11679 | 0 | ctxt->input->cur, tmp); |
11680 | 4.11k | else if (ctxt->sax->characters != NULL) |
11681 | 4.11k | ctxt->sax->characters(ctxt->userData, |
11682 | 4.11k | ctxt->input->cur, tmp); |
11683 | 4.11k | } |
11684 | 4.11k | if (ctxt->instate == XML_PARSER_EOF) |
11685 | 0 | goto done; |
11686 | 4.11k | SKIPL(tmp); |
11687 | 4.11k | ctxt->checkIndex = 0; |
11688 | 4.11k | } |
11689 | 4.59k | goto done; |
11690 | 122k | } else { |
11691 | 122k | int tmp; |
11692 | | |
11693 | 122k | tmp = xmlCheckCdataPush(ctxt->input->cur, base, 1); |
11694 | 122k | if ((tmp < 0) || (tmp != base)) { |
11695 | 104 | tmp = -tmp; |
11696 | 104 | ctxt->input->cur += tmp; |
11697 | 104 | goto encoding_error; |
11698 | 104 | } |
11699 | 122k | if ((ctxt->sax != NULL) && (base == 0) && |
11700 | 122k | (ctxt->sax->cdataBlock != NULL) && |
11701 | 122k | (!ctxt->disableSAX)) { |
11702 | | /* |
11703 | | * Special case to provide identical behaviour |
11704 | | * between pull and push parsers on enpty CDATA |
11705 | | * sections |
11706 | | */ |
11707 | 0 | if ((ctxt->input->cur - ctxt->input->base >= 9) && |
11708 | 0 | (!strncmp((const char *)&ctxt->input->cur[-9], |
11709 | 0 | "<![CDATA[", 9))) |
11710 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11711 | 0 | BAD_CAST "", 0); |
11712 | 122k | } else if ((ctxt->sax != NULL) && (base > 0) && |
11713 | 122k | (!ctxt->disableSAX)) { |
11714 | 95.7k | if (ctxt->sax->cdataBlock != NULL) |
11715 | 0 | ctxt->sax->cdataBlock(ctxt->userData, |
11716 | 0 | ctxt->input->cur, base); |
11717 | 95.7k | else if (ctxt->sax->characters != NULL) |
11718 | 95.7k | ctxt->sax->characters(ctxt->userData, |
11719 | 95.7k | ctxt->input->cur, base); |
11720 | 95.7k | } |
11721 | 122k | if (ctxt->instate == XML_PARSER_EOF) |
11722 | 0 | goto done; |
11723 | 122k | SKIPL(base + 3); |
11724 | 122k | ctxt->checkIndex = 0; |
11725 | 122k | ctxt->instate = XML_PARSER_CONTENT; |
11726 | | #ifdef DEBUG_PUSH |
11727 | | xmlGenericError(xmlGenericErrorContext, |
11728 | | "PP: entering CONTENT\n"); |
11729 | | #endif |
11730 | 122k | } |
11731 | 122k | break; |
11732 | 127k | } |
11733 | 122k | case XML_PARSER_MISC: |
11734 | 36.0k | SKIP_BLANKS; |
11735 | 36.0k | if (ctxt->input->buf == NULL) |
11736 | 0 | avail = ctxt->input->length - |
11737 | 0 | (ctxt->input->cur - ctxt->input->base); |
11738 | 36.0k | else |
11739 | 36.0k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11740 | 36.0k | (ctxt->input->cur - ctxt->input->base); |
11741 | 36.0k | if (avail < 2) |
11742 | 363 | goto done; |
11743 | 35.7k | cur = ctxt->input->cur[0]; |
11744 | 35.7k | next = ctxt->input->cur[1]; |
11745 | 35.7k | if ((cur == '<') && (next == '?')) { |
11746 | 14.7k | if ((!terminate) && |
11747 | 14.7k | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11748 | 362 | ctxt->progressive = XML_PARSER_PI; |
11749 | 362 | goto done; |
11750 | 362 | } |
11751 | | #ifdef DEBUG_PUSH |
11752 | | xmlGenericError(xmlGenericErrorContext, |
11753 | | "PP: Parsing PI\n"); |
11754 | | #endif |
11755 | 14.3k | xmlParsePI(ctxt); |
11756 | 14.3k | if (ctxt->instate == XML_PARSER_EOF) |
11757 | 0 | goto done; |
11758 | 14.3k | ctxt->instate = XML_PARSER_MISC; |
11759 | 14.3k | ctxt->progressive = 1; |
11760 | 14.3k | ctxt->checkIndex = 0; |
11761 | 20.9k | } else if ((cur == '<') && (next == '!') && |
11762 | 20.9k | (ctxt->input->cur[2] == '-') && |
11763 | 20.9k | (ctxt->input->cur[3] == '-')) { |
11764 | 2.33k | if ((!terminate) && |
11765 | 2.33k | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11766 | 289 | ctxt->progressive = XML_PARSER_COMMENT; |
11767 | 289 | goto done; |
11768 | 289 | } |
11769 | | #ifdef DEBUG_PUSH |
11770 | | xmlGenericError(xmlGenericErrorContext, |
11771 | | "PP: Parsing Comment\n"); |
11772 | | #endif |
11773 | 2.04k | xmlParseComment(ctxt); |
11774 | 2.04k | if (ctxt->instate == XML_PARSER_EOF) |
11775 | 0 | goto done; |
11776 | 2.04k | ctxt->instate = XML_PARSER_MISC; |
11777 | 2.04k | ctxt->progressive = 1; |
11778 | 2.04k | ctxt->checkIndex = 0; |
11779 | 18.6k | } else if ((cur == '<') && (next == '!') && |
11780 | 18.6k | (ctxt->input->cur[2] == 'D') && |
11781 | 18.6k | (ctxt->input->cur[3] == 'O') && |
11782 | 18.6k | (ctxt->input->cur[4] == 'C') && |
11783 | 18.6k | (ctxt->input->cur[5] == 'T') && |
11784 | 18.6k | (ctxt->input->cur[6] == 'Y') && |
11785 | 18.6k | (ctxt->input->cur[7] == 'P') && |
11786 | 18.6k | (ctxt->input->cur[8] == 'E')) { |
11787 | 6.05k | if ((!terminate) && |
11788 | 6.05k | (xmlParseLookupSequence(ctxt, '>', 0, 0) < 0)) { |
11789 | 551 | ctxt->progressive = XML_PARSER_DTD; |
11790 | 551 | goto done; |
11791 | 551 | } |
11792 | | #ifdef DEBUG_PUSH |
11793 | | xmlGenericError(xmlGenericErrorContext, |
11794 | | "PP: Parsing internal subset\n"); |
11795 | | #endif |
11796 | 5.50k | ctxt->inSubset = 1; |
11797 | 5.50k | ctxt->progressive = 0; |
11798 | 5.50k | ctxt->checkIndex = 0; |
11799 | 5.50k | xmlParseDocTypeDecl(ctxt); |
11800 | 5.50k | if (ctxt->instate == XML_PARSER_EOF) |
11801 | 0 | goto done; |
11802 | 5.50k | if (RAW == '[') { |
11803 | 4.84k | ctxt->instate = XML_PARSER_DTD; |
11804 | | #ifdef DEBUG_PUSH |
11805 | | xmlGenericError(xmlGenericErrorContext, |
11806 | | "PP: entering DTD\n"); |
11807 | | #endif |
11808 | 4.84k | } else { |
11809 | | /* |
11810 | | * Create and update the external subset. |
11811 | | */ |
11812 | 656 | ctxt->inSubset = 2; |
11813 | 656 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11814 | 656 | (ctxt->sax->externalSubset != NULL)) |
11815 | 0 | ctxt->sax->externalSubset(ctxt->userData, |
11816 | 0 | ctxt->intSubName, ctxt->extSubSystem, |
11817 | 0 | ctxt->extSubURI); |
11818 | 656 | ctxt->inSubset = 0; |
11819 | 656 | xmlCleanSpecialAttr(ctxt); |
11820 | 656 | ctxt->instate = XML_PARSER_PROLOG; |
11821 | | #ifdef DEBUG_PUSH |
11822 | | xmlGenericError(xmlGenericErrorContext, |
11823 | | "PP: entering PROLOG\n"); |
11824 | | #endif |
11825 | 656 | } |
11826 | 12.5k | } else if ((cur == '<') && (next == '!') && |
11827 | 12.5k | (avail < 9)) { |
11828 | 2.76k | goto done; |
11829 | 9.82k | } else { |
11830 | 9.82k | ctxt->instate = XML_PARSER_START_TAG; |
11831 | 9.82k | ctxt->progressive = XML_PARSER_START_TAG; |
11832 | 9.82k | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11833 | | #ifdef DEBUG_PUSH |
11834 | | xmlGenericError(xmlGenericErrorContext, |
11835 | | "PP: entering START_TAG\n"); |
11836 | | #endif |
11837 | 9.82k | } |
11838 | 31.7k | break; |
11839 | 31.7k | case XML_PARSER_PROLOG: |
11840 | 3.73k | SKIP_BLANKS; |
11841 | 3.73k | if (ctxt->input->buf == NULL) |
11842 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11843 | 3.73k | else |
11844 | 3.73k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11845 | 3.73k | (ctxt->input->cur - ctxt->input->base); |
11846 | 3.73k | if (avail < 2) |
11847 | 324 | goto done; |
11848 | 3.41k | cur = ctxt->input->cur[0]; |
11849 | 3.41k | next = ctxt->input->cur[1]; |
11850 | 3.41k | if ((cur == '<') && (next == '?')) { |
11851 | 871 | if ((!terminate) && |
11852 | 871 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11853 | 264 | ctxt->progressive = XML_PARSER_PI; |
11854 | 264 | goto done; |
11855 | 264 | } |
11856 | | #ifdef DEBUG_PUSH |
11857 | | xmlGenericError(xmlGenericErrorContext, |
11858 | | "PP: Parsing PI\n"); |
11859 | | #endif |
11860 | 607 | xmlParsePI(ctxt); |
11861 | 607 | if (ctxt->instate == XML_PARSER_EOF) |
11862 | 0 | goto done; |
11863 | 607 | ctxt->instate = XML_PARSER_PROLOG; |
11864 | 607 | ctxt->progressive = 1; |
11865 | 2.54k | } else if ((cur == '<') && (next == '!') && |
11866 | 2.54k | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
11867 | 1.18k | if ((!terminate) && |
11868 | 1.18k | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11869 | 228 | ctxt->progressive = XML_PARSER_COMMENT; |
11870 | 228 | goto done; |
11871 | 228 | } |
11872 | | #ifdef DEBUG_PUSH |
11873 | | xmlGenericError(xmlGenericErrorContext, |
11874 | | "PP: Parsing Comment\n"); |
11875 | | #endif |
11876 | 953 | xmlParseComment(ctxt); |
11877 | 953 | if (ctxt->instate == XML_PARSER_EOF) |
11878 | 0 | goto done; |
11879 | 953 | ctxt->instate = XML_PARSER_PROLOG; |
11880 | 953 | ctxt->progressive = 1; |
11881 | 1.36k | } else if ((cur == '<') && (next == '!') && |
11882 | 1.36k | (avail < 4)) { |
11883 | 536 | goto done; |
11884 | 824 | } else { |
11885 | 824 | ctxt->instate = XML_PARSER_START_TAG; |
11886 | 824 | if (ctxt->progressive == 0) |
11887 | 773 | ctxt->progressive = XML_PARSER_START_TAG; |
11888 | 824 | xmlParseGetLasts(ctxt, &lastlt, &lastgt); |
11889 | | #ifdef DEBUG_PUSH |
11890 | | xmlGenericError(xmlGenericErrorContext, |
11891 | | "PP: entering START_TAG\n"); |
11892 | | #endif |
11893 | 824 | } |
11894 | 2.38k | break; |
11895 | 2.47k | case XML_PARSER_EPILOG: |
11896 | 2.47k | SKIP_BLANKS; |
11897 | 2.47k | if (ctxt->input->buf == NULL) |
11898 | 0 | avail = ctxt->input->length - (ctxt->input->cur - ctxt->input->base); |
11899 | 2.47k | else |
11900 | 2.47k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11901 | 2.47k | (ctxt->input->cur - ctxt->input->base); |
11902 | 2.47k | if (avail < 2) |
11903 | 588 | goto done; |
11904 | 1.88k | cur = ctxt->input->cur[0]; |
11905 | 1.88k | next = ctxt->input->cur[1]; |
11906 | 1.88k | if ((cur == '<') && (next == '?')) { |
11907 | 753 | if ((!terminate) && |
11908 | 753 | (xmlParseLookupSequence(ctxt, '?', '>', 0) < 0)) { |
11909 | 227 | ctxt->progressive = XML_PARSER_PI; |
11910 | 227 | goto done; |
11911 | 227 | } |
11912 | | #ifdef DEBUG_PUSH |
11913 | | xmlGenericError(xmlGenericErrorContext, |
11914 | | "PP: Parsing PI\n"); |
11915 | | #endif |
11916 | 526 | xmlParsePI(ctxt); |
11917 | 526 | if (ctxt->instate == XML_PARSER_EOF) |
11918 | 0 | goto done; |
11919 | 526 | ctxt->instate = XML_PARSER_EPILOG; |
11920 | 526 | ctxt->progressive = 1; |
11921 | 1.13k | } else if ((cur == '<') && (next == '!') && |
11922 | 1.13k | (ctxt->input->cur[2] == '-') && (ctxt->input->cur[3] == '-')) { |
11923 | 691 | if ((!terminate) && |
11924 | 691 | (xmlParseLookupSequence(ctxt, '-', '-', '>') < 0)) { |
11925 | 194 | ctxt->progressive = XML_PARSER_COMMENT; |
11926 | 194 | goto done; |
11927 | 194 | } |
11928 | | #ifdef DEBUG_PUSH |
11929 | | xmlGenericError(xmlGenericErrorContext, |
11930 | | "PP: Parsing Comment\n"); |
11931 | | #endif |
11932 | 497 | xmlParseComment(ctxt); |
11933 | 497 | if (ctxt->instate == XML_PARSER_EOF) |
11934 | 0 | goto done; |
11935 | 497 | ctxt->instate = XML_PARSER_EPILOG; |
11936 | 497 | ctxt->progressive = 1; |
11937 | 497 | } else if ((cur == '<') && (next == '!') && |
11938 | 445 | (avail < 4)) { |
11939 | 412 | goto done; |
11940 | 412 | } else { |
11941 | 33 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11942 | 33 | xmlHaltParser(ctxt); |
11943 | | #ifdef DEBUG_PUSH |
11944 | | xmlGenericError(xmlGenericErrorContext, |
11945 | | "PP: entering EOF\n"); |
11946 | | #endif |
11947 | 33 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11948 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11949 | 33 | goto done; |
11950 | 33 | } |
11951 | 1.02k | break; |
11952 | 7.13k | case XML_PARSER_DTD: { |
11953 | | /* |
11954 | | * Sorry but progressive parsing of the internal subset |
11955 | | * is not expected to be supported. We first check that |
11956 | | * the full content of the internal subset is available and |
11957 | | * the parsing is launched only at that point. |
11958 | | * Internal subset ends up with "']' S? '>'" in an unescaped |
11959 | | * section and not in a ']]>' sequence which are conditional |
11960 | | * sections (whoever argued to keep that crap in XML deserve |
11961 | | * a place in hell !). |
11962 | | */ |
11963 | 7.13k | int base, i; |
11964 | 7.13k | xmlChar *buf; |
11965 | 7.13k | xmlChar quote = 0; |
11966 | 7.13k | size_t use; |
11967 | | |
11968 | 7.13k | base = ctxt->input->cur - ctxt->input->base; |
11969 | 7.13k | if (base < 0) return(0); |
11970 | 7.13k | if (ctxt->checkIndex > base) |
11971 | 1.41k | base = ctxt->checkIndex; |
11972 | 7.13k | buf = xmlBufContent(ctxt->input->buf->buffer); |
11973 | 7.13k | use = xmlBufUse(ctxt->input->buf->buffer); |
11974 | 2.51G | for (;(unsigned int) base < use; base++) { |
11975 | 2.51G | if (quote != 0) { |
11976 | 1.56G | if (buf[base] == quote) |
11977 | 58.4k | quote = 0; |
11978 | 1.56G | continue; |
11979 | 1.56G | } |
11980 | 952M | if ((quote == 0) && (buf[base] == '<')) { |
11981 | 383k | int found = 0; |
11982 | | /* special handling of comments */ |
11983 | 383k | if (((unsigned int) base + 4 < use) && |
11984 | 383k | (buf[base + 1] == '!') && |
11985 | 383k | (buf[base + 2] == '-') && |
11986 | 383k | (buf[base + 3] == '-')) { |
11987 | 297M | for (;(unsigned int) base + 3 < use; base++) { |
11988 | 297M | if ((buf[base] == '-') && |
11989 | 297M | (buf[base + 1] == '-') && |
11990 | 297M | (buf[base + 2] == '>')) { |
11991 | 9.16k | found = 1; |
11992 | 9.16k | base += 2; |
11993 | 9.16k | break; |
11994 | 9.16k | } |
11995 | 297M | } |
11996 | 9.58k | if (!found) { |
11997 | | #if 0 |
11998 | | fprintf(stderr, "unfinished comment\n"); |
11999 | | #endif |
12000 | 412 | break; /* for */ |
12001 | 412 | } |
12002 | 9.16k | continue; |
12003 | 9.58k | } |
12004 | 383k | } |
12005 | 952M | if (buf[base] == '"') { |
12006 | 15.8k | quote = '"'; |
12007 | 15.8k | continue; |
12008 | 15.8k | } |
12009 | 952M | if (buf[base] == '\'') { |
12010 | 43.5k | quote = '\''; |
12011 | 43.5k | continue; |
12012 | 43.5k | } |
12013 | 952M | if (buf[base] == ']') { |
12014 | | #if 0 |
12015 | | fprintf(stderr, "%c%c%c%c: ", buf[base], |
12016 | | buf[base + 1], buf[base + 2], buf[base + 3]); |
12017 | | #endif |
12018 | 130k | if ((unsigned int) base +1 >= use) |
12019 | 70 | break; |
12020 | 130k | if (buf[base + 1] == ']') { |
12021 | | /* conditional crap, skip both ']' ! */ |
12022 | 113k | base++; |
12023 | 113k | continue; |
12024 | 113k | } |
12025 | 1.11M | for (i = 1; (unsigned int) base + i < use; i++) { |
12026 | 1.11M | if (buf[base + i] == '>') { |
12027 | | #if 0 |
12028 | | fprintf(stderr, "found\n"); |
12029 | | #endif |
12030 | 4.50k | goto found_end_int_subset; |
12031 | 4.50k | } |
12032 | 1.11M | if (!IS_BLANK_CH(buf[base + i])) { |
12033 | | #if 0 |
12034 | | fprintf(stderr, "not found\n"); |
12035 | | #endif |
12036 | 12.4k | goto not_end_of_int_subset; |
12037 | 12.4k | } |
12038 | 1.11M | } |
12039 | | #if 0 |
12040 | | fprintf(stderr, "end of stream\n"); |
12041 | | #endif |
12042 | 214 | break; |
12043 | | |
12044 | 17.1k | } |
12045 | 951M | not_end_of_int_subset: |
12046 | 951M | continue; /* for */ |
12047 | 952M | } |
12048 | | /* |
12049 | | * We didn't found the end of the Internal subset |
12050 | | */ |
12051 | 2.62k | if (quote == 0) |
12052 | 1.68k | ctxt->checkIndex = base; |
12053 | 940 | else |
12054 | 940 | ctxt->checkIndex = 0; |
12055 | | #ifdef DEBUG_PUSH |
12056 | | if (next == 0) |
12057 | | xmlGenericError(xmlGenericErrorContext, |
12058 | | "PP: lookup of int subset end filed\n"); |
12059 | | #endif |
12060 | 2.62k | goto done; |
12061 | | |
12062 | 4.50k | found_end_int_subset: |
12063 | 4.50k | ctxt->checkIndex = 0; |
12064 | 4.50k | xmlParseInternalSubset(ctxt); |
12065 | 4.50k | if (ctxt->instate == XML_PARSER_EOF) |
12066 | 992 | goto done; |
12067 | 3.51k | ctxt->inSubset = 2; |
12068 | 3.51k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
12069 | 3.51k | (ctxt->sax->externalSubset != NULL)) |
12070 | 0 | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
12071 | 0 | ctxt->extSubSystem, ctxt->extSubURI); |
12072 | 3.51k | ctxt->inSubset = 0; |
12073 | 3.51k | xmlCleanSpecialAttr(ctxt); |
12074 | 3.51k | if (ctxt->instate == XML_PARSER_EOF) |
12075 | 0 | goto done; |
12076 | 3.51k | ctxt->instate = XML_PARSER_PROLOG; |
12077 | 3.51k | ctxt->checkIndex = 0; |
12078 | | #ifdef DEBUG_PUSH |
12079 | | xmlGenericError(xmlGenericErrorContext, |
12080 | | "PP: entering PROLOG\n"); |
12081 | | #endif |
12082 | 3.51k | break; |
12083 | 3.51k | } |
12084 | 0 | case XML_PARSER_COMMENT: |
12085 | 0 | xmlGenericError(xmlGenericErrorContext, |
12086 | 0 | "PP: internal error, state == COMMENT\n"); |
12087 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12088 | | #ifdef DEBUG_PUSH |
12089 | | xmlGenericError(xmlGenericErrorContext, |
12090 | | "PP: entering CONTENT\n"); |
12091 | | #endif |
12092 | 0 | break; |
12093 | 0 | case XML_PARSER_IGNORE: |
12094 | 0 | xmlGenericError(xmlGenericErrorContext, |
12095 | 0 | "PP: internal error, state == IGNORE"); |
12096 | 0 | ctxt->instate = XML_PARSER_DTD; |
12097 | | #ifdef DEBUG_PUSH |
12098 | | xmlGenericError(xmlGenericErrorContext, |
12099 | | "PP: entering DTD\n"); |
12100 | | #endif |
12101 | 0 | break; |
12102 | 0 | case XML_PARSER_PI: |
12103 | 0 | xmlGenericError(xmlGenericErrorContext, |
12104 | 0 | "PP: internal error, state == PI\n"); |
12105 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12106 | | #ifdef DEBUG_PUSH |
12107 | | xmlGenericError(xmlGenericErrorContext, |
12108 | | "PP: entering CONTENT\n"); |
12109 | | #endif |
12110 | 0 | break; |
12111 | 0 | case XML_PARSER_ENTITY_DECL: |
12112 | 0 | xmlGenericError(xmlGenericErrorContext, |
12113 | 0 | "PP: internal error, state == ENTITY_DECL\n"); |
12114 | 0 | ctxt->instate = XML_PARSER_DTD; |
12115 | | #ifdef DEBUG_PUSH |
12116 | | xmlGenericError(xmlGenericErrorContext, |
12117 | | "PP: entering DTD\n"); |
12118 | | #endif |
12119 | 0 | break; |
12120 | 0 | case XML_PARSER_ENTITY_VALUE: |
12121 | 0 | xmlGenericError(xmlGenericErrorContext, |
12122 | 0 | "PP: internal error, state == ENTITY_VALUE\n"); |
12123 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12124 | | #ifdef DEBUG_PUSH |
12125 | | xmlGenericError(xmlGenericErrorContext, |
12126 | | "PP: entering DTD\n"); |
12127 | | #endif |
12128 | 0 | break; |
12129 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
12130 | 0 | xmlGenericError(xmlGenericErrorContext, |
12131 | 0 | "PP: internal error, state == ATTRIBUTE_VALUE\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_SYSTEM_LITERAL: |
12139 | 0 | xmlGenericError(xmlGenericErrorContext, |
12140 | 0 | "PP: internal error, state == SYSTEM_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 | case XML_PARSER_PUBLIC_LITERAL: |
12148 | 0 | xmlGenericError(xmlGenericErrorContext, |
12149 | 0 | "PP: internal error, state == PUBLIC_LITERAL\n"); |
12150 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12151 | | #ifdef DEBUG_PUSH |
12152 | | xmlGenericError(xmlGenericErrorContext, |
12153 | | "PP: entering START_TAG\n"); |
12154 | | #endif |
12155 | 0 | break; |
12156 | 4.28M | } |
12157 | 4.28M | } |
12158 | 29.7k | done: |
12159 | | #ifdef DEBUG_PUSH |
12160 | | xmlGenericError(xmlGenericErrorContext, "PP: done %d\n", ret); |
12161 | | #endif |
12162 | 29.7k | return(ret); |
12163 | 116 | encoding_error: |
12164 | 116 | { |
12165 | 116 | char buffer[150]; |
12166 | | |
12167 | 116 | snprintf(buffer, 149, "Bytes: 0x%02X 0x%02X 0x%02X 0x%02X\n", |
12168 | 116 | ctxt->input->cur[0], ctxt->input->cur[1], |
12169 | 116 | ctxt->input->cur[2], ctxt->input->cur[3]); |
12170 | 116 | __xmlErrEncoding(ctxt, XML_ERR_INVALID_CHAR, |
12171 | 116 | "Input is not proper UTF-8, indicate encoding !\n%s", |
12172 | 116 | BAD_CAST buffer, NULL); |
12173 | 116 | } |
12174 | 116 | return(0); |
12175 | 104k | } |
12176 | | |
12177 | | /** |
12178 | | * xmlParseCheckTransition: |
12179 | | * @ctxt: an XML parser context |
12180 | | * @chunk: a char array |
12181 | | * @size: the size in byte of the chunk |
12182 | | * |
12183 | | * Check depending on the current parser state if the chunk given must be |
12184 | | * processed immediately or one need more data to advance on parsing. |
12185 | | * |
12186 | | * Returns -1 in case of error, 0 if the push is not needed and 1 if needed |
12187 | | */ |
12188 | | static int |
12189 | 84.3k | xmlParseCheckTransition(xmlParserCtxtPtr ctxt, const char *chunk, int size) { |
12190 | 84.3k | if ((ctxt == NULL) || (chunk == NULL) || (size < 0)) |
12191 | 0 | return(-1); |
12192 | 84.3k | if (ctxt->instate == XML_PARSER_START_TAG) { |
12193 | 14.0k | if (memchr(chunk, '>', size) != NULL) |
12194 | 3.01k | return(1); |
12195 | 10.9k | return(0); |
12196 | 14.0k | } |
12197 | 70.3k | if (ctxt->progressive == XML_PARSER_COMMENT) { |
12198 | 3.87k | if (memchr(chunk, '>', size) != NULL) |
12199 | 1.01k | return(1); |
12200 | 2.85k | return(0); |
12201 | 3.87k | } |
12202 | 66.4k | if (ctxt->instate == XML_PARSER_CDATA_SECTION) { |
12203 | 5.27k | if (memchr(chunk, '>', size) != NULL) |
12204 | 4.24k | return(1); |
12205 | 1.02k | return(0); |
12206 | 5.27k | } |
12207 | 61.1k | if (ctxt->progressive == XML_PARSER_PI) { |
12208 | 3.78k | if (memchr(chunk, '>', size) != NULL) |
12209 | 1.04k | return(1); |
12210 | 2.73k | return(0); |
12211 | 3.78k | } |
12212 | 57.3k | if (ctxt->instate == XML_PARSER_END_TAG) { |
12213 | 650 | if (memchr(chunk, '>', size) != NULL) |
12214 | 418 | return(1); |
12215 | 232 | return(0); |
12216 | 650 | } |
12217 | 56.7k | if ((ctxt->progressive == XML_PARSER_DTD) || |
12218 | 56.7k | (ctxt->instate == XML_PARSER_DTD)) { |
12219 | 40.1k | if (memchr(chunk, '>', size) != NULL) |
12220 | 2.39k | return(1); |
12221 | 37.7k | return(0); |
12222 | 40.1k | } |
12223 | 16.6k | return(1); |
12224 | 56.7k | } |
12225 | | |
12226 | | /** |
12227 | | * xmlParseChunk: |
12228 | | * @ctxt: an XML parser context |
12229 | | * @chunk: an char array |
12230 | | * @size: the size in byte of the chunk |
12231 | | * @terminate: last chunk indicator |
12232 | | * |
12233 | | * Parse a Chunk of memory |
12234 | | * |
12235 | | * Returns zero if no error, the xmlParserErrors otherwise. |
12236 | | */ |
12237 | | int |
12238 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
12239 | 102k | int terminate) { |
12240 | 102k | int end_in_lf = 0; |
12241 | 102k | int remain = 0; |
12242 | 102k | size_t old_avail = 0; |
12243 | 102k | size_t avail = 0; |
12244 | | |
12245 | 102k | if (ctxt == NULL) |
12246 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12247 | 102k | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12248 | 718 | return(ctxt->errNo); |
12249 | 102k | if (ctxt->instate == XML_PARSER_EOF) |
12250 | 1 | return(-1); |
12251 | 102k | if (ctxt->instate == XML_PARSER_START) |
12252 | 22.1k | xmlDetectSAX2(ctxt); |
12253 | 102k | if ((size > 0) && (chunk != NULL) && (!terminate) && |
12254 | 102k | (chunk[size - 1] == '\r')) { |
12255 | 2.49k | end_in_lf = 1; |
12256 | 2.49k | size--; |
12257 | 2.49k | } |
12258 | | |
12259 | 160k | xmldecl_done: |
12260 | | |
12261 | 160k | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
12262 | 160k | (ctxt->input->buf != NULL) && (ctxt->instate != XML_PARSER_EOF)) { |
12263 | 142k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12264 | 142k | size_t cur = ctxt->input->cur - ctxt->input->base; |
12265 | 142k | int res; |
12266 | | |
12267 | 142k | old_avail = xmlBufUse(ctxt->input->buf->buffer); |
12268 | | /* |
12269 | | * Specific handling if we autodetected an encoding, we should not |
12270 | | * push more than the first line ... which depend on the encoding |
12271 | | * And only push the rest once the final encoding was detected |
12272 | | */ |
12273 | 142k | if ((ctxt->instate == XML_PARSER_START) && (ctxt->input != NULL) && |
12274 | 142k | (ctxt->input->buf != NULL) && (ctxt->input->buf->encoder != NULL)) { |
12275 | 59.4k | unsigned int len = 45; |
12276 | | |
12277 | 59.4k | if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12278 | 59.4k | BAD_CAST "UTF-16")) || |
12279 | 59.4k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12280 | 1.49k | BAD_CAST "UTF16"))) |
12281 | 57.9k | len = 90; |
12282 | 1.49k | else if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12283 | 1.49k | BAD_CAST "UCS-4")) || |
12284 | 1.49k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12285 | 1.25k | BAD_CAST "UCS4"))) |
12286 | 237 | len = 180; |
12287 | | |
12288 | 59.4k | if (ctxt->input->buf->rawconsumed < len) |
12289 | 10 | len -= ctxt->input->buf->rawconsumed; |
12290 | | |
12291 | | /* |
12292 | | * Change size for reading the initial declaration only |
12293 | | * if size is greater than len. Otherwise, memmove in xmlBufferAdd |
12294 | | * will blindly copy extra bytes from memory. |
12295 | | */ |
12296 | 59.4k | if ((unsigned int) size > len) { |
12297 | 58.0k | remain = size - len; |
12298 | 58.0k | size = len; |
12299 | 58.0k | } else { |
12300 | 1.35k | remain = 0; |
12301 | 1.35k | } |
12302 | 59.4k | } |
12303 | 142k | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12304 | 142k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12305 | 142k | if (res < 0) { |
12306 | 91 | ctxt->errNo = XML_PARSER_EOF; |
12307 | 91 | xmlHaltParser(ctxt); |
12308 | 91 | return (XML_PARSER_EOF); |
12309 | 91 | } |
12310 | | #ifdef DEBUG_PUSH |
12311 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12312 | | #endif |
12313 | | |
12314 | 142k | } else if (ctxt->instate != XML_PARSER_EOF) { |
12315 | 17.3k | if ((ctxt->input != NULL) && ctxt->input->buf != NULL) { |
12316 | 17.3k | xmlParserInputBufferPtr in = ctxt->input->buf; |
12317 | 17.3k | if ((in->encoder != NULL) && (in->buffer != NULL) && |
12318 | 17.3k | (in->raw != NULL)) { |
12319 | 1.36k | int nbchars; |
12320 | 1.36k | size_t base = xmlBufGetInputBase(in->buffer, ctxt->input); |
12321 | 1.36k | size_t current = ctxt->input->cur - ctxt->input->base; |
12322 | | |
12323 | 1.36k | nbchars = xmlCharEncInput(in, terminate); |
12324 | 1.36k | xmlBufSetInputBaseCur(in->buffer, ctxt->input, base, current); |
12325 | 1.36k | if (nbchars < 0) { |
12326 | | /* TODO 2.6.0 */ |
12327 | 26 | xmlGenericError(xmlGenericErrorContext, |
12328 | 26 | "xmlParseChunk: encoder error\n"); |
12329 | 26 | xmlHaltParser(ctxt); |
12330 | 26 | return(XML_ERR_INVALID_ENCODING); |
12331 | 26 | } |
12332 | 1.36k | } |
12333 | 17.3k | } |
12334 | 17.3k | } |
12335 | 160k | if (remain != 0) { |
12336 | 58.0k | xmlParseTryOrFinish(ctxt, 0); |
12337 | 102k | } else { |
12338 | 102k | if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) |
12339 | 102k | avail = xmlBufUse(ctxt->input->buf->buffer); |
12340 | | /* |
12341 | | * Depending on the current state it may not be such |
12342 | | * a good idea to try parsing if there is nothing in the chunk |
12343 | | * which would be worth doing a parser state transition and we |
12344 | | * need to wait for more data |
12345 | | */ |
12346 | 102k | if ((terminate) || (avail > XML_MAX_TEXT_LENGTH) || |
12347 | 102k | (old_avail == 0) || (avail == 0) || |
12348 | 102k | (xmlParseCheckTransition(ctxt, |
12349 | 84.3k | (const char *)&ctxt->input->base[old_avail], |
12350 | 84.3k | avail - old_avail))) |
12351 | 46.5k | xmlParseTryOrFinish(ctxt, terminate); |
12352 | 102k | } |
12353 | 160k | if (ctxt->instate == XML_PARSER_EOF) |
12354 | 2.17k | return(ctxt->errNo); |
12355 | | |
12356 | 158k | if ((ctxt->input != NULL) && |
12357 | 158k | (((ctxt->input->end - ctxt->input->cur) > XML_MAX_LOOKUP_LIMIT) || |
12358 | 158k | ((ctxt->input->cur - ctxt->input->base) > XML_MAX_LOOKUP_LIMIT)) && |
12359 | 158k | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
12360 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
12361 | 0 | xmlHaltParser(ctxt); |
12362 | 0 | } |
12363 | 158k | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12364 | 12.6k | return(ctxt->errNo); |
12365 | | |
12366 | 145k | if (remain != 0) { |
12367 | 58.0k | chunk += size; |
12368 | 58.0k | size = remain; |
12369 | 58.0k | remain = 0; |
12370 | 58.0k | goto xmldecl_done; |
12371 | 58.0k | } |
12372 | 87.2k | if ((end_in_lf == 1) && (ctxt->input != NULL) && |
12373 | 87.2k | (ctxt->input->buf != NULL)) { |
12374 | 2.48k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
12375 | 2.48k | ctxt->input); |
12376 | 2.48k | size_t current = ctxt->input->cur - ctxt->input->base; |
12377 | | |
12378 | 2.48k | xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
12379 | | |
12380 | 2.48k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
12381 | 2.48k | base, current); |
12382 | 2.48k | } |
12383 | 87.2k | if (terminate) { |
12384 | | /* |
12385 | | * Check for termination |
12386 | | */ |
12387 | 3.33k | int cur_avail = 0; |
12388 | | |
12389 | 3.33k | if (ctxt->input != NULL) { |
12390 | 3.33k | if (ctxt->input->buf == NULL) |
12391 | 0 | cur_avail = ctxt->input->length - |
12392 | 0 | (ctxt->input->cur - ctxt->input->base); |
12393 | 3.33k | else |
12394 | 3.33k | cur_avail = xmlBufUse(ctxt->input->buf->buffer) - |
12395 | 3.33k | (ctxt->input->cur - ctxt->input->base); |
12396 | 3.33k | } |
12397 | | |
12398 | 3.33k | if ((ctxt->instate != XML_PARSER_EOF) && |
12399 | 3.33k | (ctxt->instate != XML_PARSER_EPILOG)) { |
12400 | 2.55k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12401 | 2.55k | } |
12402 | 3.33k | if ((ctxt->instate == XML_PARSER_EPILOG) && (cur_avail > 0)) { |
12403 | 40 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12404 | 40 | } |
12405 | 3.33k | if (ctxt->instate != XML_PARSER_EOF) { |
12406 | 3.33k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
12407 | 0 | ctxt->sax->endDocument(ctxt->userData); |
12408 | 3.33k | } |
12409 | 3.33k | ctxt->instate = XML_PARSER_EOF; |
12410 | 3.33k | } |
12411 | 87.2k | if (ctxt->wellFormed == 0) |
12412 | 2.59k | return((xmlParserErrors) ctxt->errNo); |
12413 | 84.6k | else |
12414 | 84.6k | return(0); |
12415 | 87.2k | } |
12416 | | |
12417 | | /************************************************************************ |
12418 | | * * |
12419 | | * I/O front end functions to the parser * |
12420 | | * * |
12421 | | ************************************************************************/ |
12422 | | |
12423 | | /** |
12424 | | * xmlCreatePushParserCtxt: |
12425 | | * @sax: a SAX handler |
12426 | | * @user_data: The user data returned on SAX callbacks |
12427 | | * @chunk: a pointer to an array of chars |
12428 | | * @size: number of chars in the array |
12429 | | * @filename: an optional file name or URI |
12430 | | * |
12431 | | * Create a parser context for using the XML parser in push mode. |
12432 | | * If @buffer and @size are non-NULL, the data is used to detect |
12433 | | * the encoding. The remaining characters will be parsed so they |
12434 | | * don't need to be fed in again through xmlParseChunk. |
12435 | | * To allow content encoding detection, @size should be >= 4 |
12436 | | * The value of @filename is used for fetching external entities |
12437 | | * and error/warning reports. |
12438 | | * |
12439 | | * Returns the new parser context or NULL |
12440 | | */ |
12441 | | |
12442 | | xmlParserCtxtPtr |
12443 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12444 | 18.3k | const char *chunk, int size, const char *filename) { |
12445 | 18.3k | xmlParserCtxtPtr ctxt; |
12446 | 18.3k | xmlParserInputPtr inputStream; |
12447 | 18.3k | xmlParserInputBufferPtr buf; |
12448 | 18.3k | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
12449 | | |
12450 | | /* |
12451 | | * plug some encoding conversion routines |
12452 | | */ |
12453 | 18.3k | if ((chunk != NULL) && (size >= 4)) |
12454 | 17.9k | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
12455 | | |
12456 | 18.3k | buf = xmlAllocParserInputBuffer(enc); |
12457 | 18.3k | if (buf == NULL) return(NULL); |
12458 | | |
12459 | 18.3k | ctxt = xmlNewParserCtxt(); |
12460 | 18.3k | if (ctxt == NULL) { |
12461 | 0 | xmlErrMemory(NULL, "creating parser: out of memory\n"); |
12462 | 0 | xmlFreeParserInputBuffer(buf); |
12463 | 0 | return(NULL); |
12464 | 0 | } |
12465 | 18.3k | ctxt->dictNames = 1; |
12466 | 18.3k | if (sax != NULL) { |
12467 | 18.3k | #ifdef LIBXML_SAX1_ENABLED |
12468 | 18.3k | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
12469 | 18.3k | #endif /* LIBXML_SAX1_ENABLED */ |
12470 | 18.3k | xmlFree(ctxt->sax); |
12471 | 18.3k | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
12472 | 18.3k | if (ctxt->sax == NULL) { |
12473 | 0 | xmlErrMemory(ctxt, NULL); |
12474 | 0 | xmlFreeParserInputBuffer(buf); |
12475 | 0 | xmlFreeParserCtxt(ctxt); |
12476 | 0 | return(NULL); |
12477 | 0 | } |
12478 | 18.3k | memset(ctxt->sax, 0, sizeof(xmlSAXHandler)); |
12479 | 18.3k | if (sax->initialized == XML_SAX2_MAGIC) |
12480 | 18.3k | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
12481 | 0 | else |
12482 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12483 | 18.3k | if (user_data != NULL) |
12484 | 18.3k | ctxt->userData = user_data; |
12485 | 18.3k | } |
12486 | 18.3k | if (filename == NULL) { |
12487 | 18.3k | ctxt->directory = NULL; |
12488 | 18.3k | } else { |
12489 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
12490 | 0 | } |
12491 | | |
12492 | 18.3k | inputStream = xmlNewInputStream(ctxt); |
12493 | 18.3k | if (inputStream == NULL) { |
12494 | 0 | xmlFreeParserCtxt(ctxt); |
12495 | 0 | xmlFreeParserInputBuffer(buf); |
12496 | 0 | return(NULL); |
12497 | 0 | } |
12498 | | |
12499 | 18.3k | if (filename == NULL) |
12500 | 18.3k | inputStream->filename = NULL; |
12501 | 0 | else { |
12502 | 0 | inputStream->filename = (char *) |
12503 | 0 | xmlCanonicPath((const xmlChar *) filename); |
12504 | 0 | if (inputStream->filename == NULL) { |
12505 | 0 | xmlFreeParserCtxt(ctxt); |
12506 | 0 | xmlFreeParserInputBuffer(buf); |
12507 | 0 | return(NULL); |
12508 | 0 | } |
12509 | 0 | } |
12510 | 18.3k | inputStream->buf = buf; |
12511 | 18.3k | xmlBufResetInput(inputStream->buf->buffer, inputStream); |
12512 | 18.3k | inputPush(ctxt, inputStream); |
12513 | | |
12514 | | /* |
12515 | | * If the caller didn't provide an initial 'chunk' for determining |
12516 | | * the encoding, we set the context to XML_CHAR_ENCODING_NONE so |
12517 | | * that it can be automatically determined later |
12518 | | */ |
12519 | 18.3k | if ((size == 0) || (chunk == NULL)) { |
12520 | 0 | ctxt->charset = XML_CHAR_ENCODING_NONE; |
12521 | 18.3k | } else if ((ctxt->input != NULL) && (ctxt->input->buf != NULL)) { |
12522 | 18.3k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12523 | 18.3k | size_t cur = ctxt->input->cur - ctxt->input->base; |
12524 | | |
12525 | 18.3k | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12526 | | |
12527 | 18.3k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12528 | | #ifdef DEBUG_PUSH |
12529 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12530 | | #endif |
12531 | 18.3k | } |
12532 | | |
12533 | 18.3k | if (enc != XML_CHAR_ENCODING_NONE) { |
12534 | 4.66k | xmlSwitchEncoding(ctxt, enc); |
12535 | 4.66k | } |
12536 | | |
12537 | 18.3k | return(ctxt); |
12538 | 18.3k | } |
12539 | | #endif /* LIBXML_PUSH_ENABLED */ |
12540 | | |
12541 | | /** |
12542 | | * xmlHaltParser: |
12543 | | * @ctxt: an XML parser context |
12544 | | * |
12545 | | * Blocks further parser processing don't override error |
12546 | | * for internal use |
12547 | | */ |
12548 | | static void |
12549 | 2.29k | xmlHaltParser(xmlParserCtxtPtr ctxt) { |
12550 | 2.29k | if (ctxt == NULL) |
12551 | 0 | return; |
12552 | 2.29k | ctxt->instate = XML_PARSER_EOF; |
12553 | 2.29k | ctxt->disableSAX = 1; |
12554 | 2.29k | while (ctxt->inputNr > 1) |
12555 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
12556 | 2.29k | if (ctxt->input != NULL) { |
12557 | | /* |
12558 | | * in case there was a specific allocation deallocate before |
12559 | | * overriding base |
12560 | | */ |
12561 | 2.29k | if (ctxt->input->free != NULL) { |
12562 | 0 | ctxt->input->free((xmlChar *) ctxt->input->base); |
12563 | 0 | ctxt->input->free = NULL; |
12564 | 0 | } |
12565 | 2.29k | if (ctxt->input->buf != NULL) { |
12566 | 2.29k | xmlFreeParserInputBuffer(ctxt->input->buf); |
12567 | 2.29k | ctxt->input->buf = NULL; |
12568 | 2.29k | } |
12569 | 2.29k | ctxt->input->cur = BAD_CAST""; |
12570 | 2.29k | ctxt->input->length = 0; |
12571 | 2.29k | ctxt->input->base = ctxt->input->cur; |
12572 | 2.29k | ctxt->input->end = ctxt->input->cur; |
12573 | 2.29k | } |
12574 | 2.29k | } |
12575 | | |
12576 | | /** |
12577 | | * xmlStopParser: |
12578 | | * @ctxt: an XML parser context |
12579 | | * |
12580 | | * Blocks further parser processing |
12581 | | */ |
12582 | | void |
12583 | 3 | xmlStopParser(xmlParserCtxtPtr ctxt) { |
12584 | 3 | if (ctxt == NULL) |
12585 | 0 | return; |
12586 | 3 | xmlHaltParser(ctxt); |
12587 | 3 | ctxt->errNo = XML_ERR_USER_STOP; |
12588 | 3 | } |
12589 | | |
12590 | | /** |
12591 | | * xmlCreateIOParserCtxt: |
12592 | | * @sax: a SAX handler |
12593 | | * @user_data: The user data returned on SAX callbacks |
12594 | | * @ioread: an I/O read function |
12595 | | * @ioclose: an I/O close function |
12596 | | * @ioctx: an I/O handler |
12597 | | * @enc: the charset encoding if known |
12598 | | * |
12599 | | * Create a parser context for using the XML parser with an existing |
12600 | | * I/O stream |
12601 | | * |
12602 | | * Returns the new parser context or NULL |
12603 | | */ |
12604 | | xmlParserCtxtPtr |
12605 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12606 | | xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
12607 | 0 | void *ioctx, xmlCharEncoding enc) { |
12608 | 0 | xmlParserCtxtPtr ctxt; |
12609 | 0 | xmlParserInputPtr inputStream; |
12610 | 0 | xmlParserInputBufferPtr buf; |
12611 | |
|
12612 | 0 | if (ioread == NULL) return(NULL); |
12613 | | |
12614 | 0 | buf = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, enc); |
12615 | 0 | if (buf == NULL) { |
12616 | 0 | if (ioclose != NULL) |
12617 | 0 | ioclose(ioctx); |
12618 | 0 | return (NULL); |
12619 | 0 | } |
12620 | | |
12621 | 0 | ctxt = xmlNewParserCtxt(); |
12622 | 0 | if (ctxt == NULL) { |
12623 | 0 | xmlFreeParserInputBuffer(buf); |
12624 | 0 | return(NULL); |
12625 | 0 | } |
12626 | 0 | if (sax != NULL) { |
12627 | 0 | #ifdef LIBXML_SAX1_ENABLED |
12628 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
12629 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
12630 | 0 | xmlFree(ctxt->sax); |
12631 | 0 | ctxt->sax = (xmlSAXHandlerPtr) xmlMalloc(sizeof(xmlSAXHandler)); |
12632 | 0 | if (ctxt->sax == NULL) { |
12633 | 0 | xmlFreeParserInputBuffer(buf); |
12634 | 0 | xmlErrMemory(ctxt, NULL); |
12635 | 0 | xmlFreeParserCtxt(ctxt); |
12636 | 0 | return(NULL); |
12637 | 0 | } |
12638 | 0 | memset(ctxt->sax, 0, sizeof(xmlSAXHandler)); |
12639 | 0 | if (sax->initialized == XML_SAX2_MAGIC) |
12640 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandler)); |
12641 | 0 | else |
12642 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12643 | 0 | if (user_data != NULL) |
12644 | 0 | ctxt->userData = user_data; |
12645 | 0 | } |
12646 | | |
12647 | 0 | inputStream = xmlNewIOInputStream(ctxt, buf, enc); |
12648 | 0 | if (inputStream == NULL) { |
12649 | 0 | xmlFreeParserCtxt(ctxt); |
12650 | 0 | return(NULL); |
12651 | 0 | } |
12652 | 0 | inputPush(ctxt, inputStream); |
12653 | |
|
12654 | 0 | return(ctxt); |
12655 | 0 | } |
12656 | | |
12657 | | #ifdef LIBXML_VALID_ENABLED |
12658 | | /************************************************************************ |
12659 | | * * |
12660 | | * Front ends when parsing a DTD * |
12661 | | * * |
12662 | | ************************************************************************/ |
12663 | | |
12664 | | /** |
12665 | | * xmlIOParseDTD: |
12666 | | * @sax: the SAX handler block or NULL |
12667 | | * @input: an Input Buffer |
12668 | | * @enc: the charset encoding if known |
12669 | | * |
12670 | | * Load and parse a DTD |
12671 | | * |
12672 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12673 | | * @input will be freed by the function in any case. |
12674 | | */ |
12675 | | |
12676 | | xmlDtdPtr |
12677 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
12678 | 0 | xmlCharEncoding enc) { |
12679 | 0 | xmlDtdPtr ret = NULL; |
12680 | 0 | xmlParserCtxtPtr ctxt; |
12681 | 0 | xmlParserInputPtr pinput = NULL; |
12682 | 0 | xmlChar start[4]; |
12683 | |
|
12684 | 0 | if (input == NULL) |
12685 | 0 | return(NULL); |
12686 | | |
12687 | 0 | ctxt = xmlNewParserCtxt(); |
12688 | 0 | if (ctxt == NULL) { |
12689 | 0 | xmlFreeParserInputBuffer(input); |
12690 | 0 | return(NULL); |
12691 | 0 | } |
12692 | | |
12693 | | /* We are loading a DTD */ |
12694 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12695 | | |
12696 | | /* |
12697 | | * Set-up the SAX context |
12698 | | */ |
12699 | 0 | if (sax != NULL) { |
12700 | 0 | if (ctxt->sax != NULL) |
12701 | 0 | xmlFree(ctxt->sax); |
12702 | 0 | ctxt->sax = sax; |
12703 | 0 | ctxt->userData = ctxt; |
12704 | 0 | } |
12705 | 0 | xmlDetectSAX2(ctxt); |
12706 | | |
12707 | | /* |
12708 | | * generate a parser input from the I/O handler |
12709 | | */ |
12710 | |
|
12711 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
12712 | 0 | if (pinput == NULL) { |
12713 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12714 | 0 | xmlFreeParserInputBuffer(input); |
12715 | 0 | xmlFreeParserCtxt(ctxt); |
12716 | 0 | return(NULL); |
12717 | 0 | } |
12718 | | |
12719 | | /* |
12720 | | * plug some encoding conversion routines here. |
12721 | | */ |
12722 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
12723 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12724 | 0 | xmlFreeParserCtxt(ctxt); |
12725 | 0 | return(NULL); |
12726 | 0 | } |
12727 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12728 | 0 | xmlSwitchEncoding(ctxt, enc); |
12729 | 0 | } |
12730 | |
|
12731 | 0 | pinput->filename = NULL; |
12732 | 0 | pinput->line = 1; |
12733 | 0 | pinput->col = 1; |
12734 | 0 | pinput->base = ctxt->input->cur; |
12735 | 0 | pinput->cur = ctxt->input->cur; |
12736 | 0 | pinput->free = NULL; |
12737 | | |
12738 | | /* |
12739 | | * let's parse that entity knowing it's an external subset. |
12740 | | */ |
12741 | 0 | ctxt->inSubset = 2; |
12742 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12743 | 0 | if (ctxt->myDoc == NULL) { |
12744 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12745 | 0 | return(NULL); |
12746 | 0 | } |
12747 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12748 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12749 | 0 | BAD_CAST "none", BAD_CAST "none"); |
12750 | |
|
12751 | 0 | if ((enc == XML_CHAR_ENCODING_NONE) && |
12752 | 0 | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
12753 | | /* |
12754 | | * Get the 4 first bytes and decode the charset |
12755 | | * if enc != XML_CHAR_ENCODING_NONE |
12756 | | * plug some encoding conversion routines. |
12757 | | */ |
12758 | 0 | start[0] = RAW; |
12759 | 0 | start[1] = NXT(1); |
12760 | 0 | start[2] = NXT(2); |
12761 | 0 | start[3] = NXT(3); |
12762 | 0 | enc = xmlDetectCharEncoding(start, 4); |
12763 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12764 | 0 | xmlSwitchEncoding(ctxt, enc); |
12765 | 0 | } |
12766 | 0 | } |
12767 | |
|
12768 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
12769 | |
|
12770 | 0 | if (ctxt->myDoc != NULL) { |
12771 | 0 | if (ctxt->wellFormed) { |
12772 | 0 | ret = ctxt->myDoc->extSubset; |
12773 | 0 | ctxt->myDoc->extSubset = NULL; |
12774 | 0 | if (ret != NULL) { |
12775 | 0 | xmlNodePtr tmp; |
12776 | |
|
12777 | 0 | ret->doc = NULL; |
12778 | 0 | tmp = ret->children; |
12779 | 0 | while (tmp != NULL) { |
12780 | 0 | tmp->doc = NULL; |
12781 | 0 | tmp = tmp->next; |
12782 | 0 | } |
12783 | 0 | } |
12784 | 0 | } else { |
12785 | 0 | ret = NULL; |
12786 | 0 | } |
12787 | 0 | xmlFreeDoc(ctxt->myDoc); |
12788 | 0 | ctxt->myDoc = NULL; |
12789 | 0 | } |
12790 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12791 | 0 | xmlFreeParserCtxt(ctxt); |
12792 | |
|
12793 | 0 | return(ret); |
12794 | 0 | } |
12795 | | |
12796 | | /** |
12797 | | * xmlSAXParseDTD: |
12798 | | * @sax: the SAX handler block |
12799 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12800 | | * @SystemID: a NAME* containing the URL to the DTD |
12801 | | * |
12802 | | * Load and parse an external subset. |
12803 | | * |
12804 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12805 | | */ |
12806 | | |
12807 | | xmlDtdPtr |
12808 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
12809 | 0 | const xmlChar *SystemID) { |
12810 | 0 | xmlDtdPtr ret = NULL; |
12811 | 0 | xmlParserCtxtPtr ctxt; |
12812 | 0 | xmlParserInputPtr input = NULL; |
12813 | 0 | xmlCharEncoding enc; |
12814 | 0 | xmlChar* systemIdCanonic; |
12815 | |
|
12816 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
12817 | | |
12818 | 0 | ctxt = xmlNewParserCtxt(); |
12819 | 0 | if (ctxt == NULL) { |
12820 | 0 | return(NULL); |
12821 | 0 | } |
12822 | | |
12823 | | /* We are loading a DTD */ |
12824 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12825 | | |
12826 | | /* |
12827 | | * Set-up the SAX context |
12828 | | */ |
12829 | 0 | if (sax != NULL) { |
12830 | 0 | if (ctxt->sax != NULL) |
12831 | 0 | xmlFree(ctxt->sax); |
12832 | 0 | ctxt->sax = sax; |
12833 | 0 | ctxt->userData = ctxt; |
12834 | 0 | } |
12835 | | |
12836 | | /* |
12837 | | * Canonicalise the system ID |
12838 | | */ |
12839 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
12840 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
12841 | 0 | xmlFreeParserCtxt(ctxt); |
12842 | 0 | return(NULL); |
12843 | 0 | } |
12844 | | |
12845 | | /* |
12846 | | * Ask the Entity resolver to load the damn thing |
12847 | | */ |
12848 | | |
12849 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
12850 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
12851 | 0 | systemIdCanonic); |
12852 | 0 | if (input == NULL) { |
12853 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12854 | 0 | xmlFreeParserCtxt(ctxt); |
12855 | 0 | if (systemIdCanonic != NULL) |
12856 | 0 | xmlFree(systemIdCanonic); |
12857 | 0 | return(NULL); |
12858 | 0 | } |
12859 | | |
12860 | | /* |
12861 | | * plug some encoding conversion routines here. |
12862 | | */ |
12863 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
12864 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12865 | 0 | xmlFreeParserCtxt(ctxt); |
12866 | 0 | if (systemIdCanonic != NULL) |
12867 | 0 | xmlFree(systemIdCanonic); |
12868 | 0 | return(NULL); |
12869 | 0 | } |
12870 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
12871 | 0 | enc = xmlDetectCharEncoding(ctxt->input->cur, 4); |
12872 | 0 | xmlSwitchEncoding(ctxt, enc); |
12873 | 0 | } |
12874 | |
|
12875 | 0 | if (input->filename == NULL) |
12876 | 0 | input->filename = (char *) systemIdCanonic; |
12877 | 0 | else |
12878 | 0 | xmlFree(systemIdCanonic); |
12879 | 0 | input->line = 1; |
12880 | 0 | input->col = 1; |
12881 | 0 | input->base = ctxt->input->cur; |
12882 | 0 | input->cur = ctxt->input->cur; |
12883 | 0 | input->free = NULL; |
12884 | | |
12885 | | /* |
12886 | | * let's parse that entity knowing it's an external subset. |
12887 | | */ |
12888 | 0 | ctxt->inSubset = 2; |
12889 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12890 | 0 | if (ctxt->myDoc == NULL) { |
12891 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12892 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12893 | 0 | xmlFreeParserCtxt(ctxt); |
12894 | 0 | return(NULL); |
12895 | 0 | } |
12896 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12897 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12898 | 0 | ExternalID, SystemID); |
12899 | 0 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
12900 | |
|
12901 | 0 | if (ctxt->myDoc != NULL) { |
12902 | 0 | if (ctxt->wellFormed) { |
12903 | 0 | ret = ctxt->myDoc->extSubset; |
12904 | 0 | ctxt->myDoc->extSubset = NULL; |
12905 | 0 | if (ret != NULL) { |
12906 | 0 | xmlNodePtr tmp; |
12907 | |
|
12908 | 0 | ret->doc = NULL; |
12909 | 0 | tmp = ret->children; |
12910 | 0 | while (tmp != NULL) { |
12911 | 0 | tmp->doc = NULL; |
12912 | 0 | tmp = tmp->next; |
12913 | 0 | } |
12914 | 0 | } |
12915 | 0 | } else { |
12916 | 0 | ret = NULL; |
12917 | 0 | } |
12918 | 0 | xmlFreeDoc(ctxt->myDoc); |
12919 | 0 | ctxt->myDoc = NULL; |
12920 | 0 | } |
12921 | 0 | if (sax != NULL) ctxt->sax = NULL; |
12922 | 0 | xmlFreeParserCtxt(ctxt); |
12923 | |
|
12924 | 0 | return(ret); |
12925 | 0 | } |
12926 | | |
12927 | | |
12928 | | /** |
12929 | | * xmlParseDTD: |
12930 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12931 | | * @SystemID: a NAME* containing the URL to the DTD |
12932 | | * |
12933 | | * Load and parse an external subset. |
12934 | | * |
12935 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12936 | | */ |
12937 | | |
12938 | | xmlDtdPtr |
12939 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
12940 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
12941 | 0 | } |
12942 | | #endif /* LIBXML_VALID_ENABLED */ |
12943 | | |
12944 | | /************************************************************************ |
12945 | | * * |
12946 | | * Front ends when parsing an Entity * |
12947 | | * * |
12948 | | ************************************************************************/ |
12949 | | |
12950 | | /** |
12951 | | * xmlParseCtxtExternalEntity: |
12952 | | * @ctx: the existing parsing context |
12953 | | * @URL: the URL for the entity to load |
12954 | | * @ID: the System ID for the entity to load |
12955 | | * @lst: the return value for the set of parsed nodes |
12956 | | * |
12957 | | * Parse an external general entity within an existing parsing context |
12958 | | * An external general parsed entity is well-formed if it matches the |
12959 | | * production labeled extParsedEnt. |
12960 | | * |
12961 | | * [78] extParsedEnt ::= TextDecl? content |
12962 | | * |
12963 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12964 | | * the parser error code otherwise |
12965 | | */ |
12966 | | |
12967 | | int |
12968 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx, const xmlChar *URL, |
12969 | 0 | const xmlChar *ID, xmlNodePtr *lst) { |
12970 | 0 | void *userData; |
12971 | |
|
12972 | 0 | if (ctx == NULL) return(-1); |
12973 | | /* |
12974 | | * If the user provided their own SAX callbacks, then reuse the |
12975 | | * userData callback field, otherwise the expected setup in a |
12976 | | * DOM builder is to have userData == ctxt |
12977 | | */ |
12978 | 0 | if (ctx->userData == ctx) |
12979 | 0 | userData = NULL; |
12980 | 0 | else |
12981 | 0 | userData = ctx->userData; |
12982 | 0 | return xmlParseExternalEntityPrivate(ctx->myDoc, ctx, ctx->sax, |
12983 | 0 | userData, ctx->depth + 1, |
12984 | 0 | URL, ID, lst); |
12985 | 0 | } |
12986 | | |
12987 | | /** |
12988 | | * xmlParseExternalEntityPrivate: |
12989 | | * @doc: the document the chunk pertains to |
12990 | | * @oldctxt: the previous parser context if available |
12991 | | * @sax: the SAX handler block (possibly NULL) |
12992 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12993 | | * @depth: Used for loop detection, use 0 |
12994 | | * @URL: the URL for the entity to load |
12995 | | * @ID: the System ID for the entity to load |
12996 | | * @list: the return value for the set of parsed nodes |
12997 | | * |
12998 | | * Private version of xmlParseExternalEntity() |
12999 | | * |
13000 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13001 | | * the parser error code otherwise |
13002 | | */ |
13003 | | |
13004 | | static xmlParserErrors |
13005 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
13006 | | xmlSAXHandlerPtr sax, |
13007 | | void *user_data, int depth, const xmlChar *URL, |
13008 | 0 | const xmlChar *ID, xmlNodePtr *list) { |
13009 | 0 | xmlParserCtxtPtr ctxt; |
13010 | 0 | xmlDocPtr newDoc; |
13011 | 0 | xmlNodePtr newRoot; |
13012 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13013 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13014 | 0 | xmlChar start[4]; |
13015 | 0 | xmlCharEncoding enc; |
13016 | |
|
13017 | 0 | if (((depth > 40) && |
13018 | 0 | ((oldctxt == NULL) || (oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13019 | 0 | (depth > 1024)) { |
13020 | 0 | return(XML_ERR_ENTITY_LOOP); |
13021 | 0 | } |
13022 | | |
13023 | 0 | if (list != NULL) |
13024 | 0 | *list = NULL; |
13025 | 0 | if ((URL == NULL) && (ID == NULL)) |
13026 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13027 | 0 | if (doc == NULL) |
13028 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13029 | | |
13030 | | |
13031 | 0 | ctxt = xmlCreateEntityParserCtxtInternal(URL, ID, NULL, oldctxt); |
13032 | 0 | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13033 | 0 | ctxt->userData = ctxt; |
13034 | 0 | if (sax != NULL) { |
13035 | 0 | oldsax = ctxt->sax; |
13036 | 0 | ctxt->sax = sax; |
13037 | 0 | if (user_data != NULL) |
13038 | 0 | ctxt->userData = user_data; |
13039 | 0 | } |
13040 | 0 | xmlDetectSAX2(ctxt); |
13041 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13042 | 0 | if (newDoc == NULL) { |
13043 | 0 | xmlFreeParserCtxt(ctxt); |
13044 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13045 | 0 | } |
13046 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13047 | 0 | if (doc) { |
13048 | 0 | newDoc->intSubset = doc->intSubset; |
13049 | 0 | newDoc->extSubset = doc->extSubset; |
13050 | 0 | if (doc->dict) { |
13051 | 0 | newDoc->dict = doc->dict; |
13052 | 0 | xmlDictReference(newDoc->dict); |
13053 | 0 | } |
13054 | 0 | if (doc->URL != NULL) { |
13055 | 0 | newDoc->URL = xmlStrdup(doc->URL); |
13056 | 0 | } |
13057 | 0 | } |
13058 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13059 | 0 | if (newRoot == NULL) { |
13060 | 0 | if (sax != NULL) |
13061 | 0 | ctxt->sax = oldsax; |
13062 | 0 | xmlFreeParserCtxt(ctxt); |
13063 | 0 | newDoc->intSubset = NULL; |
13064 | 0 | newDoc->extSubset = NULL; |
13065 | 0 | xmlFreeDoc(newDoc); |
13066 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13067 | 0 | } |
13068 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13069 | 0 | nodePush(ctxt, newDoc->children); |
13070 | 0 | if (doc == NULL) { |
13071 | 0 | ctxt->myDoc = newDoc; |
13072 | 0 | } else { |
13073 | 0 | ctxt->myDoc = doc; |
13074 | 0 | newRoot->doc = doc; |
13075 | 0 | } |
13076 | | |
13077 | | /* |
13078 | | * Get the 4 first bytes and decode the charset |
13079 | | * if enc != XML_CHAR_ENCODING_NONE |
13080 | | * plug some encoding conversion routines. |
13081 | | */ |
13082 | 0 | GROW; |
13083 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
13084 | 0 | start[0] = RAW; |
13085 | 0 | start[1] = NXT(1); |
13086 | 0 | start[2] = NXT(2); |
13087 | 0 | start[3] = NXT(3); |
13088 | 0 | enc = xmlDetectCharEncoding(start, 4); |
13089 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
13090 | 0 | xmlSwitchEncoding(ctxt, enc); |
13091 | 0 | } |
13092 | 0 | } |
13093 | | |
13094 | | /* |
13095 | | * Parse a possible text declaration first |
13096 | | */ |
13097 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
13098 | 0 | xmlParseTextDecl(ctxt); |
13099 | | /* |
13100 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
13101 | | */ |
13102 | 0 | if ((xmlStrEqual(oldctxt->version, BAD_CAST "1.0")) && |
13103 | 0 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
13104 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
13105 | 0 | "Version mismatch between document and entity\n"); |
13106 | 0 | } |
13107 | 0 | } |
13108 | |
|
13109 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13110 | 0 | ctxt->depth = depth; |
13111 | 0 | if (oldctxt != NULL) { |
13112 | 0 | ctxt->_private = oldctxt->_private; |
13113 | 0 | ctxt->loadsubset = oldctxt->loadsubset; |
13114 | 0 | ctxt->validate = oldctxt->validate; |
13115 | 0 | ctxt->valid = oldctxt->valid; |
13116 | 0 | ctxt->replaceEntities = oldctxt->replaceEntities; |
13117 | 0 | if (oldctxt->validate) { |
13118 | 0 | ctxt->vctxt.error = oldctxt->vctxt.error; |
13119 | 0 | ctxt->vctxt.warning = oldctxt->vctxt.warning; |
13120 | 0 | ctxt->vctxt.userData = oldctxt->vctxt.userData; |
13121 | 0 | } |
13122 | 0 | ctxt->external = oldctxt->external; |
13123 | 0 | if (ctxt->dict) xmlDictFree(ctxt->dict); |
13124 | 0 | ctxt->dict = oldctxt->dict; |
13125 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13126 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13127 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13128 | 0 | ctxt->dictNames = oldctxt->dictNames; |
13129 | 0 | ctxt->attsDefault = oldctxt->attsDefault; |
13130 | 0 | ctxt->attsSpecial = oldctxt->attsSpecial; |
13131 | 0 | ctxt->linenumbers = oldctxt->linenumbers; |
13132 | 0 | ctxt->record_info = oldctxt->record_info; |
13133 | 0 | ctxt->node_seq.maximum = oldctxt->node_seq.maximum; |
13134 | 0 | ctxt->node_seq.length = oldctxt->node_seq.length; |
13135 | 0 | ctxt->node_seq.buffer = oldctxt->node_seq.buffer; |
13136 | 0 | } else { |
13137 | | /* |
13138 | | * Doing validity checking on chunk without context |
13139 | | * doesn't make sense |
13140 | | */ |
13141 | 0 | ctxt->_private = NULL; |
13142 | 0 | ctxt->validate = 0; |
13143 | 0 | ctxt->external = 2; |
13144 | 0 | ctxt->loadsubset = 0; |
13145 | 0 | } |
13146 | |
|
13147 | 0 | xmlParseContent(ctxt); |
13148 | |
|
13149 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13150 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13151 | 0 | } else if (RAW != 0) { |
13152 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13153 | 0 | } |
13154 | 0 | if (ctxt->node != newDoc->children) { |
13155 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13156 | 0 | } |
13157 | |
|
13158 | 0 | if (!ctxt->wellFormed) { |
13159 | 0 | if (ctxt->errNo == 0) |
13160 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13161 | 0 | else |
13162 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13163 | 0 | } else { |
13164 | 0 | if (list != NULL) { |
13165 | 0 | xmlNodePtr cur; |
13166 | | |
13167 | | /* |
13168 | | * Return the newly created nodeset after unlinking it from |
13169 | | * they pseudo parent. |
13170 | | */ |
13171 | 0 | cur = newDoc->children->children; |
13172 | 0 | *list = cur; |
13173 | 0 | while (cur != NULL) { |
13174 | 0 | cur->parent = NULL; |
13175 | 0 | cur = cur->next; |
13176 | 0 | } |
13177 | 0 | newDoc->children->children = NULL; |
13178 | 0 | } |
13179 | 0 | ret = XML_ERR_OK; |
13180 | 0 | } |
13181 | | |
13182 | | /* |
13183 | | * Record in the parent context the number of entities replacement |
13184 | | * done when parsing that reference. |
13185 | | */ |
13186 | 0 | if (oldctxt != NULL) |
13187 | 0 | oldctxt->nbentities += ctxt->nbentities; |
13188 | | |
13189 | | /* |
13190 | | * Also record the size of the entity parsed |
13191 | | */ |
13192 | 0 | if (ctxt->input != NULL && oldctxt != NULL) { |
13193 | 0 | oldctxt->sizeentities += ctxt->input->consumed; |
13194 | 0 | oldctxt->sizeentities += (ctxt->input->cur - ctxt->input->base); |
13195 | 0 | } |
13196 | | /* |
13197 | | * And record the last error if any |
13198 | | */ |
13199 | 0 | if ((oldctxt != NULL) && (ctxt->lastError.code != XML_ERR_OK)) |
13200 | 0 | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13201 | |
|
13202 | 0 | if (sax != NULL) |
13203 | 0 | ctxt->sax = oldsax; |
13204 | 0 | if (oldctxt != NULL) { |
13205 | 0 | ctxt->dict = NULL; |
13206 | 0 | ctxt->attsDefault = NULL; |
13207 | 0 | ctxt->attsSpecial = NULL; |
13208 | 0 | oldctxt->validate = ctxt->validate; |
13209 | 0 | oldctxt->valid = ctxt->valid; |
13210 | 0 | oldctxt->node_seq.maximum = ctxt->node_seq.maximum; |
13211 | 0 | oldctxt->node_seq.length = ctxt->node_seq.length; |
13212 | 0 | oldctxt->node_seq.buffer = ctxt->node_seq.buffer; |
13213 | 0 | } |
13214 | 0 | ctxt->node_seq.maximum = 0; |
13215 | 0 | ctxt->node_seq.length = 0; |
13216 | 0 | ctxt->node_seq.buffer = NULL; |
13217 | 0 | xmlFreeParserCtxt(ctxt); |
13218 | 0 | newDoc->intSubset = NULL; |
13219 | 0 | newDoc->extSubset = NULL; |
13220 | 0 | xmlFreeDoc(newDoc); |
13221 | |
|
13222 | 0 | return(ret); |
13223 | 0 | } |
13224 | | |
13225 | | #ifdef LIBXML_SAX1_ENABLED |
13226 | | /** |
13227 | | * xmlParseExternalEntity: |
13228 | | * @doc: the document the chunk pertains to |
13229 | | * @sax: the SAX handler block (possibly NULL) |
13230 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13231 | | * @depth: Used for loop detection, use 0 |
13232 | | * @URL: the URL for the entity to load |
13233 | | * @ID: the System ID for the entity to load |
13234 | | * @lst: the return value for the set of parsed nodes |
13235 | | * |
13236 | | * Parse an external general entity |
13237 | | * An external general parsed entity is well-formed if it matches the |
13238 | | * production labeled extParsedEnt. |
13239 | | * |
13240 | | * [78] extParsedEnt ::= TextDecl? content |
13241 | | * |
13242 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13243 | | * the parser error code otherwise |
13244 | | */ |
13245 | | |
13246 | | int |
13247 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
13248 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *lst) { |
13249 | 0 | return(xmlParseExternalEntityPrivate(doc, NULL, sax, user_data, depth, URL, |
13250 | 0 | ID, lst)); |
13251 | 0 | } |
13252 | | |
13253 | | /** |
13254 | | * xmlParseBalancedChunkMemory: |
13255 | | * @doc: the document the chunk pertains to (must not be NULL) |
13256 | | * @sax: the SAX handler block (possibly NULL) |
13257 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13258 | | * @depth: Used for loop detection, use 0 |
13259 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13260 | | * @lst: the return value for the set of parsed nodes |
13261 | | * |
13262 | | * Parse a well-balanced chunk of an XML document |
13263 | | * called by the parser |
13264 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13265 | | * the content production in the XML grammar: |
13266 | | * |
13267 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13268 | | * |
13269 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13270 | | * the parser error code otherwise |
13271 | | */ |
13272 | | |
13273 | | int |
13274 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13275 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
13276 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
13277 | 0 | depth, string, lst, 0 ); |
13278 | 0 | } |
13279 | | #endif /* LIBXML_SAX1_ENABLED */ |
13280 | | |
13281 | | /** |
13282 | | * xmlParseBalancedChunkMemoryInternal: |
13283 | | * @oldctxt: the existing parsing context |
13284 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13285 | | * @user_data: the user data field for the parser context |
13286 | | * @lst: the return value for the set of parsed nodes |
13287 | | * |
13288 | | * |
13289 | | * Parse a well-balanced chunk of an XML document |
13290 | | * called by the parser |
13291 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13292 | | * the content production in the XML grammar: |
13293 | | * |
13294 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13295 | | * |
13296 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13297 | | * error code otherwise |
13298 | | * |
13299 | | * In case recover is set to 1, the nodelist will not be empty even if |
13300 | | * the parsed chunk is not well balanced. |
13301 | | */ |
13302 | | static xmlParserErrors |
13303 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
13304 | 0 | const xmlChar *string, void *user_data, xmlNodePtr *lst) { |
13305 | 0 | xmlParserCtxtPtr ctxt; |
13306 | 0 | xmlDocPtr newDoc = NULL; |
13307 | 0 | xmlNodePtr newRoot; |
13308 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13309 | 0 | xmlNodePtr content = NULL; |
13310 | 0 | xmlNodePtr last = NULL; |
13311 | 0 | int size; |
13312 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13313 | 0 | #ifdef SAX2 |
13314 | 0 | int i; |
13315 | 0 | #endif |
13316 | |
|
13317 | 0 | if (((oldctxt->depth > 40) && ((oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13318 | 0 | (oldctxt->depth > 1024)) { |
13319 | 0 | return(XML_ERR_ENTITY_LOOP); |
13320 | 0 | } |
13321 | | |
13322 | | |
13323 | 0 | if (lst != NULL) |
13324 | 0 | *lst = NULL; |
13325 | 0 | if (string == NULL) |
13326 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13327 | | |
13328 | 0 | size = xmlStrlen(string); |
13329 | |
|
13330 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13331 | 0 | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13332 | 0 | if (user_data != NULL) |
13333 | 0 | ctxt->userData = user_data; |
13334 | 0 | else |
13335 | 0 | ctxt->userData = ctxt; |
13336 | 0 | if (ctxt->dict != NULL) xmlDictFree(ctxt->dict); |
13337 | 0 | ctxt->dict = oldctxt->dict; |
13338 | 0 | ctxt->input_id = oldctxt->input_id + 1; |
13339 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13340 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13341 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13342 | |
|
13343 | 0 | #ifdef SAX2 |
13344 | | /* propagate namespaces down the entity */ |
13345 | 0 | for (i = 0;i < oldctxt->nsNr;i += 2) { |
13346 | 0 | nsPush(ctxt, oldctxt->nsTab[i], oldctxt->nsTab[i+1]); |
13347 | 0 | } |
13348 | 0 | #endif |
13349 | |
|
13350 | 0 | oldsax = ctxt->sax; |
13351 | 0 | ctxt->sax = oldctxt->sax; |
13352 | 0 | xmlDetectSAX2(ctxt); |
13353 | 0 | ctxt->replaceEntities = oldctxt->replaceEntities; |
13354 | 0 | ctxt->options = oldctxt->options; |
13355 | |
|
13356 | 0 | ctxt->_private = oldctxt->_private; |
13357 | 0 | if (oldctxt->myDoc == NULL) { |
13358 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13359 | 0 | if (newDoc == NULL) { |
13360 | 0 | ctxt->sax = oldsax; |
13361 | 0 | ctxt->dict = NULL; |
13362 | 0 | xmlFreeParserCtxt(ctxt); |
13363 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13364 | 0 | } |
13365 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13366 | 0 | newDoc->dict = ctxt->dict; |
13367 | 0 | xmlDictReference(newDoc->dict); |
13368 | 0 | ctxt->myDoc = newDoc; |
13369 | 0 | } else { |
13370 | 0 | ctxt->myDoc = oldctxt->myDoc; |
13371 | 0 | content = ctxt->myDoc->children; |
13372 | 0 | last = ctxt->myDoc->last; |
13373 | 0 | } |
13374 | 0 | newRoot = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13375 | 0 | if (newRoot == NULL) { |
13376 | 0 | ctxt->sax = oldsax; |
13377 | 0 | ctxt->dict = NULL; |
13378 | 0 | xmlFreeParserCtxt(ctxt); |
13379 | 0 | if (newDoc != NULL) { |
13380 | 0 | xmlFreeDoc(newDoc); |
13381 | 0 | } |
13382 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13383 | 0 | } |
13384 | 0 | ctxt->myDoc->children = NULL; |
13385 | 0 | ctxt->myDoc->last = NULL; |
13386 | 0 | xmlAddChild((xmlNodePtr) ctxt->myDoc, newRoot); |
13387 | 0 | nodePush(ctxt, ctxt->myDoc->children); |
13388 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13389 | 0 | ctxt->depth = oldctxt->depth + 1; |
13390 | |
|
13391 | 0 | ctxt->validate = 0; |
13392 | 0 | ctxt->loadsubset = oldctxt->loadsubset; |
13393 | 0 | if ((oldctxt->validate) || (oldctxt->replaceEntities != 0)) { |
13394 | | /* |
13395 | | * ID/IDREF registration will be done in xmlValidateElement below |
13396 | | */ |
13397 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13398 | 0 | } |
13399 | 0 | ctxt->dictNames = oldctxt->dictNames; |
13400 | 0 | ctxt->attsDefault = oldctxt->attsDefault; |
13401 | 0 | ctxt->attsSpecial = oldctxt->attsSpecial; |
13402 | |
|
13403 | 0 | xmlParseContent(ctxt); |
13404 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13405 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13406 | 0 | } else if (RAW != 0) { |
13407 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13408 | 0 | } |
13409 | 0 | if (ctxt->node != ctxt->myDoc->children) { |
13410 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13411 | 0 | } |
13412 | |
|
13413 | 0 | if (!ctxt->wellFormed) { |
13414 | 0 | if (ctxt->errNo == 0) |
13415 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13416 | 0 | else |
13417 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13418 | 0 | } else { |
13419 | 0 | ret = XML_ERR_OK; |
13420 | 0 | } |
13421 | |
|
13422 | 0 | if ((lst != NULL) && (ret == XML_ERR_OK)) { |
13423 | 0 | xmlNodePtr cur; |
13424 | | |
13425 | | /* |
13426 | | * Return the newly created nodeset after unlinking it from |
13427 | | * they pseudo parent. |
13428 | | */ |
13429 | 0 | cur = ctxt->myDoc->children->children; |
13430 | 0 | *lst = cur; |
13431 | 0 | while (cur != NULL) { |
13432 | 0 | #ifdef LIBXML_VALID_ENABLED |
13433 | 0 | if ((oldctxt->validate) && (oldctxt->wellFormed) && |
13434 | 0 | (oldctxt->myDoc) && (oldctxt->myDoc->intSubset) && |
13435 | 0 | (cur->type == XML_ELEMENT_NODE)) { |
13436 | 0 | oldctxt->valid &= xmlValidateElement(&oldctxt->vctxt, |
13437 | 0 | oldctxt->myDoc, cur); |
13438 | 0 | } |
13439 | 0 | #endif /* LIBXML_VALID_ENABLED */ |
13440 | 0 | cur->parent = NULL; |
13441 | 0 | cur = cur->next; |
13442 | 0 | } |
13443 | 0 | ctxt->myDoc->children->children = NULL; |
13444 | 0 | } |
13445 | 0 | if (ctxt->myDoc != NULL) { |
13446 | 0 | xmlFreeNode(ctxt->myDoc->children); |
13447 | 0 | ctxt->myDoc->children = content; |
13448 | 0 | ctxt->myDoc->last = last; |
13449 | 0 | } |
13450 | | |
13451 | | /* |
13452 | | * Record in the parent context the number of entities replacement |
13453 | | * done when parsing that reference. |
13454 | | */ |
13455 | 0 | if (oldctxt != NULL) |
13456 | 0 | oldctxt->nbentities += ctxt->nbentities; |
13457 | | |
13458 | | /* |
13459 | | * Also record the last error if any |
13460 | | */ |
13461 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
13462 | 0 | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13463 | |
|
13464 | 0 | ctxt->sax = oldsax; |
13465 | 0 | ctxt->dict = NULL; |
13466 | 0 | ctxt->attsDefault = NULL; |
13467 | 0 | ctxt->attsSpecial = NULL; |
13468 | 0 | xmlFreeParserCtxt(ctxt); |
13469 | 0 | if (newDoc != NULL) { |
13470 | 0 | xmlFreeDoc(newDoc); |
13471 | 0 | } |
13472 | |
|
13473 | 0 | return(ret); |
13474 | 0 | } |
13475 | | |
13476 | | /** |
13477 | | * xmlParseInNodeContext: |
13478 | | * @node: the context node |
13479 | | * @data: the input string |
13480 | | * @datalen: the input string length in bytes |
13481 | | * @options: a combination of xmlParserOption |
13482 | | * @lst: the return value for the set of parsed nodes |
13483 | | * |
13484 | | * Parse a well-balanced chunk of an XML document |
13485 | | * within the context (DTD, namespaces, etc ...) of the given node. |
13486 | | * |
13487 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
13488 | | * the content production in the XML grammar: |
13489 | | * |
13490 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13491 | | * |
13492 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13493 | | * error code otherwise |
13494 | | */ |
13495 | | xmlParserErrors |
13496 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
13497 | 0 | int options, xmlNodePtr *lst) { |
13498 | 0 | #ifdef SAX2 |
13499 | 0 | xmlParserCtxtPtr ctxt; |
13500 | 0 | xmlDocPtr doc = NULL; |
13501 | 0 | xmlNodePtr fake, cur; |
13502 | 0 | int nsnr = 0; |
13503 | |
|
13504 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13505 | | |
13506 | | /* |
13507 | | * check all input parameters, grab the document |
13508 | | */ |
13509 | 0 | if ((lst == NULL) || (node == NULL) || (data == NULL) || (datalen < 0)) |
13510 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13511 | 0 | switch (node->type) { |
13512 | 0 | case XML_ELEMENT_NODE: |
13513 | 0 | case XML_ATTRIBUTE_NODE: |
13514 | 0 | case XML_TEXT_NODE: |
13515 | 0 | case XML_CDATA_SECTION_NODE: |
13516 | 0 | case XML_ENTITY_REF_NODE: |
13517 | 0 | case XML_PI_NODE: |
13518 | 0 | case XML_COMMENT_NODE: |
13519 | 0 | case XML_DOCUMENT_NODE: |
13520 | 0 | case XML_HTML_DOCUMENT_NODE: |
13521 | 0 | break; |
13522 | 0 | default: |
13523 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13524 | |
|
13525 | 0 | } |
13526 | 0 | while ((node != NULL) && (node->type != XML_ELEMENT_NODE) && |
13527 | 0 | (node->type != XML_DOCUMENT_NODE) && |
13528 | 0 | (node->type != XML_HTML_DOCUMENT_NODE)) |
13529 | 0 | node = node->parent; |
13530 | 0 | if (node == NULL) |
13531 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13532 | 0 | if (node->type == XML_ELEMENT_NODE) |
13533 | 0 | doc = node->doc; |
13534 | 0 | else |
13535 | 0 | doc = (xmlDocPtr) node; |
13536 | 0 | if (doc == NULL) |
13537 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13538 | | |
13539 | | /* |
13540 | | * allocate a context and set-up everything not related to the |
13541 | | * node position in the tree |
13542 | | */ |
13543 | 0 | if (doc->type == XML_DOCUMENT_NODE) |
13544 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) data, datalen); |
13545 | 0 | #ifdef LIBXML_HTML_ENABLED |
13546 | 0 | else if (doc->type == XML_HTML_DOCUMENT_NODE) { |
13547 | 0 | ctxt = htmlCreateMemoryParserCtxt((char *) data, datalen); |
13548 | | /* |
13549 | | * When parsing in context, it makes no sense to add implied |
13550 | | * elements like html/body/etc... |
13551 | | */ |
13552 | 0 | options |= HTML_PARSE_NOIMPLIED; |
13553 | 0 | } |
13554 | 0 | #endif |
13555 | 0 | else |
13556 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13557 | | |
13558 | 0 | if (ctxt == NULL) |
13559 | 0 | return(XML_ERR_NO_MEMORY); |
13560 | | |
13561 | | /* |
13562 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
13563 | | * We need a dictionary for xmlDetectSAX2, so if there's no doc dict |
13564 | | * we must wait until the last moment to free the original one. |
13565 | | */ |
13566 | 0 | if (doc->dict != NULL) { |
13567 | 0 | if (ctxt->dict != NULL) |
13568 | 0 | xmlDictFree(ctxt->dict); |
13569 | 0 | ctxt->dict = doc->dict; |
13570 | 0 | } else |
13571 | 0 | options |= XML_PARSE_NODICT; |
13572 | |
|
13573 | 0 | if (doc->encoding != NULL) { |
13574 | 0 | xmlCharEncodingHandlerPtr hdlr; |
13575 | |
|
13576 | 0 | if (ctxt->encoding != NULL) |
13577 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
13578 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) doc->encoding); |
13579 | |
|
13580 | 0 | hdlr = xmlFindCharEncodingHandler((const char *) doc->encoding); |
13581 | 0 | if (hdlr != NULL) { |
13582 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
13583 | 0 | } else { |
13584 | 0 | return(XML_ERR_UNSUPPORTED_ENCODING); |
13585 | 0 | } |
13586 | 0 | } |
13587 | | |
13588 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
13589 | 0 | xmlDetectSAX2(ctxt); |
13590 | 0 | ctxt->myDoc = doc; |
13591 | | /* parsing in context, i.e. as within existing content */ |
13592 | 0 | ctxt->input_id = 2; |
13593 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13594 | |
|
13595 | 0 | fake = xmlNewDocComment(node->doc, NULL); |
13596 | 0 | if (fake == NULL) { |
13597 | 0 | xmlFreeParserCtxt(ctxt); |
13598 | 0 | return(XML_ERR_NO_MEMORY); |
13599 | 0 | } |
13600 | 0 | xmlAddChild(node, fake); |
13601 | |
|
13602 | 0 | if (node->type == XML_ELEMENT_NODE) { |
13603 | 0 | nodePush(ctxt, node); |
13604 | | /* |
13605 | | * initialize the SAX2 namespaces stack |
13606 | | */ |
13607 | 0 | cur = node; |
13608 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
13609 | 0 | xmlNsPtr ns = cur->nsDef; |
13610 | 0 | const xmlChar *iprefix, *ihref; |
13611 | |
|
13612 | 0 | while (ns != NULL) { |
13613 | 0 | if (ctxt->dict) { |
13614 | 0 | iprefix = xmlDictLookup(ctxt->dict, ns->prefix, -1); |
13615 | 0 | ihref = xmlDictLookup(ctxt->dict, ns->href, -1); |
13616 | 0 | } else { |
13617 | 0 | iprefix = ns->prefix; |
13618 | 0 | ihref = ns->href; |
13619 | 0 | } |
13620 | |
|
13621 | 0 | if (xmlGetNamespace(ctxt, iprefix) == NULL) { |
13622 | 0 | nsPush(ctxt, iprefix, ihref); |
13623 | 0 | nsnr++; |
13624 | 0 | } |
13625 | 0 | ns = ns->next; |
13626 | 0 | } |
13627 | 0 | cur = cur->parent; |
13628 | 0 | } |
13629 | 0 | } |
13630 | |
|
13631 | 0 | if ((ctxt->validate) || (ctxt->replaceEntities != 0)) { |
13632 | | /* |
13633 | | * ID/IDREF registration will be done in xmlValidateElement below |
13634 | | */ |
13635 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13636 | 0 | } |
13637 | |
|
13638 | 0 | #ifdef LIBXML_HTML_ENABLED |
13639 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) |
13640 | 0 | __htmlParseContent(ctxt); |
13641 | 0 | else |
13642 | 0 | #endif |
13643 | 0 | xmlParseContent(ctxt); |
13644 | |
|
13645 | 0 | nsPop(ctxt, nsnr); |
13646 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13647 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13648 | 0 | } else if (RAW != 0) { |
13649 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13650 | 0 | } |
13651 | 0 | if ((ctxt->node != NULL) && (ctxt->node != node)) { |
13652 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13653 | 0 | ctxt->wellFormed = 0; |
13654 | 0 | } |
13655 | |
|
13656 | 0 | if (!ctxt->wellFormed) { |
13657 | 0 | if (ctxt->errNo == 0) |
13658 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13659 | 0 | else |
13660 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13661 | 0 | } else { |
13662 | 0 | ret = XML_ERR_OK; |
13663 | 0 | } |
13664 | | |
13665 | | /* |
13666 | | * Return the newly created nodeset after unlinking it from |
13667 | | * the pseudo sibling. |
13668 | | */ |
13669 | |
|
13670 | 0 | cur = fake->next; |
13671 | 0 | fake->next = NULL; |
13672 | 0 | node->last = fake; |
13673 | |
|
13674 | 0 | if (cur != NULL) { |
13675 | 0 | cur->prev = NULL; |
13676 | 0 | } |
13677 | |
|
13678 | 0 | *lst = cur; |
13679 | |
|
13680 | 0 | while (cur != NULL) { |
13681 | 0 | cur->parent = NULL; |
13682 | 0 | cur = cur->next; |
13683 | 0 | } |
13684 | |
|
13685 | 0 | xmlUnlinkNode(fake); |
13686 | 0 | xmlFreeNode(fake); |
13687 | | |
13688 | |
|
13689 | 0 | if (ret != XML_ERR_OK) { |
13690 | 0 | xmlFreeNodeList(*lst); |
13691 | 0 | *lst = NULL; |
13692 | 0 | } |
13693 | |
|
13694 | 0 | if (doc->dict != NULL) |
13695 | 0 | ctxt->dict = NULL; |
13696 | 0 | xmlFreeParserCtxt(ctxt); |
13697 | |
|
13698 | 0 | return(ret); |
13699 | | #else /* !SAX2 */ |
13700 | | return(XML_ERR_INTERNAL_ERROR); |
13701 | | #endif |
13702 | 0 | } |
13703 | | |
13704 | | #ifdef LIBXML_SAX1_ENABLED |
13705 | | /** |
13706 | | * xmlParseBalancedChunkMemoryRecover: |
13707 | | * @doc: the document the chunk pertains to (must not be NULL) |
13708 | | * @sax: the SAX handler block (possibly NULL) |
13709 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13710 | | * @depth: Used for loop detection, use 0 |
13711 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13712 | | * @lst: the return value for the set of parsed nodes |
13713 | | * @recover: return nodes even if the data is broken (use 0) |
13714 | | * |
13715 | | * |
13716 | | * Parse a well-balanced chunk of an XML document |
13717 | | * called by the parser |
13718 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13719 | | * the content production in the XML grammar: |
13720 | | * |
13721 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13722 | | * |
13723 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13724 | | * the parser error code otherwise |
13725 | | * |
13726 | | * In case recover is set to 1, the nodelist will not be empty even if |
13727 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
13728 | | * some extent. |
13729 | | */ |
13730 | | int |
13731 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13732 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst, |
13733 | 0 | int recover) { |
13734 | 0 | xmlParserCtxtPtr ctxt; |
13735 | 0 | xmlDocPtr newDoc; |
13736 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13737 | 0 | xmlNodePtr content, newRoot; |
13738 | 0 | int size; |
13739 | 0 | int ret = 0; |
13740 | |
|
13741 | 0 | if (depth > 40) { |
13742 | 0 | return(XML_ERR_ENTITY_LOOP); |
13743 | 0 | } |
13744 | | |
13745 | | |
13746 | 0 | if (lst != NULL) |
13747 | 0 | *lst = NULL; |
13748 | 0 | if (string == NULL) |
13749 | 0 | return(-1); |
13750 | | |
13751 | 0 | size = xmlStrlen(string); |
13752 | |
|
13753 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13754 | 0 | if (ctxt == NULL) return(-1); |
13755 | 0 | ctxt->userData = ctxt; |
13756 | 0 | if (sax != NULL) { |
13757 | 0 | oldsax = ctxt->sax; |
13758 | 0 | ctxt->sax = sax; |
13759 | 0 | if (user_data != NULL) |
13760 | 0 | ctxt->userData = user_data; |
13761 | 0 | } |
13762 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13763 | 0 | if (newDoc == NULL) { |
13764 | 0 | xmlFreeParserCtxt(ctxt); |
13765 | 0 | return(-1); |
13766 | 0 | } |
13767 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13768 | 0 | if ((doc != NULL) && (doc->dict != NULL)) { |
13769 | 0 | xmlDictFree(ctxt->dict); |
13770 | 0 | ctxt->dict = doc->dict; |
13771 | 0 | xmlDictReference(ctxt->dict); |
13772 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13773 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13774 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13775 | 0 | ctxt->dictNames = 1; |
13776 | 0 | } else { |
13777 | 0 | xmlCtxtUseOptionsInternal(ctxt, XML_PARSE_NODICT, NULL); |
13778 | 0 | } |
13779 | | /* doc == NULL is only supported for historic reasons */ |
13780 | 0 | if (doc != NULL) { |
13781 | 0 | newDoc->intSubset = doc->intSubset; |
13782 | 0 | newDoc->extSubset = doc->extSubset; |
13783 | 0 | } |
13784 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13785 | 0 | if (newRoot == NULL) { |
13786 | 0 | if (sax != NULL) |
13787 | 0 | ctxt->sax = oldsax; |
13788 | 0 | xmlFreeParserCtxt(ctxt); |
13789 | 0 | newDoc->intSubset = NULL; |
13790 | 0 | newDoc->extSubset = NULL; |
13791 | 0 | xmlFreeDoc(newDoc); |
13792 | 0 | return(-1); |
13793 | 0 | } |
13794 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13795 | 0 | nodePush(ctxt, newRoot); |
13796 | | /* doc == NULL is only supported for historic reasons */ |
13797 | 0 | if (doc == NULL) { |
13798 | 0 | ctxt->myDoc = newDoc; |
13799 | 0 | } else { |
13800 | 0 | ctxt->myDoc = newDoc; |
13801 | 0 | newDoc->children->doc = doc; |
13802 | | /* Ensure that doc has XML spec namespace */ |
13803 | 0 | xmlSearchNsByHref(doc, (xmlNodePtr)doc, XML_XML_NAMESPACE); |
13804 | 0 | newDoc->oldNs = doc->oldNs; |
13805 | 0 | } |
13806 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13807 | 0 | ctxt->input_id = 2; |
13808 | 0 | ctxt->depth = depth; |
13809 | | |
13810 | | /* |
13811 | | * Doing validity checking on chunk doesn't make sense |
13812 | | */ |
13813 | 0 | ctxt->validate = 0; |
13814 | 0 | ctxt->loadsubset = 0; |
13815 | 0 | xmlDetectSAX2(ctxt); |
13816 | |
|
13817 | 0 | if ( doc != NULL ){ |
13818 | 0 | content = doc->children; |
13819 | 0 | doc->children = NULL; |
13820 | 0 | xmlParseContent(ctxt); |
13821 | 0 | doc->children = content; |
13822 | 0 | } |
13823 | 0 | else { |
13824 | 0 | xmlParseContent(ctxt); |
13825 | 0 | } |
13826 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13827 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13828 | 0 | } else if (RAW != 0) { |
13829 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13830 | 0 | } |
13831 | 0 | if (ctxt->node != newDoc->children) { |
13832 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13833 | 0 | } |
13834 | |
|
13835 | 0 | if (!ctxt->wellFormed) { |
13836 | 0 | if (ctxt->errNo == 0) |
13837 | 0 | ret = 1; |
13838 | 0 | else |
13839 | 0 | ret = ctxt->errNo; |
13840 | 0 | } else { |
13841 | 0 | ret = 0; |
13842 | 0 | } |
13843 | |
|
13844 | 0 | if ((lst != NULL) && ((ret == 0) || (recover == 1))) { |
13845 | 0 | xmlNodePtr cur; |
13846 | | |
13847 | | /* |
13848 | | * Return the newly created nodeset after unlinking it from |
13849 | | * they pseudo parent. |
13850 | | */ |
13851 | 0 | cur = newDoc->children->children; |
13852 | 0 | *lst = cur; |
13853 | 0 | while (cur != NULL) { |
13854 | 0 | xmlSetTreeDoc(cur, doc); |
13855 | 0 | cur->parent = NULL; |
13856 | 0 | cur = cur->next; |
13857 | 0 | } |
13858 | 0 | newDoc->children->children = NULL; |
13859 | 0 | } |
13860 | |
|
13861 | 0 | if (sax != NULL) |
13862 | 0 | ctxt->sax = oldsax; |
13863 | 0 | xmlFreeParserCtxt(ctxt); |
13864 | 0 | newDoc->intSubset = NULL; |
13865 | 0 | newDoc->extSubset = NULL; |
13866 | | /* This leaks the namespace list if doc == NULL */ |
13867 | 0 | newDoc->oldNs = NULL; |
13868 | 0 | xmlFreeDoc(newDoc); |
13869 | |
|
13870 | 0 | return(ret); |
13871 | 0 | } |
13872 | | |
13873 | | /** |
13874 | | * xmlSAXParseEntity: |
13875 | | * @sax: the SAX handler block |
13876 | | * @filename: the filename |
13877 | | * |
13878 | | * parse an XML external entity out of context and build a tree. |
13879 | | * It use the given SAX function block to handle the parsing callback. |
13880 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13881 | | * |
13882 | | * [78] extParsedEnt ::= TextDecl? content |
13883 | | * |
13884 | | * This correspond to a "Well Balanced" chunk |
13885 | | * |
13886 | | * Returns the resulting document tree |
13887 | | */ |
13888 | | |
13889 | | xmlDocPtr |
13890 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
13891 | 0 | xmlDocPtr ret; |
13892 | 0 | xmlParserCtxtPtr ctxt; |
13893 | |
|
13894 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13895 | 0 | if (ctxt == NULL) { |
13896 | 0 | return(NULL); |
13897 | 0 | } |
13898 | 0 | if (sax != NULL) { |
13899 | 0 | if (ctxt->sax != NULL) |
13900 | 0 | xmlFree(ctxt->sax); |
13901 | 0 | ctxt->sax = sax; |
13902 | 0 | ctxt->userData = NULL; |
13903 | 0 | } |
13904 | |
|
13905 | 0 | xmlParseExtParsedEnt(ctxt); |
13906 | |
|
13907 | 0 | if (ctxt->wellFormed) |
13908 | 0 | ret = ctxt->myDoc; |
13909 | 0 | else { |
13910 | 0 | ret = NULL; |
13911 | 0 | xmlFreeDoc(ctxt->myDoc); |
13912 | 0 | ctxt->myDoc = NULL; |
13913 | 0 | } |
13914 | 0 | if (sax != NULL) |
13915 | 0 | ctxt->sax = NULL; |
13916 | 0 | xmlFreeParserCtxt(ctxt); |
13917 | |
|
13918 | 0 | return(ret); |
13919 | 0 | } |
13920 | | |
13921 | | /** |
13922 | | * xmlParseEntity: |
13923 | | * @filename: the filename |
13924 | | * |
13925 | | * parse an XML external entity out of context and build a tree. |
13926 | | * |
13927 | | * [78] extParsedEnt ::= TextDecl? content |
13928 | | * |
13929 | | * This correspond to a "Well Balanced" chunk |
13930 | | * |
13931 | | * Returns the resulting document tree |
13932 | | */ |
13933 | | |
13934 | | xmlDocPtr |
13935 | 0 | xmlParseEntity(const char *filename) { |
13936 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
13937 | 0 | } |
13938 | | #endif /* LIBXML_SAX1_ENABLED */ |
13939 | | |
13940 | | /** |
13941 | | * xmlCreateEntityParserCtxtInternal: |
13942 | | * @URL: the entity URL |
13943 | | * @ID: the entity PUBLIC ID |
13944 | | * @base: a possible base for the target URI |
13945 | | * @pctx: parser context used to set options on new context |
13946 | | * |
13947 | | * Create a parser context for an external entity |
13948 | | * Automatic support for ZLIB/Compress compressed document is provided |
13949 | | * by default if found at compile-time. |
13950 | | * |
13951 | | * Returns the new parser context or NULL |
13952 | | */ |
13953 | | static xmlParserCtxtPtr |
13954 | | xmlCreateEntityParserCtxtInternal(const xmlChar *URL, const xmlChar *ID, |
13955 | 0 | const xmlChar *base, xmlParserCtxtPtr pctx) { |
13956 | 0 | xmlParserCtxtPtr ctxt; |
13957 | 0 | xmlParserInputPtr inputStream; |
13958 | 0 | char *directory = NULL; |
13959 | 0 | xmlChar *uri; |
13960 | |
|
13961 | 0 | ctxt = xmlNewParserCtxt(); |
13962 | 0 | if (ctxt == NULL) { |
13963 | 0 | return(NULL); |
13964 | 0 | } |
13965 | | |
13966 | 0 | if (pctx != NULL) { |
13967 | 0 | ctxt->options = pctx->options; |
13968 | 0 | ctxt->_private = pctx->_private; |
13969 | | /* |
13970 | | * this is a subparser of pctx, so the input_id should be |
13971 | | * incremented to distinguish from main entity |
13972 | | */ |
13973 | 0 | ctxt->input_id = pctx->input_id + 1; |
13974 | 0 | } |
13975 | | |
13976 | | /* Don't read from stdin. */ |
13977 | 0 | if (xmlStrcmp(URL, BAD_CAST "-") == 0) |
13978 | 0 | URL = BAD_CAST "./-"; |
13979 | |
|
13980 | 0 | uri = xmlBuildURI(URL, base); |
13981 | |
|
13982 | 0 | if (uri == NULL) { |
13983 | 0 | inputStream = xmlLoadExternalEntity((char *)URL, (char *)ID, ctxt); |
13984 | 0 | if (inputStream == NULL) { |
13985 | 0 | xmlFreeParserCtxt(ctxt); |
13986 | 0 | return(NULL); |
13987 | 0 | } |
13988 | | |
13989 | 0 | inputPush(ctxt, inputStream); |
13990 | |
|
13991 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
13992 | 0 | directory = xmlParserGetDirectory((char *)URL); |
13993 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
13994 | 0 | ctxt->directory = directory; |
13995 | 0 | } else { |
13996 | 0 | inputStream = xmlLoadExternalEntity((char *)uri, (char *)ID, ctxt); |
13997 | 0 | if (inputStream == NULL) { |
13998 | 0 | xmlFree(uri); |
13999 | 0 | xmlFreeParserCtxt(ctxt); |
14000 | 0 | return(NULL); |
14001 | 0 | } |
14002 | | |
14003 | 0 | inputPush(ctxt, inputStream); |
14004 | |
|
14005 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
14006 | 0 | directory = xmlParserGetDirectory((char *)uri); |
14007 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
14008 | 0 | ctxt->directory = directory; |
14009 | 0 | xmlFree(uri); |
14010 | 0 | } |
14011 | 0 | return(ctxt); |
14012 | 0 | } |
14013 | | |
14014 | | /** |
14015 | | * xmlCreateEntityParserCtxt: |
14016 | | * @URL: the entity URL |
14017 | | * @ID: the entity PUBLIC ID |
14018 | | * @base: a possible base for the target URI |
14019 | | * |
14020 | | * Create a parser context for an external entity |
14021 | | * Automatic support for ZLIB/Compress compressed document is provided |
14022 | | * by default if found at compile-time. |
14023 | | * |
14024 | | * Returns the new parser context or NULL |
14025 | | */ |
14026 | | xmlParserCtxtPtr |
14027 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
14028 | 0 | const xmlChar *base) { |
14029 | 0 | return xmlCreateEntityParserCtxtInternal(URL, ID, base, NULL); |
14030 | |
|
14031 | 0 | } |
14032 | | |
14033 | | /************************************************************************ |
14034 | | * * |
14035 | | * Front ends when parsing from a file * |
14036 | | * * |
14037 | | ************************************************************************/ |
14038 | | |
14039 | | /** |
14040 | | * xmlCreateURLParserCtxt: |
14041 | | * @filename: the filename or URL |
14042 | | * @options: a combination of xmlParserOption |
14043 | | * |
14044 | | * Create a parser context for a file or URL content. |
14045 | | * Automatic support for ZLIB/Compress compressed document is provided |
14046 | | * by default if found at compile-time and for file accesses |
14047 | | * |
14048 | | * Returns the new parser context or NULL |
14049 | | */ |
14050 | | xmlParserCtxtPtr |
14051 | | xmlCreateURLParserCtxt(const char *filename, int options) |
14052 | 0 | { |
14053 | 0 | xmlParserCtxtPtr ctxt; |
14054 | 0 | xmlParserInputPtr inputStream; |
14055 | 0 | char *directory = NULL; |
14056 | |
|
14057 | 0 | ctxt = xmlNewParserCtxt(); |
14058 | 0 | if (ctxt == NULL) { |
14059 | 0 | xmlErrMemory(NULL, "cannot allocate parser context"); |
14060 | 0 | return(NULL); |
14061 | 0 | } |
14062 | | |
14063 | 0 | if (options) |
14064 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
14065 | 0 | ctxt->linenumbers = 1; |
14066 | |
|
14067 | 0 | inputStream = xmlLoadExternalEntity(filename, NULL, ctxt); |
14068 | 0 | if (inputStream == NULL) { |
14069 | 0 | xmlFreeParserCtxt(ctxt); |
14070 | 0 | return(NULL); |
14071 | 0 | } |
14072 | | |
14073 | 0 | inputPush(ctxt, inputStream); |
14074 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
14075 | 0 | directory = xmlParserGetDirectory(filename); |
14076 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
14077 | 0 | ctxt->directory = directory; |
14078 | |
|
14079 | 0 | return(ctxt); |
14080 | 0 | } |
14081 | | |
14082 | | /** |
14083 | | * xmlCreateFileParserCtxt: |
14084 | | * @filename: the filename |
14085 | | * |
14086 | | * Create a parser context for a file content. |
14087 | | * Automatic support for ZLIB/Compress compressed document is provided |
14088 | | * by default if found at compile-time. |
14089 | | * |
14090 | | * Returns the new parser context or NULL |
14091 | | */ |
14092 | | xmlParserCtxtPtr |
14093 | | xmlCreateFileParserCtxt(const char *filename) |
14094 | 0 | { |
14095 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
14096 | 0 | } |
14097 | | |
14098 | | #ifdef LIBXML_SAX1_ENABLED |
14099 | | /** |
14100 | | * xmlSAXParseFileWithData: |
14101 | | * @sax: the SAX handler block |
14102 | | * @filename: the filename |
14103 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14104 | | * documents |
14105 | | * @data: the userdata |
14106 | | * |
14107 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14108 | | * compressed document is provided by default if found at compile-time. |
14109 | | * It use the given SAX function block to handle the parsing callback. |
14110 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14111 | | * |
14112 | | * User data (void *) is stored within the parser context in the |
14113 | | * context's _private member, so it is available nearly everywhere in libxml |
14114 | | * |
14115 | | * Returns the resulting document tree |
14116 | | */ |
14117 | | |
14118 | | xmlDocPtr |
14119 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
14120 | 0 | int recovery, void *data) { |
14121 | 0 | xmlDocPtr ret; |
14122 | 0 | xmlParserCtxtPtr ctxt; |
14123 | |
|
14124 | 0 | xmlInitParser(); |
14125 | |
|
14126 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14127 | 0 | if (ctxt == NULL) { |
14128 | 0 | return(NULL); |
14129 | 0 | } |
14130 | 0 | if (sax != NULL) { |
14131 | 0 | if (ctxt->sax != NULL) |
14132 | 0 | xmlFree(ctxt->sax); |
14133 | 0 | ctxt->sax = sax; |
14134 | 0 | } |
14135 | 0 | xmlDetectSAX2(ctxt); |
14136 | 0 | if (data!=NULL) { |
14137 | 0 | ctxt->_private = data; |
14138 | 0 | } |
14139 | |
|
14140 | 0 | if (ctxt->directory == NULL) |
14141 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14142 | |
|
14143 | 0 | ctxt->recovery = recovery; |
14144 | |
|
14145 | 0 | xmlParseDocument(ctxt); |
14146 | |
|
14147 | 0 | if ((ctxt->wellFormed) || recovery) { |
14148 | 0 | ret = ctxt->myDoc; |
14149 | 0 | if ((ret != NULL) && (ctxt->input->buf != NULL)) { |
14150 | 0 | if (ctxt->input->buf->compressed > 0) |
14151 | 0 | ret->compression = 9; |
14152 | 0 | else |
14153 | 0 | ret->compression = ctxt->input->buf->compressed; |
14154 | 0 | } |
14155 | 0 | } |
14156 | 0 | else { |
14157 | 0 | ret = NULL; |
14158 | 0 | xmlFreeDoc(ctxt->myDoc); |
14159 | 0 | ctxt->myDoc = NULL; |
14160 | 0 | } |
14161 | 0 | if (sax != NULL) |
14162 | 0 | ctxt->sax = NULL; |
14163 | 0 | xmlFreeParserCtxt(ctxt); |
14164 | |
|
14165 | 0 | return(ret); |
14166 | 0 | } |
14167 | | |
14168 | | /** |
14169 | | * xmlSAXParseFile: |
14170 | | * @sax: the SAX handler block |
14171 | | * @filename: the filename |
14172 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14173 | | * documents |
14174 | | * |
14175 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14176 | | * compressed document is provided by default if found at compile-time. |
14177 | | * It use the given SAX function block to handle the parsing callback. |
14178 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14179 | | * |
14180 | | * Returns the resulting document tree |
14181 | | */ |
14182 | | |
14183 | | xmlDocPtr |
14184 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
14185 | 0 | int recovery) { |
14186 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
14187 | 0 | } |
14188 | | |
14189 | | /** |
14190 | | * xmlRecoverDoc: |
14191 | | * @cur: a pointer to an array of xmlChar |
14192 | | * |
14193 | | * parse an XML in-memory document and build a tree. |
14194 | | * In the case the document is not Well Formed, a attempt to build a |
14195 | | * tree is tried anyway |
14196 | | * |
14197 | | * Returns the resulting document tree or NULL in case of failure |
14198 | | */ |
14199 | | |
14200 | | xmlDocPtr |
14201 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
14202 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
14203 | 0 | } |
14204 | | |
14205 | | /** |
14206 | | * xmlParseFile: |
14207 | | * @filename: the filename |
14208 | | * |
14209 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14210 | | * compressed document is provided by default if found at compile-time. |
14211 | | * |
14212 | | * Returns the resulting document tree if the file was wellformed, |
14213 | | * NULL otherwise. |
14214 | | */ |
14215 | | |
14216 | | xmlDocPtr |
14217 | 0 | xmlParseFile(const char *filename) { |
14218 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
14219 | 0 | } |
14220 | | |
14221 | | /** |
14222 | | * xmlRecoverFile: |
14223 | | * @filename: the filename |
14224 | | * |
14225 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14226 | | * compressed document is provided by default if found at compile-time. |
14227 | | * In the case the document is not Well Formed, it attempts to build |
14228 | | * a tree anyway |
14229 | | * |
14230 | | * Returns the resulting document tree or NULL in case of failure |
14231 | | */ |
14232 | | |
14233 | | xmlDocPtr |
14234 | 0 | xmlRecoverFile(const char *filename) { |
14235 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
14236 | 0 | } |
14237 | | |
14238 | | |
14239 | | /** |
14240 | | * xmlSetupParserForBuffer: |
14241 | | * @ctxt: an XML parser context |
14242 | | * @buffer: a xmlChar * buffer |
14243 | | * @filename: a file name |
14244 | | * |
14245 | | * Setup the parser context to parse a new buffer; Clears any prior |
14246 | | * contents from the parser context. The buffer parameter must not be |
14247 | | * NULL, but the filename parameter can be |
14248 | | */ |
14249 | | void |
14250 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
14251 | | const char* filename) |
14252 | 0 | { |
14253 | 0 | xmlParserInputPtr input; |
14254 | |
|
14255 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
14256 | 0 | return; |
14257 | | |
14258 | 0 | input = xmlNewInputStream(ctxt); |
14259 | 0 | if (input == NULL) { |
14260 | 0 | xmlErrMemory(NULL, "parsing new buffer: out of memory\n"); |
14261 | 0 | xmlClearParserCtxt(ctxt); |
14262 | 0 | return; |
14263 | 0 | } |
14264 | | |
14265 | 0 | xmlClearParserCtxt(ctxt); |
14266 | 0 | if (filename != NULL) |
14267 | 0 | input->filename = (char *) xmlCanonicPath((const xmlChar *)filename); |
14268 | 0 | input->base = buffer; |
14269 | 0 | input->cur = buffer; |
14270 | 0 | input->end = &buffer[xmlStrlen(buffer)]; |
14271 | 0 | inputPush(ctxt, input); |
14272 | 0 | } |
14273 | | |
14274 | | /** |
14275 | | * xmlSAXUserParseFile: |
14276 | | * @sax: a SAX handler |
14277 | | * @user_data: The user data returned on SAX callbacks |
14278 | | * @filename: a file name |
14279 | | * |
14280 | | * parse an XML file and call the given SAX handler routines. |
14281 | | * Automatic support for ZLIB/Compress compressed document is provided |
14282 | | * |
14283 | | * Returns 0 in case of success or a error number otherwise |
14284 | | */ |
14285 | | int |
14286 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
14287 | 0 | const char *filename) { |
14288 | 0 | int ret = 0; |
14289 | 0 | xmlParserCtxtPtr ctxt; |
14290 | |
|
14291 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14292 | 0 | if (ctxt == NULL) return -1; |
14293 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14294 | 0 | xmlFree(ctxt->sax); |
14295 | 0 | ctxt->sax = sax; |
14296 | 0 | xmlDetectSAX2(ctxt); |
14297 | |
|
14298 | 0 | if (user_data != NULL) |
14299 | 0 | ctxt->userData = user_data; |
14300 | |
|
14301 | 0 | xmlParseDocument(ctxt); |
14302 | |
|
14303 | 0 | if (ctxt->wellFormed) |
14304 | 0 | ret = 0; |
14305 | 0 | else { |
14306 | 0 | if (ctxt->errNo != 0) |
14307 | 0 | ret = ctxt->errNo; |
14308 | 0 | else |
14309 | 0 | ret = -1; |
14310 | 0 | } |
14311 | 0 | if (sax != NULL) |
14312 | 0 | ctxt->sax = NULL; |
14313 | 0 | if (ctxt->myDoc != NULL) { |
14314 | 0 | xmlFreeDoc(ctxt->myDoc); |
14315 | 0 | ctxt->myDoc = NULL; |
14316 | 0 | } |
14317 | 0 | xmlFreeParserCtxt(ctxt); |
14318 | |
|
14319 | 0 | return ret; |
14320 | 0 | } |
14321 | | #endif /* LIBXML_SAX1_ENABLED */ |
14322 | | |
14323 | | /************************************************************************ |
14324 | | * * |
14325 | | * Front ends when parsing from memory * |
14326 | | * * |
14327 | | ************************************************************************/ |
14328 | | |
14329 | | /** |
14330 | | * xmlCreateMemoryParserCtxt: |
14331 | | * @buffer: a pointer to a char array |
14332 | | * @size: the size of the array |
14333 | | * |
14334 | | * Create a parser context for an XML in-memory document. |
14335 | | * |
14336 | | * Returns the new parser context or NULL |
14337 | | */ |
14338 | | xmlParserCtxtPtr |
14339 | 0 | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
14340 | 0 | xmlParserCtxtPtr ctxt; |
14341 | 0 | xmlParserInputPtr input; |
14342 | 0 | xmlParserInputBufferPtr buf; |
14343 | |
|
14344 | 0 | if (buffer == NULL) |
14345 | 0 | return(NULL); |
14346 | 0 | if (size <= 0) |
14347 | 0 | return(NULL); |
14348 | | |
14349 | 0 | ctxt = xmlNewParserCtxt(); |
14350 | 0 | if (ctxt == NULL) |
14351 | 0 | return(NULL); |
14352 | | |
14353 | | /* TODO: xmlParserInputBufferCreateStatic, requires some serious changes */ |
14354 | 0 | buf = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
14355 | 0 | if (buf == NULL) { |
14356 | 0 | xmlFreeParserCtxt(ctxt); |
14357 | 0 | return(NULL); |
14358 | 0 | } |
14359 | | |
14360 | 0 | input = xmlNewInputStream(ctxt); |
14361 | 0 | if (input == NULL) { |
14362 | 0 | xmlFreeParserInputBuffer(buf); |
14363 | 0 | xmlFreeParserCtxt(ctxt); |
14364 | 0 | return(NULL); |
14365 | 0 | } |
14366 | | |
14367 | 0 | input->filename = NULL; |
14368 | 0 | input->buf = buf; |
14369 | 0 | xmlBufResetInput(input->buf->buffer, input); |
14370 | |
|
14371 | 0 | inputPush(ctxt, input); |
14372 | 0 | return(ctxt); |
14373 | 0 | } |
14374 | | |
14375 | | #ifdef LIBXML_SAX1_ENABLED |
14376 | | /** |
14377 | | * xmlSAXParseMemoryWithData: |
14378 | | * @sax: the SAX handler block |
14379 | | * @buffer: an pointer to a char array |
14380 | | * @size: the size of the array |
14381 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14382 | | * documents |
14383 | | * @data: the userdata |
14384 | | * |
14385 | | * parse an XML in-memory block and use the given SAX function block |
14386 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14387 | | * DOM tree building routines. |
14388 | | * |
14389 | | * User data (void *) is stored within the parser context in the |
14390 | | * context's _private member, so it is available nearly everywhere in libxml |
14391 | | * |
14392 | | * Returns the resulting document tree |
14393 | | */ |
14394 | | |
14395 | | xmlDocPtr |
14396 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
14397 | 0 | int size, int recovery, void *data) { |
14398 | 0 | xmlDocPtr ret; |
14399 | 0 | xmlParserCtxtPtr ctxt; |
14400 | |
|
14401 | 0 | xmlInitParser(); |
14402 | |
|
14403 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14404 | 0 | if (ctxt == NULL) return(NULL); |
14405 | 0 | if (sax != NULL) { |
14406 | 0 | if (ctxt->sax != NULL) |
14407 | 0 | xmlFree(ctxt->sax); |
14408 | 0 | ctxt->sax = sax; |
14409 | 0 | } |
14410 | 0 | xmlDetectSAX2(ctxt); |
14411 | 0 | if (data!=NULL) { |
14412 | 0 | ctxt->_private=data; |
14413 | 0 | } |
14414 | |
|
14415 | 0 | ctxt->recovery = recovery; |
14416 | |
|
14417 | 0 | xmlParseDocument(ctxt); |
14418 | |
|
14419 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14420 | 0 | else { |
14421 | 0 | ret = NULL; |
14422 | 0 | xmlFreeDoc(ctxt->myDoc); |
14423 | 0 | ctxt->myDoc = NULL; |
14424 | 0 | } |
14425 | 0 | if (sax != NULL) |
14426 | 0 | ctxt->sax = NULL; |
14427 | 0 | xmlFreeParserCtxt(ctxt); |
14428 | |
|
14429 | 0 | return(ret); |
14430 | 0 | } |
14431 | | |
14432 | | /** |
14433 | | * xmlSAXParseMemory: |
14434 | | * @sax: the SAX handler block |
14435 | | * @buffer: an pointer to a char array |
14436 | | * @size: the size of the array |
14437 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
14438 | | * documents |
14439 | | * |
14440 | | * parse an XML in-memory block and use the given SAX function block |
14441 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14442 | | * DOM tree building routines. |
14443 | | * |
14444 | | * Returns the resulting document tree |
14445 | | */ |
14446 | | xmlDocPtr |
14447 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
14448 | 0 | int size, int recovery) { |
14449 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
14450 | 0 | } |
14451 | | |
14452 | | /** |
14453 | | * xmlParseMemory: |
14454 | | * @buffer: an pointer to a char array |
14455 | | * @size: the size of the array |
14456 | | * |
14457 | | * parse an XML in-memory block and build a tree. |
14458 | | * |
14459 | | * Returns the resulting document tree |
14460 | | */ |
14461 | | |
14462 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
14463 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
14464 | 0 | } |
14465 | | |
14466 | | /** |
14467 | | * xmlRecoverMemory: |
14468 | | * @buffer: an pointer to a char array |
14469 | | * @size: the size of the array |
14470 | | * |
14471 | | * parse an XML in-memory block and build a tree. |
14472 | | * In the case the document is not Well Formed, an attempt to |
14473 | | * build a tree is tried anyway |
14474 | | * |
14475 | | * Returns the resulting document tree or NULL in case of error |
14476 | | */ |
14477 | | |
14478 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
14479 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
14480 | 0 | } |
14481 | | |
14482 | | /** |
14483 | | * xmlSAXUserParseMemory: |
14484 | | * @sax: a SAX handler |
14485 | | * @user_data: The user data returned on SAX callbacks |
14486 | | * @buffer: an in-memory XML document input |
14487 | | * @size: the length of the XML document in bytes |
14488 | | * |
14489 | | * A better SAX parsing routine. |
14490 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
14491 | | * |
14492 | | * Returns 0 in case of success or a error number otherwise |
14493 | | */ |
14494 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
14495 | 0 | const char *buffer, int size) { |
14496 | 0 | int ret = 0; |
14497 | 0 | xmlParserCtxtPtr ctxt; |
14498 | |
|
14499 | 0 | xmlInitParser(); |
14500 | |
|
14501 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14502 | 0 | if (ctxt == NULL) return -1; |
14503 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14504 | 0 | xmlFree(ctxt->sax); |
14505 | 0 | ctxt->sax = sax; |
14506 | 0 | xmlDetectSAX2(ctxt); |
14507 | |
|
14508 | 0 | if (user_data != NULL) |
14509 | 0 | ctxt->userData = user_data; |
14510 | |
|
14511 | 0 | xmlParseDocument(ctxt); |
14512 | |
|
14513 | 0 | if (ctxt->wellFormed) |
14514 | 0 | ret = 0; |
14515 | 0 | else { |
14516 | 0 | if (ctxt->errNo != 0) |
14517 | 0 | ret = ctxt->errNo; |
14518 | 0 | else |
14519 | 0 | ret = -1; |
14520 | 0 | } |
14521 | 0 | if (sax != NULL) |
14522 | 0 | ctxt->sax = NULL; |
14523 | 0 | if (ctxt->myDoc != NULL) { |
14524 | 0 | xmlFreeDoc(ctxt->myDoc); |
14525 | 0 | ctxt->myDoc = NULL; |
14526 | 0 | } |
14527 | 0 | xmlFreeParserCtxt(ctxt); |
14528 | |
|
14529 | 0 | return ret; |
14530 | 0 | } |
14531 | | #endif /* LIBXML_SAX1_ENABLED */ |
14532 | | |
14533 | | /** |
14534 | | * xmlCreateDocParserCtxt: |
14535 | | * @cur: a pointer to an array of xmlChar |
14536 | | * |
14537 | | * Creates a parser context for an XML in-memory document. |
14538 | | * |
14539 | | * Returns the new parser context or NULL |
14540 | | */ |
14541 | | xmlParserCtxtPtr |
14542 | 0 | xmlCreateDocParserCtxt(const xmlChar *cur) { |
14543 | 0 | int len; |
14544 | |
|
14545 | 0 | if (cur == NULL) |
14546 | 0 | return(NULL); |
14547 | 0 | len = xmlStrlen(cur); |
14548 | 0 | return(xmlCreateMemoryParserCtxt((const char *)cur, len)); |
14549 | 0 | } |
14550 | | |
14551 | | #ifdef LIBXML_SAX1_ENABLED |
14552 | | /** |
14553 | | * xmlSAXParseDoc: |
14554 | | * @sax: the SAX handler block |
14555 | | * @cur: a pointer to an array of xmlChar |
14556 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14557 | | * documents |
14558 | | * |
14559 | | * parse an XML in-memory document and build a tree. |
14560 | | * It use the given SAX function block to handle the parsing callback. |
14561 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14562 | | * |
14563 | | * Returns the resulting document tree |
14564 | | */ |
14565 | | |
14566 | | xmlDocPtr |
14567 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
14568 | 0 | xmlDocPtr ret; |
14569 | 0 | xmlParserCtxtPtr ctxt; |
14570 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
14571 | |
|
14572 | 0 | if (cur == NULL) return(NULL); |
14573 | | |
14574 | | |
14575 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
14576 | 0 | if (ctxt == NULL) return(NULL); |
14577 | 0 | if (sax != NULL) { |
14578 | 0 | oldsax = ctxt->sax; |
14579 | 0 | ctxt->sax = sax; |
14580 | 0 | ctxt->userData = NULL; |
14581 | 0 | } |
14582 | 0 | xmlDetectSAX2(ctxt); |
14583 | |
|
14584 | 0 | xmlParseDocument(ctxt); |
14585 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14586 | 0 | else { |
14587 | 0 | ret = NULL; |
14588 | 0 | xmlFreeDoc(ctxt->myDoc); |
14589 | 0 | ctxt->myDoc = NULL; |
14590 | 0 | } |
14591 | 0 | if (sax != NULL) |
14592 | 0 | ctxt->sax = oldsax; |
14593 | 0 | xmlFreeParserCtxt(ctxt); |
14594 | |
|
14595 | 0 | return(ret); |
14596 | 0 | } |
14597 | | |
14598 | | /** |
14599 | | * xmlParseDoc: |
14600 | | * @cur: a pointer to an array of xmlChar |
14601 | | * |
14602 | | * parse an XML in-memory document and build a tree. |
14603 | | * |
14604 | | * Returns the resulting document tree |
14605 | | */ |
14606 | | |
14607 | | xmlDocPtr |
14608 | 0 | xmlParseDoc(const xmlChar *cur) { |
14609 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
14610 | 0 | } |
14611 | | #endif /* LIBXML_SAX1_ENABLED */ |
14612 | | |
14613 | | #ifdef LIBXML_LEGACY_ENABLED |
14614 | | /************************************************************************ |
14615 | | * * |
14616 | | * Specific function to keep track of entities references * |
14617 | | * and used by the XSLT debugger * |
14618 | | * * |
14619 | | ************************************************************************/ |
14620 | | |
14621 | | static xmlEntityReferenceFunc xmlEntityRefFunc = NULL; |
14622 | | |
14623 | | /** |
14624 | | * xmlAddEntityReference: |
14625 | | * @ent : A valid entity |
14626 | | * @firstNode : A valid first node for children of entity |
14627 | | * @lastNode : A valid last node of children entity |
14628 | | * |
14629 | | * Notify of a reference to an entity of type XML_EXTERNAL_GENERAL_PARSED_ENTITY |
14630 | | */ |
14631 | | static void |
14632 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
14633 | | xmlNodePtr lastNode) |
14634 | | { |
14635 | | if (xmlEntityRefFunc != NULL) { |
14636 | | (*xmlEntityRefFunc) (ent, firstNode, lastNode); |
14637 | | } |
14638 | | } |
14639 | | |
14640 | | |
14641 | | /** |
14642 | | * xmlSetEntityReferenceFunc: |
14643 | | * @func: A valid function |
14644 | | * |
14645 | | * Set the function to call call back when a xml reference has been made |
14646 | | */ |
14647 | | void |
14648 | | xmlSetEntityReferenceFunc(xmlEntityReferenceFunc func) |
14649 | | { |
14650 | | xmlEntityRefFunc = func; |
14651 | | } |
14652 | | #endif /* LIBXML_LEGACY_ENABLED */ |
14653 | | |
14654 | | /************************************************************************ |
14655 | | * * |
14656 | | * Miscellaneous * |
14657 | | * * |
14658 | | ************************************************************************/ |
14659 | | |
14660 | | #ifdef LIBXML_XPATH_ENABLED |
14661 | | #include <libxml/xpath.h> |
14662 | | #endif |
14663 | | |
14664 | | extern void XMLCDECL xmlGenericErrorDefaultFunc(void *ctx, const char *msg, ...); |
14665 | | static int xmlParserInitialized = 0; |
14666 | | |
14667 | | /** |
14668 | | * xmlInitParser: |
14669 | | * |
14670 | | * Initialization function for the XML parser. |
14671 | | * This is not reentrant. Call once before processing in case of |
14672 | | * use in multithreaded programs. |
14673 | | */ |
14674 | | |
14675 | | void |
14676 | 36.6k | xmlInitParser(void) { |
14677 | 36.6k | if (xmlParserInitialized != 0) |
14678 | 36.6k | return; |
14679 | | |
14680 | | #if defined(_WIN32) && (!defined(LIBXML_STATIC) || defined(LIBXML_STATIC_FOR_DLL)) |
14681 | | if (xmlFree == free) |
14682 | | atexit(xmlCleanupParser); |
14683 | | #endif |
14684 | | |
14685 | 1 | #ifdef LIBXML_THREAD_ENABLED |
14686 | 1 | __xmlGlobalInitMutexLock(); |
14687 | 1 | if (xmlParserInitialized == 0) { |
14688 | 1 | #endif |
14689 | 1 | xmlInitThreads(); |
14690 | 1 | xmlInitGlobals(); |
14691 | 1 | if ((xmlGenericError == xmlGenericErrorDefaultFunc) || |
14692 | 1 | (xmlGenericError == NULL)) |
14693 | 1 | initGenericErrorDefaultFunc(NULL); |
14694 | 1 | xmlInitMemory(); |
14695 | 1 | xmlInitializeDict(); |
14696 | 1 | xmlInitCharEncodingHandlers(); |
14697 | 1 | xmlDefaultSAXHandlerInit(); |
14698 | 1 | xmlRegisterDefaultInputCallbacks(); |
14699 | 1 | #ifdef LIBXML_OUTPUT_ENABLED |
14700 | 1 | xmlRegisterDefaultOutputCallbacks(); |
14701 | 1 | #endif /* LIBXML_OUTPUT_ENABLED */ |
14702 | 1 | #ifdef LIBXML_HTML_ENABLED |
14703 | 1 | htmlInitAutoClose(); |
14704 | 1 | htmlDefaultSAXHandlerInit(); |
14705 | 1 | #endif |
14706 | 1 | #ifdef LIBXML_XPATH_ENABLED |
14707 | 1 | xmlXPathInit(); |
14708 | 1 | #endif |
14709 | 1 | xmlParserInitialized = 1; |
14710 | 1 | #ifdef LIBXML_THREAD_ENABLED |
14711 | 1 | } |
14712 | 1 | __xmlGlobalInitMutexUnlock(); |
14713 | 1 | #endif |
14714 | 1 | } |
14715 | | |
14716 | | /** |
14717 | | * xmlCleanupParser: |
14718 | | * |
14719 | | * This function name is somewhat misleading. It does not clean up |
14720 | | * parser state, it cleans up memory allocated by the library itself. |
14721 | | * It is a cleanup function for the XML library. It tries to reclaim all |
14722 | | * related global memory allocated for the library processing. |
14723 | | * It doesn't deallocate any document related memory. One should |
14724 | | * call xmlCleanupParser() only when the process has finished using |
14725 | | * the library and all XML/HTML documents built with it. |
14726 | | * See also xmlInitParser() which has the opposite function of preparing |
14727 | | * the library for operations. |
14728 | | * |
14729 | | * WARNING: if your application is multithreaded or has plugin support |
14730 | | * calling this may crash the application if another thread or |
14731 | | * a plugin is still using libxml2. It's sometimes very hard to |
14732 | | * guess if libxml2 is in use in the application, some libraries |
14733 | | * or plugins may use it without notice. In case of doubt abstain |
14734 | | * from calling this function or do it just before calling exit() |
14735 | | * to avoid leak reports from valgrind ! |
14736 | | */ |
14737 | | |
14738 | | void |
14739 | 0 | xmlCleanupParser(void) { |
14740 | 0 | if (!xmlParserInitialized) |
14741 | 0 | return; |
14742 | | |
14743 | 0 | xmlCleanupCharEncodingHandlers(); |
14744 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14745 | 0 | xmlCatalogCleanup(); |
14746 | 0 | #endif |
14747 | 0 | xmlDictCleanup(); |
14748 | 0 | xmlCleanupInputCallbacks(); |
14749 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
14750 | 0 | xmlCleanupOutputCallbacks(); |
14751 | 0 | #endif |
14752 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
14753 | 0 | xmlSchemaCleanupTypes(); |
14754 | 0 | xmlRelaxNGCleanupTypes(); |
14755 | 0 | #endif |
14756 | 0 | xmlCleanupGlobals(); |
14757 | 0 | xmlCleanupThreads(); /* must be last if called not from the main thread */ |
14758 | 0 | xmlCleanupMemory(); |
14759 | 0 | xmlParserInitialized = 0; |
14760 | 0 | } |
14761 | | |
14762 | | #if defined(HAVE_ATTRIBUTE_DESTRUCTOR) && !defined(LIBXML_STATIC) && \ |
14763 | | !defined(_WIN32) |
14764 | | static void |
14765 | | ATTRIBUTE_DESTRUCTOR |
14766 | | xmlDestructor(void) { |
14767 | | /* |
14768 | | * Calling custom deallocation functions in a destructor can cause |
14769 | | * problems, for example with Nokogiri. |
14770 | | */ |
14771 | | if (xmlFree == free) |
14772 | | xmlCleanupParser(); |
14773 | | } |
14774 | | #endif |
14775 | | |
14776 | | /************************************************************************ |
14777 | | * * |
14778 | | * New set (2.6.0) of simpler and more flexible APIs * |
14779 | | * * |
14780 | | ************************************************************************/ |
14781 | | |
14782 | | /** |
14783 | | * DICT_FREE: |
14784 | | * @str: a string |
14785 | | * |
14786 | | * Free a string if it is not owned by the "dict" dictionary in the |
14787 | | * current scope |
14788 | | */ |
14789 | | #define DICT_FREE(str) \ |
14790 | 0 | if ((str) && ((!dict) || \ |
14791 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
14792 | 0 | xmlFree((char *)(str)); |
14793 | | |
14794 | | /** |
14795 | | * xmlCtxtReset: |
14796 | | * @ctxt: an XML parser context |
14797 | | * |
14798 | | * Reset a parser context |
14799 | | */ |
14800 | | void |
14801 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
14802 | 0 | { |
14803 | 0 | xmlParserInputPtr input; |
14804 | 0 | xmlDictPtr dict; |
14805 | |
|
14806 | 0 | if (ctxt == NULL) |
14807 | 0 | return; |
14808 | | |
14809 | 0 | dict = ctxt->dict; |
14810 | |
|
14811 | 0 | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
14812 | 0 | xmlFreeInputStream(input); |
14813 | 0 | } |
14814 | 0 | ctxt->inputNr = 0; |
14815 | 0 | ctxt->input = NULL; |
14816 | |
|
14817 | 0 | ctxt->spaceNr = 0; |
14818 | 0 | if (ctxt->spaceTab != NULL) { |
14819 | 0 | ctxt->spaceTab[0] = -1; |
14820 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
14821 | 0 | } else { |
14822 | 0 | ctxt->space = NULL; |
14823 | 0 | } |
14824 | | |
14825 | |
|
14826 | 0 | ctxt->nodeNr = 0; |
14827 | 0 | ctxt->node = NULL; |
14828 | |
|
14829 | 0 | ctxt->nameNr = 0; |
14830 | 0 | ctxt->name = NULL; |
14831 | |
|
14832 | 0 | ctxt->nsNr = 0; |
14833 | |
|
14834 | 0 | DICT_FREE(ctxt->version); |
14835 | 0 | ctxt->version = NULL; |
14836 | 0 | DICT_FREE(ctxt->encoding); |
14837 | 0 | ctxt->encoding = NULL; |
14838 | 0 | DICT_FREE(ctxt->directory); |
14839 | 0 | ctxt->directory = NULL; |
14840 | 0 | DICT_FREE(ctxt->extSubURI); |
14841 | 0 | ctxt->extSubURI = NULL; |
14842 | 0 | DICT_FREE(ctxt->extSubSystem); |
14843 | 0 | ctxt->extSubSystem = NULL; |
14844 | 0 | if (ctxt->myDoc != NULL) |
14845 | 0 | xmlFreeDoc(ctxt->myDoc); |
14846 | 0 | ctxt->myDoc = NULL; |
14847 | |
|
14848 | 0 | ctxt->standalone = -1; |
14849 | 0 | ctxt->hasExternalSubset = 0; |
14850 | 0 | ctxt->hasPErefs = 0; |
14851 | 0 | ctxt->html = 0; |
14852 | 0 | ctxt->external = 0; |
14853 | 0 | ctxt->instate = XML_PARSER_START; |
14854 | 0 | ctxt->token = 0; |
14855 | |
|
14856 | 0 | ctxt->wellFormed = 1; |
14857 | 0 | ctxt->nsWellFormed = 1; |
14858 | 0 | ctxt->disableSAX = 0; |
14859 | 0 | ctxt->valid = 1; |
14860 | | #if 0 |
14861 | | ctxt->vctxt.userData = ctxt; |
14862 | | ctxt->vctxt.error = xmlParserValidityError; |
14863 | | ctxt->vctxt.warning = xmlParserValidityWarning; |
14864 | | #endif |
14865 | 0 | ctxt->record_info = 0; |
14866 | 0 | ctxt->checkIndex = 0; |
14867 | 0 | ctxt->inSubset = 0; |
14868 | 0 | ctxt->errNo = XML_ERR_OK; |
14869 | 0 | ctxt->depth = 0; |
14870 | 0 | ctxt->charset = XML_CHAR_ENCODING_UTF8; |
14871 | 0 | ctxt->catalogs = NULL; |
14872 | 0 | ctxt->nbentities = 0; |
14873 | 0 | ctxt->sizeentities = 0; |
14874 | 0 | ctxt->sizeentcopy = 0; |
14875 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
14876 | |
|
14877 | 0 | if (ctxt->attsDefault != NULL) { |
14878 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
14879 | 0 | ctxt->attsDefault = NULL; |
14880 | 0 | } |
14881 | 0 | if (ctxt->attsSpecial != NULL) { |
14882 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
14883 | 0 | ctxt->attsSpecial = NULL; |
14884 | 0 | } |
14885 | |
|
14886 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14887 | 0 | if (ctxt->catalogs != NULL) |
14888 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
14889 | 0 | #endif |
14890 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
14891 | 0 | xmlResetError(&ctxt->lastError); |
14892 | 0 | } |
14893 | | |
14894 | | /** |
14895 | | * xmlCtxtResetPush: |
14896 | | * @ctxt: an XML parser context |
14897 | | * @chunk: a pointer to an array of chars |
14898 | | * @size: number of chars in the array |
14899 | | * @filename: an optional file name or URI |
14900 | | * @encoding: the document encoding, or NULL |
14901 | | * |
14902 | | * Reset a push parser context |
14903 | | * |
14904 | | * Returns 0 in case of success and 1 in case of error |
14905 | | */ |
14906 | | int |
14907 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
14908 | | int size, const char *filename, const char *encoding) |
14909 | 0 | { |
14910 | 0 | xmlParserInputPtr inputStream; |
14911 | 0 | xmlParserInputBufferPtr buf; |
14912 | 0 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
14913 | |
|
14914 | 0 | if (ctxt == NULL) |
14915 | 0 | return(1); |
14916 | | |
14917 | 0 | if ((encoding == NULL) && (chunk != NULL) && (size >= 4)) |
14918 | 0 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
14919 | |
|
14920 | 0 | buf = xmlAllocParserInputBuffer(enc); |
14921 | 0 | if (buf == NULL) |
14922 | 0 | return(1); |
14923 | | |
14924 | 0 | if (ctxt == NULL) { |
14925 | 0 | xmlFreeParserInputBuffer(buf); |
14926 | 0 | return(1); |
14927 | 0 | } |
14928 | | |
14929 | 0 | xmlCtxtReset(ctxt); |
14930 | |
|
14931 | 0 | if (filename == NULL) { |
14932 | 0 | ctxt->directory = NULL; |
14933 | 0 | } else { |
14934 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14935 | 0 | } |
14936 | |
|
14937 | 0 | inputStream = xmlNewInputStream(ctxt); |
14938 | 0 | if (inputStream == NULL) { |
14939 | 0 | xmlFreeParserInputBuffer(buf); |
14940 | 0 | return(1); |
14941 | 0 | } |
14942 | | |
14943 | 0 | if (filename == NULL) |
14944 | 0 | inputStream->filename = NULL; |
14945 | 0 | else |
14946 | 0 | inputStream->filename = (char *) |
14947 | 0 | xmlCanonicPath((const xmlChar *) filename); |
14948 | 0 | inputStream->buf = buf; |
14949 | 0 | xmlBufResetInput(buf->buffer, inputStream); |
14950 | |
|
14951 | 0 | inputPush(ctxt, inputStream); |
14952 | |
|
14953 | 0 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
14954 | 0 | (ctxt->input->buf != NULL)) { |
14955 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
14956 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
14957 | |
|
14958 | 0 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
14959 | |
|
14960 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
14961 | | #ifdef DEBUG_PUSH |
14962 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
14963 | | #endif |
14964 | 0 | } |
14965 | |
|
14966 | 0 | if (encoding != NULL) { |
14967 | 0 | xmlCharEncodingHandlerPtr hdlr; |
14968 | |
|
14969 | 0 | if (ctxt->encoding != NULL) |
14970 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
14971 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
14972 | |
|
14973 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
14974 | 0 | if (hdlr != NULL) { |
14975 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
14976 | 0 | } else { |
14977 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
14978 | 0 | "Unsupported encoding %s\n", BAD_CAST encoding); |
14979 | 0 | } |
14980 | 0 | } else if (enc != XML_CHAR_ENCODING_NONE) { |
14981 | 0 | xmlSwitchEncoding(ctxt, enc); |
14982 | 0 | } |
14983 | |
|
14984 | 0 | return(0); |
14985 | 0 | } |
14986 | | |
14987 | | |
14988 | | /** |
14989 | | * xmlCtxtUseOptionsInternal: |
14990 | | * @ctxt: an XML parser context |
14991 | | * @options: a combination of xmlParserOption |
14992 | | * @encoding: the user provided encoding to use |
14993 | | * |
14994 | | * Applies the options to the parser context |
14995 | | * |
14996 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
14997 | | * in case of error. |
14998 | | */ |
14999 | | static int |
15000 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, const char *encoding) |
15001 | 18.3k | { |
15002 | 18.3k | if (ctxt == NULL) |
15003 | 0 | return(-1); |
15004 | 18.3k | if (encoding != NULL) { |
15005 | 0 | if (ctxt->encoding != NULL) |
15006 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
15007 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
15008 | 0 | } |
15009 | 18.3k | if (options & XML_PARSE_RECOVER) { |
15010 | 0 | ctxt->recovery = 1; |
15011 | 0 | options -= XML_PARSE_RECOVER; |
15012 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
15013 | 0 | } else |
15014 | 18.3k | ctxt->recovery = 0; |
15015 | 18.3k | if (options & XML_PARSE_DTDLOAD) { |
15016 | 0 | ctxt->loadsubset = XML_DETECT_IDS; |
15017 | 0 | options -= XML_PARSE_DTDLOAD; |
15018 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
15019 | 0 | } else |
15020 | 18.3k | ctxt->loadsubset = 0; |
15021 | 18.3k | if (options & XML_PARSE_DTDATTR) { |
15022 | 0 | ctxt->loadsubset |= XML_COMPLETE_ATTRS; |
15023 | 0 | options -= XML_PARSE_DTDATTR; |
15024 | 0 | ctxt->options |= XML_PARSE_DTDATTR; |
15025 | 0 | } |
15026 | 18.3k | if (options & XML_PARSE_NOENT) { |
15027 | 0 | ctxt->replaceEntities = 1; |
15028 | | /* ctxt->loadsubset |= XML_DETECT_IDS; */ |
15029 | 0 | options -= XML_PARSE_NOENT; |
15030 | 0 | ctxt->options |= XML_PARSE_NOENT; |
15031 | 0 | } else |
15032 | 18.3k | ctxt->replaceEntities = 0; |
15033 | 18.3k | if (options & XML_PARSE_PEDANTIC) { |
15034 | 0 | ctxt->pedantic = 1; |
15035 | 0 | options -= XML_PARSE_PEDANTIC; |
15036 | 0 | ctxt->options |= XML_PARSE_PEDANTIC; |
15037 | 0 | } else |
15038 | 18.3k | ctxt->pedantic = 0; |
15039 | 18.3k | if (options & XML_PARSE_NOBLANKS) { |
15040 | 18.3k | ctxt->keepBlanks = 0; |
15041 | 18.3k | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
15042 | 18.3k | options -= XML_PARSE_NOBLANKS; |
15043 | 18.3k | ctxt->options |= XML_PARSE_NOBLANKS; |
15044 | 18.3k | } else |
15045 | 0 | ctxt->keepBlanks = 1; |
15046 | 18.3k | if (options & XML_PARSE_DTDVALID) { |
15047 | 0 | ctxt->validate = 1; |
15048 | 0 | if (options & XML_PARSE_NOWARNING) |
15049 | 0 | ctxt->vctxt.warning = NULL; |
15050 | 0 | if (options & XML_PARSE_NOERROR) |
15051 | 0 | ctxt->vctxt.error = NULL; |
15052 | 0 | options -= XML_PARSE_DTDVALID; |
15053 | 0 | ctxt->options |= XML_PARSE_DTDVALID; |
15054 | 0 | } else |
15055 | 18.3k | ctxt->validate = 0; |
15056 | 18.3k | if (options & XML_PARSE_NOWARNING) { |
15057 | 0 | ctxt->sax->warning = NULL; |
15058 | 0 | options -= XML_PARSE_NOWARNING; |
15059 | 0 | } |
15060 | 18.3k | if (options & XML_PARSE_NOERROR) { |
15061 | 0 | ctxt->sax->error = NULL; |
15062 | 0 | ctxt->sax->fatalError = NULL; |
15063 | 0 | options -= XML_PARSE_NOERROR; |
15064 | 0 | } |
15065 | 18.3k | #ifdef LIBXML_SAX1_ENABLED |
15066 | 18.3k | if (options & XML_PARSE_SAX1) { |
15067 | 0 | ctxt->sax->startElement = xmlSAX2StartElement; |
15068 | 0 | ctxt->sax->endElement = xmlSAX2EndElement; |
15069 | 0 | ctxt->sax->startElementNs = NULL; |
15070 | 0 | ctxt->sax->endElementNs = NULL; |
15071 | 0 | ctxt->sax->initialized = 1; |
15072 | 0 | options -= XML_PARSE_SAX1; |
15073 | 0 | ctxt->options |= XML_PARSE_SAX1; |
15074 | 0 | } |
15075 | 18.3k | #endif /* LIBXML_SAX1_ENABLED */ |
15076 | 18.3k | if (options & XML_PARSE_NODICT) { |
15077 | 0 | ctxt->dictNames = 0; |
15078 | 0 | options -= XML_PARSE_NODICT; |
15079 | 0 | ctxt->options |= XML_PARSE_NODICT; |
15080 | 18.3k | } else { |
15081 | 18.3k | ctxt->dictNames = 1; |
15082 | 18.3k | } |
15083 | 18.3k | if (options & XML_PARSE_NOCDATA) { |
15084 | 18.3k | ctxt->sax->cdataBlock = NULL; |
15085 | 18.3k | options -= XML_PARSE_NOCDATA; |
15086 | 18.3k | ctxt->options |= XML_PARSE_NOCDATA; |
15087 | 18.3k | } |
15088 | 18.3k | if (options & XML_PARSE_NSCLEAN) { |
15089 | 18.3k | ctxt->options |= XML_PARSE_NSCLEAN; |
15090 | 18.3k | options -= XML_PARSE_NSCLEAN; |
15091 | 18.3k | } |
15092 | 18.3k | if (options & XML_PARSE_NONET) { |
15093 | 18.3k | ctxt->options |= XML_PARSE_NONET; |
15094 | 18.3k | options -= XML_PARSE_NONET; |
15095 | 18.3k | } |
15096 | 18.3k | if (options & XML_PARSE_COMPACT) { |
15097 | 0 | ctxt->options |= XML_PARSE_COMPACT; |
15098 | 0 | options -= XML_PARSE_COMPACT; |
15099 | 0 | } |
15100 | 18.3k | if (options & XML_PARSE_OLD10) { |
15101 | 0 | ctxt->options |= XML_PARSE_OLD10; |
15102 | 0 | options -= XML_PARSE_OLD10; |
15103 | 0 | } |
15104 | 18.3k | if (options & XML_PARSE_NOBASEFIX) { |
15105 | 0 | ctxt->options |= XML_PARSE_NOBASEFIX; |
15106 | 0 | options -= XML_PARSE_NOBASEFIX; |
15107 | 0 | } |
15108 | 18.3k | if (options & XML_PARSE_HUGE) { |
15109 | 18.3k | ctxt->options |= XML_PARSE_HUGE; |
15110 | 18.3k | options -= XML_PARSE_HUGE; |
15111 | 18.3k | if (ctxt->dict != NULL) |
15112 | 18.3k | xmlDictSetLimit(ctxt->dict, 0); |
15113 | 18.3k | } |
15114 | 18.3k | if (options & XML_PARSE_OLDSAX) { |
15115 | 0 | ctxt->options |= XML_PARSE_OLDSAX; |
15116 | 0 | options -= XML_PARSE_OLDSAX; |
15117 | 0 | } |
15118 | 18.3k | if (options & XML_PARSE_IGNORE_ENC) { |
15119 | 0 | ctxt->options |= XML_PARSE_IGNORE_ENC; |
15120 | 0 | options -= XML_PARSE_IGNORE_ENC; |
15121 | 0 | } |
15122 | 18.3k | if (options & XML_PARSE_BIG_LINES) { |
15123 | 0 | ctxt->options |= XML_PARSE_BIG_LINES; |
15124 | 0 | options -= XML_PARSE_BIG_LINES; |
15125 | 0 | } |
15126 | 18.3k | ctxt->linenumbers = 1; |
15127 | 18.3k | return (options); |
15128 | 18.3k | } |
15129 | | |
15130 | | /** |
15131 | | * xmlCtxtUseOptions: |
15132 | | * @ctxt: an XML parser context |
15133 | | * @options: a combination of xmlParserOption |
15134 | | * |
15135 | | * Applies the options to the parser context |
15136 | | * |
15137 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
15138 | | * in case of error. |
15139 | | */ |
15140 | | int |
15141 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
15142 | 18.3k | { |
15143 | 18.3k | return(xmlCtxtUseOptionsInternal(ctxt, options, NULL)); |
15144 | 18.3k | } |
15145 | | |
15146 | | /** |
15147 | | * xmlDoRead: |
15148 | | * @ctxt: an XML parser context |
15149 | | * @URL: the base URL to use for the document |
15150 | | * @encoding: the document encoding, or NULL |
15151 | | * @options: a combination of xmlParserOption |
15152 | | * @reuse: keep the context for reuse |
15153 | | * |
15154 | | * Common front-end for the xmlRead functions |
15155 | | * |
15156 | | * Returns the resulting document tree or NULL |
15157 | | */ |
15158 | | static xmlDocPtr |
15159 | | xmlDoRead(xmlParserCtxtPtr ctxt, const char *URL, const char *encoding, |
15160 | | int options, int reuse) |
15161 | 0 | { |
15162 | 0 | xmlDocPtr ret; |
15163 | |
|
15164 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, encoding); |
15165 | 0 | if (encoding != NULL) { |
15166 | 0 | xmlCharEncodingHandlerPtr hdlr; |
15167 | |
|
15168 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
15169 | 0 | if (hdlr != NULL) |
15170 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
15171 | 0 | } |
15172 | 0 | if ((URL != NULL) && (ctxt->input != NULL) && |
15173 | 0 | (ctxt->input->filename == NULL)) |
15174 | 0 | ctxt->input->filename = (char *) xmlStrdup((const xmlChar *) URL); |
15175 | 0 | xmlParseDocument(ctxt); |
15176 | 0 | if ((ctxt->wellFormed) || ctxt->recovery) |
15177 | 0 | ret = ctxt->myDoc; |
15178 | 0 | else { |
15179 | 0 | ret = NULL; |
15180 | 0 | if (ctxt->myDoc != NULL) { |
15181 | 0 | xmlFreeDoc(ctxt->myDoc); |
15182 | 0 | } |
15183 | 0 | } |
15184 | 0 | ctxt->myDoc = NULL; |
15185 | 0 | if (!reuse) { |
15186 | 0 | xmlFreeParserCtxt(ctxt); |
15187 | 0 | } |
15188 | |
|
15189 | 0 | return (ret); |
15190 | 0 | } |
15191 | | |
15192 | | /** |
15193 | | * xmlReadDoc: |
15194 | | * @cur: a pointer to a zero terminated string |
15195 | | * @URL: the base URL to use for the document |
15196 | | * @encoding: the document encoding, or NULL |
15197 | | * @options: a combination of xmlParserOption |
15198 | | * |
15199 | | * parse an XML in-memory document and build a tree. |
15200 | | * |
15201 | | * Returns the resulting document tree |
15202 | | */ |
15203 | | xmlDocPtr |
15204 | | xmlReadDoc(const xmlChar * cur, const char *URL, const char *encoding, int options) |
15205 | 0 | { |
15206 | 0 | xmlParserCtxtPtr ctxt; |
15207 | |
|
15208 | 0 | if (cur == NULL) |
15209 | 0 | return (NULL); |
15210 | 0 | xmlInitParser(); |
15211 | |
|
15212 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
15213 | 0 | if (ctxt == NULL) |
15214 | 0 | return (NULL); |
15215 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15216 | 0 | } |
15217 | | |
15218 | | /** |
15219 | | * xmlReadFile: |
15220 | | * @filename: a file or URL |
15221 | | * @encoding: the document encoding, or NULL |
15222 | | * @options: a combination of xmlParserOption |
15223 | | * |
15224 | | * parse an XML file from the filesystem or the network. |
15225 | | * |
15226 | | * Returns the resulting document tree |
15227 | | */ |
15228 | | xmlDocPtr |
15229 | | xmlReadFile(const char *filename, const char *encoding, int options) |
15230 | 0 | { |
15231 | 0 | xmlParserCtxtPtr ctxt; |
15232 | |
|
15233 | 0 | xmlInitParser(); |
15234 | 0 | ctxt = xmlCreateURLParserCtxt(filename, options); |
15235 | 0 | if (ctxt == NULL) |
15236 | 0 | return (NULL); |
15237 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 0)); |
15238 | 0 | } |
15239 | | |
15240 | | /** |
15241 | | * xmlReadMemory: |
15242 | | * @buffer: a pointer to a char array |
15243 | | * @size: the size of the array |
15244 | | * @URL: the base URL to use for the document |
15245 | | * @encoding: the document encoding, or NULL |
15246 | | * @options: a combination of xmlParserOption |
15247 | | * |
15248 | | * parse an XML in-memory document and build a tree. |
15249 | | * |
15250 | | * Returns the resulting document tree |
15251 | | */ |
15252 | | xmlDocPtr |
15253 | | xmlReadMemory(const char *buffer, int size, const char *URL, const char *encoding, int options) |
15254 | 0 | { |
15255 | 0 | xmlParserCtxtPtr ctxt; |
15256 | |
|
15257 | 0 | xmlInitParser(); |
15258 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
15259 | 0 | if (ctxt == NULL) |
15260 | 0 | return (NULL); |
15261 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15262 | 0 | } |
15263 | | |
15264 | | /** |
15265 | | * xmlReadFd: |
15266 | | * @fd: an open file descriptor |
15267 | | * @URL: the base URL to use for the document |
15268 | | * @encoding: the document encoding, or NULL |
15269 | | * @options: a combination of xmlParserOption |
15270 | | * |
15271 | | * parse an XML from a file descriptor and build a tree. |
15272 | | * NOTE that the file descriptor will not be closed when the |
15273 | | * reader is closed or reset. |
15274 | | * |
15275 | | * Returns the resulting document tree |
15276 | | */ |
15277 | | xmlDocPtr |
15278 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
15279 | 0 | { |
15280 | 0 | xmlParserCtxtPtr ctxt; |
15281 | 0 | xmlParserInputBufferPtr input; |
15282 | 0 | xmlParserInputPtr stream; |
15283 | |
|
15284 | 0 | if (fd < 0) |
15285 | 0 | return (NULL); |
15286 | 0 | xmlInitParser(); |
15287 | |
|
15288 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15289 | 0 | if (input == NULL) |
15290 | 0 | return (NULL); |
15291 | 0 | input->closecallback = NULL; |
15292 | 0 | ctxt = xmlNewParserCtxt(); |
15293 | 0 | if (ctxt == NULL) { |
15294 | 0 | xmlFreeParserInputBuffer(input); |
15295 | 0 | return (NULL); |
15296 | 0 | } |
15297 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15298 | 0 | if (stream == NULL) { |
15299 | 0 | xmlFreeParserInputBuffer(input); |
15300 | 0 | xmlFreeParserCtxt(ctxt); |
15301 | 0 | return (NULL); |
15302 | 0 | } |
15303 | 0 | inputPush(ctxt, stream); |
15304 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15305 | 0 | } |
15306 | | |
15307 | | /** |
15308 | | * xmlReadIO: |
15309 | | * @ioread: an I/O read function |
15310 | | * @ioclose: an I/O close function |
15311 | | * @ioctx: an I/O handler |
15312 | | * @URL: the base URL to use for the document |
15313 | | * @encoding: the document encoding, or NULL |
15314 | | * @options: a combination of xmlParserOption |
15315 | | * |
15316 | | * parse an XML document from I/O functions and source and build a tree. |
15317 | | * |
15318 | | * Returns the resulting document tree |
15319 | | */ |
15320 | | xmlDocPtr |
15321 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
15322 | | void *ioctx, const char *URL, const char *encoding, int options) |
15323 | 0 | { |
15324 | 0 | xmlParserCtxtPtr ctxt; |
15325 | 0 | xmlParserInputBufferPtr input; |
15326 | 0 | xmlParserInputPtr stream; |
15327 | |
|
15328 | 0 | if (ioread == NULL) |
15329 | 0 | return (NULL); |
15330 | 0 | xmlInitParser(); |
15331 | |
|
15332 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15333 | 0 | XML_CHAR_ENCODING_NONE); |
15334 | 0 | if (input == NULL) { |
15335 | 0 | if (ioclose != NULL) |
15336 | 0 | ioclose(ioctx); |
15337 | 0 | return (NULL); |
15338 | 0 | } |
15339 | 0 | ctxt = xmlNewParserCtxt(); |
15340 | 0 | if (ctxt == NULL) { |
15341 | 0 | xmlFreeParserInputBuffer(input); |
15342 | 0 | return (NULL); |
15343 | 0 | } |
15344 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15345 | 0 | if (stream == NULL) { |
15346 | 0 | xmlFreeParserInputBuffer(input); |
15347 | 0 | xmlFreeParserCtxt(ctxt); |
15348 | 0 | return (NULL); |
15349 | 0 | } |
15350 | 0 | inputPush(ctxt, stream); |
15351 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15352 | 0 | } |
15353 | | |
15354 | | /** |
15355 | | * xmlCtxtReadDoc: |
15356 | | * @ctxt: an XML parser context |
15357 | | * @cur: a pointer to a zero terminated string |
15358 | | * @URL: the base URL to use for the document |
15359 | | * @encoding: the document encoding, or NULL |
15360 | | * @options: a combination of xmlParserOption |
15361 | | * |
15362 | | * parse an XML in-memory document and build a tree. |
15363 | | * This reuses the existing @ctxt parser context |
15364 | | * |
15365 | | * Returns the resulting document tree |
15366 | | */ |
15367 | | xmlDocPtr |
15368 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar * cur, |
15369 | | const char *URL, const char *encoding, int options) |
15370 | 0 | { |
15371 | 0 | if (cur == NULL) |
15372 | 0 | return (NULL); |
15373 | 0 | return (xmlCtxtReadMemory(ctxt, (const char *) cur, xmlStrlen(cur), URL, |
15374 | 0 | encoding, options)); |
15375 | 0 | } |
15376 | | |
15377 | | /** |
15378 | | * xmlCtxtReadFile: |
15379 | | * @ctxt: an XML parser context |
15380 | | * @filename: a file or URL |
15381 | | * @encoding: the document encoding, or NULL |
15382 | | * @options: a combination of xmlParserOption |
15383 | | * |
15384 | | * parse an XML file from the filesystem or the network. |
15385 | | * This reuses the existing @ctxt parser context |
15386 | | * |
15387 | | * Returns the resulting document tree |
15388 | | */ |
15389 | | xmlDocPtr |
15390 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
15391 | | const char *encoding, int options) |
15392 | 0 | { |
15393 | 0 | xmlParserInputPtr stream; |
15394 | |
|
15395 | 0 | if (filename == NULL) |
15396 | 0 | return (NULL); |
15397 | 0 | if (ctxt == NULL) |
15398 | 0 | return (NULL); |
15399 | 0 | xmlInitParser(); |
15400 | |
|
15401 | 0 | xmlCtxtReset(ctxt); |
15402 | |
|
15403 | 0 | stream = xmlLoadExternalEntity(filename, NULL, ctxt); |
15404 | 0 | if (stream == NULL) { |
15405 | 0 | return (NULL); |
15406 | 0 | } |
15407 | 0 | inputPush(ctxt, stream); |
15408 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 1)); |
15409 | 0 | } |
15410 | | |
15411 | | /** |
15412 | | * xmlCtxtReadMemory: |
15413 | | * @ctxt: an XML parser context |
15414 | | * @buffer: a pointer to a char array |
15415 | | * @size: the size of the array |
15416 | | * @URL: the base URL to use for the document |
15417 | | * @encoding: the document encoding, or NULL |
15418 | | * @options: a combination of xmlParserOption |
15419 | | * |
15420 | | * parse an XML in-memory document and build a tree. |
15421 | | * This reuses the existing @ctxt parser context |
15422 | | * |
15423 | | * Returns the resulting document tree |
15424 | | */ |
15425 | | xmlDocPtr |
15426 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
15427 | | const char *URL, const char *encoding, int options) |
15428 | 0 | { |
15429 | 0 | xmlParserInputBufferPtr input; |
15430 | 0 | xmlParserInputPtr stream; |
15431 | |
|
15432 | 0 | if (ctxt == NULL) |
15433 | 0 | return (NULL); |
15434 | 0 | if (buffer == NULL) |
15435 | 0 | return (NULL); |
15436 | 0 | xmlInitParser(); |
15437 | |
|
15438 | 0 | xmlCtxtReset(ctxt); |
15439 | |
|
15440 | 0 | input = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
15441 | 0 | if (input == NULL) { |
15442 | 0 | return(NULL); |
15443 | 0 | } |
15444 | | |
15445 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15446 | 0 | if (stream == NULL) { |
15447 | 0 | xmlFreeParserInputBuffer(input); |
15448 | 0 | return(NULL); |
15449 | 0 | } |
15450 | | |
15451 | 0 | inputPush(ctxt, stream); |
15452 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15453 | 0 | } |
15454 | | |
15455 | | /** |
15456 | | * xmlCtxtReadFd: |
15457 | | * @ctxt: an XML parser context |
15458 | | * @fd: an open file descriptor |
15459 | | * @URL: the base URL to use for the document |
15460 | | * @encoding: the document encoding, or NULL |
15461 | | * @options: a combination of xmlParserOption |
15462 | | * |
15463 | | * parse an XML from a file descriptor and build a tree. |
15464 | | * This reuses the existing @ctxt parser context |
15465 | | * NOTE that the file descriptor will not be closed when the |
15466 | | * reader is closed or reset. |
15467 | | * |
15468 | | * Returns the resulting document tree |
15469 | | */ |
15470 | | xmlDocPtr |
15471 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
15472 | | const char *URL, const char *encoding, int options) |
15473 | 0 | { |
15474 | 0 | xmlParserInputBufferPtr input; |
15475 | 0 | xmlParserInputPtr stream; |
15476 | |
|
15477 | 0 | if (fd < 0) |
15478 | 0 | return (NULL); |
15479 | 0 | if (ctxt == NULL) |
15480 | 0 | return (NULL); |
15481 | 0 | xmlInitParser(); |
15482 | |
|
15483 | 0 | xmlCtxtReset(ctxt); |
15484 | | |
15485 | |
|
15486 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15487 | 0 | if (input == NULL) |
15488 | 0 | return (NULL); |
15489 | 0 | input->closecallback = NULL; |
15490 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15491 | 0 | if (stream == NULL) { |
15492 | 0 | xmlFreeParserInputBuffer(input); |
15493 | 0 | return (NULL); |
15494 | 0 | } |
15495 | 0 | inputPush(ctxt, stream); |
15496 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15497 | 0 | } |
15498 | | |
15499 | | /** |
15500 | | * xmlCtxtReadIO: |
15501 | | * @ctxt: an XML parser context |
15502 | | * @ioread: an I/O read function |
15503 | | * @ioclose: an I/O close function |
15504 | | * @ioctx: an I/O handler |
15505 | | * @URL: the base URL to use for the document |
15506 | | * @encoding: the document encoding, or NULL |
15507 | | * @options: a combination of xmlParserOption |
15508 | | * |
15509 | | * parse an XML document from I/O functions and source and build a tree. |
15510 | | * This reuses the existing @ctxt parser context |
15511 | | * |
15512 | | * Returns the resulting document tree |
15513 | | */ |
15514 | | xmlDocPtr |
15515 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
15516 | | xmlInputCloseCallback ioclose, void *ioctx, |
15517 | | const char *URL, |
15518 | | const char *encoding, int options) |
15519 | 0 | { |
15520 | 0 | xmlParserInputBufferPtr input; |
15521 | 0 | xmlParserInputPtr stream; |
15522 | |
|
15523 | 0 | if (ioread == NULL) |
15524 | 0 | return (NULL); |
15525 | 0 | if (ctxt == NULL) |
15526 | 0 | return (NULL); |
15527 | 0 | xmlInitParser(); |
15528 | |
|
15529 | 0 | xmlCtxtReset(ctxt); |
15530 | |
|
15531 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15532 | 0 | XML_CHAR_ENCODING_NONE); |
15533 | 0 | if (input == NULL) { |
15534 | 0 | if (ioclose != NULL) |
15535 | 0 | ioclose(ioctx); |
15536 | 0 | return (NULL); |
15537 | 0 | } |
15538 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15539 | 0 | if (stream == NULL) { |
15540 | 0 | xmlFreeParserInputBuffer(input); |
15541 | 0 | return (NULL); |
15542 | 0 | } |
15543 | 0 | inputPush(ctxt, stream); |
15544 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15545 | 0 | } |
15546 | | |