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 SAX2.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 | | * Author: Daniel Veillard |
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/parser.h> |
55 | | #include <libxml/xmlmemory.h> |
56 | | #include <libxml/tree.h> |
57 | | #include <libxml/parserInternals.h> |
58 | | #include <libxml/valid.h> |
59 | | #include <libxml/entities.h> |
60 | | #include <libxml/xmlerror.h> |
61 | | #include <libxml/encoding.h> |
62 | | #include <libxml/xmlIO.h> |
63 | | #include <libxml/uri.h> |
64 | | #include <libxml/SAX2.h> |
65 | | #include <libxml/HTMLparser.h> |
66 | | #ifdef LIBXML_CATALOG_ENABLED |
67 | | #include <libxml/catalog.h> |
68 | | #endif |
69 | | |
70 | | #include "private/buf.h" |
71 | | #include "private/dict.h" |
72 | | #include "private/entities.h" |
73 | | #include "private/error.h" |
74 | | #include "private/html.h" |
75 | | #include "private/io.h" |
76 | | #include "private/memory.h" |
77 | | #include "private/parser.h" |
78 | | #include "private/tree.h" |
79 | | |
80 | 404k | #define NS_INDEX_EMPTY INT_MAX |
81 | 66.7k | #define NS_INDEX_XML (INT_MAX - 1) |
82 | 194k | #define URI_HASH_EMPTY 0xD943A04E |
83 | 24.6k | #define URI_HASH_XML 0xF0451F02 |
84 | | |
85 | | #ifndef STDIN_FILENO |
86 | 0 | #define STDIN_FILENO 0 |
87 | | #endif |
88 | | |
89 | | #ifndef SIZE_MAX |
90 | | #define SIZE_MAX ((size_t) -1) |
91 | | #endif |
92 | | |
93 | 200k | #define XML_MAX_ATTRS 100000000 /* 100 million */ |
94 | | |
95 | 374k | #define XML_SPECIAL_EXTERNAL (1 << 20) |
96 | 331k | #define XML_SPECIAL_TYPE_MASK (XML_SPECIAL_EXTERNAL - 1) |
97 | | |
98 | 374k | #define XML_ATTVAL_ALLOC (1 << 0) |
99 | 977k | #define XML_ATTVAL_NORM_CHANGE (1 << 1) |
100 | | |
101 | | struct _xmlStartTag { |
102 | | const xmlChar *prefix; |
103 | | const xmlChar *URI; |
104 | | int line; |
105 | | int nsNr; |
106 | | }; |
107 | | |
108 | | typedef struct { |
109 | | void *saxData; |
110 | | unsigned prefixHashValue; |
111 | | unsigned uriHashValue; |
112 | | unsigned elementId; |
113 | | int oldIndex; |
114 | | } xmlParserNsExtra; |
115 | | |
116 | | typedef struct { |
117 | | unsigned hashValue; |
118 | | int index; |
119 | | } xmlParserNsBucket; |
120 | | |
121 | | struct _xmlParserNsData { |
122 | | xmlParserNsExtra *extra; |
123 | | |
124 | | unsigned hashSize; |
125 | | unsigned hashElems; |
126 | | xmlParserNsBucket *hash; |
127 | | |
128 | | unsigned elementId; |
129 | | int defaultNsIndex; |
130 | | int minNsIndex; |
131 | | }; |
132 | | |
133 | | static int |
134 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
135 | | |
136 | | static void |
137 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
138 | | |
139 | | static xmlEntityPtr |
140 | | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr); |
141 | | |
142 | | static const xmlChar * |
143 | | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt); |
144 | | |
145 | | /************************************************************************ |
146 | | * * |
147 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
148 | | * * |
149 | | ************************************************************************/ |
150 | | |
151 | | #define XML_PARSER_BIG_ENTITY 1000 |
152 | | #define XML_PARSER_LOT_ENTITY 5000 |
153 | | |
154 | | /* |
155 | | * Constants for protection against abusive entity expansion |
156 | | * ("billion laughs"). |
157 | | */ |
158 | | |
159 | | /* |
160 | | * A certain amount of entity expansion which is always allowed. |
161 | | */ |
162 | 1.91M | #define XML_PARSER_ALLOWED_EXPANSION 1000000 |
163 | | |
164 | | /* |
165 | | * Fixed cost for each entity reference. This crudely models processing time |
166 | | * as well to protect, for example, against exponential expansion of empty |
167 | | * or very short entities. |
168 | | */ |
169 | 1.92M | #define XML_ENT_FIXED_COST 20 |
170 | | |
171 | 68.6M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
172 | 515k | #define XML_PARSER_BUFFER_SIZE 100 |
173 | 74.9k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
174 | | |
175 | | /** |
176 | | * XML_PARSER_CHUNK_SIZE |
177 | | * |
178 | | * When calling GROW that's the minimal amount of data |
179 | | * the parser expected to have received. It is not a hard |
180 | | * limit but an optimization when reading strings like Names |
181 | | * It is not strictly needed as long as inputs available characters |
182 | | * are followed by 0, which should be provided by the I/O level |
183 | | */ |
184 | | #define XML_PARSER_CHUNK_SIZE 100 |
185 | | |
186 | | /** |
187 | | * Constant string describing the version of the library used at |
188 | | * run-time. |
189 | | */ |
190 | | const char *const |
191 | | xmlParserVersion = LIBXML_VERSION_STRING LIBXML_VERSION_EXTRA; |
192 | | |
193 | | /* |
194 | | * List of XML prefixed PI allowed by W3C specs |
195 | | */ |
196 | | |
197 | | static const char* const xmlW3CPIs[] = { |
198 | | "xml-stylesheet", |
199 | | "xml-model", |
200 | | NULL |
201 | | }; |
202 | | |
203 | | |
204 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
205 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
206 | | const xmlChar **str); |
207 | | |
208 | | static void |
209 | | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent); |
210 | | |
211 | | static int |
212 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
213 | | |
214 | | static void |
215 | | xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl); |
216 | | |
217 | | /************************************************************************ |
218 | | * * |
219 | | * Some factorized error routines * |
220 | | * * |
221 | | ************************************************************************/ |
222 | | |
223 | | static void |
224 | 2.12k | xmlErrMemory(xmlParserCtxtPtr ctxt) { |
225 | 2.12k | xmlCtxtErrMemory(ctxt); |
226 | 2.12k | } |
227 | | |
228 | | /** |
229 | | * Handle a redefinition of attribute error |
230 | | * |
231 | | * @param ctxt an XML parser context |
232 | | * @param prefix the attribute prefix |
233 | | * @param localname the attribute localname |
234 | | */ |
235 | | static void |
236 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
237 | | const xmlChar * localname) |
238 | 17.7k | { |
239 | 17.7k | if (prefix == NULL) |
240 | 11.8k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
241 | 11.8k | XML_ERR_FATAL, localname, NULL, NULL, 0, |
242 | 11.8k | "Attribute %s redefined\n", localname); |
243 | 5.90k | else |
244 | 5.90k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, XML_ERR_ATTRIBUTE_REDEFINED, |
245 | 5.90k | XML_ERR_FATAL, prefix, localname, NULL, 0, |
246 | 5.90k | "Attribute %s:%s redefined\n", prefix, localname); |
247 | 17.7k | } |
248 | | |
249 | | /** |
250 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
251 | | * |
252 | | * @param ctxt an XML parser context |
253 | | * @param error the error number |
254 | | * @param msg the error message |
255 | | */ |
256 | | static void LIBXML_ATTR_FORMAT(3,0) |
257 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
258 | | const char *msg) |
259 | 45.4M | { |
260 | 45.4M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
261 | 45.4M | NULL, NULL, NULL, 0, "%s", msg); |
262 | 45.4M | } |
263 | | |
264 | | /** |
265 | | * Handle a warning. |
266 | | * |
267 | | * @param ctxt an XML parser context |
268 | | * @param error the error number |
269 | | * @param msg the error message |
270 | | * @param str1 extra data |
271 | | * @param str2 extra data |
272 | | */ |
273 | | void LIBXML_ATTR_FORMAT(3,0) |
274 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
275 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
276 | 16.2k | { |
277 | 16.2k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_WARNING, |
278 | 16.2k | str1, str2, NULL, 0, msg, str1, str2); |
279 | 16.2k | } |
280 | | |
281 | | #ifdef LIBXML_VALID_ENABLED |
282 | | /** |
283 | | * Handle a validity error. |
284 | | * |
285 | | * @param ctxt an XML parser context |
286 | | * @param error the error number |
287 | | * @param msg the error message |
288 | | * @param str1 extra data |
289 | | * @param str2 extra data |
290 | | */ |
291 | | static void LIBXML_ATTR_FORMAT(3,0) |
292 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
293 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
294 | 0 | { |
295 | 0 | ctxt->valid = 0; |
296 | |
|
297 | 0 | xmlCtxtErr(ctxt, NULL, XML_FROM_DTD, error, XML_ERR_ERROR, |
298 | 0 | str1, str2, NULL, 0, msg, str1, str2); |
299 | 0 | } |
300 | | #endif |
301 | | |
302 | | /** |
303 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
304 | | * |
305 | | * @param ctxt an XML parser context |
306 | | * @param error the error number |
307 | | * @param msg the error message |
308 | | * @param val an integer value |
309 | | */ |
310 | | static void LIBXML_ATTR_FORMAT(3,0) |
311 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
312 | | const char *msg, int val) |
313 | 3.03M | { |
314 | 3.03M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
315 | 3.03M | NULL, NULL, NULL, val, msg, val); |
316 | 3.03M | } |
317 | | |
318 | | /** |
319 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
320 | | * |
321 | | * @param ctxt an XML parser context |
322 | | * @param error the error number |
323 | | * @param msg the error message |
324 | | * @param str1 an string info |
325 | | * @param val an integer value |
326 | | * @param str2 an string info |
327 | | */ |
328 | | static void LIBXML_ATTR_FORMAT(3,0) |
329 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
330 | | const char *msg, const xmlChar *str1, int val, |
331 | | const xmlChar *str2) |
332 | 228k | { |
333 | 228k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
334 | 228k | str1, str2, NULL, val, msg, str1, val, str2); |
335 | 228k | } |
336 | | |
337 | | /** |
338 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
339 | | * |
340 | | * @param ctxt an XML parser context |
341 | | * @param error the error number |
342 | | * @param msg the error message |
343 | | * @param val a string value |
344 | | */ |
345 | | static void LIBXML_ATTR_FORMAT(3,0) |
346 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
347 | | const char *msg, const xmlChar * val) |
348 | 2.32M | { |
349 | 2.32M | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
350 | 2.32M | val, NULL, NULL, 0, msg, val); |
351 | 2.32M | } |
352 | | |
353 | | /** |
354 | | * Handle a non fatal parser error |
355 | | * |
356 | | * @param ctxt an XML parser context |
357 | | * @param error the error number |
358 | | * @param msg the error message |
359 | | * @param val a string value |
360 | | */ |
361 | | static void LIBXML_ATTR_FORMAT(3,0) |
362 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
363 | | const char *msg, const xmlChar * val) |
364 | 73.9k | { |
365 | 73.9k | xmlCtxtErr(ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_ERROR, |
366 | 73.9k | val, NULL, NULL, 0, msg, val); |
367 | 73.9k | } |
368 | | |
369 | | /** |
370 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
371 | | * |
372 | | * @param ctxt an XML parser context |
373 | | * @param error the error number |
374 | | * @param msg the message |
375 | | * @param info1 extra information string |
376 | | * @param info2 extra information string |
377 | | * @param info3 extra information string |
378 | | */ |
379 | | static void LIBXML_ATTR_FORMAT(3,0) |
380 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
381 | | const char *msg, |
382 | | const xmlChar * info1, const xmlChar * info2, |
383 | | const xmlChar * info3) |
384 | 173k | { |
385 | 173k | ctxt->nsWellFormed = 0; |
386 | | |
387 | 173k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_ERROR, |
388 | 173k | info1, info2, info3, 0, msg, info1, info2, info3); |
389 | 173k | } |
390 | | |
391 | | /** |
392 | | * Handle a namespace warning error |
393 | | * |
394 | | * @param ctxt an XML parser context |
395 | | * @param error the error number |
396 | | * @param msg the message |
397 | | * @param info1 extra information string |
398 | | * @param info2 extra information string |
399 | | * @param info3 extra information string |
400 | | */ |
401 | | static void LIBXML_ATTR_FORMAT(3,0) |
402 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
403 | | const char *msg, |
404 | | const xmlChar * info1, const xmlChar * info2, |
405 | | const xmlChar * info3) |
406 | 11.1k | { |
407 | 11.1k | xmlCtxtErr(ctxt, NULL, XML_FROM_NAMESPACE, error, XML_ERR_WARNING, |
408 | 11.1k | info1, info2, info3, 0, msg, info1, info2, info3); |
409 | 11.1k | } |
410 | | |
411 | | /** |
412 | | * Check for non-linear entity expansion behaviour. |
413 | | * |
414 | | * In some cases like xmlExpandEntityInAttValue, this function is called |
415 | | * for each, possibly nested entity and its unexpanded content length. |
416 | | * |
417 | | * In other cases like #xmlParseReference, it's only called for each |
418 | | * top-level entity with its unexpanded content length plus the sum of |
419 | | * the unexpanded content lengths (plus fixed cost) of all nested |
420 | | * entities. |
421 | | * |
422 | | * Summing the unexpanded lengths also adds the length of the reference. |
423 | | * This is by design. Taking the length of the entity name into account |
424 | | * discourages attacks that try to waste CPU time with abusively long |
425 | | * entity names. See test/recurse/lol6.xml for example. Each call also |
426 | | * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with |
427 | | * short entities. |
428 | | * |
429 | | * @param ctxt parser context |
430 | | * @param extra sum of unexpanded entity sizes |
431 | | * @returns 1 on error, 0 on success. |
432 | | */ |
433 | | static int |
434 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra) |
435 | 2.03M | { |
436 | 2.03M | unsigned long consumed; |
437 | 2.03M | unsigned long *expandedSize; |
438 | 2.03M | xmlParserInputPtr input = ctxt->input; |
439 | 2.03M | xmlEntityPtr entity = input->entity; |
440 | | |
441 | 2.03M | if ((entity) && (entity->flags & XML_ENT_CHECKED)) |
442 | 111k | return(0); |
443 | | |
444 | | /* |
445 | | * Compute total consumed bytes so far, including input streams of |
446 | | * external entities. |
447 | | */ |
448 | 1.91M | consumed = input->consumed; |
449 | 1.91M | xmlSaturatedAddSizeT(&consumed, input->cur - input->base); |
450 | 1.91M | xmlSaturatedAdd(&consumed, ctxt->sizeentities); |
451 | | |
452 | 1.91M | if (entity) |
453 | 36.1k | expandedSize = &entity->expandedSize; |
454 | 1.88M | else |
455 | 1.88M | expandedSize = &ctxt->sizeentcopy; |
456 | | |
457 | | /* |
458 | | * Add extra cost and some fixed cost. |
459 | | */ |
460 | 1.91M | xmlSaturatedAdd(expandedSize, extra); |
461 | 1.91M | xmlSaturatedAdd(expandedSize, XML_ENT_FIXED_COST); |
462 | | |
463 | | /* |
464 | | * It's important to always use saturation arithmetic when tracking |
465 | | * entity sizes to make the size checks reliable. If "sizeentcopy" |
466 | | * overflows, we have to abort. |
467 | | */ |
468 | 1.91M | if ((*expandedSize > XML_PARSER_ALLOWED_EXPANSION) && |
469 | 1.91M | ((*expandedSize >= ULONG_MAX) || |
470 | 89.9k | (*expandedSize / ctxt->maxAmpl > consumed))) { |
471 | 951 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
472 | 951 | "Maximum entity amplification factor exceeded, see " |
473 | 951 | "xmlCtxtSetMaxAmplification.\n"); |
474 | 951 | return(1); |
475 | 951 | } |
476 | | |
477 | 1.91M | return(0); |
478 | 1.91M | } |
479 | | |
480 | | /************************************************************************ |
481 | | * * |
482 | | * Library wide options * |
483 | | * * |
484 | | ************************************************************************/ |
485 | | |
486 | | /** |
487 | | * Examines if the library has been compiled with a given feature. |
488 | | * |
489 | | * @param feature the feature to be examined |
490 | | * @returns zero (0) if the feature does not exist or an unknown |
491 | | * feature is requested, non-zero otherwise. |
492 | | */ |
493 | | int |
494 | | xmlHasFeature(xmlFeature feature) |
495 | 0 | { |
496 | 0 | switch (feature) { |
497 | 0 | case XML_WITH_THREAD: |
498 | 0 | #ifdef LIBXML_THREAD_ENABLED |
499 | 0 | return(1); |
500 | | #else |
501 | | return(0); |
502 | | #endif |
503 | 0 | case XML_WITH_TREE: |
504 | 0 | return(1); |
505 | 0 | case XML_WITH_OUTPUT: |
506 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
507 | 0 | return(1); |
508 | | #else |
509 | | return(0); |
510 | | #endif |
511 | 0 | case XML_WITH_PUSH: |
512 | 0 | #ifdef LIBXML_PUSH_ENABLED |
513 | 0 | return(1); |
514 | | #else |
515 | | return(0); |
516 | | #endif |
517 | 0 | case XML_WITH_READER: |
518 | 0 | #ifdef LIBXML_READER_ENABLED |
519 | 0 | return(1); |
520 | | #else |
521 | | return(0); |
522 | | #endif |
523 | 0 | case XML_WITH_PATTERN: |
524 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
525 | 0 | return(1); |
526 | | #else |
527 | | return(0); |
528 | | #endif |
529 | 0 | case XML_WITH_WRITER: |
530 | 0 | #ifdef LIBXML_WRITER_ENABLED |
531 | 0 | return(1); |
532 | | #else |
533 | | return(0); |
534 | | #endif |
535 | 0 | case XML_WITH_SAX1: |
536 | 0 | #ifdef LIBXML_SAX1_ENABLED |
537 | 0 | return(1); |
538 | | #else |
539 | | return(0); |
540 | | #endif |
541 | 0 | case XML_WITH_HTTP: |
542 | 0 | return(0); |
543 | 0 | case XML_WITH_VALID: |
544 | 0 | #ifdef LIBXML_VALID_ENABLED |
545 | 0 | return(1); |
546 | | #else |
547 | | return(0); |
548 | | #endif |
549 | 0 | case XML_WITH_HTML: |
550 | 0 | #ifdef LIBXML_HTML_ENABLED |
551 | 0 | return(1); |
552 | | #else |
553 | | return(0); |
554 | | #endif |
555 | 0 | case XML_WITH_LEGACY: |
556 | 0 | return(0); |
557 | 0 | case XML_WITH_C14N: |
558 | 0 | #ifdef LIBXML_C14N_ENABLED |
559 | 0 | return(1); |
560 | | #else |
561 | | return(0); |
562 | | #endif |
563 | 0 | case XML_WITH_CATALOG: |
564 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
565 | 0 | return(1); |
566 | | #else |
567 | | return(0); |
568 | | #endif |
569 | 0 | case XML_WITH_XPATH: |
570 | 0 | #ifdef LIBXML_XPATH_ENABLED |
571 | 0 | return(1); |
572 | | #else |
573 | | return(0); |
574 | | #endif |
575 | 0 | case XML_WITH_XPTR: |
576 | 0 | #ifdef LIBXML_XPTR_ENABLED |
577 | 0 | return(1); |
578 | | #else |
579 | | return(0); |
580 | | #endif |
581 | 0 | case XML_WITH_XINCLUDE: |
582 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
583 | 0 | return(1); |
584 | | #else |
585 | | return(0); |
586 | | #endif |
587 | 0 | case XML_WITH_ICONV: |
588 | 0 | #ifdef LIBXML_ICONV_ENABLED |
589 | 0 | return(1); |
590 | | #else |
591 | | return(0); |
592 | | #endif |
593 | 0 | case XML_WITH_ISO8859X: |
594 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
595 | 0 | return(1); |
596 | | #else |
597 | | return(0); |
598 | | #endif |
599 | 0 | case XML_WITH_UNICODE: |
600 | 0 | return(0); |
601 | 0 | case XML_WITH_REGEXP: |
602 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
603 | 0 | return(1); |
604 | | #else |
605 | | return(0); |
606 | | #endif |
607 | 0 | case XML_WITH_AUTOMATA: |
608 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
609 | 0 | return(1); |
610 | | #else |
611 | | return(0); |
612 | | #endif |
613 | 0 | case XML_WITH_EXPR: |
614 | 0 | return(0); |
615 | 0 | case XML_WITH_RELAXNG: |
616 | 0 | #ifdef LIBXML_RELAXNG_ENABLED |
617 | 0 | return(1); |
618 | | #else |
619 | | return(0); |
620 | | #endif |
621 | 0 | case XML_WITH_SCHEMAS: |
622 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
623 | 0 | return(1); |
624 | | #else |
625 | | return(0); |
626 | | #endif |
627 | 0 | case XML_WITH_SCHEMATRON: |
628 | | #ifdef LIBXML_SCHEMATRON_ENABLED |
629 | | return(1); |
630 | | #else |
631 | 0 | return(0); |
632 | 0 | #endif |
633 | 0 | case XML_WITH_MODULES: |
634 | 0 | #ifdef LIBXML_MODULES_ENABLED |
635 | 0 | return(1); |
636 | | #else |
637 | | return(0); |
638 | | #endif |
639 | 0 | case XML_WITH_DEBUG: |
640 | | #ifdef LIBXML_DEBUG_ENABLED |
641 | | return(1); |
642 | | #else |
643 | 0 | return(0); |
644 | 0 | #endif |
645 | 0 | case XML_WITH_DEBUG_MEM: |
646 | 0 | return(0); |
647 | 0 | case XML_WITH_ZLIB: |
648 | 0 | #ifdef LIBXML_ZLIB_ENABLED |
649 | 0 | return(1); |
650 | | #else |
651 | | return(0); |
652 | | #endif |
653 | 0 | case XML_WITH_LZMA: |
654 | 0 | #ifdef LIBXML_LZMA_ENABLED |
655 | 0 | return(1); |
656 | | #else |
657 | | return(0); |
658 | | #endif |
659 | 0 | case XML_WITH_ICU: |
660 | | #ifdef LIBXML_ICU_ENABLED |
661 | | return(1); |
662 | | #else |
663 | 0 | return(0); |
664 | 0 | #endif |
665 | 0 | default: |
666 | 0 | break; |
667 | 0 | } |
668 | 0 | return(0); |
669 | 0 | } |
670 | | |
671 | | /************************************************************************ |
672 | | * * |
673 | | * Simple string buffer * |
674 | | * * |
675 | | ************************************************************************/ |
676 | | |
677 | | typedef struct { |
678 | | xmlChar *mem; |
679 | | unsigned size; |
680 | | unsigned cap; /* size < cap */ |
681 | | unsigned max; /* size <= max */ |
682 | | xmlParserErrors code; |
683 | | } xmlSBuf; |
684 | | |
685 | | static void |
686 | 377k | xmlSBufInit(xmlSBuf *buf, unsigned max) { |
687 | 377k | buf->mem = NULL; |
688 | 377k | buf->size = 0; |
689 | 377k | buf->cap = 0; |
690 | 377k | buf->max = max; |
691 | 377k | buf->code = XML_ERR_OK; |
692 | 377k | } |
693 | | |
694 | | static int |
695 | 188k | xmlSBufGrow(xmlSBuf *buf, unsigned len) { |
696 | 188k | xmlChar *mem; |
697 | 188k | unsigned cap; |
698 | | |
699 | 188k | if (len >= UINT_MAX / 2 - buf->size) { |
700 | 0 | if (buf->code == XML_ERR_OK) |
701 | 0 | buf->code = XML_ERR_RESOURCE_LIMIT; |
702 | 0 | return(-1); |
703 | 0 | } |
704 | | |
705 | 188k | cap = (buf->size + len) * 2; |
706 | 188k | if (cap < 240) |
707 | 145k | cap = 240; |
708 | | |
709 | 188k | mem = xmlRealloc(buf->mem, cap); |
710 | 188k | if (mem == NULL) { |
711 | 341 | buf->code = XML_ERR_NO_MEMORY; |
712 | 341 | return(-1); |
713 | 341 | } |
714 | | |
715 | 188k | buf->mem = mem; |
716 | 188k | buf->cap = cap; |
717 | | |
718 | 188k | return(0); |
719 | 188k | } |
720 | | |
721 | | static void |
722 | 125M | xmlSBufAddString(xmlSBuf *buf, const xmlChar *str, unsigned len) { |
723 | 125M | if (buf->max - buf->size < len) { |
724 | 600k | if (buf->code == XML_ERR_OK) |
725 | 201 | buf->code = XML_ERR_RESOURCE_LIMIT; |
726 | 600k | return; |
727 | 600k | } |
728 | | |
729 | 124M | if (buf->cap - buf->size <= len) { |
730 | 179k | if (xmlSBufGrow(buf, len) < 0) |
731 | 312 | return; |
732 | 179k | } |
733 | | |
734 | 124M | if (len > 0) |
735 | 124M | memcpy(buf->mem + buf->size, str, len); |
736 | 124M | buf->size += len; |
737 | 124M | } |
738 | | |
739 | | static void |
740 | 122M | xmlSBufAddCString(xmlSBuf *buf, const char *str, unsigned len) { |
741 | 122M | xmlSBufAddString(buf, (const xmlChar *) str, len); |
742 | 122M | } |
743 | | |
744 | | static void |
745 | 332k | xmlSBufAddChar(xmlSBuf *buf, int c) { |
746 | 332k | xmlChar *end; |
747 | | |
748 | 332k | if (buf->max - buf->size < 4) { |
749 | 422 | if (buf->code == XML_ERR_OK) |
750 | 6 | buf->code = XML_ERR_RESOURCE_LIMIT; |
751 | 422 | return; |
752 | 422 | } |
753 | | |
754 | 332k | if (buf->cap - buf->size <= 4) { |
755 | 9.53k | if (xmlSBufGrow(buf, 4) < 0) |
756 | 29 | return; |
757 | 9.53k | } |
758 | | |
759 | 332k | end = buf->mem + buf->size; |
760 | | |
761 | 332k | if (c < 0x80) { |
762 | 132k | *end = (xmlChar) c; |
763 | 132k | buf->size += 1; |
764 | 200k | } else { |
765 | 200k | buf->size += xmlCopyCharMultiByte(end, c); |
766 | 200k | } |
767 | 332k | } |
768 | | |
769 | | static void |
770 | 106M | xmlSBufAddReplChar(xmlSBuf *buf) { |
771 | 106M | xmlSBufAddCString(buf, "\xEF\xBF\xBD", 3); |
772 | 106M | } |
773 | | |
774 | | static void |
775 | 570 | xmlSBufReportError(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
776 | 570 | if (buf->code == XML_ERR_NO_MEMORY) |
777 | 363 | xmlCtxtErrMemory(ctxt); |
778 | 207 | else |
779 | 207 | xmlFatalErr(ctxt, buf->code, errMsg); |
780 | 570 | } |
781 | | |
782 | | static xmlChar * |
783 | | xmlSBufFinish(xmlSBuf *buf, int *sizeOut, xmlParserCtxtPtr ctxt, |
784 | 155k | const char *errMsg) { |
785 | 155k | if (buf->mem == NULL) { |
786 | 16.6k | buf->mem = xmlMalloc(1); |
787 | 16.6k | if (buf->mem == NULL) { |
788 | 22 | buf->code = XML_ERR_NO_MEMORY; |
789 | 16.6k | } else { |
790 | 16.6k | buf->mem[0] = 0; |
791 | 16.6k | } |
792 | 139k | } else { |
793 | 139k | buf->mem[buf->size] = 0; |
794 | 139k | } |
795 | | |
796 | 155k | if (buf->code == XML_ERR_OK) { |
797 | 155k | if (sizeOut != NULL) |
798 | 62.2k | *sizeOut = buf->size; |
799 | 155k | return(buf->mem); |
800 | 155k | } |
801 | | |
802 | 345 | xmlSBufReportError(buf, ctxt, errMsg); |
803 | | |
804 | 345 | xmlFree(buf->mem); |
805 | | |
806 | 345 | if (sizeOut != NULL) |
807 | 166 | *sizeOut = 0; |
808 | 345 | return(NULL); |
809 | 155k | } |
810 | | |
811 | | static void |
812 | 214k | xmlSBufCleanup(xmlSBuf *buf, xmlParserCtxtPtr ctxt, const char *errMsg) { |
813 | 214k | if (buf->code != XML_ERR_OK) |
814 | 225 | xmlSBufReportError(buf, ctxt, errMsg); |
815 | | |
816 | 214k | xmlFree(buf->mem); |
817 | 214k | } |
818 | | |
819 | | static int |
820 | | xmlUTF8MultibyteLen(xmlParserCtxtPtr ctxt, const xmlChar *str, |
821 | 442M | const char *errMsg) { |
822 | 442M | int c = str[0]; |
823 | 442M | int c1 = str[1]; |
824 | | |
825 | 442M | if ((c1 & 0xC0) != 0x80) |
826 | 48.8M | goto encoding_error; |
827 | | |
828 | 393M | if (c < 0xE0) { |
829 | | /* 2-byte sequence */ |
830 | 18.8M | if (c < 0xC2) |
831 | 13.2M | goto encoding_error; |
832 | | |
833 | 5.54M | return(2); |
834 | 374M | } else { |
835 | 374M | int c2 = str[2]; |
836 | | |
837 | 374M | if ((c2 & 0xC0) != 0x80) |
838 | 13.2k | goto encoding_error; |
839 | | |
840 | 374M | if (c < 0xF0) { |
841 | | /* 3-byte sequence */ |
842 | 374M | if (c == 0xE0) { |
843 | | /* overlong */ |
844 | 4.21k | if (c1 < 0xA0) |
845 | 260 | goto encoding_error; |
846 | 374M | } else if (c == 0xED) { |
847 | | /* surrogate */ |
848 | 780 | if (c1 >= 0xA0) |
849 | 274 | goto encoding_error; |
850 | 374M | } else if (c == 0xEF) { |
851 | | /* U+FFFE and U+FFFF are invalid Chars */ |
852 | 39.3M | if ((c1 == 0xBF) && (c2 >= 0xBE)) |
853 | 332 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, errMsg); |
854 | 39.3M | } |
855 | | |
856 | 374M | return(3); |
857 | 374M | } else { |
858 | | /* 4-byte sequence */ |
859 | 14.5k | if ((str[3] & 0xC0) != 0x80) |
860 | 2.07k | goto encoding_error; |
861 | 12.4k | if (c == 0xF0) { |
862 | | /* overlong */ |
863 | 1.11k | if (c1 < 0x90) |
864 | 323 | goto encoding_error; |
865 | 11.3k | } else if (c >= 0xF4) { |
866 | | /* greater than 0x10FFFF */ |
867 | 5.28k | if ((c > 0xF4) || (c1 >= 0x90)) |
868 | 4.13k | goto encoding_error; |
869 | 5.28k | } |
870 | | |
871 | 8.02k | return(4); |
872 | 12.4k | } |
873 | 374M | } |
874 | | |
875 | 62.1M | encoding_error: |
876 | | /* Only report the first error */ |
877 | 62.1M | if ((ctxt->input->flags & XML_INPUT_ENCODING_ERROR) == 0) { |
878 | 13.3k | xmlCtxtErrIO(ctxt, XML_ERR_INVALID_ENCODING, NULL); |
879 | 13.3k | ctxt->input->flags |= XML_INPUT_ENCODING_ERROR; |
880 | 13.3k | } |
881 | | |
882 | 62.1M | return(0); |
883 | 393M | } |
884 | | |
885 | | /************************************************************************ |
886 | | * * |
887 | | * SAX2 defaulted attributes handling * |
888 | | * * |
889 | | ************************************************************************/ |
890 | | |
891 | | /** |
892 | | * Final initialization of the parser context before starting to parse. |
893 | | * |
894 | | * This accounts for users modifying struct members of parser context |
895 | | * directly. |
896 | | * |
897 | | * @param ctxt an XML parser context |
898 | | */ |
899 | | static void |
900 | 60.0k | xmlCtxtInitializeLate(xmlParserCtxtPtr ctxt) { |
901 | 60.0k | xmlSAXHandlerPtr sax; |
902 | | |
903 | | /* Avoid unused variable warning if features are disabled. */ |
904 | 60.0k | (void) sax; |
905 | | |
906 | | /* |
907 | | * Changing the SAX struct directly is still widespread practice |
908 | | * in internal and external code. |
909 | | */ |
910 | 60.0k | if (ctxt == NULL) return; |
911 | 60.0k | sax = ctxt->sax; |
912 | 60.0k | #ifdef LIBXML_SAX1_ENABLED |
913 | | /* |
914 | | * Only enable SAX2 if there SAX2 element handlers, except when there |
915 | | * are no element handlers at all. |
916 | | */ |
917 | 60.0k | if (((ctxt->options & XML_PARSE_SAX1) == 0) && |
918 | 60.0k | (sax) && |
919 | 60.0k | (sax->initialized == XML_SAX2_MAGIC) && |
920 | 60.0k | ((sax->startElementNs != NULL) || |
921 | 60.0k | (sax->endElementNs != NULL) || |
922 | 60.0k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
923 | 60.0k | ctxt->sax2 = 1; |
924 | | #else |
925 | | ctxt->sax2 = 1; |
926 | | #endif /* LIBXML_SAX1_ENABLED */ |
927 | | |
928 | | /* |
929 | | * Some users replace the dictionary directly in the context struct. |
930 | | * We really need an API function to do that cleanly. |
931 | | */ |
932 | 60.0k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
933 | 60.0k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
934 | 60.0k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
935 | 60.0k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
936 | 60.0k | (ctxt->str_xml_ns == NULL)) { |
937 | 491 | xmlErrMemory(ctxt); |
938 | 491 | } |
939 | | |
940 | 60.0k | xmlDictSetLimit(ctxt->dict, |
941 | 60.0k | (ctxt->options & XML_PARSE_HUGE) ? |
942 | 26.4k | 0 : |
943 | 60.0k | XML_MAX_DICTIONARY_LIMIT); |
944 | | |
945 | 60.0k | #ifdef LIBXML_VALID_ENABLED |
946 | 60.0k | if (ctxt->validate) |
947 | 0 | ctxt->vctxt.flags |= XML_VCTXT_VALIDATE; |
948 | 60.0k | else |
949 | 60.0k | ctxt->vctxt.flags &= ~XML_VCTXT_VALIDATE; |
950 | 60.0k | #endif /* LIBXML_VALID_ENABLED */ |
951 | 60.0k | } |
952 | | |
953 | | typedef struct { |
954 | | xmlHashedString prefix; |
955 | | xmlHashedString name; |
956 | | xmlHashedString value; |
957 | | const xmlChar *valueEnd; |
958 | | int external; |
959 | | int expandedSize; |
960 | | } xmlDefAttr; |
961 | | |
962 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
963 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
964 | | struct _xmlDefAttrs { |
965 | | int nbAttrs; /* number of defaulted attributes on that element */ |
966 | | int maxAttrs; /* the size of the array */ |
967 | | #if __STDC_VERSION__ >= 199901L |
968 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
969 | | xmlDefAttr attrs[] ATTRIBUTE_COUNTED_BY(maxAttrs); |
970 | | #else |
971 | | xmlDefAttr attrs[1]; |
972 | | #endif |
973 | | }; |
974 | | |
975 | | /** |
976 | | * Normalize the space in non CDATA attribute values: |
977 | | * If the attribute type is not CDATA, then the XML processor MUST further |
978 | | * process the normalized attribute value by discarding any leading and |
979 | | * trailing space (\#x20) characters, and by replacing sequences of space |
980 | | * (\#x20) characters by a single space (\#x20) character. |
981 | | * Note that the size of dst need to be at least src, and if one doesn't need |
982 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
983 | | * passing src as dst is just fine. |
984 | | * |
985 | | * @param src the source string |
986 | | * @param dst the target string |
987 | | * @returns a pointer to the normalized value (dst) or NULL if no conversion |
988 | | * is needed. |
989 | | */ |
990 | | static xmlChar * |
991 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
992 | 47.4k | { |
993 | 47.4k | if ((src == NULL) || (dst == NULL)) |
994 | 0 | return(NULL); |
995 | | |
996 | 48.5k | while (*src == 0x20) src++; |
997 | 51.8M | while (*src != 0) { |
998 | 51.8M | if (*src == 0x20) { |
999 | 307k | while (*src == 0x20) src++; |
1000 | 12.1k | if (*src != 0) |
1001 | 11.5k | *dst++ = 0x20; |
1002 | 51.8M | } else { |
1003 | 51.8M | *dst++ = *src++; |
1004 | 51.8M | } |
1005 | 51.8M | } |
1006 | 47.4k | *dst = 0; |
1007 | 47.4k | if (dst == src) |
1008 | 45.3k | return(NULL); |
1009 | 2.06k | return(dst); |
1010 | 47.4k | } |
1011 | | |
1012 | | /** |
1013 | | * Add a defaulted attribute for an element |
1014 | | * |
1015 | | * @param ctxt an XML parser context |
1016 | | * @param fullname the element fullname |
1017 | | * @param fullattr the attribute fullname |
1018 | | * @param value the attribute value |
1019 | | */ |
1020 | | static void |
1021 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1022 | | const xmlChar *fullname, |
1023 | | const xmlChar *fullattr, |
1024 | 43.7k | const xmlChar *value) { |
1025 | 43.7k | xmlDefAttrsPtr defaults; |
1026 | 43.7k | xmlDefAttr *attr; |
1027 | 43.7k | int len, expandedSize; |
1028 | 43.7k | xmlHashedString name; |
1029 | 43.7k | xmlHashedString prefix; |
1030 | 43.7k | xmlHashedString hvalue; |
1031 | 43.7k | const xmlChar *localname; |
1032 | | |
1033 | | /* |
1034 | | * Allows to detect attribute redefinitions |
1035 | | */ |
1036 | 43.7k | if (ctxt->attsSpecial != NULL) { |
1037 | 39.8k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1038 | 25.9k | return; |
1039 | 39.8k | } |
1040 | | |
1041 | 17.8k | if (ctxt->attsDefault == NULL) { |
1042 | 4.10k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1043 | 4.10k | if (ctxt->attsDefault == NULL) |
1044 | 11 | goto mem_error; |
1045 | 4.10k | } |
1046 | | |
1047 | | /* |
1048 | | * split the element name into prefix:localname , the string found |
1049 | | * are within the DTD and then not associated to namespace names. |
1050 | | */ |
1051 | 17.8k | localname = xmlSplitQName3(fullname, &len); |
1052 | 17.8k | if (localname == NULL) { |
1053 | 17.2k | name = xmlDictLookupHashed(ctxt->dict, fullname, -1); |
1054 | 17.2k | prefix.name = NULL; |
1055 | 17.2k | } else { |
1056 | 556 | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1057 | 556 | prefix = xmlDictLookupHashed(ctxt->dict, fullname, len); |
1058 | 556 | if (prefix.name == NULL) |
1059 | 1 | goto mem_error; |
1060 | 556 | } |
1061 | 17.8k | if (name.name == NULL) |
1062 | 3 | goto mem_error; |
1063 | | |
1064 | | /* |
1065 | | * make sure there is some storage |
1066 | | */ |
1067 | 17.8k | defaults = xmlHashLookup2(ctxt->attsDefault, name.name, prefix.name); |
1068 | 17.8k | if ((defaults == NULL) || |
1069 | 17.8k | (defaults->nbAttrs >= defaults->maxAttrs)) { |
1070 | 5.92k | xmlDefAttrsPtr temp; |
1071 | 5.92k | int newSize; |
1072 | | |
1073 | 5.92k | if (defaults == NULL) { |
1074 | 4.89k | newSize = 4; |
1075 | 4.89k | } else { |
1076 | 1.03k | if ((defaults->maxAttrs >= XML_MAX_ATTRS) || |
1077 | 1.03k | ((size_t) defaults->maxAttrs > |
1078 | 1.03k | SIZE_MAX / 2 / sizeof(temp[0]) - sizeof(*defaults))) |
1079 | 0 | goto mem_error; |
1080 | | |
1081 | 1.03k | if (defaults->maxAttrs > XML_MAX_ATTRS / 2) |
1082 | 0 | newSize = XML_MAX_ATTRS; |
1083 | 1.03k | else |
1084 | 1.03k | newSize = defaults->maxAttrs * 2; |
1085 | 1.03k | } |
1086 | 5.92k | temp = xmlRealloc(defaults, |
1087 | 5.92k | sizeof(*defaults) + newSize * sizeof(xmlDefAttr)); |
1088 | 5.92k | if (temp == NULL) |
1089 | 11 | goto mem_error; |
1090 | 5.91k | if (defaults == NULL) |
1091 | 4.88k | temp->nbAttrs = 0; |
1092 | 5.91k | temp->maxAttrs = newSize; |
1093 | 5.91k | defaults = temp; |
1094 | 5.91k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name.name, prefix.name, |
1095 | 5.91k | defaults, NULL) < 0) { |
1096 | 1 | xmlFree(defaults); |
1097 | 1 | goto mem_error; |
1098 | 1 | } |
1099 | 5.91k | } |
1100 | | |
1101 | | /* |
1102 | | * Split the attribute name into prefix:localname , the string found |
1103 | | * are within the DTD and hen not associated to namespace names. |
1104 | | */ |
1105 | 17.8k | localname = xmlSplitQName3(fullattr, &len); |
1106 | 17.8k | if (localname == NULL) { |
1107 | 13.7k | name = xmlDictLookupHashed(ctxt->dict, fullattr, -1); |
1108 | 13.7k | prefix.name = NULL; |
1109 | 13.7k | } else { |
1110 | 4.07k | name = xmlDictLookupHashed(ctxt->dict, localname, -1); |
1111 | 4.07k | prefix = xmlDictLookupHashed(ctxt->dict, fullattr, len); |
1112 | 4.07k | if (prefix.name == NULL) |
1113 | 3 | goto mem_error; |
1114 | 4.07k | } |
1115 | 17.8k | if (name.name == NULL) |
1116 | 3 | goto mem_error; |
1117 | | |
1118 | | /* intern the string and precompute the end */ |
1119 | 17.8k | len = strlen((const char *) value); |
1120 | 17.8k | hvalue = xmlDictLookupHashed(ctxt->dict, value, len); |
1121 | 17.8k | if (hvalue.name == NULL) |
1122 | 5 | goto mem_error; |
1123 | | |
1124 | 17.8k | expandedSize = strlen((const char *) name.name); |
1125 | 17.8k | if (prefix.name != NULL) |
1126 | 4.06k | expandedSize += strlen((const char *) prefix.name); |
1127 | 17.8k | expandedSize += len; |
1128 | | |
1129 | 17.8k | attr = &defaults->attrs[defaults->nbAttrs++]; |
1130 | 17.8k | attr->name = name; |
1131 | 17.8k | attr->prefix = prefix; |
1132 | 17.8k | attr->value = hvalue; |
1133 | 17.8k | attr->valueEnd = hvalue.name + len; |
1134 | 17.8k | attr->external = PARSER_EXTERNAL(ctxt); |
1135 | 17.8k | attr->expandedSize = expandedSize; |
1136 | | |
1137 | 17.8k | return; |
1138 | | |
1139 | 38 | mem_error: |
1140 | 38 | xmlErrMemory(ctxt); |
1141 | 38 | } |
1142 | | |
1143 | | /** |
1144 | | * Register this attribute type |
1145 | | * |
1146 | | * @param ctxt an XML parser context |
1147 | | * @param fullname the element fullname |
1148 | | * @param fullattr the attribute fullname |
1149 | | * @param type the attribute type |
1150 | | */ |
1151 | | static void |
1152 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1153 | | const xmlChar *fullname, |
1154 | | const xmlChar *fullattr, |
1155 | | int type) |
1156 | 81.5k | { |
1157 | 81.5k | if (ctxt->attsSpecial == NULL) { |
1158 | 5.93k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1159 | 5.93k | if (ctxt->attsSpecial == NULL) |
1160 | 13 | goto mem_error; |
1161 | 5.93k | } |
1162 | | |
1163 | 81.5k | if (PARSER_EXTERNAL(ctxt)) |
1164 | 42.4k | type |= XML_SPECIAL_EXTERNAL; |
1165 | | |
1166 | 81.5k | if (xmlHashAdd2(ctxt->attsSpecial, fullname, fullattr, |
1167 | 81.5k | XML_INT_TO_PTR(type)) < 0) |
1168 | 5 | goto mem_error; |
1169 | 81.5k | return; |
1170 | | |
1171 | 81.5k | mem_error: |
1172 | 18 | xmlErrMemory(ctxt); |
1173 | 18 | } |
1174 | | |
1175 | | /** |
1176 | | * Removes CDATA attributes from the special attribute table |
1177 | | */ |
1178 | | static void |
1179 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1180 | | const xmlChar *fullname, const xmlChar *fullattr, |
1181 | 46.3k | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1182 | 46.3k | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1183 | | |
1184 | 46.3k | if (XML_PTR_TO_INT(payload) == XML_ATTRIBUTE_CDATA) { |
1185 | 4.79k | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1186 | 4.79k | } |
1187 | 46.3k | } |
1188 | | |
1189 | | /** |
1190 | | * Trim the list of attributes defined to remove all those of type |
1191 | | * CDATA as they are not special. This call should be done when finishing |
1192 | | * to parse the DTD and before starting to parse the document root. |
1193 | | * |
1194 | | * @param ctxt an XML parser context |
1195 | | */ |
1196 | | static void |
1197 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1198 | 26.4k | { |
1199 | 26.4k | if (ctxt->attsSpecial == NULL) |
1200 | 20.5k | return; |
1201 | | |
1202 | 5.91k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1203 | | |
1204 | 5.91k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1205 | 166 | xmlHashFree(ctxt->attsSpecial, NULL); |
1206 | 166 | ctxt->attsSpecial = NULL; |
1207 | 166 | } |
1208 | 5.91k | } |
1209 | | |
1210 | | /** |
1211 | | * Checks that the value conforms to the LanguageID production: |
1212 | | * |
1213 | | * @deprecated Internal function, do not use. |
1214 | | * |
1215 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1216 | | * the XML Second edition. |
1217 | | * |
1218 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1219 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1220 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1221 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1222 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1223 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1224 | | * |
1225 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1226 | | * |
1227 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1228 | | * |
1229 | | * langtag = language |
1230 | | * ["-" script] |
1231 | | * ["-" region] |
1232 | | * *("-" variant) |
1233 | | * *("-" extension) |
1234 | | * ["-" privateuse] |
1235 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1236 | | * ["-" extlang] ; sometimes followed by |
1237 | | * ; extended language subtags |
1238 | | * / 4ALPHA ; or reserved for future use |
1239 | | * / 5*8ALPHA ; or registered language subtag |
1240 | | * |
1241 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1242 | | * *2("-" 3ALPHA) ; permanently reserved |
1243 | | * |
1244 | | * script = 4ALPHA ; ISO 15924 code |
1245 | | * |
1246 | | * region = 2ALPHA ; ISO 3166-1 code |
1247 | | * / 3DIGIT ; UN M.49 code |
1248 | | * |
1249 | | * variant = 5*8alphanum ; registered variants |
1250 | | * / (DIGIT 3alphanum) |
1251 | | * |
1252 | | * extension = singleton 1*("-" (2*8alphanum)) |
1253 | | * |
1254 | | * ; Single alphanumerics |
1255 | | * ; "x" reserved for private use |
1256 | | * singleton = DIGIT ; 0 - 9 |
1257 | | * / %x41-57 ; A - W |
1258 | | * / %x59-5A ; Y - Z |
1259 | | * / %x61-77 ; a - w |
1260 | | * / %x79-7A ; y - z |
1261 | | * |
1262 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1263 | | * The parser below doesn't try to cope with extension or privateuse |
1264 | | * that could be added but that's not interoperable anyway |
1265 | | * |
1266 | | * @param lang pointer to the string value |
1267 | | * @returns 1 if correct 0 otherwise |
1268 | | **/ |
1269 | | int |
1270 | | xmlCheckLanguageID(const xmlChar * lang) |
1271 | 11.5k | { |
1272 | 11.5k | const xmlChar *cur = lang, *nxt; |
1273 | | |
1274 | 11.5k | if (cur == NULL) |
1275 | 0 | return (0); |
1276 | 11.5k | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1277 | 11.5k | ((cur[0] == 'I') && (cur[1] == '-')) || |
1278 | 11.5k | ((cur[0] == 'x') && (cur[1] == '-')) || |
1279 | 11.5k | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1280 | | /* |
1281 | | * Still allow IANA code and user code which were coming |
1282 | | * from the previous version of the XML-1.0 specification |
1283 | | * it's deprecated but we should not fail |
1284 | | */ |
1285 | 1.35k | cur += 2; |
1286 | 3.73k | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1287 | 3.73k | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1288 | 2.37k | cur++; |
1289 | 1.35k | return(cur[0] == 0); |
1290 | 1.35k | } |
1291 | 10.2k | nxt = cur; |
1292 | 35.4k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1293 | 35.4k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1294 | 25.2k | nxt++; |
1295 | 10.2k | if (nxt - cur >= 4) { |
1296 | | /* |
1297 | | * Reserved |
1298 | | */ |
1299 | 1.06k | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1300 | 856 | return(0); |
1301 | 211 | return(1); |
1302 | 1.06k | } |
1303 | 9.15k | if (nxt - cur < 2) |
1304 | 770 | return(0); |
1305 | | /* we got an ISO 639 code */ |
1306 | 8.38k | if (nxt[0] == 0) |
1307 | 431 | return(1); |
1308 | 7.95k | if (nxt[0] != '-') |
1309 | 484 | return(0); |
1310 | | |
1311 | 7.46k | nxt++; |
1312 | 7.46k | cur = nxt; |
1313 | | /* now we can have extlang or script or region or variant */ |
1314 | 7.46k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1315 | 810 | goto region_m49; |
1316 | | |
1317 | 37.4k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1318 | 37.4k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1319 | 30.7k | nxt++; |
1320 | 6.65k | if (nxt - cur == 4) |
1321 | 1.88k | goto script; |
1322 | 4.76k | if (nxt - cur == 2) |
1323 | 984 | goto region; |
1324 | 3.78k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1325 | 911 | goto variant; |
1326 | 2.87k | if (nxt - cur != 3) |
1327 | 457 | return(0); |
1328 | | /* we parsed an extlang */ |
1329 | 2.41k | if (nxt[0] == 0) |
1330 | 294 | return(1); |
1331 | 2.12k | if (nxt[0] != '-') |
1332 | 253 | return(0); |
1333 | | |
1334 | 1.87k | nxt++; |
1335 | 1.87k | cur = nxt; |
1336 | | /* now we can have script or region or variant */ |
1337 | 1.87k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1338 | 219 | goto region_m49; |
1339 | | |
1340 | 9.26k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1341 | 9.26k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1342 | 7.61k | nxt++; |
1343 | 1.65k | if (nxt - cur == 2) |
1344 | 275 | goto region; |
1345 | 1.37k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1346 | 340 | goto variant; |
1347 | 1.03k | if (nxt - cur != 4) |
1348 | 486 | return(0); |
1349 | | /* we parsed a script */ |
1350 | 2.43k | script: |
1351 | 2.43k | if (nxt[0] == 0) |
1352 | 477 | return(1); |
1353 | 1.96k | if (nxt[0] != '-') |
1354 | 266 | return(0); |
1355 | | |
1356 | 1.69k | nxt++; |
1357 | 1.69k | cur = nxt; |
1358 | | /* now we can have region or variant */ |
1359 | 1.69k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1360 | 407 | goto region_m49; |
1361 | | |
1362 | 8.10k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1363 | 8.10k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1364 | 6.81k | nxt++; |
1365 | | |
1366 | 1.28k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1367 | 340 | goto variant; |
1368 | 948 | if (nxt - cur != 2) |
1369 | 617 | return(0); |
1370 | | /* we parsed a region */ |
1371 | 1.82k | region: |
1372 | 1.82k | if (nxt[0] == 0) |
1373 | 328 | return(1); |
1374 | 1.49k | if (nxt[0] != '-') |
1375 | 848 | return(0); |
1376 | | |
1377 | 647 | nxt++; |
1378 | 647 | cur = nxt; |
1379 | | /* now we can just have a variant */ |
1380 | 4.43k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1381 | 4.43k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1382 | 3.78k | nxt++; |
1383 | | |
1384 | 647 | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1385 | 439 | return(0); |
1386 | | |
1387 | | /* we parsed a variant */ |
1388 | 1.79k | variant: |
1389 | 1.79k | if (nxt[0] == 0) |
1390 | 256 | return(1); |
1391 | 1.54k | if (nxt[0] != '-') |
1392 | 1.08k | return(0); |
1393 | | /* extensions and private use subtags not checked */ |
1394 | 462 | return (1); |
1395 | | |
1396 | 1.43k | region_m49: |
1397 | 1.43k | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1398 | 1.43k | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1399 | 233 | nxt += 3; |
1400 | 233 | goto region; |
1401 | 233 | } |
1402 | 1.20k | return(0); |
1403 | 1.43k | } |
1404 | | |
1405 | | /************************************************************************ |
1406 | | * * |
1407 | | * Parser stacks related functions and macros * |
1408 | | * * |
1409 | | ************************************************************************/ |
1410 | | |
1411 | | static xmlChar * |
1412 | | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar **str); |
1413 | | |
1414 | | /** |
1415 | | * Create a new namespace database. |
1416 | | * |
1417 | | * @returns the new obejct. |
1418 | | */ |
1419 | | xmlParserNsData * |
1420 | 49.6k | xmlParserNsCreate(void) { |
1421 | 49.6k | xmlParserNsData *nsdb = xmlMalloc(sizeof(*nsdb)); |
1422 | | |
1423 | 49.6k | if (nsdb == NULL) |
1424 | 10 | return(NULL); |
1425 | 49.6k | memset(nsdb, 0, sizeof(*nsdb)); |
1426 | 49.6k | nsdb->defaultNsIndex = INT_MAX; |
1427 | | |
1428 | 49.6k | return(nsdb); |
1429 | 49.6k | } |
1430 | | |
1431 | | /** |
1432 | | * Free a namespace database. |
1433 | | * |
1434 | | * @param nsdb namespace database |
1435 | | */ |
1436 | | void |
1437 | 49.6k | xmlParserNsFree(xmlParserNsData *nsdb) { |
1438 | 49.6k | if (nsdb == NULL) |
1439 | 0 | return; |
1440 | | |
1441 | 49.6k | xmlFree(nsdb->extra); |
1442 | 49.6k | xmlFree(nsdb->hash); |
1443 | 49.6k | xmlFree(nsdb); |
1444 | 49.6k | } |
1445 | | |
1446 | | /** |
1447 | | * Reset a namespace database. |
1448 | | * |
1449 | | * @param nsdb namespace database |
1450 | | */ |
1451 | | static void |
1452 | 24.8k | xmlParserNsReset(xmlParserNsData *nsdb) { |
1453 | 24.8k | if (nsdb == NULL) |
1454 | 0 | return; |
1455 | | |
1456 | 24.8k | nsdb->hashElems = 0; |
1457 | 24.8k | nsdb->elementId = 0; |
1458 | 24.8k | nsdb->defaultNsIndex = INT_MAX; |
1459 | | |
1460 | 24.8k | if (nsdb->hash) |
1461 | 0 | memset(nsdb->hash, 0, nsdb->hashSize * sizeof(nsdb->hash[0])); |
1462 | 24.8k | } |
1463 | | |
1464 | | /** |
1465 | | * Signal that a new element has started. |
1466 | | * |
1467 | | * @param nsdb namespace database |
1468 | | * @returns 0 on success, -1 if the element counter overflowed. |
1469 | | */ |
1470 | | static int |
1471 | 1.02M | xmlParserNsStartElement(xmlParserNsData *nsdb) { |
1472 | 1.02M | if (nsdb->elementId == UINT_MAX) |
1473 | 0 | return(-1); |
1474 | 1.02M | nsdb->elementId++; |
1475 | | |
1476 | 1.02M | return(0); |
1477 | 1.02M | } |
1478 | | |
1479 | | /** |
1480 | | * Lookup namespace with given prefix. If `bucketPtr` is non-NULL, it will |
1481 | | * be set to the matching bucket, or the first empty bucket if no match |
1482 | | * was found. |
1483 | | * |
1484 | | * @param ctxt parser context |
1485 | | * @param prefix namespace prefix |
1486 | | * @param bucketPtr optional bucket (return value) |
1487 | | * @returns the namespace index on success, INT_MAX if no namespace was |
1488 | | * found. |
1489 | | */ |
1490 | | static int |
1491 | | xmlParserNsLookup(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1492 | 1.70M | xmlParserNsBucket **bucketPtr) { |
1493 | 1.70M | xmlParserNsBucket *bucket, *tombstone; |
1494 | 1.70M | unsigned index, hashValue; |
1495 | | |
1496 | 1.70M | if (prefix->name == NULL) |
1497 | 781k | return(ctxt->nsdb->defaultNsIndex); |
1498 | | |
1499 | 927k | if (ctxt->nsdb->hashSize == 0) |
1500 | 23.4k | return(INT_MAX); |
1501 | | |
1502 | 903k | hashValue = prefix->hashValue; |
1503 | 903k | index = hashValue & (ctxt->nsdb->hashSize - 1); |
1504 | 903k | bucket = &ctxt->nsdb->hash[index]; |
1505 | 903k | tombstone = NULL; |
1506 | | |
1507 | 1.16M | while (bucket->hashValue) { |
1508 | 1.06M | if (bucket->index == INT_MAX) { |
1509 | 55.2k | if (tombstone == NULL) |
1510 | 43.9k | tombstone = bucket; |
1511 | 1.01M | } else if (bucket->hashValue == hashValue) { |
1512 | 804k | if (ctxt->nsTab[bucket->index * 2] == prefix->name) { |
1513 | 804k | if (bucketPtr != NULL) |
1514 | 530k | *bucketPtr = bucket; |
1515 | 804k | return(bucket->index); |
1516 | 804k | } |
1517 | 804k | } |
1518 | | |
1519 | 262k | index++; |
1520 | 262k | bucket++; |
1521 | 262k | if (index == ctxt->nsdb->hashSize) { |
1522 | 16.8k | index = 0; |
1523 | 16.8k | bucket = ctxt->nsdb->hash; |
1524 | 16.8k | } |
1525 | 262k | } |
1526 | | |
1527 | 99.3k | if (bucketPtr != NULL) |
1528 | 16.9k | *bucketPtr = tombstone ? tombstone : bucket; |
1529 | 99.3k | return(INT_MAX); |
1530 | 903k | } |
1531 | | |
1532 | | /** |
1533 | | * Lookup namespace URI with given prefix. |
1534 | | * |
1535 | | * @param ctxt parser context |
1536 | | * @param prefix namespace prefix |
1537 | | * @returns the namespace URI on success, NULL if no namespace was found. |
1538 | | */ |
1539 | | static const xmlChar * |
1540 | 673k | xmlParserNsLookupUri(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix) { |
1541 | 673k | const xmlChar *ret; |
1542 | 673k | int nsIndex; |
1543 | | |
1544 | 673k | if (prefix->name == ctxt->str_xml) |
1545 | 680 | return(ctxt->str_xml_ns); |
1546 | | |
1547 | | /* |
1548 | | * minNsIndex is used when building an entity tree. We must |
1549 | | * ignore namespaces declared outside the entity. |
1550 | | */ |
1551 | 672k | nsIndex = xmlParserNsLookup(ctxt, prefix, NULL); |
1552 | 672k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1553 | 460k | return(NULL); |
1554 | | |
1555 | 212k | ret = ctxt->nsTab[nsIndex * 2 + 1]; |
1556 | 212k | if (ret[0] == 0) |
1557 | 13.2k | ret = NULL; |
1558 | 212k | return(ret); |
1559 | 672k | } |
1560 | | |
1561 | | /** |
1562 | | * Lookup extra data for the given prefix. This returns data stored |
1563 | | * with xmlParserNsUdpateSax. |
1564 | | * |
1565 | | * @param ctxt parser context |
1566 | | * @param prefix namespace prefix |
1567 | | * @returns the data on success, NULL if no namespace was found. |
1568 | | */ |
1569 | | void * |
1570 | 197k | xmlParserNsLookupSax(xmlParserCtxt *ctxt, const xmlChar *prefix) { |
1571 | 197k | xmlHashedString hprefix; |
1572 | 197k | int nsIndex; |
1573 | | |
1574 | 197k | if (prefix == ctxt->str_xml) |
1575 | 39.0k | return(NULL); |
1576 | | |
1577 | 158k | hprefix.name = prefix; |
1578 | 158k | if (prefix != NULL) |
1579 | 31.6k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1580 | 126k | else |
1581 | 126k | hprefix.hashValue = 0; |
1582 | 158k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1583 | 158k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1584 | 0 | return(NULL); |
1585 | | |
1586 | 158k | return(ctxt->nsdb->extra[nsIndex].saxData); |
1587 | 158k | } |
1588 | | |
1589 | | /** |
1590 | | * Sets or updates extra data for the given prefix. This value will be |
1591 | | * returned by xmlParserNsLookupSax as long as the namespace with the |
1592 | | * given prefix is in scope. |
1593 | | * |
1594 | | * @param ctxt parser context |
1595 | | * @param prefix namespace prefix |
1596 | | * @param saxData extra data for SAX handler |
1597 | | * @returns the data on success, NULL if no namespace was found. |
1598 | | */ |
1599 | | int |
1600 | | xmlParserNsUpdateSax(xmlParserCtxt *ctxt, const xmlChar *prefix, |
1601 | 242k | void *saxData) { |
1602 | 242k | xmlHashedString hprefix; |
1603 | 242k | int nsIndex; |
1604 | | |
1605 | 242k | if (prefix == ctxt->str_xml) |
1606 | 0 | return(-1); |
1607 | | |
1608 | 242k | hprefix.name = prefix; |
1609 | 242k | if (prefix != NULL) |
1610 | 197k | hprefix.hashValue = xmlDictComputeHash(ctxt->dict, prefix); |
1611 | 45.0k | else |
1612 | 45.0k | hprefix.hashValue = 0; |
1613 | 242k | nsIndex = xmlParserNsLookup(ctxt, &hprefix, NULL); |
1614 | 242k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) |
1615 | 0 | return(-1); |
1616 | | |
1617 | 242k | ctxt->nsdb->extra[nsIndex].saxData = saxData; |
1618 | 242k | return(0); |
1619 | 242k | } |
1620 | | |
1621 | | /** |
1622 | | * Grows the namespace tables. |
1623 | | * |
1624 | | * @param ctxt parser context |
1625 | | * @returns 0 on success, -1 if a memory allocation failed. |
1626 | | */ |
1627 | | static int |
1628 | 20.4k | xmlParserNsGrow(xmlParserCtxtPtr ctxt) { |
1629 | 20.4k | const xmlChar **table; |
1630 | 20.4k | xmlParserNsExtra *extra; |
1631 | 20.4k | int newSize; |
1632 | | |
1633 | 20.4k | newSize = xmlGrowCapacity(ctxt->nsMax, |
1634 | 20.4k | sizeof(table[0]) + sizeof(extra[0]), |
1635 | 20.4k | 16, XML_MAX_ITEMS); |
1636 | 20.4k | if (newSize < 0) |
1637 | 0 | goto error; |
1638 | | |
1639 | 20.4k | table = xmlRealloc(ctxt->nsTab, 2 * newSize * sizeof(table[0])); |
1640 | 20.4k | if (table == NULL) |
1641 | 38 | goto error; |
1642 | 20.4k | ctxt->nsTab = table; |
1643 | | |
1644 | 20.4k | extra = xmlRealloc(ctxt->nsdb->extra, newSize * sizeof(extra[0])); |
1645 | 20.4k | if (extra == NULL) |
1646 | 35 | goto error; |
1647 | 20.3k | ctxt->nsdb->extra = extra; |
1648 | | |
1649 | 20.3k | ctxt->nsMax = newSize; |
1650 | 20.3k | return(0); |
1651 | | |
1652 | 73 | error: |
1653 | 73 | xmlErrMemory(ctxt); |
1654 | 73 | return(-1); |
1655 | 20.4k | } |
1656 | | |
1657 | | /** |
1658 | | * Push a new namespace on the table. |
1659 | | * |
1660 | | * @param ctxt parser context |
1661 | | * @param prefix prefix with hash value |
1662 | | * @param uri uri with hash value |
1663 | | * @param saxData extra data for SAX handler |
1664 | | * @param defAttr whether the namespace comes from a default attribute |
1665 | | * @returns 1 if the namespace was pushed, 0 if the namespace was ignored, |
1666 | | * -1 if a memory allocation failed. |
1667 | | */ |
1668 | | static int |
1669 | | xmlParserNsPush(xmlParserCtxtPtr ctxt, const xmlHashedString *prefix, |
1670 | 355k | const xmlHashedString *uri, void *saxData, int defAttr) { |
1671 | 355k | xmlParserNsBucket *bucket = NULL; |
1672 | 355k | xmlParserNsExtra *extra; |
1673 | 355k | const xmlChar **ns; |
1674 | 355k | unsigned hashValue, nsIndex, oldIndex; |
1675 | | |
1676 | 355k | if ((prefix != NULL) && (prefix->name == ctxt->str_xml)) |
1677 | 194 | return(0); |
1678 | | |
1679 | 355k | if ((ctxt->nsNr >= ctxt->nsMax) && (xmlParserNsGrow(ctxt) < 0)) { |
1680 | 73 | xmlErrMemory(ctxt); |
1681 | 73 | return(-1); |
1682 | 73 | } |
1683 | | |
1684 | | /* |
1685 | | * Default namespace and 'xml' namespace |
1686 | | */ |
1687 | 354k | if ((prefix == NULL) || (prefix->name == NULL)) { |
1688 | 63.0k | oldIndex = ctxt->nsdb->defaultNsIndex; |
1689 | | |
1690 | 63.0k | if (oldIndex != INT_MAX) { |
1691 | 53.1k | extra = &ctxt->nsdb->extra[oldIndex]; |
1692 | | |
1693 | 53.1k | if (extra->elementId == ctxt->nsdb->elementId) { |
1694 | 981 | if (defAttr == 0) |
1695 | 763 | xmlErrAttributeDup(ctxt, NULL, BAD_CAST "xmlns"); |
1696 | 981 | return(0); |
1697 | 981 | } |
1698 | | |
1699 | 52.1k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1700 | 52.1k | (uri->name == ctxt->nsTab[oldIndex * 2 + 1])) |
1701 | 10.0k | return(0); |
1702 | 52.1k | } |
1703 | | |
1704 | 51.9k | ctxt->nsdb->defaultNsIndex = ctxt->nsNr; |
1705 | 51.9k | goto populate_entry; |
1706 | 63.0k | } |
1707 | | |
1708 | | /* |
1709 | | * Hash table lookup |
1710 | | */ |
1711 | 291k | oldIndex = xmlParserNsLookup(ctxt, prefix, &bucket); |
1712 | 291k | if (oldIndex != INT_MAX) { |
1713 | 271k | extra = &ctxt->nsdb->extra[oldIndex]; |
1714 | | |
1715 | | /* |
1716 | | * Check for duplicate definitions on the same element. |
1717 | | */ |
1718 | 271k | if (extra->elementId == ctxt->nsdb->elementId) { |
1719 | 755 | if (defAttr == 0) |
1720 | 515 | xmlErrAttributeDup(ctxt, BAD_CAST "xmlns", prefix->name); |
1721 | 755 | return(0); |
1722 | 755 | } |
1723 | | |
1724 | 271k | if ((ctxt->options & XML_PARSE_NSCLEAN) && |
1725 | 271k | (uri->name == ctxt->nsTab[bucket->index * 2 + 1])) |
1726 | 12.1k | return(0); |
1727 | | |
1728 | 259k | bucket->index = ctxt->nsNr; |
1729 | 259k | goto populate_entry; |
1730 | 271k | } |
1731 | | |
1732 | | /* |
1733 | | * Insert new bucket |
1734 | | */ |
1735 | | |
1736 | 20.0k | hashValue = prefix->hashValue; |
1737 | | |
1738 | | /* |
1739 | | * Grow hash table, 50% fill factor |
1740 | | */ |
1741 | 20.0k | if (ctxt->nsdb->hashElems + 1 > ctxt->nsdb->hashSize / 2) { |
1742 | 3.91k | xmlParserNsBucket *newHash; |
1743 | 3.91k | unsigned newSize, i, index; |
1744 | | |
1745 | 3.91k | if (ctxt->nsdb->hashSize > UINT_MAX / 2) { |
1746 | 0 | xmlErrMemory(ctxt); |
1747 | 0 | return(-1); |
1748 | 0 | } |
1749 | 3.91k | newSize = ctxt->nsdb->hashSize ? ctxt->nsdb->hashSize * 2 : 16; |
1750 | 3.91k | newHash = xmlMalloc(newSize * sizeof(newHash[0])); |
1751 | 3.91k | if (newHash == NULL) { |
1752 | 7 | xmlErrMemory(ctxt); |
1753 | 7 | return(-1); |
1754 | 7 | } |
1755 | 3.91k | memset(newHash, 0, newSize * sizeof(newHash[0])); |
1756 | | |
1757 | 44.8k | for (i = 0; i < ctxt->nsdb->hashSize; i++) { |
1758 | 40.9k | unsigned hv = ctxt->nsdb->hash[i].hashValue; |
1759 | 40.9k | unsigned newIndex; |
1760 | | |
1761 | 40.9k | if ((hv == 0) || (ctxt->nsdb->hash[i].index == INT_MAX)) |
1762 | 38.5k | continue; |
1763 | 2.39k | newIndex = hv & (newSize - 1); |
1764 | | |
1765 | 3.46k | while (newHash[newIndex].hashValue != 0) { |
1766 | 1.07k | newIndex++; |
1767 | 1.07k | if (newIndex == newSize) |
1768 | 46 | newIndex = 0; |
1769 | 1.07k | } |
1770 | | |
1771 | 2.39k | newHash[newIndex] = ctxt->nsdb->hash[i]; |
1772 | 2.39k | } |
1773 | | |
1774 | 3.91k | xmlFree(ctxt->nsdb->hash); |
1775 | 3.91k | ctxt->nsdb->hash = newHash; |
1776 | 3.91k | ctxt->nsdb->hashSize = newSize; |
1777 | | |
1778 | | /* |
1779 | | * Relookup |
1780 | | */ |
1781 | 3.91k | index = hashValue & (newSize - 1); |
1782 | | |
1783 | 4.29k | while (newHash[index].hashValue != 0) { |
1784 | 384 | index++; |
1785 | 384 | if (index == newSize) |
1786 | 26 | index = 0; |
1787 | 384 | } |
1788 | | |
1789 | 3.91k | bucket = &newHash[index]; |
1790 | 3.91k | } |
1791 | | |
1792 | 19.9k | bucket->hashValue = hashValue; |
1793 | 19.9k | bucket->index = ctxt->nsNr; |
1794 | 19.9k | ctxt->nsdb->hashElems++; |
1795 | 19.9k | oldIndex = INT_MAX; |
1796 | | |
1797 | 330k | populate_entry: |
1798 | 330k | nsIndex = ctxt->nsNr; |
1799 | | |
1800 | 330k | ns = &ctxt->nsTab[nsIndex * 2]; |
1801 | 330k | ns[0] = prefix ? prefix->name : NULL; |
1802 | 330k | ns[1] = uri->name; |
1803 | | |
1804 | 330k | extra = &ctxt->nsdb->extra[nsIndex]; |
1805 | 330k | extra->saxData = saxData; |
1806 | 330k | extra->prefixHashValue = prefix ? prefix->hashValue : 0; |
1807 | 330k | extra->uriHashValue = uri->hashValue; |
1808 | 330k | extra->elementId = ctxt->nsdb->elementId; |
1809 | 330k | extra->oldIndex = oldIndex; |
1810 | | |
1811 | 330k | ctxt->nsNr++; |
1812 | | |
1813 | 330k | return(1); |
1814 | 19.9k | } |
1815 | | |
1816 | | /** |
1817 | | * Pops the top `nr` namespaces and restores the hash table. |
1818 | | * |
1819 | | * @param ctxt an XML parser context |
1820 | | * @param nr the number to pop |
1821 | | * @returns the number of namespaces popped. |
1822 | | */ |
1823 | | static int |
1824 | | xmlParserNsPop(xmlParserCtxtPtr ctxt, int nr) |
1825 | 97.3k | { |
1826 | 97.3k | int i; |
1827 | | |
1828 | | /* assert(nr <= ctxt->nsNr); */ |
1829 | | |
1830 | 401k | for (i = ctxt->nsNr - 1; i >= ctxt->nsNr - nr; i--) { |
1831 | 304k | const xmlChar *prefix = ctxt->nsTab[i * 2]; |
1832 | 304k | xmlParserNsExtra *extra = &ctxt->nsdb->extra[i]; |
1833 | | |
1834 | 304k | if (prefix == NULL) { |
1835 | 45.5k | ctxt->nsdb->defaultNsIndex = extra->oldIndex; |
1836 | 258k | } else { |
1837 | 258k | xmlHashedString hprefix; |
1838 | 258k | xmlParserNsBucket *bucket = NULL; |
1839 | | |
1840 | 258k | hprefix.name = prefix; |
1841 | 258k | hprefix.hashValue = extra->prefixHashValue; |
1842 | 258k | xmlParserNsLookup(ctxt, &hprefix, &bucket); |
1843 | | /* assert(bucket && bucket->hashValue); */ |
1844 | 258k | bucket->index = extra->oldIndex; |
1845 | 258k | } |
1846 | 304k | } |
1847 | | |
1848 | 97.3k | ctxt->nsNr -= nr; |
1849 | 97.3k | return(nr); |
1850 | 97.3k | } |
1851 | | |
1852 | | static int |
1853 | 13.6k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt) { |
1854 | 13.6k | const xmlChar **atts; |
1855 | 13.6k | unsigned *attallocs; |
1856 | 13.6k | int newSize; |
1857 | | |
1858 | 13.6k | newSize = xmlGrowCapacity(ctxt->maxatts / 5, |
1859 | 13.6k | sizeof(atts[0]) * 5 + sizeof(attallocs[0]), |
1860 | 13.6k | 10, XML_MAX_ATTRS); |
1861 | 13.6k | if (newSize < 0) { |
1862 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
1863 | 0 | "Maximum number of attributes exceeded"); |
1864 | 0 | return(-1); |
1865 | 0 | } |
1866 | | |
1867 | 13.6k | atts = xmlRealloc(ctxt->atts, newSize * sizeof(atts[0]) * 5); |
1868 | 13.6k | if (atts == NULL) |
1869 | 13 | goto mem_error; |
1870 | 13.6k | ctxt->atts = atts; |
1871 | | |
1872 | 13.6k | attallocs = xmlRealloc(ctxt->attallocs, |
1873 | 13.6k | newSize * sizeof(attallocs[0])); |
1874 | 13.6k | if (attallocs == NULL) |
1875 | 8 | goto mem_error; |
1876 | 13.6k | ctxt->attallocs = attallocs; |
1877 | | |
1878 | 13.6k | ctxt->maxatts = newSize * 5; |
1879 | | |
1880 | 13.6k | return(0); |
1881 | | |
1882 | 21 | mem_error: |
1883 | 21 | xmlErrMemory(ctxt); |
1884 | 21 | return(-1); |
1885 | 13.6k | } |
1886 | | |
1887 | | /** |
1888 | | * Pushes a new parser input on top of the input stack |
1889 | | * |
1890 | | * @param ctxt an XML parser context |
1891 | | * @param value the parser input |
1892 | | * @returns -1 in case of error, the index in the stack otherwise |
1893 | | */ |
1894 | | int |
1895 | | xmlCtxtPushInput(xmlParserCtxt *ctxt, xmlParserInput *value) |
1896 | 226k | { |
1897 | 226k | char *directory = NULL; |
1898 | 226k | int maxDepth; |
1899 | | |
1900 | 226k | if ((ctxt == NULL) || (value == NULL)) |
1901 | 4.73k | return(-1); |
1902 | | |
1903 | 221k | maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
1904 | | |
1905 | 221k | if (ctxt->inputNr >= ctxt->inputMax) { |
1906 | 11.1k | xmlParserInputPtr *tmp; |
1907 | 11.1k | int newSize; |
1908 | | |
1909 | 11.1k | newSize = xmlGrowCapacity(ctxt->inputMax, sizeof(tmp[0]), |
1910 | 11.1k | 5, maxDepth); |
1911 | 11.1k | if (newSize < 0) { |
1912 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
1913 | 3 | "Maximum entity nesting depth exceeded"); |
1914 | 3 | return(-1); |
1915 | 3 | } |
1916 | 11.1k | tmp = xmlRealloc(ctxt->inputTab, newSize * sizeof(tmp[0])); |
1917 | 11.1k | if (tmp == NULL) { |
1918 | 51 | xmlErrMemory(ctxt); |
1919 | 51 | return(-1); |
1920 | 51 | } |
1921 | 11.1k | ctxt->inputTab = tmp; |
1922 | 11.1k | ctxt->inputMax = newSize; |
1923 | 11.1k | } |
1924 | | |
1925 | 221k | if ((ctxt->inputNr == 0) && (value->filename != NULL)) { |
1926 | 70.1k | directory = xmlParserGetDirectory(value->filename); |
1927 | 70.1k | if (directory == NULL) { |
1928 | 92 | xmlErrMemory(ctxt); |
1929 | 92 | return(-1); |
1930 | 92 | } |
1931 | 70.1k | } |
1932 | | |
1933 | 221k | if (ctxt->input_id >= INT_MAX) { |
1934 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, "Input ID overflow\n"); |
1935 | 0 | return(-1); |
1936 | 0 | } |
1937 | | |
1938 | 221k | ctxt->inputTab[ctxt->inputNr] = value; |
1939 | 221k | ctxt->input = value; |
1940 | | |
1941 | 221k | if (ctxt->inputNr == 0) { |
1942 | 70.0k | xmlFree(ctxt->directory); |
1943 | 70.0k | ctxt->directory = directory; |
1944 | 70.0k | } |
1945 | | |
1946 | | /* |
1947 | | * The input ID is unused internally, but there are entity |
1948 | | * loaders in downstream code that detect the main document |
1949 | | * by checking for "input_id == 1". |
1950 | | */ |
1951 | 221k | value->id = ctxt->input_id++; |
1952 | | |
1953 | 221k | return(ctxt->inputNr++); |
1954 | 221k | } |
1955 | | |
1956 | | /** |
1957 | | * Pops the top parser input from the input stack |
1958 | | * |
1959 | | * @param ctxt an XML parser context |
1960 | | * @returns the input just removed |
1961 | | */ |
1962 | | xmlParserInput * |
1963 | | xmlCtxtPopInput(xmlParserCtxt *ctxt) |
1964 | 343k | { |
1965 | 343k | xmlParserInputPtr ret; |
1966 | | |
1967 | 343k | if (ctxt == NULL) |
1968 | 0 | return(NULL); |
1969 | 343k | if (ctxt->inputNr <= 0) |
1970 | 124k | return (NULL); |
1971 | 219k | ctxt->inputNr--; |
1972 | 219k | if (ctxt->inputNr > 0) |
1973 | 151k | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1974 | 68.5k | else |
1975 | 68.5k | ctxt->input = NULL; |
1976 | 219k | ret = ctxt->inputTab[ctxt->inputNr]; |
1977 | 219k | ctxt->inputTab[ctxt->inputNr] = NULL; |
1978 | 219k | return (ret); |
1979 | 343k | } |
1980 | | |
1981 | | /** |
1982 | | * Pushes a new element node on top of the node stack |
1983 | | * |
1984 | | * @deprecated Internal function, do not use. |
1985 | | * |
1986 | | * @param ctxt an XML parser context |
1987 | | * @param value the element node |
1988 | | * @returns -1 in case of error, the index in the stack otherwise |
1989 | | */ |
1990 | | int |
1991 | | nodePush(xmlParserCtxt *ctxt, xmlNode *value) |
1992 | 674k | { |
1993 | 674k | if (ctxt == NULL) |
1994 | 0 | return(0); |
1995 | | |
1996 | 674k | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1997 | 41.5k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
1998 | 41.5k | xmlNodePtr *tmp; |
1999 | 41.5k | int newSize; |
2000 | | |
2001 | 41.5k | newSize = xmlGrowCapacity(ctxt->nodeMax, sizeof(tmp[0]), |
2002 | 41.5k | 10, maxDepth); |
2003 | 41.5k | if (newSize < 0) { |
2004 | 20 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
2005 | 20 | "Excessive depth in document: %d," |
2006 | 20 | " use XML_PARSE_HUGE option\n", |
2007 | 20 | ctxt->nodeNr); |
2008 | 20 | return(-1); |
2009 | 20 | } |
2010 | | |
2011 | 41.4k | tmp = xmlRealloc(ctxt->nodeTab, newSize * sizeof(tmp[0])); |
2012 | 41.4k | if (tmp == NULL) { |
2013 | 39 | xmlErrMemory(ctxt); |
2014 | 39 | return (-1); |
2015 | 39 | } |
2016 | 41.4k | ctxt->nodeTab = tmp; |
2017 | 41.4k | ctxt->nodeMax = newSize; |
2018 | 41.4k | } |
2019 | | |
2020 | 674k | ctxt->nodeTab[ctxt->nodeNr] = value; |
2021 | 674k | ctxt->node = value; |
2022 | 674k | return (ctxt->nodeNr++); |
2023 | 674k | } |
2024 | | |
2025 | | /** |
2026 | | * Pops the top element node from the node stack |
2027 | | * |
2028 | | * @deprecated Internal function, do not use. |
2029 | | * |
2030 | | * @param ctxt an XML parser context |
2031 | | * @returns the node just removed |
2032 | | */ |
2033 | | xmlNode * |
2034 | | nodePop(xmlParserCtxt *ctxt) |
2035 | 612k | { |
2036 | 612k | xmlNodePtr ret; |
2037 | | |
2038 | 612k | if (ctxt == NULL) return(NULL); |
2039 | 612k | if (ctxt->nodeNr <= 0) |
2040 | 14.4k | return (NULL); |
2041 | 598k | ctxt->nodeNr--; |
2042 | 598k | if (ctxt->nodeNr > 0) |
2043 | 591k | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
2044 | 7.22k | else |
2045 | 7.22k | ctxt->node = NULL; |
2046 | 598k | ret = ctxt->nodeTab[ctxt->nodeNr]; |
2047 | 598k | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
2048 | 598k | return (ret); |
2049 | 612k | } |
2050 | | |
2051 | | /** |
2052 | | * Pushes a new element name/prefix/URL on top of the name stack |
2053 | | * |
2054 | | * @param ctxt an XML parser context |
2055 | | * @param value the element name |
2056 | | * @param prefix the element prefix |
2057 | | * @param URI the element namespace name |
2058 | | * @param line the current line number for error messages |
2059 | | * @param nsNr the number of namespaces pushed on the namespace table |
2060 | | * @returns -1 in case of error, the index in the stack otherwise |
2061 | | */ |
2062 | | static int |
2063 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
2064 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
2065 | 708k | { |
2066 | 708k | xmlStartTag *tag; |
2067 | | |
2068 | 708k | if (ctxt->nameNr >= ctxt->nameMax) { |
2069 | 42.6k | const xmlChar **tmp; |
2070 | 42.6k | xmlStartTag *tmp2; |
2071 | 42.6k | int newSize; |
2072 | | |
2073 | 42.6k | newSize = xmlGrowCapacity(ctxt->nameMax, |
2074 | 42.6k | sizeof(tmp[0]) + sizeof(tmp2[0]), |
2075 | 42.6k | 10, XML_MAX_ITEMS); |
2076 | 42.6k | if (newSize < 0) |
2077 | 0 | goto mem_error; |
2078 | | |
2079 | 42.6k | tmp = xmlRealloc(ctxt->nameTab, newSize * sizeof(tmp[0])); |
2080 | 42.6k | if (tmp == NULL) |
2081 | 25 | goto mem_error; |
2082 | 42.6k | ctxt->nameTab = tmp; |
2083 | | |
2084 | 42.6k | tmp2 = xmlRealloc(ctxt->pushTab, newSize * sizeof(tmp2[0])); |
2085 | 42.6k | if (tmp2 == NULL) |
2086 | 35 | goto mem_error; |
2087 | 42.6k | ctxt->pushTab = tmp2; |
2088 | | |
2089 | 42.6k | ctxt->nameMax = newSize; |
2090 | 665k | } else if (ctxt->pushTab == NULL) { |
2091 | 23.5k | ctxt->pushTab = xmlMalloc(ctxt->nameMax * sizeof(ctxt->pushTab[0])); |
2092 | 23.5k | if (ctxt->pushTab == NULL) |
2093 | 44 | goto mem_error; |
2094 | 23.5k | } |
2095 | 708k | ctxt->nameTab[ctxt->nameNr] = value; |
2096 | 708k | ctxt->name = value; |
2097 | 708k | tag = &ctxt->pushTab[ctxt->nameNr]; |
2098 | 708k | tag->prefix = prefix; |
2099 | 708k | tag->URI = URI; |
2100 | 708k | tag->line = line; |
2101 | 708k | tag->nsNr = nsNr; |
2102 | 708k | return (ctxt->nameNr++); |
2103 | 104 | mem_error: |
2104 | 104 | xmlErrMemory(ctxt); |
2105 | 104 | return (-1); |
2106 | 708k | } |
2107 | | #ifdef LIBXML_PUSH_ENABLED |
2108 | | /** |
2109 | | * Pops the top element/prefix/URI name from the name stack |
2110 | | * |
2111 | | * @param ctxt an XML parser context |
2112 | | * @returns the name just removed |
2113 | | */ |
2114 | | static const xmlChar * |
2115 | | nameNsPop(xmlParserCtxtPtr ctxt) |
2116 | 6.18k | { |
2117 | 6.18k | const xmlChar *ret; |
2118 | | |
2119 | 6.18k | if (ctxt->nameNr <= 0) |
2120 | 0 | return (NULL); |
2121 | 6.18k | ctxt->nameNr--; |
2122 | 6.18k | if (ctxt->nameNr > 0) |
2123 | 5.90k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2124 | 282 | else |
2125 | 282 | ctxt->name = NULL; |
2126 | 6.18k | ret = ctxt->nameTab[ctxt->nameNr]; |
2127 | 6.18k | ctxt->nameTab[ctxt->nameNr] = NULL; |
2128 | 6.18k | return (ret); |
2129 | 6.18k | } |
2130 | | #endif /* LIBXML_PUSH_ENABLED */ |
2131 | | |
2132 | | /** |
2133 | | * Pops the top element name from the name stack |
2134 | | * |
2135 | | * @deprecated Internal function, do not use. |
2136 | | * |
2137 | | * @param ctxt an XML parser context |
2138 | | * @returns the name just removed |
2139 | | */ |
2140 | | static const xmlChar * |
2141 | | namePop(xmlParserCtxtPtr ctxt) |
2142 | 624k | { |
2143 | 624k | const xmlChar *ret; |
2144 | | |
2145 | 624k | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
2146 | 26 | return (NULL); |
2147 | 624k | ctxt->nameNr--; |
2148 | 624k | if (ctxt->nameNr > 0) |
2149 | 618k | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
2150 | 6.25k | else |
2151 | 6.25k | ctxt->name = NULL; |
2152 | 624k | ret = ctxt->nameTab[ctxt->nameNr]; |
2153 | 624k | ctxt->nameTab[ctxt->nameNr] = NULL; |
2154 | 624k | return (ret); |
2155 | 624k | } |
2156 | | |
2157 | 1.09M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
2158 | 1.09M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
2159 | 58.1k | int *tmp; |
2160 | 58.1k | int newSize; |
2161 | | |
2162 | 58.1k | newSize = xmlGrowCapacity(ctxt->spaceMax, sizeof(tmp[0]), |
2163 | 58.1k | 10, XML_MAX_ITEMS); |
2164 | 58.1k | if (newSize < 0) { |
2165 | 0 | xmlErrMemory(ctxt); |
2166 | 0 | return(-1); |
2167 | 0 | } |
2168 | | |
2169 | 58.1k | tmp = xmlRealloc(ctxt->spaceTab, newSize * sizeof(tmp[0])); |
2170 | 58.1k | if (tmp == NULL) { |
2171 | 46 | xmlErrMemory(ctxt); |
2172 | 46 | return(-1); |
2173 | 46 | } |
2174 | 58.1k | ctxt->spaceTab = tmp; |
2175 | | |
2176 | 58.1k | ctxt->spaceMax = newSize; |
2177 | 58.1k | } |
2178 | 1.09M | ctxt->spaceTab[ctxt->spaceNr] = val; |
2179 | 1.09M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
2180 | 1.09M | return(ctxt->spaceNr++); |
2181 | 1.09M | } |
2182 | | |
2183 | 1.01M | static int spacePop(xmlParserCtxtPtr ctxt) { |
2184 | 1.01M | int ret; |
2185 | 1.01M | if (ctxt->spaceNr <= 0) return(0); |
2186 | 1.01M | ctxt->spaceNr--; |
2187 | 1.01M | if (ctxt->spaceNr > 0) |
2188 | 1.01M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
2189 | 6.15k | else |
2190 | 6.15k | ctxt->space = &ctxt->spaceTab[0]; |
2191 | 1.01M | ret = ctxt->spaceTab[ctxt->spaceNr]; |
2192 | 1.01M | ctxt->spaceTab[ctxt->spaceNr] = -1; |
2193 | 1.01M | return(ret); |
2194 | 1.01M | } |
2195 | | |
2196 | | /* |
2197 | | * Macros for accessing the content. Those should be used only by the parser, |
2198 | | * and not exported. |
2199 | | * |
2200 | | * Dirty macros, i.e. one often need to make assumption on the context to |
2201 | | * use them |
2202 | | * |
2203 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
2204 | | * To be used with extreme caution since operations consuming |
2205 | | * characters may move the input buffer to a different location ! |
2206 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
2207 | | * This should be used internally by the parser |
2208 | | * only to compare to ASCII values otherwise it would break when |
2209 | | * running with UTF-8 encoding. |
2210 | | * RAW same as CUR but in the input buffer, bypass any token |
2211 | | * extraction that may have been done |
2212 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
2213 | | * to compare on ASCII based substring. |
2214 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
2215 | | * strings without newlines within the parser. |
2216 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
2217 | | * defined char within the parser. |
2218 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2219 | | * |
2220 | | * NEXT Skip to the next character, this does the proper decoding |
2221 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2222 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2223 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2224 | | * the index |
2225 | | * GROW, SHRINK handling of input buffers |
2226 | | */ |
2227 | | |
2228 | 13.2M | #define RAW (*ctxt->input->cur) |
2229 | 1.53G | #define CUR (*ctxt->input->cur) |
2230 | 12.3M | #define NXT(val) ctxt->input->cur[(val)] |
2231 | 2.40G | #define CUR_PTR ctxt->input->cur |
2232 | 2.40M | #define BASE_PTR ctxt->input->base |
2233 | | |
2234 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2235 | 11.6M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2236 | 5.87M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2237 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2238 | 11.2M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2239 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2240 | 10.5M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2241 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2242 | 9.92M | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2243 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2244 | 9.48M | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2245 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2246 | 4.63M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2247 | 4.63M | ((unsigned char *) s)[ 8 ] == c9 ) |
2248 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2249 | 3.45k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2250 | 3.45k | ((unsigned char *) s)[ 9 ] == c10 ) |
2251 | | |
2252 | 2.08M | #define SKIP(val) do { \ |
2253 | 2.08M | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2254 | 2.08M | if (*ctxt->input->cur == 0) \ |
2255 | 2.08M | xmlParserGrow(ctxt); \ |
2256 | 2.08M | } while (0) |
2257 | | |
2258 | | #define SKIPL(val) do { \ |
2259 | | int skipl; \ |
2260 | | for(skipl=0; skipl<val; skipl++) { \ |
2261 | | if (*(ctxt->input->cur) == '\n') { \ |
2262 | | ctxt->input->line++; ctxt->input->col = 1; \ |
2263 | | } else ctxt->input->col++; \ |
2264 | | ctxt->input->cur++; \ |
2265 | | } \ |
2266 | | if (*ctxt->input->cur == 0) \ |
2267 | | xmlParserGrow(ctxt); \ |
2268 | | } while (0) |
2269 | | |
2270 | | #define SHRINK \ |
2271 | 12.5M | if (!PARSER_PROGRESSIVE(ctxt)) \ |
2272 | 12.5M | xmlParserShrink(ctxt); |
2273 | | |
2274 | | #define GROW \ |
2275 | 24.8M | if ((!PARSER_PROGRESSIVE(ctxt)) && \ |
2276 | 24.8M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2277 | 2.95M | xmlParserGrow(ctxt); |
2278 | | |
2279 | 3.10M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2280 | | |
2281 | 842k | #define SKIP_BLANKS_PE xmlSkipBlankCharsPE(ctxt) |
2282 | | |
2283 | 376M | #define NEXT xmlNextChar(ctxt) |
2284 | | |
2285 | 1.40M | #define NEXT1 { \ |
2286 | 1.40M | ctxt->input->col++; \ |
2287 | 1.40M | ctxt->input->cur++; \ |
2288 | 1.40M | if (*ctxt->input->cur == 0) \ |
2289 | 1.40M | xmlParserGrow(ctxt); \ |
2290 | 1.40M | } |
2291 | | |
2292 | 1.30G | #define NEXTL(l) do { \ |
2293 | 1.30G | if (*(ctxt->input->cur) == '\n') { \ |
2294 | 14.4M | ctxt->input->line++; ctxt->input->col = 1; \ |
2295 | 1.28G | } else ctxt->input->col++; \ |
2296 | 1.30G | ctxt->input->cur += l; \ |
2297 | 1.30G | } while (0) |
2298 | | |
2299 | | #define COPY_BUF(b, i, v) \ |
2300 | 123M | if (v < 0x80) b[i++] = v; \ |
2301 | 123M | else i += xmlCopyCharMultiByte(&b[i],v) |
2302 | | |
2303 | | static int |
2304 | 120M | xmlCurrentCharRecover(xmlParserCtxtPtr ctxt, int *len) { |
2305 | 120M | int c = xmlCurrentChar(ctxt, len); |
2306 | | |
2307 | 120M | if (c == XML_INVALID_CHAR) |
2308 | 21.3M | c = 0xFFFD; /* replacement character */ |
2309 | | |
2310 | 120M | return(c); |
2311 | 120M | } |
2312 | | |
2313 | | /** |
2314 | | * Skip whitespace in the input stream. |
2315 | | * |
2316 | | * @deprecated Internal function, do not use. |
2317 | | * |
2318 | | * @param ctxt the XML parser context |
2319 | | * @returns the number of space chars skipped |
2320 | | */ |
2321 | | int |
2322 | 3.33M | xmlSkipBlankChars(xmlParserCtxt *ctxt) { |
2323 | 3.33M | const xmlChar *cur; |
2324 | 3.33M | int res = 0; |
2325 | | |
2326 | 3.33M | cur = ctxt->input->cur; |
2327 | 3.98M | while (IS_BLANK_CH(*cur)) { |
2328 | 3.98M | if (*cur == '\n') { |
2329 | 3.02M | ctxt->input->line++; ctxt->input->col = 1; |
2330 | 3.02M | } else { |
2331 | 953k | ctxt->input->col++; |
2332 | 953k | } |
2333 | 3.98M | cur++; |
2334 | 3.98M | if (res < INT_MAX) |
2335 | 3.98M | res++; |
2336 | 3.98M | if (*cur == 0) { |
2337 | 12.1k | ctxt->input->cur = cur; |
2338 | 12.1k | xmlParserGrow(ctxt); |
2339 | 12.1k | cur = ctxt->input->cur; |
2340 | 12.1k | } |
2341 | 3.98M | } |
2342 | 3.33M | ctxt->input->cur = cur; |
2343 | | |
2344 | 3.33M | if (res > 4) |
2345 | 31.7k | GROW; |
2346 | | |
2347 | 3.33M | return(res); |
2348 | 3.33M | } |
2349 | | |
2350 | | static void |
2351 | 76.4k | xmlPopPE(xmlParserCtxtPtr ctxt) { |
2352 | 76.4k | unsigned long consumed; |
2353 | 76.4k | xmlEntityPtr ent; |
2354 | | |
2355 | 76.4k | ent = ctxt->input->entity; |
2356 | | |
2357 | 76.4k | ent->flags &= ~XML_ENT_EXPANDING; |
2358 | | |
2359 | 76.4k | if ((ent->flags & XML_ENT_CHECKED) == 0) { |
2360 | 7.40k | int result; |
2361 | | |
2362 | | /* |
2363 | | * Read the rest of the stream in case of errors. We want |
2364 | | * to account for the whole entity size. |
2365 | | */ |
2366 | 7.65k | do { |
2367 | 7.65k | ctxt->input->cur = ctxt->input->end; |
2368 | 7.65k | xmlParserShrink(ctxt); |
2369 | 7.65k | result = xmlParserGrow(ctxt); |
2370 | 7.65k | } while (result > 0); |
2371 | | |
2372 | 7.40k | consumed = ctxt->input->consumed; |
2373 | 7.40k | xmlSaturatedAddSizeT(&consumed, |
2374 | 7.40k | ctxt->input->end - ctxt->input->base); |
2375 | | |
2376 | 7.40k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
2377 | | |
2378 | | /* |
2379 | | * Add to sizeentities when parsing an external entity |
2380 | | * for the first time. |
2381 | | */ |
2382 | 7.40k | if (ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
2383 | 4.48k | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
2384 | 4.48k | } |
2385 | | |
2386 | 7.40k | ent->flags |= XML_ENT_CHECKED; |
2387 | 7.40k | } |
2388 | | |
2389 | 76.4k | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
2390 | | |
2391 | 76.4k | xmlParserEntityCheck(ctxt, ent->expandedSize); |
2392 | | |
2393 | 76.4k | GROW; |
2394 | 76.4k | } |
2395 | | |
2396 | | /** |
2397 | | * Skip whitespace in the input stream, also handling parameter |
2398 | | * entities. |
2399 | | * |
2400 | | * @param ctxt the XML parser context |
2401 | | * @returns the number of space chars skipped |
2402 | | */ |
2403 | | static int |
2404 | 842k | xmlSkipBlankCharsPE(xmlParserCtxtPtr ctxt) { |
2405 | 842k | int res = 0; |
2406 | 842k | int inParam; |
2407 | 842k | int expandParam; |
2408 | | |
2409 | 842k | inParam = PARSER_IN_PE(ctxt); |
2410 | 842k | expandParam = PARSER_EXTERNAL(ctxt); |
2411 | | |
2412 | 842k | if (!inParam && !expandParam) |
2413 | 234k | return(xmlSkipBlankChars(ctxt)); |
2414 | | |
2415 | | /* |
2416 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2417 | | * the ASCII range. |
2418 | | */ |
2419 | 1.72M | while (PARSER_STOPPED(ctxt) == 0) { |
2420 | 1.72M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2421 | 1.09M | NEXT; |
2422 | 1.09M | } else if (CUR == '%') { |
2423 | 26.4k | if ((expandParam == 0) || |
2424 | 26.4k | (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2425 | 15.4k | break; |
2426 | | |
2427 | | /* |
2428 | | * Expand parameter entity. We continue to consume |
2429 | | * whitespace at the start of the entity and possible |
2430 | | * even consume the whole entity and pop it. We might |
2431 | | * even pop multiple PEs in this loop. |
2432 | | */ |
2433 | 11.0k | xmlParsePERefInternal(ctxt, 0); |
2434 | | |
2435 | 11.0k | inParam = PARSER_IN_PE(ctxt); |
2436 | 11.0k | expandParam = PARSER_EXTERNAL(ctxt); |
2437 | 600k | } else if (CUR == 0) { |
2438 | 31.3k | if (inParam == 0) |
2439 | 123 | break; |
2440 | | |
2441 | | /* |
2442 | | * Don't pop parameter entities that start a markup |
2443 | | * declaration to detect Well-formedness constraint: |
2444 | | * PE Between Declarations. |
2445 | | */ |
2446 | 31.2k | if (ctxt->input->flags & XML_INPUT_MARKUP_DECL) |
2447 | 22.6k | break; |
2448 | | |
2449 | 8.60k | xmlPopPE(ctxt); |
2450 | | |
2451 | 8.60k | inParam = PARSER_IN_PE(ctxt); |
2452 | 8.60k | expandParam = PARSER_EXTERNAL(ctxt); |
2453 | 569k | } else { |
2454 | 569k | break; |
2455 | 569k | } |
2456 | | |
2457 | | /* |
2458 | | * Also increase the counter when entering or exiting a PERef. |
2459 | | * The spec says: "When a parameter-entity reference is recognized |
2460 | | * in the DTD and included, its replacement text MUST be enlarged |
2461 | | * by the attachment of one leading and one following space (#x20) |
2462 | | * character." |
2463 | | */ |
2464 | 1.11M | if (res < INT_MAX) |
2465 | 1.11M | res++; |
2466 | 1.11M | } |
2467 | | |
2468 | 608k | return(res); |
2469 | 842k | } |
2470 | | |
2471 | | /************************************************************************ |
2472 | | * * |
2473 | | * Commodity functions to handle entities * |
2474 | | * * |
2475 | | ************************************************************************/ |
2476 | | |
2477 | | /** |
2478 | | * @deprecated Internal function, don't use. |
2479 | | * |
2480 | | * @param ctxt an XML parser context |
2481 | | * @returns the current xmlChar in the parser context |
2482 | | */ |
2483 | | xmlChar |
2484 | 0 | xmlPopInput(xmlParserCtxt *ctxt) { |
2485 | 0 | xmlParserInputPtr input; |
2486 | |
|
2487 | 0 | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2488 | 0 | input = xmlCtxtPopInput(ctxt); |
2489 | 0 | xmlFreeInputStream(input); |
2490 | 0 | if (*ctxt->input->cur == 0) |
2491 | 0 | xmlParserGrow(ctxt); |
2492 | 0 | return(CUR); |
2493 | 0 | } |
2494 | | |
2495 | | /** |
2496 | | * Push an input stream onto the stack. |
2497 | | * |
2498 | | * @deprecated Internal function, don't use. |
2499 | | * |
2500 | | * @param ctxt an XML parser context |
2501 | | * @param input an XML parser input fragment (entity, XML fragment ...). |
2502 | | * @returns -1 in case of error or the index in the input stack |
2503 | | */ |
2504 | | int |
2505 | 0 | xmlPushInput(xmlParserCtxt *ctxt, xmlParserInput *input) { |
2506 | 0 | int ret; |
2507 | |
|
2508 | 0 | if ((ctxt == NULL) || (input == NULL)) |
2509 | 0 | return(-1); |
2510 | | |
2511 | 0 | ret = xmlCtxtPushInput(ctxt, input); |
2512 | 0 | if (ret >= 0) |
2513 | 0 | GROW; |
2514 | 0 | return(ret); |
2515 | 0 | } |
2516 | | |
2517 | | /** |
2518 | | * Parse a numeric character reference. Always consumes '&'. |
2519 | | * |
2520 | | * @deprecated Internal function, don't use. |
2521 | | * |
2522 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2523 | | * '&#x' [0-9a-fA-F]+ ';' |
2524 | | * |
2525 | | * [ WFC: Legal Character ] |
2526 | | * Characters referred to using character references must match the |
2527 | | * production for Char. |
2528 | | * |
2529 | | * @param ctxt an XML parser context |
2530 | | * @returns the value parsed (as an int), 0 in case of error |
2531 | | */ |
2532 | | int |
2533 | 63.5k | xmlParseCharRef(xmlParserCtxt *ctxt) { |
2534 | 63.5k | int val = 0; |
2535 | 63.5k | int count = 0; |
2536 | | |
2537 | | /* |
2538 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2539 | | */ |
2540 | 63.5k | if ((RAW == '&') && (NXT(1) == '#') && |
2541 | 63.5k | (NXT(2) == 'x')) { |
2542 | 31.1k | SKIP(3); |
2543 | 31.1k | GROW; |
2544 | 127k | while ((RAW != ';') && (PARSER_STOPPED(ctxt) == 0)) { |
2545 | 107k | if (count++ > 20) { |
2546 | 4.21k | count = 0; |
2547 | 4.21k | GROW; |
2548 | 4.21k | } |
2549 | 107k | if ((RAW >= '0') && (RAW <= '9')) |
2550 | 64.7k | val = val * 16 + (CUR - '0'); |
2551 | 42.5k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2552 | 21.0k | val = val * 16 + (CUR - 'a') + 10; |
2553 | 21.5k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2554 | 11.0k | val = val * 16 + (CUR - 'A') + 10; |
2555 | 10.4k | else { |
2556 | 10.4k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2557 | 10.4k | val = 0; |
2558 | 10.4k | break; |
2559 | 10.4k | } |
2560 | 96.8k | if (val > 0x110000) |
2561 | 49.7k | val = 0x110000; |
2562 | | |
2563 | 96.8k | NEXT; |
2564 | 96.8k | count++; |
2565 | 96.8k | } |
2566 | 31.1k | if (RAW == ';') { |
2567 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2568 | 20.6k | ctxt->input->col++; |
2569 | 20.6k | ctxt->input->cur++; |
2570 | 20.6k | } |
2571 | 32.4k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2572 | 32.4k | SKIP(2); |
2573 | 32.4k | GROW; |
2574 | 127k | while (RAW != ';') { /* loop blocked by count */ |
2575 | 99.9k | if (count++ > 20) { |
2576 | 2.77k | count = 0; |
2577 | 2.77k | GROW; |
2578 | 2.77k | } |
2579 | 99.9k | if ((RAW >= '0') && (RAW <= '9')) |
2580 | 94.8k | val = val * 10 + (CUR - '0'); |
2581 | 5.06k | else { |
2582 | 5.06k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2583 | 5.06k | val = 0; |
2584 | 5.06k | break; |
2585 | 5.06k | } |
2586 | 94.8k | if (val > 0x110000) |
2587 | 30.0k | val = 0x110000; |
2588 | | |
2589 | 94.8k | NEXT; |
2590 | 94.8k | count++; |
2591 | 94.8k | } |
2592 | 32.4k | if (RAW == ';') { |
2593 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2594 | 27.4k | ctxt->input->col++; |
2595 | 27.4k | ctxt->input->cur++; |
2596 | 27.4k | } |
2597 | 32.4k | } else { |
2598 | 0 | if (RAW == '&') |
2599 | 0 | SKIP(1); |
2600 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2601 | 0 | } |
2602 | | |
2603 | | /* |
2604 | | * [ WFC: Legal Character ] |
2605 | | * Characters referred to using character references must match the |
2606 | | * production for Char. |
2607 | | */ |
2608 | 63.5k | if (val >= 0x110000) { |
2609 | 664 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2610 | 664 | "xmlParseCharRef: character reference out of bounds\n", |
2611 | 664 | val); |
2612 | 664 | val = 0xFFFD; |
2613 | 62.9k | } else if (!IS_CHAR(val)) { |
2614 | 17.8k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2615 | 17.8k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2616 | 17.8k | val); |
2617 | 17.8k | } |
2618 | 63.5k | return(val); |
2619 | 63.5k | } |
2620 | | |
2621 | | /** |
2622 | | * Parse Reference declarations, variant parsing from a string rather |
2623 | | * than an an input flow. |
2624 | | * |
2625 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2626 | | * '&#x' [0-9a-fA-F]+ ';' |
2627 | | * |
2628 | | * [ WFC: Legal Character ] |
2629 | | * Characters referred to using character references must match the |
2630 | | * production for Char. |
2631 | | * |
2632 | | * @param ctxt an XML parser context |
2633 | | * @param str a pointer to an index in the string |
2634 | | * @returns the value parsed (as an int), 0 in case of error, str will be |
2635 | | * updated to the current value of the index |
2636 | | */ |
2637 | | static int |
2638 | 327k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2639 | 327k | const xmlChar *ptr; |
2640 | 327k | xmlChar cur; |
2641 | 327k | int val = 0; |
2642 | | |
2643 | 327k | if ((str == NULL) || (*str == NULL)) return(0); |
2644 | 327k | ptr = *str; |
2645 | 327k | cur = *ptr; |
2646 | 327k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2647 | 197k | ptr += 3; |
2648 | 197k | cur = *ptr; |
2649 | 1.17M | while (cur != ';') { /* Non input consuming loop */ |
2650 | 979k | if ((cur >= '0') && (cur <= '9')) |
2651 | 199k | val = val * 16 + (cur - '0'); |
2652 | 779k | else if ((cur >= 'a') && (cur <= 'f')) |
2653 | 389k | val = val * 16 + (cur - 'a') + 10; |
2654 | 390k | else if ((cur >= 'A') && (cur <= 'F')) |
2655 | 389k | val = val * 16 + (cur - 'A') + 10; |
2656 | 499 | else { |
2657 | 499 | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2658 | 499 | val = 0; |
2659 | 499 | break; |
2660 | 499 | } |
2661 | 979k | if (val > 0x110000) |
2662 | 760 | val = 0x110000; |
2663 | | |
2664 | 979k | ptr++; |
2665 | 979k | cur = *ptr; |
2666 | 979k | } |
2667 | 197k | if (cur == ';') |
2668 | 197k | ptr++; |
2669 | 197k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2670 | 129k | ptr += 2; |
2671 | 129k | cur = *ptr; |
2672 | 410k | while (cur != ';') { /* Non input consuming loops */ |
2673 | 281k | if ((cur >= '0') && (cur <= '9')) |
2674 | 280k | val = val * 10 + (cur - '0'); |
2675 | 1.18k | else { |
2676 | 1.18k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2677 | 1.18k | val = 0; |
2678 | 1.18k | break; |
2679 | 1.18k | } |
2680 | 280k | if (val > 0x110000) |
2681 | 5.77k | val = 0x110000; |
2682 | | |
2683 | 280k | ptr++; |
2684 | 280k | cur = *ptr; |
2685 | 280k | } |
2686 | 129k | if (cur == ';') |
2687 | 128k | ptr++; |
2688 | 129k | } else { |
2689 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2690 | 0 | return(0); |
2691 | 0 | } |
2692 | 327k | *str = ptr; |
2693 | | |
2694 | | /* |
2695 | | * [ WFC: Legal Character ] |
2696 | | * Characters referred to using character references must match the |
2697 | | * production for Char. |
2698 | | */ |
2699 | 327k | if (val >= 0x110000) { |
2700 | 236 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2701 | 236 | "xmlParseStringCharRef: character reference out of bounds\n", |
2702 | 236 | val); |
2703 | 327k | } else if (IS_CHAR(val)) { |
2704 | 324k | return(val); |
2705 | 324k | } else { |
2706 | 2.63k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2707 | 2.63k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2708 | 2.63k | val); |
2709 | 2.63k | } |
2710 | 2.87k | return(0); |
2711 | 327k | } |
2712 | | |
2713 | | /** |
2714 | | * [69] PEReference ::= '%' Name ';' |
2715 | | * |
2716 | | * @deprecated Internal function, do not use. |
2717 | | * |
2718 | | * [ WFC: No Recursion ] |
2719 | | * A parsed entity must not contain a recursive |
2720 | | * reference to itself, either directly or indirectly. |
2721 | | * |
2722 | | * [ WFC: Entity Declared ] |
2723 | | * In a document without any DTD, a document with only an internal DTD |
2724 | | * subset which contains no parameter entity references, or a document |
2725 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2726 | | * entity must precede any reference to it... |
2727 | | * |
2728 | | * [ VC: Entity Declared ] |
2729 | | * In a document with an external subset or external parameter entities |
2730 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2731 | | * must precede any reference to it... |
2732 | | * |
2733 | | * [ WFC: In DTD ] |
2734 | | * Parameter-entity references may only appear in the DTD. |
2735 | | * NOTE: misleading but this is handled. |
2736 | | * |
2737 | | * A PEReference may have been detected in the current input stream |
2738 | | * the handling is done accordingly to |
2739 | | * http://www.w3.org/TR/REC-xml#entproc |
2740 | | * i.e. |
2741 | | * - Included in literal in entity values |
2742 | | * - Included as Parameter Entity reference within DTDs |
2743 | | * @param ctxt the parser context |
2744 | | */ |
2745 | | void |
2746 | 0 | xmlParserHandlePEReference(xmlParserCtxt *ctxt) { |
2747 | 0 | xmlParsePERefInternal(ctxt, 0); |
2748 | 0 | } |
2749 | | |
2750 | | /** |
2751 | | * @deprecated Internal function, don't use. |
2752 | | * |
2753 | | * @param ctxt the parser context |
2754 | | * @param str the input string |
2755 | | * @param len the string length |
2756 | | * @param what combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2757 | | * @param end an end marker xmlChar, 0 if none |
2758 | | * @param end2 an end marker xmlChar, 0 if none |
2759 | | * @param end3 an end marker xmlChar, 0 if none |
2760 | | * @returns A newly allocated string with the substitution done. The caller |
2761 | | * must deallocate it ! |
2762 | | */ |
2763 | | xmlChar * |
2764 | | xmlStringLenDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str, int len, |
2765 | | int what ATTRIBUTE_UNUSED, |
2766 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2767 | 0 | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2768 | 0 | return(NULL); |
2769 | | |
2770 | 0 | if ((str[len] != 0) || |
2771 | 0 | (end != 0) || (end2 != 0) || (end3 != 0)) |
2772 | 0 | return(NULL); |
2773 | | |
2774 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2775 | 0 | } |
2776 | | |
2777 | | /** |
2778 | | * @deprecated Internal function, don't use. |
2779 | | * |
2780 | | * @param ctxt the parser context |
2781 | | * @param str the input string |
2782 | | * @param what combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2783 | | * @param end an end marker xmlChar, 0 if none |
2784 | | * @param end2 an end marker xmlChar, 0 if none |
2785 | | * @param end3 an end marker xmlChar, 0 if none |
2786 | | * @returns A newly allocated string with the substitution done. The caller |
2787 | | * must deallocate it ! |
2788 | | */ |
2789 | | xmlChar * |
2790 | | xmlStringDecodeEntities(xmlParserCtxt *ctxt, const xmlChar *str, |
2791 | | int what ATTRIBUTE_UNUSED, |
2792 | 0 | xmlChar end, xmlChar end2, xmlChar end3) { |
2793 | 0 | if ((ctxt == NULL) || (str == NULL)) |
2794 | 0 | return(NULL); |
2795 | | |
2796 | 0 | if ((end != 0) || (end2 != 0) || (end3 != 0)) |
2797 | 0 | return(NULL); |
2798 | | |
2799 | 0 | return(xmlExpandEntitiesInAttValue(ctxt, str, 0)); |
2800 | 0 | } |
2801 | | |
2802 | | /************************************************************************ |
2803 | | * * |
2804 | | * Commodity functions, cleanup needed ? * |
2805 | | * * |
2806 | | ************************************************************************/ |
2807 | | |
2808 | | /** |
2809 | | * Is this a sequence of blank chars that one can ignore ? |
2810 | | * |
2811 | | * @param ctxt an XML parser context |
2812 | | * @param str a xmlChar * |
2813 | | * @param len the size of `str` |
2814 | | * @param blank_chars we know the chars are blanks |
2815 | | * @returns 1 if ignorable 0 otherwise. |
2816 | | */ |
2817 | | |
2818 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2819 | 659k | int blank_chars) { |
2820 | 659k | int i; |
2821 | 659k | xmlNodePtr lastChild; |
2822 | | |
2823 | | /* |
2824 | | * Check for xml:space value. |
2825 | | */ |
2826 | 659k | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2827 | 659k | (*(ctxt->space) == -2)) |
2828 | 578k | return(0); |
2829 | | |
2830 | | /* |
2831 | | * Check that the string is made of blanks |
2832 | | */ |
2833 | 81.1k | if (blank_chars == 0) { |
2834 | 145k | for (i = 0;i < len;i++) |
2835 | 141k | if (!(IS_BLANK_CH(str[i]))) return(0); |
2836 | 67.2k | } |
2837 | | |
2838 | | /* |
2839 | | * Look if the element is mixed content in the DTD if available |
2840 | | */ |
2841 | 18.0k | if (ctxt->node == NULL) return(0); |
2842 | 18.0k | if (ctxt->myDoc != NULL) { |
2843 | 18.0k | xmlElementPtr elemDecl = NULL; |
2844 | 18.0k | xmlDocPtr doc = ctxt->myDoc; |
2845 | 18.0k | const xmlChar *prefix = NULL; |
2846 | | |
2847 | 18.0k | if (ctxt->node->ns) |
2848 | 8.60k | prefix = ctxt->node->ns->prefix; |
2849 | 18.0k | if (doc->intSubset != NULL) |
2850 | 12.9k | elemDecl = xmlHashLookup2(doc->intSubset->elements, ctxt->node->name, |
2851 | 12.9k | prefix); |
2852 | 18.0k | if ((elemDecl == NULL) && (doc->extSubset != NULL)) |
2853 | 787 | elemDecl = xmlHashLookup2(doc->extSubset->elements, ctxt->node->name, |
2854 | 787 | prefix); |
2855 | 18.0k | if (elemDecl != NULL) { |
2856 | 1.71k | if (elemDecl->etype == XML_ELEMENT_TYPE_ELEMENT) |
2857 | 743 | return(1); |
2858 | 973 | if ((elemDecl->etype == XML_ELEMENT_TYPE_ANY) || |
2859 | 973 | (elemDecl->etype == XML_ELEMENT_TYPE_MIXED)) |
2860 | 392 | return(0); |
2861 | 973 | } |
2862 | 18.0k | } |
2863 | | |
2864 | | /* |
2865 | | * Otherwise, heuristic :-\ |
2866 | | * |
2867 | | * When push parsing, we could be at the end of a chunk. |
2868 | | * This makes the look-ahead and consequently the NOBLANKS |
2869 | | * option unreliable. |
2870 | | */ |
2871 | 16.9k | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2872 | 15.0k | if ((ctxt->node->children == NULL) && |
2873 | 15.0k | (RAW == '<') && (NXT(1) == '/')) return(0); |
2874 | | |
2875 | 14.6k | lastChild = xmlGetLastChild(ctxt->node); |
2876 | 14.6k | if (lastChild == NULL) { |
2877 | 7.85k | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2878 | 7.85k | (ctxt->node->content != NULL)) return(0); |
2879 | 7.85k | } else if (xmlNodeIsText(lastChild)) |
2880 | 366 | return(0); |
2881 | 6.40k | else if ((ctxt->node->children != NULL) && |
2882 | 6.40k | (xmlNodeIsText(ctxt->node->children))) |
2883 | 272 | return(0); |
2884 | 13.9k | return(1); |
2885 | 14.6k | } |
2886 | | |
2887 | | /************************************************************************ |
2888 | | * * |
2889 | | * Extra stuff for namespace support * |
2890 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2891 | | * * |
2892 | | ************************************************************************/ |
2893 | | |
2894 | | /** |
2895 | | * Parse an UTF8 encoded XML qualified name string |
2896 | | * |
2897 | | * @deprecated Don't use. |
2898 | | * |
2899 | | * @param ctxt an XML parser context |
2900 | | * @param name an XML parser context |
2901 | | * @param prefixOut a xmlChar ** |
2902 | | * @returns the local part, and prefix is updated |
2903 | | * to get the Prefix if any. |
2904 | | */ |
2905 | | |
2906 | | xmlChar * |
2907 | 0 | xmlSplitQName(xmlParserCtxt *ctxt, const xmlChar *name, xmlChar **prefixOut) { |
2908 | 0 | xmlChar *ret; |
2909 | 0 | const xmlChar *localname; |
2910 | |
|
2911 | 0 | localname = xmlSplitQName4(name, prefixOut); |
2912 | 0 | if (localname == NULL) { |
2913 | 0 | xmlCtxtErrMemory(ctxt); |
2914 | 0 | return(NULL); |
2915 | 0 | } |
2916 | | |
2917 | 0 | ret = xmlStrdup(localname); |
2918 | 0 | if (ret == NULL) { |
2919 | 0 | xmlCtxtErrMemory(ctxt); |
2920 | 0 | xmlFree(*prefixOut); |
2921 | 0 | } |
2922 | |
|
2923 | 0 | return(ret); |
2924 | 0 | } |
2925 | | |
2926 | | /************************************************************************ |
2927 | | * * |
2928 | | * The parser itself * |
2929 | | * Relates to http://www.w3.org/TR/REC-xml * |
2930 | | * * |
2931 | | ************************************************************************/ |
2932 | | |
2933 | | /************************************************************************ |
2934 | | * * |
2935 | | * Routines to parse Name, NCName and NmToken * |
2936 | | * * |
2937 | | ************************************************************************/ |
2938 | | |
2939 | | /* |
2940 | | * The two following functions are related to the change of accepted |
2941 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
2942 | | * They correspond to the modified production [4] and the new production [4a] |
2943 | | * changes in that revision. Also note that the macros used for the |
2944 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
2945 | | * anymore. |
2946 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
2947 | | * new XML_PARSE_OLD10 option is given to the parser. |
2948 | | */ |
2949 | | |
2950 | | static int |
2951 | 1.06M | xmlIsNameStartCharNew(int c) { |
2952 | | /* |
2953 | | * Use the new checks of production [4] [4a] amd [5] of the |
2954 | | * Update 5 of XML-1.0 |
2955 | | */ |
2956 | 1.06M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
2957 | 1.06M | (((c >= 'a') && (c <= 'z')) || |
2958 | 1.06M | ((c >= 'A') && (c <= 'Z')) || |
2959 | 1.06M | (c == '_') || (c == ':') || |
2960 | 1.06M | ((c >= 0xC0) && (c <= 0xD6)) || |
2961 | 1.06M | ((c >= 0xD8) && (c <= 0xF6)) || |
2962 | 1.06M | ((c >= 0xF8) && (c <= 0x2FF)) || |
2963 | 1.06M | ((c >= 0x370) && (c <= 0x37D)) || |
2964 | 1.06M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
2965 | 1.06M | ((c >= 0x200C) && (c <= 0x200D)) || |
2966 | 1.06M | ((c >= 0x2070) && (c <= 0x218F)) || |
2967 | 1.06M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
2968 | 1.06M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
2969 | 1.06M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
2970 | 1.06M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
2971 | 1.06M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
2972 | 637k | return(1); |
2973 | 432k | return(0); |
2974 | 1.06M | } |
2975 | | |
2976 | | static int |
2977 | 22.4M | xmlIsNameCharNew(int c) { |
2978 | | /* |
2979 | | * Use the new checks of production [4] [4a] amd [5] of the |
2980 | | * Update 5 of XML-1.0 |
2981 | | */ |
2982 | 22.4M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
2983 | 22.4M | (((c >= 'a') && (c <= 'z')) || |
2984 | 22.3M | ((c >= 'A') && (c <= 'Z')) || |
2985 | 22.3M | ((c >= '0') && (c <= '9')) || /* !start */ |
2986 | 22.3M | (c == '_') || (c == ':') || |
2987 | 22.3M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
2988 | 22.3M | ((c >= 0xC0) && (c <= 0xD6)) || |
2989 | 22.3M | ((c >= 0xD8) && (c <= 0xF6)) || |
2990 | 22.3M | ((c >= 0xF8) && (c <= 0x2FF)) || |
2991 | 22.3M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
2992 | 22.3M | ((c >= 0x370) && (c <= 0x37D)) || |
2993 | 22.3M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
2994 | 22.3M | ((c >= 0x200C) && (c <= 0x200D)) || |
2995 | 22.3M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
2996 | 22.3M | ((c >= 0x2070) && (c <= 0x218F)) || |
2997 | 22.3M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
2998 | 22.3M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
2999 | 22.3M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3000 | 22.3M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3001 | 22.3M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3002 | 21.7M | return(1); |
3003 | 650k | return(0); |
3004 | 22.4M | } |
3005 | | |
3006 | | static int |
3007 | 410k | xmlIsNameStartCharOld(int c) { |
3008 | 410k | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3009 | 410k | ((IS_LETTER(c) || (c == '_') || (c == ':')))) |
3010 | 300k | return(1); |
3011 | 110k | return(0); |
3012 | 410k | } |
3013 | | |
3014 | | static int |
3015 | 6.38M | xmlIsNameCharOld(int c) { |
3016 | 6.38M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3017 | 6.38M | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3018 | 6.38M | (c == '.') || (c == '-') || |
3019 | 6.38M | (c == '_') || (c == ':') || |
3020 | 6.38M | (IS_COMBINING(c)) || |
3021 | 6.38M | (IS_EXTENDER(c)))) |
3022 | 6.08M | return(1); |
3023 | 309k | return(0); |
3024 | 6.38M | } |
3025 | | |
3026 | | static int |
3027 | 1.48M | xmlIsNameStartChar(int c, int old10) { |
3028 | 1.48M | if (!old10) |
3029 | 1.06M | return(xmlIsNameStartCharNew(c)); |
3030 | 410k | else |
3031 | 410k | return(xmlIsNameStartCharOld(c)); |
3032 | 1.48M | } |
3033 | | |
3034 | | static int |
3035 | 28.7M | xmlIsNameChar(int c, int old10) { |
3036 | 28.7M | if (!old10) |
3037 | 22.4M | return(xmlIsNameCharNew(c)); |
3038 | 6.38M | else |
3039 | 6.38M | return(xmlIsNameCharOld(c)); |
3040 | 28.7M | } |
3041 | | |
3042 | | /* |
3043 | | * Scan an XML Name, NCName or Nmtoken. |
3044 | | * |
3045 | | * Returns a pointer to the end of the name on success. If the |
3046 | | * name is invalid, returns `ptr`. If the name is longer than |
3047 | | * `maxSize` bytes, returns NULL. |
3048 | | * |
3049 | | * @param ptr pointer to the start of the name |
3050 | | * @param maxSize maximum size in bytes |
3051 | | * @param flags XML_SCAN_* flags |
3052 | | * @returns a pointer to the end of the name or NULL |
3053 | | */ |
3054 | | const xmlChar * |
3055 | 814k | xmlScanName(const xmlChar *ptr, size_t maxSize, int flags) { |
3056 | 814k | int stop = flags & XML_SCAN_NC ? ':' : 0; |
3057 | 814k | int old10 = flags & XML_SCAN_OLD10 ? 1 : 0; |
3058 | | |
3059 | 8.30M | while (1) { |
3060 | 8.30M | int c, len; |
3061 | | |
3062 | 8.30M | c = *ptr; |
3063 | 8.30M | if (c < 0x80) { |
3064 | 3.75M | if (c == stop) |
3065 | 1.12k | break; |
3066 | 3.75M | len = 1; |
3067 | 4.55M | } else { |
3068 | 4.55M | len = 4; |
3069 | 4.55M | c = xmlGetUTF8Char(ptr, &len); |
3070 | 4.55M | if (c < 0) |
3071 | 2.81k | break; |
3072 | 4.55M | } |
3073 | | |
3074 | 8.30M | if (flags & XML_SCAN_NMTOKEN ? |
3075 | 7.48M | !xmlIsNameChar(c, old10) : |
3076 | 8.30M | !xmlIsNameStartChar(c, old10)) |
3077 | 810k | break; |
3078 | | |
3079 | 7.49M | if ((size_t) len > maxSize) |
3080 | 106 | return(NULL); |
3081 | 7.49M | ptr += len; |
3082 | 7.49M | maxSize -= len; |
3083 | 7.49M | flags |= XML_SCAN_NMTOKEN; |
3084 | 7.49M | } |
3085 | | |
3086 | 814k | return(ptr); |
3087 | 814k | } |
3088 | | |
3089 | | static const xmlChar * |
3090 | 149k | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3091 | 149k | const xmlChar *ret; |
3092 | 149k | int len = 0, l; |
3093 | 149k | int c; |
3094 | 149k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3095 | 74.6k | XML_MAX_TEXT_LENGTH : |
3096 | 149k | XML_MAX_NAME_LENGTH; |
3097 | 149k | int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0; |
3098 | | |
3099 | | /* |
3100 | | * Handler for more complex cases |
3101 | | */ |
3102 | 149k | c = xmlCurrentChar(ctxt, &l); |
3103 | 149k | if (!xmlIsNameStartChar(c, old10)) |
3104 | 99.9k | return(NULL); |
3105 | 50.0k | len += l; |
3106 | 50.0k | NEXTL(l); |
3107 | 50.0k | c = xmlCurrentChar(ctxt, &l); |
3108 | 3.46M | while (xmlIsNameChar(c, old10)) { |
3109 | 3.41M | if (len <= INT_MAX - l) |
3110 | 3.41M | len += l; |
3111 | 3.41M | NEXTL(l); |
3112 | 3.41M | c = xmlCurrentChar(ctxt, &l); |
3113 | 3.41M | } |
3114 | 50.0k | if (len > maxLength) { |
3115 | 63 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3116 | 63 | return(NULL); |
3117 | 63 | } |
3118 | 49.9k | if (ctxt->input->cur - ctxt->input->base < len) { |
3119 | | /* |
3120 | | * There were a couple of bugs where PERefs lead to to a change |
3121 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3122 | | * pointer to xmlDictLookup. |
3123 | | */ |
3124 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3125 | 0 | "unexpected change of input buffer"); |
3126 | 0 | return (NULL); |
3127 | 0 | } |
3128 | 49.9k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3129 | 464 | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len); |
3130 | 49.4k | else |
3131 | 49.4k | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len); |
3132 | 49.9k | if (ret == NULL) |
3133 | 6 | xmlErrMemory(ctxt); |
3134 | 49.9k | return(ret); |
3135 | 49.9k | } |
3136 | | |
3137 | | /** |
3138 | | * Parse an XML name. |
3139 | | * |
3140 | | * @deprecated Internal function, don't use. |
3141 | | * |
3142 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3143 | | * CombiningChar | Extender |
3144 | | * |
3145 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3146 | | * |
3147 | | * [6] Names ::= Name (#x20 Name)* |
3148 | | * |
3149 | | * @param ctxt an XML parser context |
3150 | | * @returns the Name parsed or NULL |
3151 | | */ |
3152 | | |
3153 | | const xmlChar * |
3154 | 1.97M | xmlParseName(xmlParserCtxt *ctxt) { |
3155 | 1.97M | const xmlChar *in; |
3156 | 1.97M | const xmlChar *ret; |
3157 | 1.97M | size_t count = 0; |
3158 | 1.97M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3159 | 1.26M | XML_MAX_TEXT_LENGTH : |
3160 | 1.97M | XML_MAX_NAME_LENGTH; |
3161 | | |
3162 | 1.97M | GROW; |
3163 | | |
3164 | | /* |
3165 | | * Accelerator for simple ASCII names |
3166 | | */ |
3167 | 1.97M | in = ctxt->input->cur; |
3168 | 1.97M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3169 | 1.97M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3170 | 1.97M | (*in == '_') || (*in == ':')) { |
3171 | 1.86M | in++; |
3172 | 10.1M | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3173 | 10.1M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3174 | 10.1M | ((*in >= 0x30) && (*in <= 0x39)) || |
3175 | 10.1M | (*in == '_') || (*in == '-') || |
3176 | 10.1M | (*in == ':') || (*in == '.')) |
3177 | 8.30M | in++; |
3178 | 1.86M | if ((*in > 0) && (*in < 0x80)) { |
3179 | 1.82M | count = in - ctxt->input->cur; |
3180 | 1.82M | if (count > maxLength) { |
3181 | 93 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3182 | 93 | return(NULL); |
3183 | 93 | } |
3184 | 1.82M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3185 | 1.82M | ctxt->input->cur = in; |
3186 | 1.82M | ctxt->input->col += count; |
3187 | 1.82M | if (ret == NULL) |
3188 | 11 | xmlErrMemory(ctxt); |
3189 | 1.82M | return(ret); |
3190 | 1.82M | } |
3191 | 1.86M | } |
3192 | | /* accelerator for special cases */ |
3193 | 149k | return(xmlParseNameComplex(ctxt)); |
3194 | 1.97M | } |
3195 | | |
3196 | | static xmlHashedString |
3197 | 534k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3198 | 534k | xmlHashedString ret; |
3199 | 534k | int len = 0, l; |
3200 | 534k | int c; |
3201 | 534k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3202 | 433k | XML_MAX_TEXT_LENGTH : |
3203 | 534k | XML_MAX_NAME_LENGTH; |
3204 | 534k | int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0; |
3205 | 534k | size_t startPosition = 0; |
3206 | | |
3207 | 534k | ret.name = NULL; |
3208 | 534k | ret.hashValue = 0; |
3209 | | |
3210 | | /* |
3211 | | * Handler for more complex cases |
3212 | | */ |
3213 | 534k | startPosition = CUR_PTR - BASE_PTR; |
3214 | 534k | c = xmlCurrentChar(ctxt, &l); |
3215 | 534k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3216 | 534k | (!xmlIsNameStartChar(c, old10) || (c == ':'))) { |
3217 | 460k | return(ret); |
3218 | 460k | } |
3219 | | |
3220 | 12.0M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3221 | 12.0M | (xmlIsNameChar(c, old10) && (c != ':'))) { |
3222 | 11.9M | if (len <= INT_MAX - l) |
3223 | 11.9M | len += l; |
3224 | 11.9M | NEXTL(l); |
3225 | 11.9M | c = xmlCurrentChar(ctxt, &l); |
3226 | 11.9M | } |
3227 | 74.1k | if (len > maxLength) { |
3228 | 156 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3229 | 156 | return(ret); |
3230 | 156 | } |
3231 | 73.9k | ret = xmlDictLookupHashed(ctxt->dict, (BASE_PTR + startPosition), len); |
3232 | 73.9k | if (ret.name == NULL) |
3233 | 3 | xmlErrMemory(ctxt); |
3234 | 73.9k | return(ret); |
3235 | 74.1k | } |
3236 | | |
3237 | | /** |
3238 | | * Parse an XML name. |
3239 | | * |
3240 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3241 | | * CombiningChar | Extender |
3242 | | * |
3243 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3244 | | * |
3245 | | * @param ctxt an XML parser context |
3246 | | * @returns the Name parsed or NULL |
3247 | | */ |
3248 | | |
3249 | | static xmlHashedString |
3250 | 1.60M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3251 | 1.60M | const xmlChar *in, *e; |
3252 | 1.60M | xmlHashedString ret; |
3253 | 1.60M | size_t count = 0; |
3254 | 1.60M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3255 | 1.10M | XML_MAX_TEXT_LENGTH : |
3256 | 1.60M | XML_MAX_NAME_LENGTH; |
3257 | | |
3258 | 1.60M | ret.name = NULL; |
3259 | | |
3260 | | /* |
3261 | | * Accelerator for simple ASCII names |
3262 | | */ |
3263 | 1.60M | in = ctxt->input->cur; |
3264 | 1.60M | e = ctxt->input->end; |
3265 | 1.60M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3266 | 1.60M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3267 | 1.60M | (*in == '_')) && (in < e)) { |
3268 | 1.12M | in++; |
3269 | 10.0M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3270 | 10.0M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3271 | 10.0M | ((*in >= 0x30) && (*in <= 0x39)) || |
3272 | 10.0M | (*in == '_') || (*in == '-') || |
3273 | 10.0M | (*in == '.')) && (in < e)) |
3274 | 8.91M | in++; |
3275 | 1.12M | if (in >= e) |
3276 | 2.25k | goto complex; |
3277 | 1.12M | if ((*in > 0) && (*in < 0x80)) { |
3278 | 1.07M | count = in - ctxt->input->cur; |
3279 | 1.07M | if (count > maxLength) { |
3280 | 81 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3281 | 81 | return(ret); |
3282 | 81 | } |
3283 | 1.07M | ret = xmlDictLookupHashed(ctxt->dict, ctxt->input->cur, count); |
3284 | 1.07M | ctxt->input->cur = in; |
3285 | 1.07M | ctxt->input->col += count; |
3286 | 1.07M | if (ret.name == NULL) { |
3287 | 6 | xmlErrMemory(ctxt); |
3288 | 6 | } |
3289 | 1.07M | return(ret); |
3290 | 1.07M | } |
3291 | 1.12M | } |
3292 | 534k | complex: |
3293 | 534k | return(xmlParseNCNameComplex(ctxt)); |
3294 | 1.60M | } |
3295 | | |
3296 | | /** |
3297 | | * Parse an XML name and compares for match |
3298 | | * (specialized for endtag parsing) |
3299 | | * |
3300 | | * @param ctxt an XML parser context |
3301 | | * @param other the name to compare with |
3302 | | * @returns NULL for an illegal name, (xmlChar*) 1 for success |
3303 | | * and the name for mismatch |
3304 | | */ |
3305 | | |
3306 | | static const xmlChar * |
3307 | 68.3k | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3308 | 68.3k | register const xmlChar *cmp = other; |
3309 | 68.3k | register const xmlChar *in; |
3310 | 68.3k | const xmlChar *ret; |
3311 | | |
3312 | 68.3k | GROW; |
3313 | | |
3314 | 68.3k | in = ctxt->input->cur; |
3315 | 169k | while (*in != 0 && *in == *cmp) { |
3316 | 100k | ++in; |
3317 | 100k | ++cmp; |
3318 | 100k | } |
3319 | 68.3k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3320 | | /* success */ |
3321 | 36.9k | ctxt->input->col += in - ctxt->input->cur; |
3322 | 36.9k | ctxt->input->cur = in; |
3323 | 36.9k | return (const xmlChar*) 1; |
3324 | 36.9k | } |
3325 | | /* failure (or end of input buffer), check with full function */ |
3326 | 31.4k | ret = xmlParseName (ctxt); |
3327 | | /* strings coming from the dictionary direct compare possible */ |
3328 | 31.4k | if (ret == other) { |
3329 | 2.36k | return (const xmlChar*) 1; |
3330 | 2.36k | } |
3331 | 29.0k | return ret; |
3332 | 31.4k | } |
3333 | | |
3334 | | /** |
3335 | | * Parse an XML name. |
3336 | | * |
3337 | | * @param ctxt an XML parser context |
3338 | | * @param str a pointer to the string pointer (IN/OUT) |
3339 | | * @returns the Name parsed or NULL. The `str` pointer |
3340 | | * is updated to the current location in the string. |
3341 | | */ |
3342 | | |
3343 | | static xmlChar * |
3344 | 810k | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3345 | 810k | xmlChar *ret; |
3346 | 810k | const xmlChar *cur = *str; |
3347 | 810k | int flags = 0; |
3348 | 810k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3349 | 256k | XML_MAX_TEXT_LENGTH : |
3350 | 810k | XML_MAX_NAME_LENGTH; |
3351 | | |
3352 | 810k | if (ctxt->options & XML_PARSE_OLD10) |
3353 | 255k | flags |= XML_SCAN_OLD10; |
3354 | | |
3355 | 810k | cur = xmlScanName(*str, maxLength, flags); |
3356 | 810k | if (cur == NULL) { |
3357 | 106 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3358 | 106 | return(NULL); |
3359 | 106 | } |
3360 | 810k | if (cur == *str) |
3361 | 5.36k | return(NULL); |
3362 | | |
3363 | 804k | ret = xmlStrndup(*str, cur - *str); |
3364 | 804k | if (ret == NULL) |
3365 | 92 | xmlErrMemory(ctxt); |
3366 | 804k | *str = cur; |
3367 | 804k | return(ret); |
3368 | 810k | } |
3369 | | |
3370 | | /** |
3371 | | * Parse an XML Nmtoken. |
3372 | | * |
3373 | | * @deprecated Internal function, don't use. |
3374 | | * |
3375 | | * [7] Nmtoken ::= (NameChar)+ |
3376 | | * |
3377 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3378 | | * |
3379 | | * @param ctxt an XML parser context |
3380 | | * @returns the Nmtoken parsed or NULL |
3381 | | */ |
3382 | | |
3383 | | xmlChar * |
3384 | 42.8k | xmlParseNmtoken(xmlParserCtxt *ctxt) { |
3385 | 42.8k | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3386 | 42.8k | xmlChar *ret; |
3387 | 42.8k | int len = 0, l; |
3388 | 42.8k | int c; |
3389 | 42.8k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3390 | 19.2k | XML_MAX_TEXT_LENGTH : |
3391 | 42.8k | XML_MAX_NAME_LENGTH; |
3392 | 42.8k | int old10 = (ctxt->options & XML_PARSE_OLD10) ? 1 : 0; |
3393 | | |
3394 | 42.8k | c = xmlCurrentChar(ctxt, &l); |
3395 | | |
3396 | 208k | while (xmlIsNameChar(c, old10)) { |
3397 | 167k | COPY_BUF(buf, len, c); |
3398 | 167k | NEXTL(l); |
3399 | 167k | c = xmlCurrentChar(ctxt, &l); |
3400 | 167k | if (len >= XML_MAX_NAMELEN) { |
3401 | | /* |
3402 | | * Okay someone managed to make a huge token, so he's ready to pay |
3403 | | * for the processing speed. |
3404 | | */ |
3405 | 1.27k | xmlChar *buffer; |
3406 | 1.27k | int max = len * 2; |
3407 | | |
3408 | 1.27k | buffer = xmlMalloc(max); |
3409 | 1.27k | if (buffer == NULL) { |
3410 | 5 | xmlErrMemory(ctxt); |
3411 | 5 | return(NULL); |
3412 | 5 | } |
3413 | 1.26k | memcpy(buffer, buf, len); |
3414 | 5.63M | while (xmlIsNameChar(c, old10)) { |
3415 | 5.63M | if (len + 10 > max) { |
3416 | 3.23k | xmlChar *tmp; |
3417 | 3.23k | int newSize; |
3418 | | |
3419 | 3.23k | newSize = xmlGrowCapacity(max, 1, 1, maxLength); |
3420 | 3.23k | if (newSize < 0) { |
3421 | 82 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3422 | 82 | xmlFree(buffer); |
3423 | 82 | return(NULL); |
3424 | 82 | } |
3425 | 3.15k | tmp = xmlRealloc(buffer, newSize); |
3426 | 3.15k | if (tmp == NULL) { |
3427 | 5 | xmlErrMemory(ctxt); |
3428 | 5 | xmlFree(buffer); |
3429 | 5 | return(NULL); |
3430 | 5 | } |
3431 | 3.15k | buffer = tmp; |
3432 | 3.15k | max = newSize; |
3433 | 3.15k | } |
3434 | 5.63M | COPY_BUF(buffer, len, c); |
3435 | 5.63M | NEXTL(l); |
3436 | 5.63M | c = xmlCurrentChar(ctxt, &l); |
3437 | 5.63M | } |
3438 | 1.18k | buffer[len] = 0; |
3439 | 1.18k | return(buffer); |
3440 | 1.26k | } |
3441 | 167k | } |
3442 | 41.5k | if (len == 0) |
3443 | 7.90k | return(NULL); |
3444 | 33.6k | if (len > maxLength) { |
3445 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3446 | 0 | return(NULL); |
3447 | 0 | } |
3448 | 33.6k | ret = xmlStrndup(buf, len); |
3449 | 33.6k | if (ret == NULL) |
3450 | 17 | xmlErrMemory(ctxt); |
3451 | 33.6k | return(ret); |
3452 | 33.6k | } |
3453 | | |
3454 | | /** |
3455 | | * Validate an entity value and expand parameter entities. |
3456 | | * |
3457 | | * @param ctxt parser context |
3458 | | * @param buf string buffer |
3459 | | * @param str entity value |
3460 | | * @param length size of entity value |
3461 | | * @param depth nesting depth |
3462 | | */ |
3463 | | static void |
3464 | | xmlExpandPEsInEntityValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
3465 | 79.5k | const xmlChar *str, int length, int depth) { |
3466 | 79.5k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3467 | 79.5k | const xmlChar *end, *chunk; |
3468 | 79.5k | int c, l; |
3469 | | |
3470 | 79.5k | if (str == NULL) |
3471 | 19.6k | return; |
3472 | | |
3473 | 59.9k | depth += 1; |
3474 | 59.9k | if (depth > maxDepth) { |
3475 | 5 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3476 | 5 | "Maximum entity nesting depth exceeded"); |
3477 | 5 | return; |
3478 | 5 | } |
3479 | | |
3480 | 59.9k | end = str + length; |
3481 | 59.9k | chunk = str; |
3482 | | |
3483 | 433M | while ((str < end) && (!PARSER_STOPPED(ctxt))) { |
3484 | 433M | c = *str; |
3485 | | |
3486 | 433M | if (c >= 0x80) { |
3487 | 362M | l = xmlUTF8MultibyteLen(ctxt, str, |
3488 | 362M | "invalid character in entity value\n"); |
3489 | 362M | if (l == 0) { |
3490 | 28.1M | if (chunk < str) |
3491 | 62.1k | xmlSBufAddString(buf, chunk, str - chunk); |
3492 | 28.1M | xmlSBufAddReplChar(buf); |
3493 | 28.1M | str += 1; |
3494 | 28.1M | chunk = str; |
3495 | 334M | } else { |
3496 | 334M | str += l; |
3497 | 334M | } |
3498 | 362M | } else if (c == '&') { |
3499 | 53.2k | if (str[1] == '#') { |
3500 | 15.7k | if (chunk < str) |
3501 | 7.22k | xmlSBufAddString(buf, chunk, str - chunk); |
3502 | | |
3503 | 15.7k | c = xmlParseStringCharRef(ctxt, &str); |
3504 | 15.7k | if (c == 0) |
3505 | 2.83k | return; |
3506 | | |
3507 | 12.9k | xmlSBufAddChar(buf, c); |
3508 | | |
3509 | 12.9k | chunk = str; |
3510 | 37.4k | } else { |
3511 | 37.4k | xmlChar *name; |
3512 | | |
3513 | | /* |
3514 | | * General entity references are checked for |
3515 | | * syntactic validity. |
3516 | | */ |
3517 | 37.4k | str++; |
3518 | 37.4k | name = xmlParseStringName(ctxt, &str); |
3519 | | |
3520 | 37.4k | if ((name == NULL) || (*str++ != ';')) { |
3521 | 1.55k | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3522 | 1.55k | "EntityValue: '&' forbidden except for entities " |
3523 | 1.55k | "references\n"); |
3524 | 1.55k | xmlFree(name); |
3525 | 1.55k | return; |
3526 | 1.55k | } |
3527 | | |
3528 | 35.9k | xmlFree(name); |
3529 | 35.9k | } |
3530 | 70.8M | } else if (c == '%') { |
3531 | 46.4k | xmlEntityPtr ent; |
3532 | | |
3533 | 46.4k | if (chunk < str) |
3534 | 7.65k | xmlSBufAddString(buf, chunk, str - chunk); |
3535 | | |
3536 | 46.4k | ent = xmlParseStringPEReference(ctxt, &str); |
3537 | 46.4k | if (ent == NULL) |
3538 | 11.5k | return; |
3539 | | |
3540 | 34.8k | if (!PARSER_EXTERNAL(ctxt)) { |
3541 | 326 | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3542 | 326 | return; |
3543 | 326 | } |
3544 | | |
3545 | 34.5k | if (ent->content == NULL) { |
3546 | | /* |
3547 | | * Note: external parsed entities will not be loaded, |
3548 | | * it is not required for a non-validating parser to |
3549 | | * complete external PEReferences coming from the |
3550 | | * internal subset |
3551 | | */ |
3552 | 19.9k | if (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
3553 | 19.9k | ((ctxt->replaceEntities) || |
3554 | 19.9k | (ctxt->validate))) { |
3555 | 19.5k | xmlLoadEntityContent(ctxt, ent); |
3556 | 19.5k | } else { |
3557 | 344 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
3558 | 344 | "not validating will not read content for " |
3559 | 344 | "PE entity %s\n", ent->name, NULL); |
3560 | 344 | } |
3561 | 19.9k | } |
3562 | | |
3563 | | /* |
3564 | | * TODO: Skip if ent->content is still NULL. |
3565 | | */ |
3566 | | |
3567 | 34.5k | if (xmlParserEntityCheck(ctxt, ent->length)) |
3568 | 11 | return; |
3569 | | |
3570 | 34.5k | if (ent->flags & XML_ENT_EXPANDING) { |
3571 | 171 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3572 | 171 | return; |
3573 | 171 | } |
3574 | | |
3575 | 34.3k | ent->flags |= XML_ENT_EXPANDING; |
3576 | 34.3k | xmlExpandPEsInEntityValue(ctxt, buf, ent->content, ent->length, |
3577 | 34.3k | depth); |
3578 | 34.3k | ent->flags &= ~XML_ENT_EXPANDING; |
3579 | | |
3580 | 34.3k | chunk = str; |
3581 | 70.7M | } else { |
3582 | | /* Normal ASCII char */ |
3583 | 70.7M | if (!IS_BYTE_CHAR(c)) { |
3584 | 7.41M | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3585 | 7.41M | "invalid character in entity value\n"); |
3586 | 7.41M | if (chunk < str) |
3587 | 19.0k | xmlSBufAddString(buf, chunk, str - chunk); |
3588 | 7.41M | xmlSBufAddReplChar(buf); |
3589 | 7.41M | str += 1; |
3590 | 7.41M | chunk = str; |
3591 | 63.3M | } else { |
3592 | 63.3M | str += 1; |
3593 | 63.3M | } |
3594 | 70.7M | } |
3595 | 433M | } |
3596 | | |
3597 | 43.4k | if (chunk < str) |
3598 | 36.0k | xmlSBufAddString(buf, chunk, str - chunk); |
3599 | 43.4k | } |
3600 | | |
3601 | | /** |
3602 | | * Parse a value for ENTITY declarations |
3603 | | * |
3604 | | * @deprecated Internal function, don't use. |
3605 | | * |
3606 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3607 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3608 | | * |
3609 | | * @param ctxt an XML parser context |
3610 | | * @param orig if non-NULL store a copy of the original entity value |
3611 | | * @returns the EntityValue parsed with reference substituted or NULL |
3612 | | */ |
3613 | | xmlChar * |
3614 | 46.1k | xmlParseEntityValue(xmlParserCtxt *ctxt, xmlChar **orig) { |
3615 | 46.1k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3616 | 21.2k | XML_MAX_HUGE_LENGTH : |
3617 | 46.1k | XML_MAX_TEXT_LENGTH; |
3618 | 46.1k | xmlSBuf buf; |
3619 | 46.1k | const xmlChar *start; |
3620 | 46.1k | int quote, length; |
3621 | | |
3622 | 46.1k | xmlSBufInit(&buf, maxLength); |
3623 | | |
3624 | 46.1k | GROW; |
3625 | | |
3626 | 46.1k | quote = CUR; |
3627 | 46.1k | if ((quote != '"') && (quote != '\'')) { |
3628 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3629 | 0 | return(NULL); |
3630 | 0 | } |
3631 | 46.1k | CUR_PTR++; |
3632 | | |
3633 | 46.1k | length = 0; |
3634 | | |
3635 | | /* |
3636 | | * Copy raw content of the entity into a buffer |
3637 | | */ |
3638 | 1.02G | while (1) { |
3639 | 1.02G | int c; |
3640 | | |
3641 | 1.02G | if (PARSER_STOPPED(ctxt)) |
3642 | 18 | goto error; |
3643 | | |
3644 | 1.02G | if (CUR_PTR >= ctxt->input->end) { |
3645 | 786 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3646 | 786 | goto error; |
3647 | 786 | } |
3648 | | |
3649 | 1.02G | c = CUR; |
3650 | | |
3651 | 1.02G | if (c == 0) { |
3652 | 87 | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
3653 | 87 | "invalid character in entity value\n"); |
3654 | 87 | goto error; |
3655 | 87 | } |
3656 | 1.02G | if (c == quote) |
3657 | 45.2k | break; |
3658 | 1.02G | NEXTL(1); |
3659 | 1.02G | length += 1; |
3660 | | |
3661 | | /* |
3662 | | * TODO: Check growth threshold |
3663 | | */ |
3664 | 1.02G | if (ctxt->input->end - CUR_PTR < 10) |
3665 | 87.3k | GROW; |
3666 | 1.02G | } |
3667 | | |
3668 | 45.2k | start = CUR_PTR - length; |
3669 | | |
3670 | 45.2k | if (orig != NULL) { |
3671 | 45.2k | *orig = xmlStrndup(start, length); |
3672 | 45.2k | if (*orig == NULL) |
3673 | 48 | xmlErrMemory(ctxt); |
3674 | 45.2k | } |
3675 | | |
3676 | 45.2k | xmlExpandPEsInEntityValue(ctxt, &buf, start, length, ctxt->inputNr); |
3677 | | |
3678 | 45.2k | NEXTL(1); |
3679 | | |
3680 | 45.2k | return(xmlSBufFinish(&buf, NULL, ctxt, "entity length too long")); |
3681 | | |
3682 | 891 | error: |
3683 | 891 | xmlSBufCleanup(&buf, ctxt, "entity length too long"); |
3684 | 891 | return(NULL); |
3685 | 46.1k | } |
3686 | | |
3687 | | /** |
3688 | | * Check an entity reference in an attribute value for validity |
3689 | | * without expanding it. |
3690 | | * |
3691 | | * @param ctxt parser context |
3692 | | * @param pent entity |
3693 | | * @param depth nesting depth |
3694 | | */ |
3695 | | static void |
3696 | 6.08k | xmlCheckEntityInAttValue(xmlParserCtxtPtr ctxt, xmlEntityPtr pent, int depth) { |
3697 | 6.08k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3698 | 6.08k | const xmlChar *str; |
3699 | 6.08k | unsigned long expandedSize = pent->length; |
3700 | 6.08k | int c, flags; |
3701 | | |
3702 | 6.08k | depth += 1; |
3703 | 6.08k | if (depth > maxDepth) { |
3704 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3705 | 3 | "Maximum entity nesting depth exceeded"); |
3706 | 3 | return; |
3707 | 3 | } |
3708 | | |
3709 | 6.08k | if (pent->flags & XML_ENT_EXPANDING) { |
3710 | 28 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3711 | 28 | return; |
3712 | 28 | } |
3713 | | |
3714 | | /* |
3715 | | * If we're parsing a default attribute value in DTD content, |
3716 | | * the entity might reference other entities which weren't |
3717 | | * defined yet, so the check isn't reliable. |
3718 | | */ |
3719 | 6.05k | if (ctxt->inSubset == 0) |
3720 | 5.92k | flags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
3721 | 133 | else |
3722 | 133 | flags = XML_ENT_VALIDATED; |
3723 | | |
3724 | 6.05k | str = pent->content; |
3725 | 6.05k | if (str == NULL) |
3726 | 3 | goto done; |
3727 | | |
3728 | | /* |
3729 | | * Note that entity values are already validated. We only check |
3730 | | * for illegal less-than signs and compute the expanded size |
3731 | | * of the entity. No special handling for multi-byte characters |
3732 | | * is needed. |
3733 | | */ |
3734 | 91.3M | while (!PARSER_STOPPED(ctxt)) { |
3735 | 91.3M | c = *str; |
3736 | | |
3737 | 91.3M | if (c != '&') { |
3738 | 91.2M | if (c == 0) |
3739 | 5.83k | break; |
3740 | | |
3741 | 91.2M | if (c == '<') |
3742 | 1.53k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
3743 | 1.53k | "'<' in entity '%s' is not allowed in attributes " |
3744 | 1.53k | "values\n", pent->name); |
3745 | | |
3746 | 91.2M | str += 1; |
3747 | 91.2M | } else if (str[1] == '#') { |
3748 | 1.87k | int val; |
3749 | | |
3750 | 1.87k | val = xmlParseStringCharRef(ctxt, &str); |
3751 | 1.87k | if (val == 0) { |
3752 | 18 | pent->content[0] = 0; |
3753 | 18 | break; |
3754 | 18 | } |
3755 | 13.2k | } else { |
3756 | 13.2k | xmlChar *name; |
3757 | 13.2k | xmlEntityPtr ent; |
3758 | | |
3759 | 13.2k | name = xmlParseStringEntityRef(ctxt, &str); |
3760 | 13.2k | if (name == NULL) { |
3761 | 31 | pent->content[0] = 0; |
3762 | 31 | break; |
3763 | 31 | } |
3764 | | |
3765 | 13.2k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
3766 | 13.2k | xmlFree(name); |
3767 | | |
3768 | 13.2k | if ((ent != NULL) && |
3769 | 13.2k | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
3770 | 10.2k | if ((ent->flags & flags) != flags) { |
3771 | 4.92k | pent->flags |= XML_ENT_EXPANDING; |
3772 | 4.92k | xmlCheckEntityInAttValue(ctxt, ent, depth); |
3773 | 4.92k | pent->flags &= ~XML_ENT_EXPANDING; |
3774 | 4.92k | } |
3775 | | |
3776 | 10.2k | xmlSaturatedAdd(&expandedSize, ent->expandedSize); |
3777 | 10.2k | xmlSaturatedAdd(&expandedSize, XML_ENT_FIXED_COST); |
3778 | 10.2k | } |
3779 | 13.2k | } |
3780 | 91.3M | } |
3781 | | |
3782 | 6.05k | done: |
3783 | 6.05k | if (ctxt->inSubset == 0) |
3784 | 5.92k | pent->expandedSize = expandedSize; |
3785 | | |
3786 | 6.05k | pent->flags |= flags; |
3787 | 6.05k | } |
3788 | | |
3789 | | /** |
3790 | | * Expand general entity references in an entity or attribute value. |
3791 | | * Perform attribute value normalization. |
3792 | | * |
3793 | | * @param ctxt parser context |
3794 | | * @param buf string buffer |
3795 | | * @param str entity or attribute value |
3796 | | * @param pent entity for entity value, NULL for attribute values |
3797 | | * @param normalize whether to collapse whitespace |
3798 | | * @param inSpace whitespace state |
3799 | | * @param depth nesting depth |
3800 | | * @param check whether to check for amplification |
3801 | | * @returns whether there was a normalization change |
3802 | | */ |
3803 | | static int |
3804 | | xmlExpandEntityInAttValue(xmlParserCtxtPtr ctxt, xmlSBuf *buf, |
3805 | | const xmlChar *str, xmlEntityPtr pent, int normalize, |
3806 | 687k | int *inSpace, int depth, int check) { |
3807 | 687k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 40 : 20; |
3808 | 687k | int c, chunkSize; |
3809 | 687k | int normChange = 0; |
3810 | | |
3811 | 687k | if (str == NULL) |
3812 | 36 | return(0); |
3813 | | |
3814 | 687k | depth += 1; |
3815 | 687k | if (depth > maxDepth) { |
3816 | 3 | xmlFatalErrMsg(ctxt, XML_ERR_RESOURCE_LIMIT, |
3817 | 3 | "Maximum entity nesting depth exceeded"); |
3818 | 3 | return(0); |
3819 | 3 | } |
3820 | | |
3821 | 687k | if (pent != NULL) { |
3822 | 687k | if (pent->flags & XML_ENT_EXPANDING) { |
3823 | 11 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
3824 | 11 | return(0); |
3825 | 11 | } |
3826 | | |
3827 | 687k | if (check) { |
3828 | 687k | if (xmlParserEntityCheck(ctxt, pent->length)) |
3829 | 357 | return(0); |
3830 | 687k | } |
3831 | 687k | } |
3832 | | |
3833 | 687k | chunkSize = 0; |
3834 | | |
3835 | | /* |
3836 | | * Note that entity values are already validated. No special |
3837 | | * handling for multi-byte characters is needed. |
3838 | | */ |
3839 | 4.10G | while (!PARSER_STOPPED(ctxt)) { |
3840 | 4.10G | c = *str; |
3841 | | |
3842 | 4.10G | if (c != '&') { |
3843 | 4.10G | if (c == 0) |
3844 | 599k | break; |
3845 | | |
3846 | | /* |
3847 | | * If this function is called without an entity, it is used to |
3848 | | * expand entities in an attribute content where less-than was |
3849 | | * already unscaped and is allowed. |
3850 | | */ |
3851 | 4.10G | if ((pent != NULL) && (c == '<')) { |
3852 | 87.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
3853 | 87.0k | "'<' in entity '%s' is not allowed in attributes " |
3854 | 87.0k | "values\n", pent->name); |
3855 | 87.0k | break; |
3856 | 87.0k | } |
3857 | | |
3858 | 4.10G | if (c <= 0x20) { |
3859 | 13.5M | if ((normalize) && (*inSpace)) { |
3860 | | /* Skip char */ |
3861 | 835k | if (chunkSize > 0) { |
3862 | 149k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
3863 | 149k | chunkSize = 0; |
3864 | 149k | } |
3865 | 835k | normChange = 1; |
3866 | 12.7M | } else if (c < 0x20) { |
3867 | 11.7M | if (chunkSize > 0) { |
3868 | 98.3k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
3869 | 98.3k | chunkSize = 0; |
3870 | 98.3k | } |
3871 | | |
3872 | 11.7M | xmlSBufAddCString(buf, " ", 1); |
3873 | 11.7M | } else { |
3874 | 968k | chunkSize += 1; |
3875 | 968k | } |
3876 | | |
3877 | 13.5M | *inSpace = 1; |
3878 | 4.09G | } else { |
3879 | 4.09G | chunkSize += 1; |
3880 | 4.09G | *inSpace = 0; |
3881 | 4.09G | } |
3882 | | |
3883 | 4.10G | str += 1; |
3884 | 4.10G | } else if (str[1] == '#') { |
3885 | 309k | int val; |
3886 | | |
3887 | 309k | if (chunkSize > 0) { |
3888 | 306k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
3889 | 306k | chunkSize = 0; |
3890 | 306k | } |
3891 | | |
3892 | 309k | val = xmlParseStringCharRef(ctxt, &str); |
3893 | 309k | if (val == 0) { |
3894 | 17 | if (pent != NULL) |
3895 | 17 | pent->content[0] = 0; |
3896 | 17 | break; |
3897 | 17 | } |
3898 | | |
3899 | 309k | if (val == ' ') { |
3900 | 4.43k | if ((normalize) && (*inSpace)) |
3901 | 221 | normChange = 1; |
3902 | 4.21k | else |
3903 | 4.21k | xmlSBufAddCString(buf, " ", 1); |
3904 | 4.43k | *inSpace = 1; |
3905 | 305k | } else { |
3906 | 305k | xmlSBufAddChar(buf, val); |
3907 | 305k | *inSpace = 0; |
3908 | 305k | } |
3909 | 713k | } else { |
3910 | 713k | xmlChar *name; |
3911 | 713k | xmlEntityPtr ent; |
3912 | | |
3913 | 713k | if (chunkSize > 0) { |
3914 | 186k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
3915 | 186k | chunkSize = 0; |
3916 | 186k | } |
3917 | | |
3918 | 713k | name = xmlParseStringEntityRef(ctxt, &str); |
3919 | 713k | if (name == NULL) { |
3920 | 43 | if (pent != NULL) |
3921 | 43 | pent->content[0] = 0; |
3922 | 43 | break; |
3923 | 43 | } |
3924 | | |
3925 | 713k | ent = xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 1); |
3926 | 713k | xmlFree(name); |
3927 | | |
3928 | 713k | if ((ent != NULL) && |
3929 | 713k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
3930 | 164k | if (ent->content == NULL) { |
3931 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
3932 | 0 | "predefined entity has no content\n"); |
3933 | 0 | break; |
3934 | 0 | } |
3935 | | |
3936 | 164k | xmlSBufAddString(buf, ent->content, ent->length); |
3937 | | |
3938 | 164k | *inSpace = 0; |
3939 | 548k | } else if ((ent != NULL) && (ent->content != NULL)) { |
3940 | 397k | if (pent != NULL) |
3941 | 397k | pent->flags |= XML_ENT_EXPANDING; |
3942 | 397k | normChange |= xmlExpandEntityInAttValue(ctxt, buf, |
3943 | 397k | ent->content, ent, normalize, inSpace, depth, check); |
3944 | 397k | if (pent != NULL) |
3945 | 397k | pent->flags &= ~XML_ENT_EXPANDING; |
3946 | 397k | } |
3947 | 713k | } |
3948 | 4.10G | } |
3949 | | |
3950 | 687k | if (chunkSize > 0) |
3951 | 242k | xmlSBufAddString(buf, str - chunkSize, chunkSize); |
3952 | | |
3953 | 687k | return(normChange); |
3954 | 687k | } |
3955 | | |
3956 | | /** |
3957 | | * Expand general entity references in an entity or attribute value. |
3958 | | * Perform attribute value normalization. |
3959 | | * |
3960 | | * @param ctxt parser context |
3961 | | * @param str entity or attribute value |
3962 | | * @param normalize whether to collapse whitespace |
3963 | | * @returns the expanded attribtue value. |
3964 | | */ |
3965 | | xmlChar * |
3966 | | xmlExpandEntitiesInAttValue(xmlParserCtxt *ctxt, const xmlChar *str, |
3967 | 0 | int normalize) { |
3968 | 0 | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3969 | 0 | XML_MAX_HUGE_LENGTH : |
3970 | 0 | XML_MAX_TEXT_LENGTH; |
3971 | 0 | xmlSBuf buf; |
3972 | 0 | int inSpace = 1; |
3973 | |
|
3974 | 0 | xmlSBufInit(&buf, maxLength); |
3975 | |
|
3976 | 0 | xmlExpandEntityInAttValue(ctxt, &buf, str, NULL, normalize, &inSpace, |
3977 | 0 | ctxt->inputNr, /* check */ 0); |
3978 | |
|
3979 | 0 | if ((normalize) && (inSpace) && (buf.size > 0)) |
3980 | 0 | buf.size--; |
3981 | |
|
3982 | 0 | return(xmlSBufFinish(&buf, NULL, ctxt, "AttValue length too long")); |
3983 | 0 | } |
3984 | | |
3985 | | /** |
3986 | | * Parse a value for an attribute. |
3987 | | * |
3988 | | * NOTE: if no normalization is needed, the routine will return pointers |
3989 | | * directly from the data buffer. |
3990 | | * |
3991 | | * 3.3.3 Attribute-Value Normalization: |
3992 | | * |
3993 | | * Before the value of an attribute is passed to the application or |
3994 | | * checked for validity, the XML processor must normalize it as follows: |
3995 | | * |
3996 | | * - a character reference is processed by appending the referenced |
3997 | | * character to the attribute value |
3998 | | * - an entity reference is processed by recursively processing the |
3999 | | * replacement text of the entity |
4000 | | * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by |
4001 | | * appending \#x20 to the normalized value, except that only a single |
4002 | | * \#x20 is appended for a "#xD#xA" sequence that is part of an external |
4003 | | * parsed entity or the literal entity value of an internal parsed entity |
4004 | | * - other characters are processed by appending them to the normalized value |
4005 | | * |
4006 | | * If the declared value is not CDATA, then the XML processor must further |
4007 | | * process the normalized attribute value by discarding any leading and |
4008 | | * trailing space (\#x20) characters, and by replacing sequences of space |
4009 | | * (\#x20) characters by a single space (\#x20) character. |
4010 | | * All attributes for which no declaration has been read should be treated |
4011 | | * by a non-validating parser as if declared CDATA. |
4012 | | * |
4013 | | * @param ctxt an XML parser context |
4014 | | * @param attlen attribute len result |
4015 | | * @param outFlags resulting XML_ATTVAL_* flags |
4016 | | * @param special value from attsSpecial |
4017 | | * @param isNamespace whether this is a namespace declaration |
4018 | | * @returns the AttValue parsed or NULL. The value has to be freed by the |
4019 | | * caller if it was copied, this can be detected by val[*len] == 0. |
4020 | | */ |
4021 | | static xmlChar * |
4022 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *attlen, int *outFlags, |
4023 | 331k | int special, int isNamespace) { |
4024 | 331k | unsigned maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4025 | 174k | XML_MAX_HUGE_LENGTH : |
4026 | 331k | XML_MAX_TEXT_LENGTH; |
4027 | 331k | xmlSBuf buf; |
4028 | 331k | xmlChar *ret; |
4029 | 331k | int c, l, quote, entFlags, chunkSize; |
4030 | 331k | int inSpace = 1; |
4031 | 331k | int replaceEntities; |
4032 | 331k | int normalize = (special & XML_SPECIAL_TYPE_MASK) != 0; |
4033 | 331k | int attvalFlags = 0; |
4034 | | |
4035 | | /* Always expand namespace URIs */ |
4036 | 331k | replaceEntities = (ctxt->replaceEntities) || (isNamespace); |
4037 | | |
4038 | 331k | xmlSBufInit(&buf, maxLength); |
4039 | | |
4040 | 331k | GROW; |
4041 | | |
4042 | 331k | quote = CUR; |
4043 | 331k | if ((quote != '"') && (quote != '\'')) { |
4044 | 6.90k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
4045 | 6.90k | return(NULL); |
4046 | 6.90k | } |
4047 | 324k | NEXTL(1); |
4048 | | |
4049 | 324k | if (ctxt->inSubset == 0) |
4050 | 269k | entFlags = XML_ENT_CHECKED | XML_ENT_VALIDATED; |
4051 | 55.1k | else |
4052 | 55.1k | entFlags = XML_ENT_VALIDATED; |
4053 | | |
4054 | 324k | inSpace = 1; |
4055 | 324k | chunkSize = 0; |
4056 | | |
4057 | 130M | while (1) { |
4058 | 130M | if (PARSER_STOPPED(ctxt)) |
4059 | 540 | goto error; |
4060 | | |
4061 | 130M | if (CUR_PTR >= ctxt->input->end) { |
4062 | 8.69k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4063 | 8.69k | "AttValue: ' expected\n"); |
4064 | 8.69k | goto error; |
4065 | 8.69k | } |
4066 | | |
4067 | | /* |
4068 | | * TODO: Check growth threshold |
4069 | | */ |
4070 | 130M | if (ctxt->input->end - CUR_PTR < 10) |
4071 | 130k | GROW; |
4072 | | |
4073 | 130M | c = CUR; |
4074 | | |
4075 | 130M | if (c >= 0x80) { |
4076 | 79.4M | l = xmlUTF8MultibyteLen(ctxt, CUR_PTR, |
4077 | 79.4M | "invalid character in attribute value\n"); |
4078 | 79.4M | if (l == 0) { |
4079 | 33.9M | if (chunkSize > 0) { |
4080 | 82.5k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4081 | 82.5k | chunkSize = 0; |
4082 | 82.5k | } |
4083 | 33.9M | xmlSBufAddReplChar(&buf); |
4084 | 33.9M | NEXTL(1); |
4085 | 45.4M | } else { |
4086 | 45.4M | chunkSize += l; |
4087 | 45.4M | NEXTL(l); |
4088 | 45.4M | } |
4089 | | |
4090 | 79.4M | inSpace = 0; |
4091 | 79.4M | } else if (c != '&') { |
4092 | 50.0M | if (c > 0x20) { |
4093 | 8.91M | if (c == quote) |
4094 | 312k | break; |
4095 | | |
4096 | 8.60M | if (c == '<') |
4097 | 142k | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4098 | | |
4099 | 8.60M | chunkSize += 1; |
4100 | 8.60M | inSpace = 0; |
4101 | 41.0M | } else if (!IS_BYTE_CHAR(c)) { |
4102 | 37.3M | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4103 | 37.3M | "invalid character in attribute value\n"); |
4104 | 37.3M | if (chunkSize > 0) { |
4105 | 62.7k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4106 | 62.7k | chunkSize = 0; |
4107 | 62.7k | } |
4108 | 37.3M | xmlSBufAddReplChar(&buf); |
4109 | 37.3M | inSpace = 0; |
4110 | 37.3M | } else { |
4111 | | /* Whitespace */ |
4112 | 3.75M | if ((normalize) && (inSpace)) { |
4113 | | /* Skip char */ |
4114 | 923k | if (chunkSize > 0) { |
4115 | 5.54k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4116 | 5.54k | chunkSize = 0; |
4117 | 5.54k | } |
4118 | 923k | attvalFlags |= XML_ATTVAL_NORM_CHANGE; |
4119 | 2.83M | } else if (c < 0x20) { |
4120 | | /* Convert to space */ |
4121 | 2.65M | if (chunkSize > 0) { |
4122 | 44.2k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4123 | 44.2k | chunkSize = 0; |
4124 | 44.2k | } |
4125 | | |
4126 | 2.65M | xmlSBufAddCString(&buf, " ", 1); |
4127 | 2.65M | } else { |
4128 | 173k | chunkSize += 1; |
4129 | 173k | } |
4130 | | |
4131 | 3.75M | inSpace = 1; |
4132 | | |
4133 | 3.75M | if ((c == 0xD) && (NXT(1) == 0xA)) |
4134 | 2.23k | CUR_PTR++; |
4135 | 3.75M | } |
4136 | | |
4137 | 49.7M | NEXTL(1); |
4138 | 49.7M | } else if (NXT(1) == '#') { |
4139 | 25.3k | int val; |
4140 | | |
4141 | 25.3k | if (chunkSize > 0) { |
4142 | 10.4k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4143 | 10.4k | chunkSize = 0; |
4144 | 10.4k | } |
4145 | | |
4146 | 25.3k | val = xmlParseCharRef(ctxt); |
4147 | 25.3k | if (val == 0) |
4148 | 2.93k | goto error; |
4149 | | |
4150 | 22.4k | if ((val == '&') && (!replaceEntities)) { |
4151 | | /* |
4152 | | * The reparsing will be done in xmlNodeParseContent() |
4153 | | * called from SAX2.c |
4154 | | */ |
4155 | 3.11k | xmlSBufAddCString(&buf, "&", 5); |
4156 | 3.11k | inSpace = 0; |
4157 | 19.3k | } else if (val == ' ') { |
4158 | 4.63k | if ((normalize) && (inSpace)) |
4159 | 237 | attvalFlags |= XML_ATTVAL_NORM_CHANGE; |
4160 | 4.40k | else |
4161 | 4.40k | xmlSBufAddCString(&buf, " ", 1); |
4162 | 4.63k | inSpace = 1; |
4163 | 14.7k | } else { |
4164 | 14.7k | xmlSBufAddChar(&buf, val); |
4165 | 14.7k | inSpace = 0; |
4166 | 14.7k | } |
4167 | 1.12M | } else { |
4168 | 1.12M | const xmlChar *name; |
4169 | 1.12M | xmlEntityPtr ent; |
4170 | | |
4171 | 1.12M | if (chunkSize > 0) { |
4172 | 83.4k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4173 | 83.4k | chunkSize = 0; |
4174 | 83.4k | } |
4175 | | |
4176 | 1.12M | name = xmlParseEntityRefInternal(ctxt); |
4177 | 1.12M | if (name == NULL) { |
4178 | | /* |
4179 | | * Probably a literal '&' which wasn't escaped. |
4180 | | * TODO: Handle gracefully in recovery mode. |
4181 | | */ |
4182 | 50.6k | continue; |
4183 | 50.6k | } |
4184 | | |
4185 | 1.07M | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 1); |
4186 | 1.07M | if (ent == NULL) |
4187 | 88.9k | continue; |
4188 | | |
4189 | 982k | if (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY) { |
4190 | 7.52k | if ((ent->content[0] == '&') && (!replaceEntities)) |
4191 | 1.72k | xmlSBufAddCString(&buf, "&", 5); |
4192 | 5.79k | else |
4193 | 5.79k | xmlSBufAddString(&buf, ent->content, ent->length); |
4194 | 7.52k | inSpace = 0; |
4195 | 974k | } else if (replaceEntities) { |
4196 | 289k | if (xmlExpandEntityInAttValue(ctxt, &buf, |
4197 | 289k | ent->content, ent, normalize, &inSpace, ctxt->inputNr, |
4198 | 289k | /* check */ 1) > 0) |
4199 | 51.0k | attvalFlags |= XML_ATTVAL_NORM_CHANGE; |
4200 | 684k | } else { |
4201 | 684k | if ((ent->flags & entFlags) != entFlags) |
4202 | 1.16k | xmlCheckEntityInAttValue(ctxt, ent, ctxt->inputNr); |
4203 | | |
4204 | 684k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) { |
4205 | 167 | ent->content[0] = 0; |
4206 | 167 | goto error; |
4207 | 167 | } |
4208 | | |
4209 | | /* |
4210 | | * Just output the reference |
4211 | | */ |
4212 | 684k | xmlSBufAddCString(&buf, "&", 1); |
4213 | 684k | xmlSBufAddString(&buf, ent->name, xmlStrlen(ent->name)); |
4214 | 684k | xmlSBufAddCString(&buf, ";", 1); |
4215 | | |
4216 | 684k | inSpace = 0; |
4217 | 684k | } |
4218 | 982k | } |
4219 | 130M | } |
4220 | | |
4221 | 312k | if ((buf.mem == NULL) && (outFlags != NULL)) { |
4222 | 201k | ret = (xmlChar *) CUR_PTR - chunkSize; |
4223 | | |
4224 | 201k | if (attlen != NULL) |
4225 | 201k | *attlen = chunkSize; |
4226 | 201k | if ((normalize) && (inSpace) && (chunkSize > 0)) { |
4227 | 275 | attvalFlags |= XML_ATTVAL_NORM_CHANGE; |
4228 | 275 | *attlen -= 1; |
4229 | 275 | } |
4230 | | |
4231 | | /* Report potential error */ |
4232 | 201k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4233 | 201k | } else { |
4234 | 110k | if (chunkSize > 0) |
4235 | 82.1k | xmlSBufAddString(&buf, CUR_PTR - chunkSize, chunkSize); |
4236 | | |
4237 | 110k | if ((normalize) && (inSpace) && (buf.size > 0)) { |
4238 | 2.20k | attvalFlags |= XML_ATTVAL_NORM_CHANGE; |
4239 | 2.20k | buf.size--; |
4240 | 2.20k | } |
4241 | | |
4242 | 110k | ret = xmlSBufFinish(&buf, attlen, ctxt, "AttValue length too long"); |
4243 | 110k | attvalFlags |= XML_ATTVAL_ALLOC; |
4244 | | |
4245 | 110k | if (ret != NULL) { |
4246 | 110k | if (attlen != NULL) |
4247 | 62.2k | *attlen = buf.size; |
4248 | 110k | } |
4249 | 110k | } |
4250 | | |
4251 | 312k | if (outFlags != NULL) |
4252 | 264k | *outFlags = attvalFlags; |
4253 | | |
4254 | 312k | NEXTL(1); |
4255 | | |
4256 | 312k | return(ret); |
4257 | | |
4258 | 12.3k | error: |
4259 | 12.3k | xmlSBufCleanup(&buf, ctxt, "AttValue length too long"); |
4260 | 12.3k | return(NULL); |
4261 | 324k | } |
4262 | | |
4263 | | /** |
4264 | | * Parse a value for an attribute |
4265 | | * Note: the parser won't do substitution of entities here, this |
4266 | | * will be handled later in #xmlStringGetNodeList |
4267 | | * |
4268 | | * @deprecated Internal function, don't use. |
4269 | | * |
4270 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4271 | | * "'" ([^<&'] | Reference)* "'" |
4272 | | * |
4273 | | * 3.3.3 Attribute-Value Normalization: |
4274 | | * |
4275 | | * Before the value of an attribute is passed to the application or |
4276 | | * checked for validity, the XML processor must normalize it as follows: |
4277 | | * |
4278 | | * - a character reference is processed by appending the referenced |
4279 | | * character to the attribute value |
4280 | | * - an entity reference is processed by recursively processing the |
4281 | | * replacement text of the entity |
4282 | | * - a whitespace character (\#x20, \#xD, \#xA, \#x9) is processed by |
4283 | | * appending \#x20 to the normalized value, except that only a single |
4284 | | * \#x20 is appended for a "#xD#xA" sequence that is part of an external |
4285 | | * parsed entity or the literal entity value of an internal parsed entity |
4286 | | * - other characters are processed by appending them to the normalized value |
4287 | | * |
4288 | | * If the declared value is not CDATA, then the XML processor must further |
4289 | | * process the normalized attribute value by discarding any leading and |
4290 | | * trailing space (\#x20) characters, and by replacing sequences of space |
4291 | | * (\#x20) characters by a single space (\#x20) character. |
4292 | | * All attributes for which no declaration has been read should be treated |
4293 | | * by a non-validating parser as if declared CDATA. |
4294 | | * |
4295 | | * @param ctxt an XML parser context |
4296 | | * @returns the AttValue parsed or NULL. The value has to be freed by the |
4297 | | * caller. |
4298 | | */ |
4299 | | xmlChar * |
4300 | 56.1k | xmlParseAttValue(xmlParserCtxt *ctxt) { |
4301 | 56.1k | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4302 | 56.1k | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0, 0)); |
4303 | 56.1k | } |
4304 | | |
4305 | | /** |
4306 | | * Parse an XML Literal |
4307 | | * |
4308 | | * @deprecated Internal function, don't use. |
4309 | | * |
4310 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4311 | | * |
4312 | | * @param ctxt an XML parser context |
4313 | | * @returns the SystemLiteral parsed or NULL |
4314 | | */ |
4315 | | |
4316 | | xmlChar * |
4317 | 29.3k | xmlParseSystemLiteral(xmlParserCtxt *ctxt) { |
4318 | 29.3k | xmlChar *buf = NULL; |
4319 | 29.3k | int len = 0; |
4320 | 29.3k | int size = XML_PARSER_BUFFER_SIZE; |
4321 | 29.3k | int cur, l; |
4322 | 29.3k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4323 | 12.1k | XML_MAX_TEXT_LENGTH : |
4324 | 29.3k | XML_MAX_NAME_LENGTH; |
4325 | 29.3k | xmlChar stop; |
4326 | | |
4327 | 29.3k | if (RAW == '"') { |
4328 | 14.6k | NEXT; |
4329 | 14.6k | stop = '"'; |
4330 | 14.6k | } else if (RAW == '\'') { |
4331 | 12.0k | NEXT; |
4332 | 12.0k | stop = '\''; |
4333 | 12.0k | } else { |
4334 | 2.59k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4335 | 2.59k | return(NULL); |
4336 | 2.59k | } |
4337 | | |
4338 | 26.7k | buf = xmlMalloc(size); |
4339 | 26.7k | if (buf == NULL) { |
4340 | 23 | xmlErrMemory(ctxt); |
4341 | 23 | return(NULL); |
4342 | 23 | } |
4343 | 26.6k | cur = xmlCurrentCharRecover(ctxt, &l); |
4344 | 12.9M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4345 | 12.8M | if (len + 5 >= size) { |
4346 | 6.09k | xmlChar *tmp; |
4347 | 6.09k | int newSize; |
4348 | | |
4349 | 6.09k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
4350 | 6.09k | if (newSize < 0) { |
4351 | 3 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4352 | 3 | xmlFree(buf); |
4353 | 3 | return(NULL); |
4354 | 3 | } |
4355 | 6.09k | tmp = xmlRealloc(buf, newSize); |
4356 | 6.09k | if (tmp == NULL) { |
4357 | 9 | xmlFree(buf); |
4358 | 9 | xmlErrMemory(ctxt); |
4359 | 9 | return(NULL); |
4360 | 9 | } |
4361 | 6.08k | buf = tmp; |
4362 | 6.08k | size = newSize; |
4363 | 6.08k | } |
4364 | 12.8M | COPY_BUF(buf, len, cur); |
4365 | 12.8M | NEXTL(l); |
4366 | 12.8M | cur = xmlCurrentCharRecover(ctxt, &l); |
4367 | 12.8M | } |
4368 | 26.6k | buf[len] = 0; |
4369 | 26.6k | if (!IS_CHAR(cur)) { |
4370 | 1.22k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4371 | 25.4k | } else { |
4372 | 25.4k | NEXT; |
4373 | 25.4k | } |
4374 | 26.6k | return(buf); |
4375 | 26.6k | } |
4376 | | |
4377 | | /** |
4378 | | * Parse an XML public literal |
4379 | | * |
4380 | | * @deprecated Internal function, don't use. |
4381 | | * |
4382 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4383 | | * |
4384 | | * @param ctxt an XML parser context |
4385 | | * @returns the PubidLiteral parsed or NULL. |
4386 | | */ |
4387 | | |
4388 | | xmlChar * |
4389 | 9.24k | xmlParsePubidLiteral(xmlParserCtxt *ctxt) { |
4390 | 9.24k | xmlChar *buf = NULL; |
4391 | 9.24k | int len = 0; |
4392 | 9.24k | int size = XML_PARSER_BUFFER_SIZE; |
4393 | 9.24k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4394 | 4.42k | XML_MAX_TEXT_LENGTH : |
4395 | 9.24k | XML_MAX_NAME_LENGTH; |
4396 | 9.24k | xmlChar cur; |
4397 | 9.24k | xmlChar stop; |
4398 | | |
4399 | 9.24k | if (RAW == '"') { |
4400 | 5.61k | NEXT; |
4401 | 5.61k | stop = '"'; |
4402 | 5.61k | } else if (RAW == '\'') { |
4403 | 3.36k | NEXT; |
4404 | 3.36k | stop = '\''; |
4405 | 3.36k | } else { |
4406 | 261 | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4407 | 261 | return(NULL); |
4408 | 261 | } |
4409 | 8.98k | buf = xmlMalloc(size); |
4410 | 8.98k | if (buf == NULL) { |
4411 | 18 | xmlErrMemory(ctxt); |
4412 | 18 | return(NULL); |
4413 | 18 | } |
4414 | 8.96k | cur = CUR; |
4415 | 777k | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop) && |
4416 | 777k | (PARSER_STOPPED(ctxt) == 0)) { /* checked */ |
4417 | 768k | if (len + 1 >= size) { |
4418 | 1.27k | xmlChar *tmp; |
4419 | 1.27k | int newSize; |
4420 | | |
4421 | 1.27k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
4422 | 1.27k | if (newSize < 0) { |
4423 | 3 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4424 | 3 | xmlFree(buf); |
4425 | 3 | return(NULL); |
4426 | 3 | } |
4427 | 1.27k | tmp = xmlRealloc(buf, newSize); |
4428 | 1.27k | if (tmp == NULL) { |
4429 | 5 | xmlErrMemory(ctxt); |
4430 | 5 | xmlFree(buf); |
4431 | 5 | return(NULL); |
4432 | 5 | } |
4433 | 1.27k | buf = tmp; |
4434 | 1.27k | size = newSize; |
4435 | 1.27k | } |
4436 | 768k | buf[len++] = cur; |
4437 | 768k | NEXT; |
4438 | 768k | cur = CUR; |
4439 | 768k | } |
4440 | 8.95k | buf[len] = 0; |
4441 | 8.95k | if (cur != stop) { |
4442 | 1.22k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4443 | 7.73k | } else { |
4444 | 7.73k | NEXTL(1); |
4445 | 7.73k | } |
4446 | 8.95k | return(buf); |
4447 | 8.96k | } |
4448 | | |
4449 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial); |
4450 | | |
4451 | | /* |
4452 | | * used for the test in the inner loop of the char data testing |
4453 | | */ |
4454 | | static const unsigned char test_char_data[256] = { |
4455 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4456 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4457 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4458 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4459 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4460 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4461 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4462 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4463 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4464 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4465 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4466 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4467 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4468 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4469 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4470 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4471 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4472 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4473 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4474 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4475 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4476 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4477 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4478 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4479 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4480 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4481 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4482 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4483 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4484 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4485 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4486 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4487 | | }; |
4488 | | |
4489 | | static void |
4490 | | xmlCharacters(xmlParserCtxtPtr ctxt, const xmlChar *buf, int size, |
4491 | 1.38M | int isBlank) { |
4492 | 1.38M | int checkBlanks; |
4493 | | |
4494 | 1.38M | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
4495 | 64.5k | return; |
4496 | | |
4497 | 1.31M | checkBlanks = (!ctxt->keepBlanks) || |
4498 | 1.31M | (ctxt->sax->ignorableWhitespace != ctxt->sax->characters); |
4499 | | |
4500 | | /* |
4501 | | * Calling areBlanks with only parts of a text node |
4502 | | * is fundamentally broken, making the NOBLANKS option |
4503 | | * essentially unusable. |
4504 | | */ |
4505 | 1.31M | if ((checkBlanks) && |
4506 | 1.31M | (areBlanks(ctxt, buf, size, isBlank))) { |
4507 | 14.7k | if ((ctxt->sax->ignorableWhitespace != NULL) && |
4508 | 14.7k | (ctxt->keepBlanks)) |
4509 | 0 | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, size); |
4510 | 1.30M | } else { |
4511 | 1.30M | if (ctxt->sax->characters != NULL) |
4512 | 1.30M | ctxt->sax->characters(ctxt->userData, buf, size); |
4513 | | |
4514 | | /* |
4515 | | * The old code used to update this value for "complex" data |
4516 | | * even if checkBlanks was false. This was probably a bug. |
4517 | | */ |
4518 | 1.30M | if ((checkBlanks) && (*ctxt->space == -1)) |
4519 | 65.0k | *ctxt->space = -2; |
4520 | 1.30M | } |
4521 | 1.31M | } |
4522 | | |
4523 | | /** |
4524 | | * Parse character data. Always makes progress if the first char isn't |
4525 | | * '<' or '&'. |
4526 | | * |
4527 | | * The right angle bracket (>) may be represented using the string ">", |
4528 | | * and must, for compatibility, be escaped using ">" or a character |
4529 | | * reference when it appears in the string "]]>" in content, when that |
4530 | | * string is not marking the end of a CDATA section. |
4531 | | * |
4532 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4533 | | * @param ctxt an XML parser context |
4534 | | * @param partial buffer may contain partial UTF-8 sequences |
4535 | | */ |
4536 | | static void |
4537 | 3.43M | xmlParseCharDataInternal(xmlParserCtxtPtr ctxt, int partial) { |
4538 | 3.43M | const xmlChar *in; |
4539 | 3.43M | int line = ctxt->input->line; |
4540 | 3.43M | int col = ctxt->input->col; |
4541 | 3.43M | int ccol; |
4542 | 3.43M | int terminate = 0; |
4543 | | |
4544 | 3.43M | GROW; |
4545 | | /* |
4546 | | * Accelerated common case where input don't need to be |
4547 | | * modified before passing it to the handler. |
4548 | | */ |
4549 | 3.43M | in = ctxt->input->cur; |
4550 | 3.46M | do { |
4551 | 3.53M | get_more_space: |
4552 | 3.64M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4553 | 3.53M | if (*in == 0xA) { |
4554 | 7.24M | do { |
4555 | 7.24M | ctxt->input->line++; ctxt->input->col = 1; |
4556 | 7.24M | in++; |
4557 | 7.24M | } while (*in == 0xA); |
4558 | 74.7k | goto get_more_space; |
4559 | 74.7k | } |
4560 | 3.46M | if (*in == '<') { |
4561 | 120k | while (in > ctxt->input->cur) { |
4562 | 60.3k | const xmlChar *tmp = ctxt->input->cur; |
4563 | 60.3k | size_t nbchar = in - tmp; |
4564 | | |
4565 | 60.3k | if (nbchar > XML_MAX_ITEMS) |
4566 | 0 | nbchar = XML_MAX_ITEMS; |
4567 | 60.3k | ctxt->input->cur += nbchar; |
4568 | | |
4569 | 60.3k | xmlCharacters(ctxt, tmp, nbchar, 1); |
4570 | 60.3k | } |
4571 | 60.3k | return; |
4572 | 60.3k | } |
4573 | | |
4574 | 5.46M | get_more: |
4575 | 5.46M | ccol = ctxt->input->col; |
4576 | 11.4M | while (test_char_data[*in]) { |
4577 | 5.98M | in++; |
4578 | 5.98M | ccol++; |
4579 | 5.98M | } |
4580 | 5.46M | ctxt->input->col = ccol; |
4581 | 5.46M | if (*in == 0xA) { |
4582 | 4.37M | do { |
4583 | 4.37M | ctxt->input->line++; ctxt->input->col = 1; |
4584 | 4.37M | in++; |
4585 | 4.37M | } while (*in == 0xA); |
4586 | 80.8k | goto get_more; |
4587 | 80.8k | } |
4588 | 5.38M | if (*in == ']') { |
4589 | 1.98M | size_t avail = ctxt->input->end - in; |
4590 | | |
4591 | 1.98M | if (partial && avail < 2) { |
4592 | 363 | terminate = 1; |
4593 | 363 | goto invoke_callback; |
4594 | 363 | } |
4595 | 1.98M | if (in[1] == ']') { |
4596 | 1.70M | if (partial && avail < 3) { |
4597 | 1.86k | terminate = 1; |
4598 | 1.86k | goto invoke_callback; |
4599 | 1.86k | } |
4600 | 1.70M | if (in[2] == '>') |
4601 | 5.31k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4602 | 1.70M | } |
4603 | | |
4604 | 1.98M | in++; |
4605 | 1.98M | ctxt->input->col++; |
4606 | 1.98M | goto get_more; |
4607 | 1.98M | } |
4608 | | |
4609 | 3.40M | invoke_callback: |
4610 | 3.84M | while (in > ctxt->input->cur) { |
4611 | 441k | const xmlChar *tmp = ctxt->input->cur; |
4612 | 441k | size_t nbchar = in - tmp; |
4613 | | |
4614 | 441k | if (nbchar > XML_MAX_ITEMS) |
4615 | 0 | nbchar = XML_MAX_ITEMS; |
4616 | 441k | ctxt->input->cur += nbchar; |
4617 | | |
4618 | 441k | xmlCharacters(ctxt, tmp, nbchar, 0); |
4619 | | |
4620 | 441k | line = ctxt->input->line; |
4621 | 441k | col = ctxt->input->col; |
4622 | 441k | } |
4623 | 3.40M | ctxt->input->cur = in; |
4624 | 3.40M | if (*in == 0xD) { |
4625 | 26.4k | in++; |
4626 | 26.4k | if (*in == 0xA) { |
4627 | 20.2k | ctxt->input->cur = in; |
4628 | 20.2k | in++; |
4629 | 20.2k | ctxt->input->line++; ctxt->input->col = 1; |
4630 | 20.2k | continue; /* while */ |
4631 | 20.2k | } |
4632 | 6.21k | in--; |
4633 | 6.21k | } |
4634 | 3.37M | if (*in == '<') { |
4635 | 191k | return; |
4636 | 191k | } |
4637 | 3.18M | if (*in == '&') { |
4638 | 58.9k | return; |
4639 | 58.9k | } |
4640 | 3.12M | if (terminate) { |
4641 | 2.23k | return; |
4642 | 2.23k | } |
4643 | 3.12M | SHRINK; |
4644 | 3.12M | GROW; |
4645 | 3.12M | in = ctxt->input->cur; |
4646 | 3.14M | } while (((*in >= 0x20) && (*in <= 0x7F)) || |
4647 | 3.14M | (*in == 0x09) || (*in == 0x0a)); |
4648 | 3.12M | ctxt->input->line = line; |
4649 | 3.12M | ctxt->input->col = col; |
4650 | 3.12M | xmlParseCharDataComplex(ctxt, partial); |
4651 | 3.12M | } |
4652 | | |
4653 | | /** |
4654 | | * Always makes progress if the first char isn't '<' or '&'. |
4655 | | * |
4656 | | * parse a CharData section.this is the fallback function |
4657 | | * of #xmlParseCharData when the parsing requires handling |
4658 | | * of non-ASCII characters. |
4659 | | * |
4660 | | * @param ctxt an XML parser context |
4661 | | * @param partial whether the input can end with truncated UTF-8 |
4662 | | */ |
4663 | | static void |
4664 | 3.12M | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt, int partial) { |
4665 | 3.12M | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4666 | 3.12M | int nbchar = 0; |
4667 | 3.12M | int cur, l; |
4668 | | |
4669 | 3.12M | cur = xmlCurrentCharRecover(ctxt, &l); |
4670 | 71.5M | while ((cur != '<') && /* checked */ |
4671 | 71.5M | (cur != '&') && |
4672 | 71.5M | (IS_CHAR(cur))) { |
4673 | 68.4M | if (cur == ']') { |
4674 | 2.38M | size_t avail = ctxt->input->end - ctxt->input->cur; |
4675 | | |
4676 | 2.38M | if (partial && avail < 2) |
4677 | 139 | break; |
4678 | 2.38M | if (NXT(1) == ']') { |
4679 | 2.13M | if (partial && avail < 3) |
4680 | 742 | break; |
4681 | 2.13M | if (NXT(2) == '>') |
4682 | 1.82k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4683 | 2.13M | } |
4684 | 2.38M | } |
4685 | | |
4686 | 68.4M | COPY_BUF(buf, nbchar, cur); |
4687 | | /* move current position before possible calling of ctxt->sax->characters */ |
4688 | 68.4M | NEXTL(l); |
4689 | 68.4M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4690 | 580k | buf[nbchar] = 0; |
4691 | | |
4692 | 580k | xmlCharacters(ctxt, buf, nbchar, 0); |
4693 | 580k | nbchar = 0; |
4694 | 580k | SHRINK; |
4695 | 580k | } |
4696 | 68.4M | cur = xmlCurrentCharRecover(ctxt, &l); |
4697 | 68.4M | } |
4698 | 3.12M | if (nbchar != 0) { |
4699 | 299k | buf[nbchar] = 0; |
4700 | | |
4701 | 299k | xmlCharacters(ctxt, buf, nbchar, 0); |
4702 | 299k | } |
4703 | | /* |
4704 | | * cur == 0 can mean |
4705 | | * |
4706 | | * - End of buffer. |
4707 | | * - An actual 0 character. |
4708 | | * - An incomplete UTF-8 sequence. This is allowed if partial is set. |
4709 | | */ |
4710 | 3.12M | if (ctxt->input->cur < ctxt->input->end) { |
4711 | 3.11M | if ((cur == 0) && (CUR != 0)) { |
4712 | 1.52k | if (partial == 0) { |
4713 | 1.31k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4714 | 1.31k | "Incomplete UTF-8 sequence starting with %02X\n", CUR); |
4715 | 1.31k | NEXTL(1); |
4716 | 1.31k | } |
4717 | 3.11M | } else if ((cur != '<') && (cur != '&') && (cur != ']')) { |
4718 | | /* Generate the error and skip the offending character */ |
4719 | 2.99M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4720 | 2.99M | "PCDATA invalid Char value %d\n", cur); |
4721 | 2.99M | NEXTL(l); |
4722 | 2.99M | } |
4723 | 3.11M | } |
4724 | 3.12M | } |
4725 | | |
4726 | | /** |
4727 | | * @deprecated Internal function, don't use. |
4728 | | * @param ctxt an XML parser context |
4729 | | * @param cdata unused |
4730 | | */ |
4731 | | void |
4732 | 0 | xmlParseCharData(xmlParserCtxt *ctxt, ATTRIBUTE_UNUSED int cdata) { |
4733 | 0 | xmlParseCharDataInternal(ctxt, 0); |
4734 | 0 | } |
4735 | | |
4736 | | /** |
4737 | | * Parse an External ID or a Public ID |
4738 | | * |
4739 | | * @deprecated Internal function, don't use. |
4740 | | * |
4741 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4742 | | * `'PUBLIC' S PubidLiteral S SystemLiteral` |
4743 | | * |
4744 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4745 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4746 | | * |
4747 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4748 | | * |
4749 | | * @param ctxt an XML parser context |
4750 | | * @param publicId a xmlChar** receiving PubidLiteral |
4751 | | * @param strict indicate whether we should restrict parsing to only |
4752 | | * production [75], see NOTE below |
4753 | | * @returns the function returns SystemLiteral and in the second |
4754 | | * case publicID receives PubidLiteral, is strict is off |
4755 | | * it is possible to return NULL and have publicID set. |
4756 | | */ |
4757 | | |
4758 | | xmlChar * |
4759 | 56.1k | xmlParseExternalID(xmlParserCtxt *ctxt, xmlChar **publicId, int strict) { |
4760 | 56.1k | xmlChar *URI = NULL; |
4761 | | |
4762 | 56.1k | *publicId = NULL; |
4763 | 56.1k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4764 | 22.3k | SKIP(6); |
4765 | 22.3k | if (SKIP_BLANKS == 0) { |
4766 | 823 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4767 | 823 | "Space required after 'SYSTEM'\n"); |
4768 | 823 | } |
4769 | 22.3k | URI = xmlParseSystemLiteral(ctxt); |
4770 | 22.3k | if (URI == NULL) { |
4771 | 760 | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4772 | 760 | } |
4773 | 33.7k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4774 | 9.24k | SKIP(6); |
4775 | 9.24k | if (SKIP_BLANKS == 0) { |
4776 | 782 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4777 | 782 | "Space required after 'PUBLIC'\n"); |
4778 | 782 | } |
4779 | 9.24k | *publicId = xmlParsePubidLiteral(ctxt); |
4780 | 9.24k | if (*publicId == NULL) { |
4781 | 287 | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4782 | 287 | } |
4783 | 9.24k | if (strict) { |
4784 | | /* |
4785 | | * We don't handle [83] so "S SystemLiteral" is required. |
4786 | | */ |
4787 | 6.50k | if (SKIP_BLANKS == 0) { |
4788 | 2.17k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4789 | 2.17k | "Space required after the Public Identifier\n"); |
4790 | 2.17k | } |
4791 | 6.50k | } else { |
4792 | | /* |
4793 | | * We handle [83] so we return immediately, if |
4794 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4795 | | * system literal was found, but this is harmless since we must |
4796 | | * be at the end of a NotationDecl. |
4797 | | */ |
4798 | 2.73k | if (SKIP_BLANKS == 0) return(NULL); |
4799 | 791 | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4800 | 791 | } |
4801 | 6.93k | URI = xmlParseSystemLiteral(ctxt); |
4802 | 6.93k | if (URI == NULL) { |
4803 | 1.86k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4804 | 1.86k | } |
4805 | 6.93k | } |
4806 | 53.8k | return(URI); |
4807 | 56.1k | } |
4808 | | |
4809 | | /** |
4810 | | * Skip an XML (SGML) comment <!-- .... --> |
4811 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4812 | | * must not occur within comments. " |
4813 | | * This is the slow routine in case the accelerator for ascii didn't work |
4814 | | * |
4815 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4816 | | * @param ctxt an XML parser context |
4817 | | * @param buf the already parsed part of the buffer |
4818 | | * @param len number of bytes in the buffer |
4819 | | * @param size allocated size of the buffer |
4820 | | */ |
4821 | | static void |
4822 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4823 | 16.9k | size_t len, size_t size) { |
4824 | 16.9k | int q, ql; |
4825 | 16.9k | int r, rl; |
4826 | 16.9k | int cur, l; |
4827 | 16.9k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4828 | 6.90k | XML_MAX_HUGE_LENGTH : |
4829 | 16.9k | XML_MAX_TEXT_LENGTH; |
4830 | | |
4831 | 16.9k | if (buf == NULL) { |
4832 | 4.42k | len = 0; |
4833 | 4.42k | size = XML_PARSER_BUFFER_SIZE; |
4834 | 4.42k | buf = xmlMalloc(size); |
4835 | 4.42k | if (buf == NULL) { |
4836 | 25 | xmlErrMemory(ctxt); |
4837 | 25 | return; |
4838 | 25 | } |
4839 | 4.42k | } |
4840 | 16.9k | q = xmlCurrentCharRecover(ctxt, &ql); |
4841 | 16.9k | if (q == 0) |
4842 | 3.63k | goto not_terminated; |
4843 | 13.2k | if (!IS_CHAR(q)) { |
4844 | 761 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4845 | 761 | "xmlParseComment: invalid xmlChar value %d\n", |
4846 | 761 | q); |
4847 | 761 | xmlFree (buf); |
4848 | 761 | return; |
4849 | 761 | } |
4850 | 12.5k | NEXTL(ql); |
4851 | 12.5k | r = xmlCurrentCharRecover(ctxt, &rl); |
4852 | 12.5k | if (r == 0) |
4853 | 376 | goto not_terminated; |
4854 | 12.1k | if (!IS_CHAR(r)) { |
4855 | 763 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4856 | 763 | "xmlParseComment: invalid xmlChar value %d\n", |
4857 | 763 | r); |
4858 | 763 | xmlFree (buf); |
4859 | 763 | return; |
4860 | 763 | } |
4861 | 11.3k | NEXTL(rl); |
4862 | 11.3k | cur = xmlCurrentCharRecover(ctxt, &l); |
4863 | 11.3k | if (cur == 0) |
4864 | 487 | goto not_terminated; |
4865 | 19.4M | while (IS_CHAR(cur) && /* checked */ |
4866 | 19.4M | ((cur != '>') || |
4867 | 19.4M | (r != '-') || (q != '-'))) { |
4868 | 19.3M | if ((r == '-') && (q == '-')) { |
4869 | 2.07M | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
4870 | 2.07M | } |
4871 | 19.3M | if (len + 5 >= size) { |
4872 | 5.44k | xmlChar *tmp; |
4873 | 5.44k | int newSize; |
4874 | | |
4875 | 5.44k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
4876 | 5.44k | if (newSize < 0) { |
4877 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4878 | 0 | "Comment too big found", NULL); |
4879 | 0 | xmlFree (buf); |
4880 | 0 | return; |
4881 | 0 | } |
4882 | 5.44k | tmp = xmlRealloc(buf, newSize); |
4883 | 5.44k | if (tmp == NULL) { |
4884 | 10 | xmlErrMemory(ctxt); |
4885 | 10 | xmlFree(buf); |
4886 | 10 | return; |
4887 | 10 | } |
4888 | 5.43k | buf = tmp; |
4889 | 5.43k | size = newSize; |
4890 | 5.43k | } |
4891 | 19.3M | COPY_BUF(buf, len, q); |
4892 | | |
4893 | 19.3M | q = r; |
4894 | 19.3M | ql = rl; |
4895 | 19.3M | r = cur; |
4896 | 19.3M | rl = l; |
4897 | | |
4898 | 19.3M | NEXTL(l); |
4899 | 19.3M | cur = xmlCurrentCharRecover(ctxt, &l); |
4900 | | |
4901 | 19.3M | } |
4902 | 10.8k | buf[len] = 0; |
4903 | 10.8k | if (cur == 0) { |
4904 | 2.84k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4905 | 2.84k | "Comment not terminated \n<!--%.50s\n", buf); |
4906 | 8.05k | } else if (!IS_CHAR(cur)) { |
4907 | 751 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4908 | 751 | "xmlParseComment: invalid xmlChar value %d\n", |
4909 | 751 | cur); |
4910 | 7.30k | } else { |
4911 | 7.30k | NEXT; |
4912 | 7.30k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
4913 | 7.30k | (!ctxt->disableSAX)) |
4914 | 6.00k | ctxt->sax->comment(ctxt->userData, buf); |
4915 | 7.30k | } |
4916 | 10.8k | xmlFree(buf); |
4917 | 10.8k | return; |
4918 | 4.50k | not_terminated: |
4919 | 4.50k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4920 | 4.50k | "Comment not terminated\n", NULL); |
4921 | 4.50k | xmlFree(buf); |
4922 | 4.50k | } |
4923 | | |
4924 | | /** |
4925 | | * Parse an XML (SGML) comment. Always consumes '<!'. |
4926 | | * |
4927 | | * @deprecated Internal function, don't use. |
4928 | | * |
4929 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4930 | | * must not occur within comments. " |
4931 | | * |
4932 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4933 | | * @param ctxt an XML parser context |
4934 | | */ |
4935 | | void |
4936 | 352k | xmlParseComment(xmlParserCtxt *ctxt) { |
4937 | 352k | xmlChar *buf = NULL; |
4938 | 352k | size_t size = XML_PARSER_BUFFER_SIZE; |
4939 | 352k | size_t len = 0; |
4940 | 352k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4941 | 88.3k | XML_MAX_HUGE_LENGTH : |
4942 | 352k | XML_MAX_TEXT_LENGTH; |
4943 | 352k | const xmlChar *in; |
4944 | 352k | size_t nbchar = 0; |
4945 | 352k | int ccol; |
4946 | | |
4947 | | /* |
4948 | | * Check that there is a comment right here. |
4949 | | */ |
4950 | 352k | if ((RAW != '<') || (NXT(1) != '!')) |
4951 | 0 | return; |
4952 | 352k | SKIP(2); |
4953 | 352k | if ((RAW != '-') || (NXT(1) != '-')) |
4954 | 35 | return; |
4955 | 352k | SKIP(2); |
4956 | 352k | GROW; |
4957 | | |
4958 | | /* |
4959 | | * Accelerated common case where input don't need to be |
4960 | | * modified before passing it to the handler. |
4961 | | */ |
4962 | 352k | in = ctxt->input->cur; |
4963 | 353k | do { |
4964 | 353k | if (*in == 0xA) { |
4965 | 4.17k | do { |
4966 | 4.17k | ctxt->input->line++; ctxt->input->col = 1; |
4967 | 4.17k | in++; |
4968 | 4.17k | } while (*in == 0xA); |
4969 | 3.93k | } |
4970 | 3.64M | get_more: |
4971 | 3.64M | ccol = ctxt->input->col; |
4972 | 11.9M | while (((*in > '-') && (*in <= 0x7F)) || |
4973 | 11.9M | ((*in >= 0x20) && (*in < '-')) || |
4974 | 11.9M | (*in == 0x09)) { |
4975 | 8.34M | in++; |
4976 | 8.34M | ccol++; |
4977 | 8.34M | } |
4978 | 3.64M | ctxt->input->col = ccol; |
4979 | 3.64M | if (*in == 0xA) { |
4980 | 86.4k | do { |
4981 | 86.4k | ctxt->input->line++; ctxt->input->col = 1; |
4982 | 86.4k | in++; |
4983 | 86.4k | } while (*in == 0xA); |
4984 | 25.0k | goto get_more; |
4985 | 25.0k | } |
4986 | 3.62M | nbchar = in - ctxt->input->cur; |
4987 | | /* |
4988 | | * save current set of data |
4989 | | */ |
4990 | 3.62M | if (nbchar > 0) { |
4991 | 3.29M | if (nbchar > maxLength - len) { |
4992 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4993 | 0 | "Comment too big found", NULL); |
4994 | 0 | xmlFree(buf); |
4995 | 0 | return; |
4996 | 0 | } |
4997 | 3.29M | if (buf == NULL) { |
4998 | 37.4k | if ((*in == '-') && (in[1] == '-')) |
4999 | 17.4k | size = nbchar + 1; |
5000 | 19.9k | else |
5001 | 19.9k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
5002 | 37.4k | buf = xmlMalloc(size); |
5003 | 37.4k | if (buf == NULL) { |
5004 | 26 | xmlErrMemory(ctxt); |
5005 | 26 | return; |
5006 | 26 | } |
5007 | 37.3k | len = 0; |
5008 | 3.25M | } else if (len + nbchar + 1 >= size) { |
5009 | 12.5k | xmlChar *new_buf; |
5010 | 12.5k | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
5011 | 12.5k | new_buf = xmlRealloc(buf, size); |
5012 | 12.5k | if (new_buf == NULL) { |
5013 | 7 | xmlErrMemory(ctxt); |
5014 | 7 | xmlFree(buf); |
5015 | 7 | return; |
5016 | 7 | } |
5017 | 12.5k | buf = new_buf; |
5018 | 12.5k | } |
5019 | 3.29M | memcpy(&buf[len], ctxt->input->cur, nbchar); |
5020 | 3.29M | len += nbchar; |
5021 | 3.29M | buf[len] = 0; |
5022 | 3.29M | } |
5023 | 3.62M | ctxt->input->cur = in; |
5024 | 3.62M | if (*in == 0xA) { |
5025 | 0 | in++; |
5026 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5027 | 0 | } |
5028 | 3.62M | if (*in == 0xD) { |
5029 | 6.93k | in++; |
5030 | 6.93k | if (*in == 0xA) { |
5031 | 6.21k | ctxt->input->cur = in; |
5032 | 6.21k | in++; |
5033 | 6.21k | ctxt->input->line++; ctxt->input->col = 1; |
5034 | 6.21k | goto get_more; |
5035 | 6.21k | } |
5036 | 724 | in--; |
5037 | 724 | } |
5038 | 3.61M | SHRINK; |
5039 | 3.61M | GROW; |
5040 | 3.61M | in = ctxt->input->cur; |
5041 | 3.61M | if (*in == '-') { |
5042 | 3.59M | if (in[1] == '-') { |
5043 | 2.16M | if (in[2] == '>') { |
5044 | 335k | SKIP(3); |
5045 | 335k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5046 | 335k | (!ctxt->disableSAX)) { |
5047 | 317k | if (buf != NULL) |
5048 | 20.2k | ctxt->sax->comment(ctxt->userData, buf); |
5049 | 297k | else |
5050 | 297k | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5051 | 317k | } |
5052 | 335k | if (buf != NULL) |
5053 | 24.8k | xmlFree(buf); |
5054 | 335k | return; |
5055 | 335k | } |
5056 | 1.83M | if (buf != NULL) { |
5057 | 1.82M | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5058 | 1.82M | "Double hyphen within comment: " |
5059 | 1.82M | "<!--%.50s\n", |
5060 | 1.82M | buf); |
5061 | 1.82M | } else |
5062 | 5.32k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5063 | 5.32k | "Double hyphen within comment\n", NULL); |
5064 | 1.83M | in++; |
5065 | 1.83M | ctxt->input->col++; |
5066 | 1.83M | } |
5067 | 3.26M | in++; |
5068 | 3.26M | ctxt->input->col++; |
5069 | 3.26M | goto get_more; |
5070 | 3.59M | } |
5071 | 3.61M | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5072 | 16.9k | xmlParseCommentComplex(ctxt, buf, len, size); |
5073 | 16.9k | } |
5074 | | |
5075 | | |
5076 | | /** |
5077 | | * Parse the name of a PI |
5078 | | * |
5079 | | * @deprecated Internal function, don't use. |
5080 | | * |
5081 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5082 | | * |
5083 | | * @param ctxt an XML parser context |
5084 | | * @returns the PITarget name or NULL |
5085 | | */ |
5086 | | |
5087 | | const xmlChar * |
5088 | 60.6k | xmlParsePITarget(xmlParserCtxt *ctxt) { |
5089 | 60.6k | const xmlChar *name; |
5090 | | |
5091 | 60.6k | name = xmlParseName(ctxt); |
5092 | 60.6k | if ((name != NULL) && |
5093 | 60.6k | ((name[0] == 'x') || (name[0] == 'X')) && |
5094 | 60.6k | ((name[1] == 'm') || (name[1] == 'M')) && |
5095 | 60.6k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5096 | 6.83k | int i; |
5097 | 6.83k | if ((name[0] == 'x') && (name[1] == 'm') && |
5098 | 6.83k | (name[2] == 'l') && (name[3] == 0)) { |
5099 | 5.19k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5100 | 5.19k | "XML declaration allowed only at the start of the document\n"); |
5101 | 5.19k | return(name); |
5102 | 5.19k | } else if (name[3] == 0) { |
5103 | 600 | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5104 | 600 | return(name); |
5105 | 600 | } |
5106 | 2.75k | for (i = 0;;i++) { |
5107 | 2.75k | if (xmlW3CPIs[i] == NULL) break; |
5108 | 1.93k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5109 | 226 | return(name); |
5110 | 1.93k | } |
5111 | 821 | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5112 | 821 | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5113 | 821 | NULL, NULL); |
5114 | 821 | } |
5115 | 54.5k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5116 | 512 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5117 | 512 | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5118 | 512 | } |
5119 | 54.5k | return(name); |
5120 | 60.6k | } |
5121 | | |
5122 | | #ifdef LIBXML_CATALOG_ENABLED |
5123 | | /** |
5124 | | * Parse an XML Catalog Processing Instruction. |
5125 | | * |
5126 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5127 | | * |
5128 | | * Occurs only if allowed by the user and if happening in the Misc |
5129 | | * part of the document before any doctype information |
5130 | | * This will add the given catalog to the parsing context in order |
5131 | | * to be used if there is a resolution need further down in the document |
5132 | | * |
5133 | | * @param ctxt an XML parser context |
5134 | | * @param catalog the PI value string |
5135 | | */ |
5136 | | |
5137 | | static void |
5138 | 0 | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5139 | 0 | xmlChar *URL = NULL; |
5140 | 0 | const xmlChar *tmp, *base; |
5141 | 0 | xmlChar marker; |
5142 | |
|
5143 | 0 | tmp = catalog; |
5144 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5145 | 0 | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5146 | 0 | goto error; |
5147 | 0 | tmp += 7; |
5148 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5149 | 0 | if (*tmp != '=') { |
5150 | 0 | return; |
5151 | 0 | } |
5152 | 0 | tmp++; |
5153 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5154 | 0 | marker = *tmp; |
5155 | 0 | if ((marker != '\'') && (marker != '"')) |
5156 | 0 | goto error; |
5157 | 0 | tmp++; |
5158 | 0 | base = tmp; |
5159 | 0 | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5160 | 0 | if (*tmp == 0) |
5161 | 0 | goto error; |
5162 | 0 | URL = xmlStrndup(base, tmp - base); |
5163 | 0 | tmp++; |
5164 | 0 | while (IS_BLANK_CH(*tmp)) tmp++; |
5165 | 0 | if (*tmp != 0) |
5166 | 0 | goto error; |
5167 | | |
5168 | 0 | if (URL != NULL) { |
5169 | | /* |
5170 | | * Unfortunately, the catalog API doesn't report OOM errors. |
5171 | | * xmlGetLastError isn't very helpful since we don't know |
5172 | | * where the last error came from. We'd have to reset it |
5173 | | * before this call and restore it afterwards. |
5174 | | */ |
5175 | 0 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5176 | 0 | xmlFree(URL); |
5177 | 0 | } |
5178 | 0 | return; |
5179 | | |
5180 | 0 | error: |
5181 | 0 | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5182 | 0 | "Catalog PI syntax error: %s\n", |
5183 | 0 | catalog, NULL); |
5184 | 0 | if (URL != NULL) |
5185 | 0 | xmlFree(URL); |
5186 | 0 | } |
5187 | | #endif |
5188 | | |
5189 | | /** |
5190 | | * Parse an XML Processing Instruction. |
5191 | | * |
5192 | | * @deprecated Internal function, don't use. |
5193 | | * |
5194 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5195 | | * |
5196 | | * The processing is transferred to SAX once parsed. |
5197 | | * |
5198 | | * @param ctxt an XML parser context |
5199 | | */ |
5200 | | |
5201 | | void |
5202 | 60.6k | xmlParsePI(xmlParserCtxt *ctxt) { |
5203 | 60.6k | xmlChar *buf = NULL; |
5204 | 60.6k | size_t len = 0; |
5205 | 60.6k | size_t size = XML_PARSER_BUFFER_SIZE; |
5206 | 60.6k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5207 | 23.7k | XML_MAX_HUGE_LENGTH : |
5208 | 60.6k | XML_MAX_TEXT_LENGTH; |
5209 | 60.6k | int cur, l; |
5210 | 60.6k | const xmlChar *target; |
5211 | | |
5212 | 60.6k | if ((RAW == '<') && (NXT(1) == '?')) { |
5213 | | /* |
5214 | | * this is a Processing Instruction. |
5215 | | */ |
5216 | 60.6k | SKIP(2); |
5217 | | |
5218 | | /* |
5219 | | * Parse the target name and check for special support like |
5220 | | * namespace. |
5221 | | */ |
5222 | 60.6k | target = xmlParsePITarget(ctxt); |
5223 | 60.6k | if (target != NULL) { |
5224 | 54.7k | if ((RAW == '?') && (NXT(1) == '>')) { |
5225 | 11.6k | SKIP(2); |
5226 | | |
5227 | | /* |
5228 | | * SAX: PI detected. |
5229 | | */ |
5230 | 11.6k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5231 | 11.6k | (ctxt->sax->processingInstruction != NULL)) |
5232 | 11.3k | ctxt->sax->processingInstruction(ctxt->userData, |
5233 | 11.3k | target, NULL); |
5234 | 11.6k | return; |
5235 | 11.6k | } |
5236 | 43.1k | buf = xmlMalloc(size); |
5237 | 43.1k | if (buf == NULL) { |
5238 | 48 | xmlErrMemory(ctxt); |
5239 | 48 | return; |
5240 | 48 | } |
5241 | 43.0k | if (SKIP_BLANKS == 0) { |
5242 | 24.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5243 | 24.7k | "ParsePI: PI %s space expected\n", target); |
5244 | 24.7k | } |
5245 | 43.0k | cur = xmlCurrentCharRecover(ctxt, &l); |
5246 | 11.8M | while (IS_CHAR(cur) && /* checked */ |
5247 | 11.8M | ((cur != '?') || (NXT(1) != '>'))) { |
5248 | 11.8M | if (len + 5 >= size) { |
5249 | 14.6k | xmlChar *tmp; |
5250 | 14.6k | int newSize; |
5251 | | |
5252 | 14.6k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
5253 | 14.6k | if (newSize < 0) { |
5254 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5255 | 0 | "PI %s too big found", target); |
5256 | 0 | xmlFree(buf); |
5257 | 0 | return; |
5258 | 0 | } |
5259 | 14.6k | tmp = xmlRealloc(buf, newSize); |
5260 | 14.6k | if (tmp == NULL) { |
5261 | 6 | xmlErrMemory(ctxt); |
5262 | 6 | xmlFree(buf); |
5263 | 6 | return; |
5264 | 6 | } |
5265 | 14.6k | buf = tmp; |
5266 | 14.6k | size = newSize; |
5267 | 14.6k | } |
5268 | 11.8M | COPY_BUF(buf, len, cur); |
5269 | 11.8M | NEXTL(l); |
5270 | 11.8M | cur = xmlCurrentCharRecover(ctxt, &l); |
5271 | 11.8M | } |
5272 | 43.0k | buf[len] = 0; |
5273 | 43.0k | if (cur != '?') { |
5274 | 11.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5275 | 11.5k | "ParsePI: PI %s never end ...\n", target); |
5276 | 31.5k | } else { |
5277 | 31.5k | SKIP(2); |
5278 | | |
5279 | 31.5k | #ifdef LIBXML_CATALOG_ENABLED |
5280 | 31.5k | if ((ctxt->inSubset == 0) && |
5281 | 31.5k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5282 | 3.60k | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5283 | | |
5284 | 3.60k | if ((ctxt->options & XML_PARSE_CATALOG_PI) && |
5285 | 3.60k | ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5286 | 1.02k | (allow == XML_CATA_ALLOW_ALL))) |
5287 | 0 | xmlParseCatalogPI(ctxt, buf); |
5288 | 3.60k | } |
5289 | 31.5k | #endif |
5290 | | |
5291 | | /* |
5292 | | * SAX: PI detected. |
5293 | | */ |
5294 | 31.5k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5295 | 31.5k | (ctxt->sax->processingInstruction != NULL)) |
5296 | 28.2k | ctxt->sax->processingInstruction(ctxt->userData, |
5297 | 28.2k | target, buf); |
5298 | 31.5k | } |
5299 | 43.0k | xmlFree(buf); |
5300 | 43.0k | } else { |
5301 | 5.81k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5302 | 5.81k | } |
5303 | 60.6k | } |
5304 | 60.6k | } |
5305 | | |
5306 | | /** |
5307 | | * Parse a notation declaration. Always consumes '<!'. |
5308 | | * |
5309 | | * @deprecated Internal function, don't use. |
5310 | | * |
5311 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) |
5312 | | * S? '>' |
5313 | | * |
5314 | | * Hence there is actually 3 choices: |
5315 | | * |
5316 | | * 'PUBLIC' S PubidLiteral |
5317 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5318 | | * 'SYSTEM' S SystemLiteral |
5319 | | * |
5320 | | * See the NOTE on #xmlParseExternalID. |
5321 | | * |
5322 | | * @param ctxt an XML parser context |
5323 | | */ |
5324 | | |
5325 | | void |
5326 | 10.0k | xmlParseNotationDecl(xmlParserCtxt *ctxt) { |
5327 | 10.0k | const xmlChar *name; |
5328 | 10.0k | xmlChar *Pubid; |
5329 | 10.0k | xmlChar *Systemid; |
5330 | | |
5331 | 10.0k | if ((CUR != '<') || (NXT(1) != '!')) |
5332 | 0 | return; |
5333 | 10.0k | SKIP(2); |
5334 | | |
5335 | 10.0k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5336 | 10.0k | #ifdef LIBXML_VALID_ENABLED |
5337 | 10.0k | int oldInputNr = ctxt->inputNr; |
5338 | 10.0k | #endif |
5339 | | |
5340 | 10.0k | SKIP(8); |
5341 | 10.0k | if (SKIP_BLANKS_PE == 0) { |
5342 | 410 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5343 | 410 | "Space required after '<!NOTATION'\n"); |
5344 | 410 | return; |
5345 | 410 | } |
5346 | | |
5347 | 9.61k | name = xmlParseName(ctxt); |
5348 | 9.61k | if (name == NULL) { |
5349 | 267 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5350 | 267 | return; |
5351 | 267 | } |
5352 | 9.35k | if (xmlStrchr(name, ':') != NULL) { |
5353 | 231 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5354 | 231 | "colons are forbidden from notation names '%s'\n", |
5355 | 231 | name, NULL, NULL); |
5356 | 231 | } |
5357 | 9.35k | if (SKIP_BLANKS_PE == 0) { |
5358 | 227 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5359 | 227 | "Space required after the NOTATION name'\n"); |
5360 | 227 | return; |
5361 | 227 | } |
5362 | | |
5363 | | /* |
5364 | | * Parse the IDs. |
5365 | | */ |
5366 | 9.12k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5367 | 9.12k | SKIP_BLANKS_PE; |
5368 | | |
5369 | 9.12k | if (RAW == '>') { |
5370 | 7.00k | #ifdef LIBXML_VALID_ENABLED |
5371 | 7.00k | if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) { |
5372 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5373 | 0 | "Notation declaration doesn't start and stop" |
5374 | 0 | " in the same entity\n", |
5375 | 0 | NULL, NULL); |
5376 | 0 | } |
5377 | 7.00k | #endif |
5378 | 7.00k | NEXT; |
5379 | 7.00k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5380 | 7.00k | (ctxt->sax->notationDecl != NULL)) |
5381 | 6.44k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5382 | 7.00k | } else { |
5383 | 2.12k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5384 | 2.12k | } |
5385 | 9.12k | if (Systemid != NULL) xmlFree(Systemid); |
5386 | 9.12k | if (Pubid != NULL) xmlFree(Pubid); |
5387 | 9.12k | } |
5388 | 10.0k | } |
5389 | | |
5390 | | /** |
5391 | | * Parse an entity declaration. Always consumes '<!'. |
5392 | | * |
5393 | | * @deprecated Internal function, don't use. |
5394 | | * |
5395 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5396 | | * |
5397 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5398 | | * |
5399 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5400 | | * |
5401 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5402 | | * |
5403 | | * [74] PEDef ::= EntityValue | ExternalID |
5404 | | * |
5405 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5406 | | * |
5407 | | * [ VC: Notation Declared ] |
5408 | | * The Name must match the declared name of a notation. |
5409 | | * |
5410 | | * @param ctxt an XML parser context |
5411 | | */ |
5412 | | |
5413 | | void |
5414 | 67.3k | xmlParseEntityDecl(xmlParserCtxt *ctxt) { |
5415 | 67.3k | const xmlChar *name = NULL; |
5416 | 67.3k | xmlChar *value = NULL; |
5417 | 67.3k | xmlChar *URI = NULL, *literal = NULL; |
5418 | 67.3k | const xmlChar *ndata = NULL; |
5419 | 67.3k | int isParameter = 0; |
5420 | 67.3k | xmlChar *orig = NULL; |
5421 | | |
5422 | 67.3k | if ((CUR != '<') || (NXT(1) != '!')) |
5423 | 0 | return; |
5424 | 67.3k | SKIP(2); |
5425 | | |
5426 | | /* GROW; done in the caller */ |
5427 | 67.3k | if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5428 | 67.2k | #ifdef LIBXML_VALID_ENABLED |
5429 | 67.2k | int oldInputNr = ctxt->inputNr; |
5430 | 67.2k | #endif |
5431 | | |
5432 | 67.2k | SKIP(6); |
5433 | 67.2k | if (SKIP_BLANKS_PE == 0) { |
5434 | 4.89k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5435 | 4.89k | "Space required after '<!ENTITY'\n"); |
5436 | 4.89k | } |
5437 | | |
5438 | 67.2k | if (RAW == '%') { |
5439 | 24.8k | NEXT; |
5440 | 24.8k | if (SKIP_BLANKS_PE == 0) { |
5441 | 632 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5442 | 632 | "Space required after '%%'\n"); |
5443 | 632 | } |
5444 | 24.8k | isParameter = 1; |
5445 | 24.8k | } |
5446 | | |
5447 | 67.2k | name = xmlParseName(ctxt); |
5448 | 67.2k | if (name == NULL) { |
5449 | 1.42k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5450 | 1.42k | "xmlParseEntityDecl: no name\n"); |
5451 | 1.42k | return; |
5452 | 1.42k | } |
5453 | 65.8k | if (xmlStrchr(name, ':') != NULL) { |
5454 | 472 | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5455 | 472 | "colons are forbidden from entities names '%s'\n", |
5456 | 472 | name, NULL, NULL); |
5457 | 472 | } |
5458 | 65.8k | if (SKIP_BLANKS_PE == 0) { |
5459 | 3.66k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5460 | 3.66k | "Space required after the entity name\n"); |
5461 | 3.66k | } |
5462 | | |
5463 | | /* |
5464 | | * handle the various case of definitions... |
5465 | | */ |
5466 | 65.8k | if (isParameter) { |
5467 | 24.3k | if ((RAW == '"') || (RAW == '\'')) { |
5468 | 15.0k | value = xmlParseEntityValue(ctxt, &orig); |
5469 | 15.0k | if (value) { |
5470 | 14.7k | if ((ctxt->sax != NULL) && |
5471 | 14.7k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5472 | 13.2k | ctxt->sax->entityDecl(ctxt->userData, name, |
5473 | 13.2k | XML_INTERNAL_PARAMETER_ENTITY, |
5474 | 13.2k | NULL, NULL, value); |
5475 | 14.7k | } |
5476 | 15.0k | } else { |
5477 | 9.25k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5478 | 9.25k | if ((URI == NULL) && (literal == NULL)) { |
5479 | 251 | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5480 | 251 | } |
5481 | 9.25k | if (URI) { |
5482 | 8.80k | if (xmlStrchr(URI, '#')) { |
5483 | 289 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5484 | 8.51k | } else { |
5485 | 8.51k | if ((ctxt->sax != NULL) && |
5486 | 8.51k | (!ctxt->disableSAX) && |
5487 | 8.51k | (ctxt->sax->entityDecl != NULL)) |
5488 | 7.06k | ctxt->sax->entityDecl(ctxt->userData, name, |
5489 | 7.06k | XML_EXTERNAL_PARAMETER_ENTITY, |
5490 | 7.06k | literal, URI, NULL); |
5491 | 8.51k | } |
5492 | 8.80k | } |
5493 | 9.25k | } |
5494 | 41.5k | } else { |
5495 | 41.5k | if ((RAW == '"') || (RAW == '\'')) { |
5496 | 31.0k | value = xmlParseEntityValue(ctxt, &orig); |
5497 | 31.0k | if ((ctxt->sax != NULL) && |
5498 | 31.0k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5499 | 28.2k | ctxt->sax->entityDecl(ctxt->userData, name, |
5500 | 28.2k | XML_INTERNAL_GENERAL_ENTITY, |
5501 | 28.2k | NULL, NULL, value); |
5502 | | /* |
5503 | | * For expat compatibility in SAX mode. |
5504 | | */ |
5505 | 31.0k | if ((ctxt->myDoc == NULL) || |
5506 | 31.0k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5507 | 1.27k | if (ctxt->myDoc == NULL) { |
5508 | 214 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5509 | 214 | if (ctxt->myDoc == NULL) { |
5510 | 1 | xmlErrMemory(ctxt); |
5511 | 1 | goto done; |
5512 | 1 | } |
5513 | 213 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5514 | 213 | } |
5515 | 1.27k | if (ctxt->myDoc->intSubset == NULL) { |
5516 | 213 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5517 | 213 | BAD_CAST "fake", NULL, NULL); |
5518 | 213 | if (ctxt->myDoc->intSubset == NULL) { |
5519 | 1 | xmlErrMemory(ctxt); |
5520 | 1 | goto done; |
5521 | 1 | } |
5522 | 213 | } |
5523 | | |
5524 | 1.27k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5525 | 1.27k | NULL, NULL, value); |
5526 | 1.27k | } |
5527 | 31.0k | } else { |
5528 | 10.4k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5529 | 10.4k | if ((URI == NULL) && (literal == NULL)) { |
5530 | 1.98k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5531 | 1.98k | } |
5532 | 10.4k | if (URI) { |
5533 | 8.07k | if (xmlStrchr(URI, '#')) { |
5534 | 363 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5535 | 363 | } |
5536 | 8.07k | } |
5537 | 10.4k | if ((RAW != '>') && (SKIP_BLANKS_PE == 0)) { |
5538 | 1.23k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5539 | 1.23k | "Space required before 'NDATA'\n"); |
5540 | 1.23k | } |
5541 | 10.4k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5542 | 851 | SKIP(5); |
5543 | 851 | if (SKIP_BLANKS_PE == 0) { |
5544 | 344 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5545 | 344 | "Space required after 'NDATA'\n"); |
5546 | 344 | } |
5547 | 851 | ndata = xmlParseName(ctxt); |
5548 | 851 | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5549 | 851 | (ctxt->sax->unparsedEntityDecl != NULL)) |
5550 | 581 | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5551 | 581 | literal, URI, ndata); |
5552 | 9.60k | } else { |
5553 | 9.60k | if ((ctxt->sax != NULL) && |
5554 | 9.60k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5555 | 9.06k | ctxt->sax->entityDecl(ctxt->userData, name, |
5556 | 9.06k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5557 | 9.06k | literal, URI, NULL); |
5558 | | /* |
5559 | | * For expat compatibility in SAX mode. |
5560 | | * assuming the entity replacement was asked for |
5561 | | */ |
5562 | 9.60k | if ((ctxt->replaceEntities != 0) && |
5563 | 9.60k | ((ctxt->myDoc == NULL) || |
5564 | 5.86k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5565 | 78 | if (ctxt->myDoc == NULL) { |
5566 | 12 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5567 | 12 | if (ctxt->myDoc == NULL) { |
5568 | 1 | xmlErrMemory(ctxt); |
5569 | 1 | goto done; |
5570 | 1 | } |
5571 | 11 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5572 | 11 | } |
5573 | | |
5574 | 77 | if (ctxt->myDoc->intSubset == NULL) { |
5575 | 11 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5576 | 11 | BAD_CAST "fake", NULL, NULL); |
5577 | 11 | if (ctxt->myDoc->intSubset == NULL) { |
5578 | 1 | xmlErrMemory(ctxt); |
5579 | 1 | goto done; |
5580 | 1 | } |
5581 | 11 | } |
5582 | 76 | xmlSAX2EntityDecl(ctxt, name, |
5583 | 76 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5584 | 76 | literal, URI, NULL); |
5585 | 76 | } |
5586 | 9.60k | } |
5587 | 10.4k | } |
5588 | 41.5k | } |
5589 | 65.8k | SKIP_BLANKS_PE; |
5590 | 65.8k | if (RAW != '>') { |
5591 | 3.78k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5592 | 3.78k | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5593 | 62.0k | } else { |
5594 | 62.0k | #ifdef LIBXML_VALID_ENABLED |
5595 | 62.0k | if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) { |
5596 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5597 | 0 | "Entity declaration doesn't start and stop in" |
5598 | 0 | " the same entity\n", |
5599 | 0 | NULL, NULL); |
5600 | 0 | } |
5601 | 62.0k | #endif |
5602 | 62.0k | NEXT; |
5603 | 62.0k | } |
5604 | 65.8k | if (orig != NULL) { |
5605 | | /* |
5606 | | * Ugly mechanism to save the raw entity value. |
5607 | | */ |
5608 | 45.2k | xmlEntityPtr cur = NULL; |
5609 | | |
5610 | 45.2k | if (isParameter) { |
5611 | 14.7k | if ((ctxt->sax != NULL) && |
5612 | 14.7k | (ctxt->sax->getParameterEntity != NULL)) |
5613 | 14.7k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5614 | 30.4k | } else { |
5615 | 30.4k | if ((ctxt->sax != NULL) && |
5616 | 30.4k | (ctxt->sax->getEntity != NULL)) |
5617 | 30.4k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5618 | 30.4k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5619 | 1.16k | cur = xmlSAX2GetEntity(ctxt, name); |
5620 | 1.16k | } |
5621 | 30.4k | } |
5622 | 45.2k | if ((cur != NULL) && (cur->orig == NULL)) { |
5623 | 23.2k | cur->orig = orig; |
5624 | 23.2k | orig = NULL; |
5625 | 23.2k | } |
5626 | 45.2k | } |
5627 | | |
5628 | 65.8k | done: |
5629 | 65.8k | if (value != NULL) xmlFree(value); |
5630 | 65.8k | if (URI != NULL) xmlFree(URI); |
5631 | 65.8k | if (literal != NULL) xmlFree(literal); |
5632 | 65.8k | if (orig != NULL) xmlFree(orig); |
5633 | 65.8k | } |
5634 | 67.3k | } |
5635 | | |
5636 | | /** |
5637 | | * Parse an attribute default declaration |
5638 | | * |
5639 | | * @deprecated Internal function, don't use. |
5640 | | * |
5641 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5642 | | * |
5643 | | * [ VC: Required Attribute ] |
5644 | | * if the default declaration is the keyword \#REQUIRED, then the |
5645 | | * attribute must be specified for all elements of the type in the |
5646 | | * attribute-list declaration. |
5647 | | * |
5648 | | * [ VC: Attribute Default Legal ] |
5649 | | * The declared default value must meet the lexical constraints of |
5650 | | * the declared attribute type c.f. #xmlValidateAttributeDecl |
5651 | | * |
5652 | | * [ VC: Fixed Attribute Default ] |
5653 | | * if an attribute has a default value declared with the \#FIXED |
5654 | | * keyword, instances of that attribute must match the default value. |
5655 | | * |
5656 | | * [ WFC: No < in Attribute Values ] |
5657 | | * handled in #xmlParseAttValue |
5658 | | * |
5659 | | * @param ctxt an XML parser context |
5660 | | * @param value Receive a possible fixed default value for the attribute |
5661 | | * @returns XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5662 | | * or XML_ATTRIBUTE_FIXED. |
5663 | | */ |
5664 | | |
5665 | | int |
5666 | 93.9k | xmlParseDefaultDecl(xmlParserCtxt *ctxt, xmlChar **value) { |
5667 | 93.9k | int val; |
5668 | 93.9k | xmlChar *ret; |
5669 | | |
5670 | 93.9k | *value = NULL; |
5671 | 93.9k | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5672 | 7.11k | SKIP(9); |
5673 | 7.11k | return(XML_ATTRIBUTE_REQUIRED); |
5674 | 7.11k | } |
5675 | 86.8k | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5676 | 30.6k | SKIP(8); |
5677 | 30.6k | return(XML_ATTRIBUTE_IMPLIED); |
5678 | 30.6k | } |
5679 | 56.1k | val = XML_ATTRIBUTE_NONE; |
5680 | 56.1k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5681 | 6.44k | SKIP(6); |
5682 | 6.44k | val = XML_ATTRIBUTE_FIXED; |
5683 | 6.44k | if (SKIP_BLANKS_PE == 0) { |
5684 | 946 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5685 | 946 | "Space required after '#FIXED'\n"); |
5686 | 946 | } |
5687 | 6.44k | } |
5688 | 56.1k | ret = xmlParseAttValue(ctxt); |
5689 | 56.1k | if (ret == NULL) { |
5690 | 7.93k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5691 | 7.93k | "Attribute default value declaration error\n"); |
5692 | 7.93k | } else |
5693 | 48.2k | *value = ret; |
5694 | 56.1k | return(val); |
5695 | 86.8k | } |
5696 | | |
5697 | | /** |
5698 | | * Parse an Notation attribute type. |
5699 | | * |
5700 | | * @deprecated Internal function, don't use. |
5701 | | * |
5702 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5703 | | * |
5704 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5705 | | * |
5706 | | * [ VC: Notation Attributes ] |
5707 | | * Values of this type must match one of the notation names included |
5708 | | * in the declaration; all notation names in the declaration must be declared. |
5709 | | * |
5710 | | * @param ctxt an XML parser context |
5711 | | * @returns the notation attribute tree built while parsing |
5712 | | */ |
5713 | | |
5714 | | xmlEnumeration * |
5715 | 1.93k | xmlParseNotationType(xmlParserCtxt *ctxt) { |
5716 | 1.93k | const xmlChar *name; |
5717 | 1.93k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5718 | | |
5719 | 1.93k | if (RAW != '(') { |
5720 | 278 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5721 | 278 | return(NULL); |
5722 | 278 | } |
5723 | 2.00k | do { |
5724 | 2.00k | NEXT; |
5725 | 2.00k | SKIP_BLANKS_PE; |
5726 | 2.00k | name = xmlParseName(ctxt); |
5727 | 2.00k | if (name == NULL) { |
5728 | 226 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5729 | 226 | "Name expected in NOTATION declaration\n"); |
5730 | 226 | xmlFreeEnumeration(ret); |
5731 | 226 | return(NULL); |
5732 | 226 | } |
5733 | 1.78k | tmp = NULL; |
5734 | 1.78k | #ifdef LIBXML_VALID_ENABLED |
5735 | 1.78k | if (ctxt->validate) { |
5736 | 0 | tmp = ret; |
5737 | 0 | while (tmp != NULL) { |
5738 | 0 | if (xmlStrEqual(name, tmp->name)) { |
5739 | 0 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5740 | 0 | "standalone: attribute notation value token %s duplicated\n", |
5741 | 0 | name, NULL); |
5742 | 0 | if (!xmlDictOwns(ctxt->dict, name)) |
5743 | 0 | xmlFree((xmlChar *) name); |
5744 | 0 | break; |
5745 | 0 | } |
5746 | 0 | tmp = tmp->next; |
5747 | 0 | } |
5748 | 0 | } |
5749 | 1.78k | #endif /* LIBXML_VALID_ENABLED */ |
5750 | 1.78k | if (tmp == NULL) { |
5751 | 1.78k | cur = xmlCreateEnumeration(name); |
5752 | 1.78k | if (cur == NULL) { |
5753 | 6 | xmlErrMemory(ctxt); |
5754 | 6 | xmlFreeEnumeration(ret); |
5755 | 6 | return(NULL); |
5756 | 6 | } |
5757 | 1.77k | if (last == NULL) ret = last = cur; |
5758 | 352 | else { |
5759 | 352 | last->next = cur; |
5760 | 352 | last = cur; |
5761 | 352 | } |
5762 | 1.77k | } |
5763 | 1.77k | SKIP_BLANKS_PE; |
5764 | 1.77k | } while (RAW == '|'); |
5765 | 1.42k | if (RAW != ')') { |
5766 | 270 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5767 | 270 | xmlFreeEnumeration(ret); |
5768 | 270 | return(NULL); |
5769 | 270 | } |
5770 | 1.15k | NEXT; |
5771 | 1.15k | return(ret); |
5772 | 1.42k | } |
5773 | | |
5774 | | /** |
5775 | | * Parse an Enumeration attribute type. |
5776 | | * |
5777 | | * @deprecated Internal function, don't use. |
5778 | | * |
5779 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5780 | | * |
5781 | | * [ VC: Enumeration ] |
5782 | | * Values of this type must match one of the Nmtoken tokens in |
5783 | | * the declaration |
5784 | | * |
5785 | | * @param ctxt an XML parser context |
5786 | | * @returns the enumeration attribute tree built while parsing |
5787 | | */ |
5788 | | |
5789 | | xmlEnumeration * |
5790 | 17.9k | xmlParseEnumerationType(xmlParserCtxt *ctxt) { |
5791 | 17.9k | xmlChar *name; |
5792 | 17.9k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5793 | | |
5794 | 17.9k | if (RAW != '(') { |
5795 | 2.16k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5796 | 2.16k | return(NULL); |
5797 | 2.16k | } |
5798 | 25.6k | do { |
5799 | 25.6k | NEXT; |
5800 | 25.6k | SKIP_BLANKS_PE; |
5801 | 25.6k | name = xmlParseNmtoken(ctxt); |
5802 | 25.6k | if (name == NULL) { |
5803 | 232 | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5804 | 232 | return(ret); |
5805 | 232 | } |
5806 | 25.4k | tmp = NULL; |
5807 | 25.4k | #ifdef LIBXML_VALID_ENABLED |
5808 | 25.4k | if (ctxt->validate) { |
5809 | 0 | tmp = ret; |
5810 | 0 | while (tmp != NULL) { |
5811 | 0 | if (xmlStrEqual(name, tmp->name)) { |
5812 | 0 | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5813 | 0 | "standalone: attribute enumeration value token %s duplicated\n", |
5814 | 0 | name, NULL); |
5815 | 0 | if (!xmlDictOwns(ctxt->dict, name)) |
5816 | 0 | xmlFree(name); |
5817 | 0 | break; |
5818 | 0 | } |
5819 | 0 | tmp = tmp->next; |
5820 | 0 | } |
5821 | 0 | } |
5822 | 25.4k | #endif /* LIBXML_VALID_ENABLED */ |
5823 | 25.4k | if (tmp == NULL) { |
5824 | 25.4k | cur = xmlCreateEnumeration(name); |
5825 | 25.4k | if (!xmlDictOwns(ctxt->dict, name)) |
5826 | 25.4k | xmlFree(name); |
5827 | 25.4k | if (cur == NULL) { |
5828 | 25 | xmlErrMemory(ctxt); |
5829 | 25 | xmlFreeEnumeration(ret); |
5830 | 25 | return(NULL); |
5831 | 25 | } |
5832 | 25.3k | if (last == NULL) ret = last = cur; |
5833 | 9.82k | else { |
5834 | 9.82k | last->next = cur; |
5835 | 9.82k | last = cur; |
5836 | 9.82k | } |
5837 | 25.3k | } |
5838 | 25.3k | SKIP_BLANKS_PE; |
5839 | 25.3k | } while (RAW == '|'); |
5840 | 15.5k | if (RAW != ')') { |
5841 | 440 | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5842 | 440 | return(ret); |
5843 | 440 | } |
5844 | 15.1k | NEXT; |
5845 | 15.1k | return(ret); |
5846 | 15.5k | } |
5847 | | |
5848 | | /** |
5849 | | * Parse an Enumerated attribute type. |
5850 | | * |
5851 | | * @deprecated Internal function, don't use. |
5852 | | * |
5853 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5854 | | * |
5855 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5856 | | * |
5857 | | * @param ctxt an XML parser context |
5858 | | * @param tree the enumeration tree built while parsing |
5859 | | * @returns XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
5860 | | */ |
5861 | | |
5862 | | int |
5863 | 20.1k | xmlParseEnumeratedType(xmlParserCtxt *ctxt, xmlEnumeration **tree) { |
5864 | 20.1k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5865 | 2.16k | SKIP(8); |
5866 | 2.16k | if (SKIP_BLANKS_PE == 0) { |
5867 | 232 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5868 | 232 | "Space required after 'NOTATION'\n"); |
5869 | 232 | return(0); |
5870 | 232 | } |
5871 | 1.93k | *tree = xmlParseNotationType(ctxt); |
5872 | 1.93k | if (*tree == NULL) return(0); |
5873 | 1.15k | return(XML_ATTRIBUTE_NOTATION); |
5874 | 1.93k | } |
5875 | 17.9k | *tree = xmlParseEnumerationType(ctxt); |
5876 | 17.9k | if (*tree == NULL) return(0); |
5877 | 15.5k | return(XML_ATTRIBUTE_ENUMERATION); |
5878 | 17.9k | } |
5879 | | |
5880 | | /** |
5881 | | * Parse the Attribute list def for an element |
5882 | | * |
5883 | | * @deprecated Internal function, don't use. |
5884 | | * |
5885 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
5886 | | * |
5887 | | * [55] StringType ::= 'CDATA' |
5888 | | * |
5889 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
5890 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
5891 | | * |
5892 | | * Validity constraints for attribute values syntax are checked in |
5893 | | * #xmlValidateAttributeValue |
5894 | | * |
5895 | | * [ VC: ID ] |
5896 | | * Values of type ID must match the Name production. A name must not |
5897 | | * appear more than once in an XML document as a value of this type; |
5898 | | * i.e., ID values must uniquely identify the elements which bear them. |
5899 | | * |
5900 | | * [ VC: One ID per Element Type ] |
5901 | | * No element type may have more than one ID attribute specified. |
5902 | | * |
5903 | | * [ VC: ID Attribute Default ] |
5904 | | * An ID attribute must have a declared default of \#IMPLIED or \#REQUIRED. |
5905 | | * |
5906 | | * [ VC: IDREF ] |
5907 | | * Values of type IDREF must match the Name production, and values |
5908 | | * of type IDREFS must match Names; each IDREF Name must match the value |
5909 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
5910 | | * values must match the value of some ID attribute. |
5911 | | * |
5912 | | * [ VC: Entity Name ] |
5913 | | * Values of type ENTITY must match the Name production, values |
5914 | | * of type ENTITIES must match Names; each Entity Name must match the |
5915 | | * name of an unparsed entity declared in the DTD. |
5916 | | * |
5917 | | * [ VC: Name Token ] |
5918 | | * Values of type NMTOKEN must match the Nmtoken production; values |
5919 | | * of type NMTOKENS must match Nmtokens. |
5920 | | * |
5921 | | * @param ctxt an XML parser context |
5922 | | * @param tree the enumeration tree built while parsing |
5923 | | * @returns the attribute type |
5924 | | */ |
5925 | | int |
5926 | 99.9k | xmlParseAttributeType(xmlParserCtxt *ctxt, xmlEnumeration **tree) { |
5927 | 99.9k | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
5928 | 15.5k | SKIP(5); |
5929 | 15.5k | return(XML_ATTRIBUTE_CDATA); |
5930 | 84.3k | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
5931 | 1.94k | SKIP(6); |
5932 | 1.94k | return(XML_ATTRIBUTE_IDREFS); |
5933 | 82.3k | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
5934 | 2.13k | SKIP(5); |
5935 | 2.13k | return(XML_ATTRIBUTE_IDREF); |
5936 | 80.2k | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
5937 | 22.4k | SKIP(2); |
5938 | 22.4k | return(XML_ATTRIBUTE_ID); |
5939 | 57.8k | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5940 | 14.0k | SKIP(6); |
5941 | 14.0k | return(XML_ATTRIBUTE_ENTITY); |
5942 | 43.7k | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
5943 | 6.14k | SKIP(8); |
5944 | 6.14k | return(XML_ATTRIBUTE_ENTITIES); |
5945 | 37.6k | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
5946 | 7.65k | SKIP(8); |
5947 | 7.65k | return(XML_ATTRIBUTE_NMTOKENS); |
5948 | 29.9k | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
5949 | 9.83k | SKIP(7); |
5950 | 9.83k | return(XML_ATTRIBUTE_NMTOKEN); |
5951 | 9.83k | } |
5952 | 20.1k | return(xmlParseEnumeratedType(ctxt, tree)); |
5953 | 99.9k | } |
5954 | | |
5955 | | /** |
5956 | | * Parse an attribute list declaration for an element. Always consumes '<!'. |
5957 | | * |
5958 | | * @deprecated Internal function, don't use. |
5959 | | * |
5960 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
5961 | | * |
5962 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
5963 | | * @param ctxt an XML parser context |
5964 | | */ |
5965 | | void |
5966 | 70.0k | xmlParseAttributeListDecl(xmlParserCtxt *ctxt) { |
5967 | 70.0k | const xmlChar *elemName; |
5968 | 70.0k | const xmlChar *attrName; |
5969 | 70.0k | xmlEnumerationPtr tree; |
5970 | | |
5971 | 70.0k | if ((CUR != '<') || (NXT(1) != '!')) |
5972 | 0 | return; |
5973 | 70.0k | SKIP(2); |
5974 | | |
5975 | 70.0k | if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
5976 | 69.9k | #ifdef LIBXML_VALID_ENABLED |
5977 | 69.9k | int oldInputNr = ctxt->inputNr; |
5978 | 69.9k | #endif |
5979 | | |
5980 | 69.9k | SKIP(7); |
5981 | 69.9k | if (SKIP_BLANKS_PE == 0) { |
5982 | 13.9k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5983 | 13.9k | "Space required after '<!ATTLIST'\n"); |
5984 | 13.9k | } |
5985 | 69.9k | elemName = xmlParseName(ctxt); |
5986 | 69.9k | if (elemName == NULL) { |
5987 | 7.68k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5988 | 7.68k | "ATTLIST: no name for Element\n"); |
5989 | 7.68k | return; |
5990 | 7.68k | } |
5991 | 62.3k | SKIP_BLANKS_PE; |
5992 | 62.3k | GROW; |
5993 | 143k | while ((RAW != '>') && (PARSER_STOPPED(ctxt) == 0)) { |
5994 | 107k | int type; |
5995 | 107k | int def; |
5996 | 107k | xmlChar *defaultValue = NULL; |
5997 | | |
5998 | 107k | GROW; |
5999 | 107k | tree = NULL; |
6000 | 107k | attrName = xmlParseName(ctxt); |
6001 | 107k | if (attrName == NULL) { |
6002 | 3.11k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6003 | 3.11k | "ATTLIST: no name for Attribute\n"); |
6004 | 3.11k | break; |
6005 | 3.11k | } |
6006 | 104k | GROW; |
6007 | 104k | if (SKIP_BLANKS_PE == 0) { |
6008 | 4.87k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6009 | 4.87k | "Space required after the attribute name\n"); |
6010 | 4.87k | break; |
6011 | 4.87k | } |
6012 | | |
6013 | 99.9k | type = xmlParseAttributeType(ctxt, &tree); |
6014 | 99.9k | if (type <= 0) { |
6015 | 3.43k | break; |
6016 | 3.43k | } |
6017 | | |
6018 | 96.4k | GROW; |
6019 | 96.4k | if (SKIP_BLANKS_PE == 0) { |
6020 | 2.50k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6021 | 2.50k | "Space required after the attribute type\n"); |
6022 | 2.50k | if (tree != NULL) |
6023 | 826 | xmlFreeEnumeration(tree); |
6024 | 2.50k | break; |
6025 | 2.50k | } |
6026 | | |
6027 | 93.9k | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6028 | 93.9k | if (def <= 0) { |
6029 | 0 | if (defaultValue != NULL) |
6030 | 0 | xmlFree(defaultValue); |
6031 | 0 | if (tree != NULL) |
6032 | 0 | xmlFreeEnumeration(tree); |
6033 | 0 | break; |
6034 | 0 | } |
6035 | 93.9k | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6036 | 47.4k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6037 | | |
6038 | 93.9k | GROW; |
6039 | 93.9k | if (RAW != '>') { |
6040 | 66.9k | if (SKIP_BLANKS_PE == 0) { |
6041 | 12.4k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6042 | 12.4k | "Space required after the attribute default value\n"); |
6043 | 12.4k | if (defaultValue != NULL) |
6044 | 4.43k | xmlFree(defaultValue); |
6045 | 12.4k | if (tree != NULL) |
6046 | 744 | xmlFreeEnumeration(tree); |
6047 | 12.4k | break; |
6048 | 12.4k | } |
6049 | 66.9k | } |
6050 | 81.5k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6051 | 81.5k | (ctxt->sax->attributeDecl != NULL)) |
6052 | 77.9k | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6053 | 77.9k | type, def, defaultValue, tree); |
6054 | 3.57k | else if (tree != NULL) |
6055 | 1.42k | xmlFreeEnumeration(tree); |
6056 | | |
6057 | 81.5k | if ((ctxt->sax2) && (defaultValue != NULL) && |
6058 | 81.5k | (def != XML_ATTRIBUTE_IMPLIED) && |
6059 | 81.5k | (def != XML_ATTRIBUTE_REQUIRED)) { |
6060 | 43.7k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6061 | 43.7k | } |
6062 | 81.5k | if (ctxt->sax2) { |
6063 | 81.5k | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6064 | 81.5k | } |
6065 | 81.5k | if (defaultValue != NULL) |
6066 | 43.7k | xmlFree(defaultValue); |
6067 | 81.5k | GROW; |
6068 | 81.5k | } |
6069 | 62.3k | if (RAW == '>') { |
6070 | 37.0k | #ifdef LIBXML_VALID_ENABLED |
6071 | 37.0k | if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) { |
6072 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6073 | 0 | "Attribute list declaration doesn't start and" |
6074 | 0 | " stop in the same entity\n", |
6075 | 0 | NULL, NULL); |
6076 | 0 | } |
6077 | 37.0k | #endif |
6078 | 37.0k | NEXT; |
6079 | 37.0k | } |
6080 | 62.3k | } |
6081 | 70.0k | } |
6082 | | |
6083 | | /** |
6084 | | * Handle PEs and check that we don't pop the entity that started |
6085 | | * a balanced group. |
6086 | | * |
6087 | | * @param ctxt parser context |
6088 | | * @param openInputNr input nr of the entity with opening '(' |
6089 | | */ |
6090 | | static void |
6091 | 292k | xmlSkipBlankCharsPEBalanced(xmlParserCtxt *ctxt, int openInputNr) { |
6092 | 292k | SKIP_BLANKS; |
6093 | 292k | GROW; |
6094 | | |
6095 | 292k | (void) openInputNr; |
6096 | | |
6097 | 292k | if (!PARSER_EXTERNAL(ctxt) && !PARSER_IN_PE(ctxt)) |
6098 | 137k | return; |
6099 | | |
6100 | 168k | while (!PARSER_STOPPED(ctxt)) { |
6101 | 168k | if (ctxt->input->cur >= ctxt->input->end) { |
6102 | 5.53k | #ifdef LIBXML_VALID_ENABLED |
6103 | 5.53k | if ((ctxt->validate) && (ctxt->inputNr <= openInputNr)) { |
6104 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6105 | 0 | "Element content declaration doesn't start " |
6106 | 0 | "and stop in the same entity\n", |
6107 | 0 | NULL, NULL); |
6108 | 0 | } |
6109 | 5.53k | #endif |
6110 | 5.53k | if (PARSER_IN_PE(ctxt)) |
6111 | 5.49k | xmlPopPE(ctxt); |
6112 | 46 | else |
6113 | 46 | break; |
6114 | 162k | } else if (RAW == '%') { |
6115 | 8.06k | xmlParsePERefInternal(ctxt, 0); |
6116 | 154k | } else { |
6117 | 154k | break; |
6118 | 154k | } |
6119 | | |
6120 | 13.5k | SKIP_BLANKS; |
6121 | 13.5k | GROW; |
6122 | 13.5k | } |
6123 | 154k | } |
6124 | | |
6125 | | /** |
6126 | | * Parse the declaration for a Mixed Element content |
6127 | | * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl |
6128 | | * |
6129 | | * @deprecated Internal function, don't use. |
6130 | | * |
6131 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6132 | | * '(' S? '#PCDATA' S? ')' |
6133 | | * |
6134 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6135 | | * |
6136 | | * [ VC: No Duplicate Types ] |
6137 | | * The same name must not appear more than once in a single |
6138 | | * mixed-content declaration. |
6139 | | * |
6140 | | * @param ctxt an XML parser context |
6141 | | * @param openInputNr the input used for the current entity, needed for |
6142 | | * boundary checks |
6143 | | * @returns the list of the xmlElementContent describing the element choices |
6144 | | */ |
6145 | | xmlElementContent * |
6146 | 10.5k | xmlParseElementMixedContentDecl(xmlParserCtxt *ctxt, int openInputNr) { |
6147 | 10.5k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6148 | 10.5k | const xmlChar *elem = NULL; |
6149 | | |
6150 | 10.5k | GROW; |
6151 | 10.5k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6152 | 10.5k | SKIP(7); |
6153 | 10.5k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6154 | 10.5k | if (RAW == ')') { |
6155 | 7.21k | #ifdef LIBXML_VALID_ENABLED |
6156 | 7.21k | if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) { |
6157 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6158 | 0 | "Element content declaration doesn't start " |
6159 | 0 | "and stop in the same entity\n", |
6160 | 0 | NULL, NULL); |
6161 | 0 | } |
6162 | 7.21k | #endif |
6163 | 7.21k | NEXT; |
6164 | 7.21k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6165 | 7.21k | if (ret == NULL) |
6166 | 10 | goto mem_error; |
6167 | 7.20k | if (RAW == '*') { |
6168 | 220 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6169 | 220 | NEXT; |
6170 | 220 | } |
6171 | 7.20k | return(ret); |
6172 | 7.21k | } |
6173 | 3.37k | if ((RAW == '(') || (RAW == '|')) { |
6174 | 2.17k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6175 | 2.17k | if (ret == NULL) |
6176 | 8 | goto mem_error; |
6177 | 2.17k | } |
6178 | 23.9k | while ((RAW == '|') && (PARSER_STOPPED(ctxt) == 0)) { |
6179 | 20.7k | NEXT; |
6180 | 20.7k | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6181 | 20.7k | if (n == NULL) |
6182 | 5 | goto mem_error; |
6183 | 20.7k | if (elem == NULL) { |
6184 | 2.15k | n->c1 = cur; |
6185 | 2.15k | if (cur != NULL) |
6186 | 2.15k | cur->parent = n; |
6187 | 2.15k | ret = cur = n; |
6188 | 18.6k | } else { |
6189 | 18.6k | cur->c2 = n; |
6190 | 18.6k | n->parent = cur; |
6191 | 18.6k | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6192 | 18.6k | if (n->c1 == NULL) |
6193 | 3 | goto mem_error; |
6194 | 18.6k | n->c1->parent = n; |
6195 | 18.6k | cur = n; |
6196 | 18.6k | } |
6197 | 20.7k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6198 | 20.7k | elem = xmlParseName(ctxt); |
6199 | 20.7k | if (elem == NULL) { |
6200 | 133 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6201 | 133 | "xmlParseElementMixedContentDecl : Name expected\n"); |
6202 | 133 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6203 | 133 | return(NULL); |
6204 | 133 | } |
6205 | 20.6k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6206 | 20.6k | } |
6207 | 3.22k | if ((RAW == ')') && (NXT(1) == '*')) { |
6208 | 1.85k | if (elem != NULL) { |
6209 | 1.85k | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6210 | 1.85k | XML_ELEMENT_CONTENT_ELEMENT); |
6211 | 1.85k | if (cur->c2 == NULL) |
6212 | 5 | goto mem_error; |
6213 | 1.85k | cur->c2->parent = cur; |
6214 | 1.85k | } |
6215 | 1.85k | if (ret != NULL) |
6216 | 1.85k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6217 | 1.85k | #ifdef LIBXML_VALID_ENABLED |
6218 | 1.85k | if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) { |
6219 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6220 | 0 | "Element content declaration doesn't start " |
6221 | 0 | "and stop in the same entity\n", |
6222 | 0 | NULL, NULL); |
6223 | 0 | } |
6224 | 1.85k | #endif |
6225 | 1.85k | SKIP(2); |
6226 | 1.85k | } else { |
6227 | 1.36k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6228 | 1.36k | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6229 | 1.36k | return(NULL); |
6230 | 1.36k | } |
6231 | | |
6232 | 3.22k | } else { |
6233 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6234 | 0 | } |
6235 | 1.85k | return(ret); |
6236 | | |
6237 | 31 | mem_error: |
6238 | 31 | xmlErrMemory(ctxt); |
6239 | 31 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6240 | 31 | return(NULL); |
6241 | 10.5k | } |
6242 | | |
6243 | | /** |
6244 | | * Parse the declaration for a Mixed Element content |
6245 | | * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl |
6246 | | * |
6247 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6248 | | * |
6249 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6250 | | * |
6251 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6252 | | * |
6253 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6254 | | * |
6255 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6256 | | * TODO Parameter-entity replacement text must be properly nested |
6257 | | * with parenthesized groups. That is to say, if either of the |
6258 | | * opening or closing parentheses in a choice, seq, or Mixed |
6259 | | * construct is contained in the replacement text for a parameter |
6260 | | * entity, both must be contained in the same replacement text. For |
6261 | | * interoperability, if a parameter-entity reference appears in a |
6262 | | * choice, seq, or Mixed construct, its replacement text should not |
6263 | | * be empty, and neither the first nor last non-blank character of |
6264 | | * the replacement text should be a connector (| or ,). |
6265 | | * |
6266 | | * @param ctxt an XML parser context |
6267 | | * @param openInputNr the input used for the current entity, needed for |
6268 | | * boundary checks |
6269 | | * @param depth the level of recursion |
6270 | | * @returns the tree of xmlElementContent describing the element |
6271 | | * hierarchy. |
6272 | | */ |
6273 | | static xmlElementContentPtr |
6274 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int openInputNr, |
6275 | 95.9k | int depth) { |
6276 | 95.9k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
6277 | 95.9k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6278 | 95.9k | const xmlChar *elem; |
6279 | 95.9k | xmlChar type = 0; |
6280 | | |
6281 | 95.9k | if (depth > maxDepth) { |
6282 | 3 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
6283 | 3 | "xmlParseElementChildrenContentDecl : depth %d too deep, " |
6284 | 3 | "use XML_PARSE_HUGE\n", depth); |
6285 | 3 | return(NULL); |
6286 | 3 | } |
6287 | 95.9k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6288 | 95.9k | if (RAW == '(') { |
6289 | 73.5k | int newInputNr = ctxt->inputNr; |
6290 | | |
6291 | | /* Recurse on first child */ |
6292 | 73.5k | NEXT; |
6293 | 73.5k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr, |
6294 | 73.5k | depth + 1); |
6295 | 73.5k | if (cur == NULL) |
6296 | 54.5k | return(NULL); |
6297 | 73.5k | } else { |
6298 | 22.4k | elem = xmlParseName(ctxt); |
6299 | 22.4k | if (elem == NULL) { |
6300 | 442 | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6301 | 442 | return(NULL); |
6302 | 442 | } |
6303 | 22.0k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6304 | 22.0k | if (cur == NULL) { |
6305 | 30 | xmlErrMemory(ctxt); |
6306 | 30 | return(NULL); |
6307 | 30 | } |
6308 | 22.0k | GROW; |
6309 | 22.0k | if (RAW == '?') { |
6310 | 3.62k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6311 | 3.62k | NEXT; |
6312 | 18.3k | } else if (RAW == '*') { |
6313 | 1.74k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6314 | 1.74k | NEXT; |
6315 | 16.6k | } else if (RAW == '+') { |
6316 | 2.03k | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6317 | 2.03k | NEXT; |
6318 | 14.5k | } else { |
6319 | 14.5k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6320 | 14.5k | } |
6321 | 22.0k | GROW; |
6322 | 22.0k | } |
6323 | 77.7k | while (!PARSER_STOPPED(ctxt)) { |
6324 | 76.4k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6325 | 76.4k | if (RAW == ')') |
6326 | 30.9k | break; |
6327 | | /* |
6328 | | * Each loop we parse one separator and one element. |
6329 | | */ |
6330 | 45.5k | if (RAW == ',') { |
6331 | 14.6k | if (type == 0) type = CUR; |
6332 | | |
6333 | | /* |
6334 | | * Detect "Name | Name , Name" error |
6335 | | */ |
6336 | 7.47k | else if (type != CUR) { |
6337 | 5 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6338 | 5 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6339 | 5 | type); |
6340 | 5 | if ((last != NULL) && (last != ret)) |
6341 | 5 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6342 | 5 | if (ret != NULL) |
6343 | 5 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6344 | 5 | return(NULL); |
6345 | 5 | } |
6346 | 14.6k | NEXT; |
6347 | | |
6348 | 14.6k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6349 | 14.6k | if (op == NULL) { |
6350 | 7 | xmlErrMemory(ctxt); |
6351 | 7 | if ((last != NULL) && (last != ret)) |
6352 | 4 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6353 | 7 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6354 | 7 | return(NULL); |
6355 | 7 | } |
6356 | 14.6k | if (last == NULL) { |
6357 | 7.16k | op->c1 = ret; |
6358 | 7.16k | if (ret != NULL) |
6359 | 7.16k | ret->parent = op; |
6360 | 7.16k | ret = cur = op; |
6361 | 7.46k | } else { |
6362 | 7.46k | cur->c2 = op; |
6363 | 7.46k | if (op != NULL) |
6364 | 7.46k | op->parent = cur; |
6365 | 7.46k | op->c1 = last; |
6366 | 7.46k | if (last != NULL) |
6367 | 7.46k | last->parent = op; |
6368 | 7.46k | cur =op; |
6369 | 7.46k | last = NULL; |
6370 | 7.46k | } |
6371 | 30.8k | } else if (RAW == '|') { |
6372 | 26.8k | if (type == 0) type = CUR; |
6373 | | |
6374 | | /* |
6375 | | * Detect "Name , Name | Name" error |
6376 | | */ |
6377 | 15.5k | else if (type != CUR) { |
6378 | 3 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6379 | 3 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6380 | 3 | type); |
6381 | 3 | if ((last != NULL) && (last != ret)) |
6382 | 3 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6383 | 3 | if (ret != NULL) |
6384 | 3 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6385 | 3 | return(NULL); |
6386 | 3 | } |
6387 | 26.8k | NEXT; |
6388 | | |
6389 | 26.8k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6390 | 26.8k | if (op == NULL) { |
6391 | 11 | xmlErrMemory(ctxt); |
6392 | 11 | if ((last != NULL) && (last != ret)) |
6393 | 4 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6394 | 11 | if (ret != NULL) |
6395 | 11 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6396 | 11 | return(NULL); |
6397 | 11 | } |
6398 | 26.8k | if (last == NULL) { |
6399 | 11.2k | op->c1 = ret; |
6400 | 11.2k | if (ret != NULL) |
6401 | 11.2k | ret->parent = op; |
6402 | 11.2k | ret = cur = op; |
6403 | 15.5k | } else { |
6404 | 15.5k | cur->c2 = op; |
6405 | 15.5k | if (op != NULL) |
6406 | 15.5k | op->parent = cur; |
6407 | 15.5k | op->c1 = last; |
6408 | 15.5k | if (last != NULL) |
6409 | 15.5k | last->parent = op; |
6410 | 15.5k | cur =op; |
6411 | 15.5k | last = NULL; |
6412 | 15.5k | } |
6413 | 26.8k | } else { |
6414 | 4.00k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6415 | 4.00k | if ((last != NULL) && (last != ret)) |
6416 | 1.15k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6417 | 4.00k | if (ret != NULL) |
6418 | 4.00k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6419 | 4.00k | return(NULL); |
6420 | 4.00k | } |
6421 | 41.4k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6422 | 41.4k | if (RAW == '(') { |
6423 | 6.59k | int newInputNr = ctxt->inputNr; |
6424 | | |
6425 | | /* Recurse on second child */ |
6426 | 6.59k | NEXT; |
6427 | 6.59k | last = xmlParseElementChildrenContentDeclPriv(ctxt, newInputNr, |
6428 | 6.59k | depth + 1); |
6429 | 6.59k | if (last == NULL) { |
6430 | 2.63k | if (ret != NULL) |
6431 | 2.63k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6432 | 2.63k | return(NULL); |
6433 | 2.63k | } |
6434 | 34.8k | } else { |
6435 | 34.8k | elem = xmlParseName(ctxt); |
6436 | 34.8k | if (elem == NULL) { |
6437 | 2.00k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6438 | 2.00k | if (ret != NULL) |
6439 | 2.00k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6440 | 2.00k | return(NULL); |
6441 | 2.00k | } |
6442 | 32.8k | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6443 | 32.8k | if (last == NULL) { |
6444 | 11 | xmlErrMemory(ctxt); |
6445 | 11 | if (ret != NULL) |
6446 | 11 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6447 | 11 | return(NULL); |
6448 | 11 | } |
6449 | 32.8k | if (RAW == '?') { |
6450 | 6.71k | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6451 | 6.71k | NEXT; |
6452 | 26.1k | } else if (RAW == '*') { |
6453 | 5.12k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6454 | 5.12k | NEXT; |
6455 | 21.0k | } else if (RAW == '+') { |
6456 | 916 | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6457 | 916 | NEXT; |
6458 | 20.1k | } else { |
6459 | 20.1k | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6460 | 20.1k | } |
6461 | 32.8k | } |
6462 | 41.4k | } |
6463 | 32.2k | if ((cur != NULL) && (last != NULL)) { |
6464 | 12.6k | cur->c2 = last; |
6465 | 12.6k | if (last != NULL) |
6466 | 12.6k | last->parent = cur; |
6467 | 12.6k | } |
6468 | 32.2k | #ifdef LIBXML_VALID_ENABLED |
6469 | 32.2k | if ((ctxt->validate) && (ctxt->inputNr > openInputNr)) { |
6470 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6471 | 0 | "Element content declaration doesn't start " |
6472 | 0 | "and stop in the same entity\n", |
6473 | 0 | NULL, NULL); |
6474 | 0 | } |
6475 | 32.2k | #endif |
6476 | 32.2k | NEXT; |
6477 | 32.2k | if (RAW == '?') { |
6478 | 773 | if (ret != NULL) { |
6479 | 773 | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6480 | 773 | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6481 | 106 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6482 | 667 | else |
6483 | 667 | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6484 | 773 | } |
6485 | 773 | NEXT; |
6486 | 31.5k | } else if (RAW == '*') { |
6487 | 3.05k | if (ret != NULL) { |
6488 | 3.05k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6489 | 3.05k | cur = ret; |
6490 | | /* |
6491 | | * Some normalization: |
6492 | | * (a | b* | c?)* == (a | b | c)* |
6493 | | */ |
6494 | 12.0k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6495 | 9.03k | if ((cur->c1 != NULL) && |
6496 | 9.03k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6497 | 9.03k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6498 | 1.53k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6499 | 9.03k | if ((cur->c2 != NULL) && |
6500 | 9.03k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6501 | 9.03k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6502 | 767 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6503 | 9.03k | cur = cur->c2; |
6504 | 9.03k | } |
6505 | 3.05k | } |
6506 | 3.05k | NEXT; |
6507 | 28.4k | } else if (RAW == '+') { |
6508 | 7.60k | if (ret != NULL) { |
6509 | 7.60k | int found = 0; |
6510 | | |
6511 | 7.60k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6512 | 7.60k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6513 | 614 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6514 | 6.98k | else |
6515 | 6.98k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6516 | | /* |
6517 | | * Some normalization: |
6518 | | * (a | b*)+ == (a | b)* |
6519 | | * (a | b?)+ == (a | b)* |
6520 | | */ |
6521 | 13.5k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6522 | 5.93k | if ((cur->c1 != NULL) && |
6523 | 5.93k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6524 | 5.93k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6525 | 3.25k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6526 | 3.25k | found = 1; |
6527 | 3.25k | } |
6528 | 5.93k | if ((cur->c2 != NULL) && |
6529 | 5.93k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6530 | 5.93k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6531 | 1.78k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6532 | 1.78k | found = 1; |
6533 | 1.78k | } |
6534 | 5.93k | cur = cur->c2; |
6535 | 5.93k | } |
6536 | 7.60k | if (found) |
6537 | 3.35k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6538 | 7.60k | } |
6539 | 7.60k | NEXT; |
6540 | 7.60k | } |
6541 | 32.2k | return(ret); |
6542 | 40.9k | } |
6543 | | |
6544 | | /** |
6545 | | * Parse the declaration for a Mixed Element content |
6546 | | * The leading '(' and spaces have been skipped in #xmlParseElementContentDecl |
6547 | | * |
6548 | | * @deprecated Internal function, don't use. |
6549 | | * |
6550 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6551 | | * |
6552 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6553 | | * |
6554 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6555 | | * |
6556 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6557 | | * |
6558 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6559 | | * TODO Parameter-entity replacement text must be properly nested |
6560 | | * with parenthesized groups. That is to say, if either of the |
6561 | | * opening or closing parentheses in a choice, seq, or Mixed |
6562 | | * construct is contained in the replacement text for a parameter |
6563 | | * entity, both must be contained in the same replacement text. For |
6564 | | * interoperability, if a parameter-entity reference appears in a |
6565 | | * choice, seq, or Mixed construct, its replacement text should not |
6566 | | * be empty, and neither the first nor last non-blank character of |
6567 | | * the replacement text should be a connector (| or ,). |
6568 | | * |
6569 | | * @param ctxt an XML parser context |
6570 | | * @param inputchk the input used for the current entity, needed for boundary checks |
6571 | | * @returns the tree of xmlElementContent describing the element |
6572 | | * hierarchy. |
6573 | | */ |
6574 | | xmlElementContent * |
6575 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxt *ctxt, int inputchk) { |
6576 | | /* stub left for API/ABI compat */ |
6577 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6578 | 0 | } |
6579 | | |
6580 | | /** |
6581 | | * Parse the declaration for an Element content either Mixed or Children, |
6582 | | * the cases EMPTY and ANY are handled directly in #xmlParseElementDecl |
6583 | | * |
6584 | | * @deprecated Internal function, don't use. |
6585 | | * |
6586 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6587 | | * |
6588 | | * @param ctxt an XML parser context |
6589 | | * @param name the name of the element being defined. |
6590 | | * @param result the Element Content pointer will be stored here if any |
6591 | | * @returns an xmlElementTypeVal value or -1 on error |
6592 | | */ |
6593 | | |
6594 | | int |
6595 | | xmlParseElementContentDecl(xmlParserCtxt *ctxt, const xmlChar *name, |
6596 | 26.4k | xmlElementContent **result) { |
6597 | | |
6598 | 26.4k | xmlElementContentPtr tree = NULL; |
6599 | 26.4k | int openInputNr = ctxt->inputNr; |
6600 | 26.4k | int res; |
6601 | | |
6602 | 26.4k | *result = NULL; |
6603 | | |
6604 | 26.4k | if (RAW != '(') { |
6605 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6606 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6607 | 0 | return(-1); |
6608 | 0 | } |
6609 | 26.4k | NEXT; |
6610 | 26.4k | xmlSkipBlankCharsPEBalanced(ctxt, openInputNr); |
6611 | 26.4k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6612 | 10.5k | tree = xmlParseElementMixedContentDecl(ctxt, openInputNr); |
6613 | 10.5k | res = XML_ELEMENT_TYPE_MIXED; |
6614 | 15.8k | } else { |
6615 | 15.8k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, openInputNr, 1); |
6616 | 15.8k | res = XML_ELEMENT_TYPE_ELEMENT; |
6617 | 15.8k | } |
6618 | 26.4k | if (tree == NULL) |
6619 | 8.04k | return(-1); |
6620 | 18.4k | SKIP_BLANKS_PE; |
6621 | 18.4k | *result = tree; |
6622 | 18.4k | return(res); |
6623 | 26.4k | } |
6624 | | |
6625 | | /** |
6626 | | * Parse an element declaration. Always consumes '<!'. |
6627 | | * |
6628 | | * @deprecated Internal function, don't use. |
6629 | | * |
6630 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6631 | | * |
6632 | | * [ VC: Unique Element Type Declaration ] |
6633 | | * No element type may be declared more than once |
6634 | | * |
6635 | | * @param ctxt an XML parser context |
6636 | | * @returns the type of the element, or -1 in case of error |
6637 | | */ |
6638 | | int |
6639 | 33.5k | xmlParseElementDecl(xmlParserCtxt *ctxt) { |
6640 | 33.5k | const xmlChar *name; |
6641 | 33.5k | int ret = -1; |
6642 | 33.5k | xmlElementContentPtr content = NULL; |
6643 | | |
6644 | 33.5k | if ((CUR != '<') || (NXT(1) != '!')) |
6645 | 0 | return(ret); |
6646 | 33.5k | SKIP(2); |
6647 | | |
6648 | | /* GROW; done in the caller */ |
6649 | 33.5k | if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6650 | 33.4k | #ifdef LIBXML_VALID_ENABLED |
6651 | 33.4k | int oldInputNr = ctxt->inputNr; |
6652 | 33.4k | #endif |
6653 | | |
6654 | 33.4k | SKIP(7); |
6655 | 33.4k | if (SKIP_BLANKS_PE == 0) { |
6656 | 257 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6657 | 257 | "Space required after 'ELEMENT'\n"); |
6658 | 257 | return(-1); |
6659 | 257 | } |
6660 | 33.2k | name = xmlParseName(ctxt); |
6661 | 33.2k | if (name == NULL) { |
6662 | 595 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6663 | 595 | "xmlParseElementDecl: no name for Element\n"); |
6664 | 595 | return(-1); |
6665 | 595 | } |
6666 | 32.6k | if (SKIP_BLANKS_PE == 0) { |
6667 | 5.73k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6668 | 5.73k | "Space required after the element name\n"); |
6669 | 5.73k | } |
6670 | 32.6k | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6671 | 4.40k | SKIP(5); |
6672 | | /* |
6673 | | * Element must always be empty. |
6674 | | */ |
6675 | 4.40k | ret = XML_ELEMENT_TYPE_EMPTY; |
6676 | 28.2k | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6677 | 28.2k | (NXT(2) == 'Y')) { |
6678 | 566 | SKIP(3); |
6679 | | /* |
6680 | | * Element is a generic container. |
6681 | | */ |
6682 | 566 | ret = XML_ELEMENT_TYPE_ANY; |
6683 | 27.6k | } else if (RAW == '(') { |
6684 | 26.4k | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6685 | 26.4k | if (ret <= 0) |
6686 | 8.04k | return(-1); |
6687 | 26.4k | } else { |
6688 | | /* |
6689 | | * [ WFC: PEs in Internal Subset ] error handling. |
6690 | | */ |
6691 | 1.18k | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6692 | 1.18k | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6693 | 1.18k | return(-1); |
6694 | 1.18k | } |
6695 | | |
6696 | 23.3k | SKIP_BLANKS_PE; |
6697 | | |
6698 | 23.3k | if (RAW != '>') { |
6699 | 958 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6700 | 958 | if (content != NULL) { |
6701 | 632 | xmlFreeDocElementContent(ctxt->myDoc, content); |
6702 | 632 | } |
6703 | 22.4k | } else { |
6704 | 22.4k | #ifdef LIBXML_VALID_ENABLED |
6705 | 22.4k | if ((ctxt->validate) && (ctxt->inputNr > oldInputNr)) { |
6706 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6707 | 0 | "Element declaration doesn't start and stop in" |
6708 | 0 | " the same entity\n", |
6709 | 0 | NULL, NULL); |
6710 | 0 | } |
6711 | 22.4k | #endif |
6712 | | |
6713 | 22.4k | NEXT; |
6714 | 22.4k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6715 | 22.4k | (ctxt->sax->elementDecl != NULL)) { |
6716 | 17.7k | if (content != NULL) |
6717 | 13.5k | content->parent = NULL; |
6718 | 17.7k | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6719 | 17.7k | content); |
6720 | 17.7k | if ((content != NULL) && (content->parent == NULL)) { |
6721 | | /* |
6722 | | * this is a trick: if xmlAddElementDecl is called, |
6723 | | * instead of copying the full tree it is plugged directly |
6724 | | * if called from the parser. Avoid duplicating the |
6725 | | * interfaces or change the API/ABI |
6726 | | */ |
6727 | 5.44k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6728 | 5.44k | } |
6729 | 17.7k | } else if (content != NULL) { |
6730 | 4.24k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6731 | 4.24k | } |
6732 | 22.4k | } |
6733 | 23.3k | } |
6734 | 23.4k | return(ret); |
6735 | 33.5k | } |
6736 | | |
6737 | | /** |
6738 | | * Parse a conditional section. Always consumes '<!['. |
6739 | | * |
6740 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6741 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6742 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6743 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' |
6744 | | * Ignore)* |
6745 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6746 | | * @param ctxt an XML parser context |
6747 | | */ |
6748 | | |
6749 | | static void |
6750 | 8.14k | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6751 | 8.14k | size_t depth = 0; |
6752 | 8.14k | int isFreshPE = 0; |
6753 | 8.14k | int oldInputNr = ctxt->inputNr; |
6754 | 8.14k | int declInputNr = ctxt->inputNr; |
6755 | | |
6756 | 32.9k | while (!PARSER_STOPPED(ctxt)) { |
6757 | 32.8k | if (ctxt->input->cur >= ctxt->input->end) { |
6758 | 1.13k | if (ctxt->inputNr <= oldInputNr) { |
6759 | 430 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6760 | 430 | return; |
6761 | 430 | } |
6762 | | |
6763 | 706 | xmlPopPE(ctxt); |
6764 | 706 | declInputNr = ctxt->inputNr; |
6765 | 31.7k | } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6766 | 9.05k | SKIP(3); |
6767 | 9.05k | SKIP_BLANKS_PE; |
6768 | | |
6769 | 9.05k | isFreshPE = 0; |
6770 | | |
6771 | 9.05k | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6772 | 4.40k | SKIP(7); |
6773 | 4.40k | SKIP_BLANKS_PE; |
6774 | 4.40k | if (RAW != '[') { |
6775 | 278 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6776 | 278 | return; |
6777 | 278 | } |
6778 | 4.13k | #ifdef LIBXML_VALID_ENABLED |
6779 | 4.13k | if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) { |
6780 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6781 | 0 | "All markup of the conditional section is" |
6782 | 0 | " not in the same entity\n", |
6783 | 0 | NULL, NULL); |
6784 | 0 | } |
6785 | 4.13k | #endif |
6786 | 4.13k | NEXT; |
6787 | | |
6788 | 4.13k | depth++; |
6789 | 4.64k | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6790 | 2.11k | size_t ignoreDepth = 0; |
6791 | | |
6792 | 2.11k | SKIP(6); |
6793 | 2.11k | SKIP_BLANKS_PE; |
6794 | 2.11k | if (RAW != '[') { |
6795 | 613 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6796 | 613 | return; |
6797 | 613 | } |
6798 | 1.50k | #ifdef LIBXML_VALID_ENABLED |
6799 | 1.50k | if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) { |
6800 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6801 | 0 | "All markup of the conditional section is" |
6802 | 0 | " not in the same entity\n", |
6803 | 0 | NULL, NULL); |
6804 | 0 | } |
6805 | 1.50k | #endif |
6806 | 1.50k | NEXT; |
6807 | | |
6808 | 19.4k | while (PARSER_STOPPED(ctxt) == 0) { |
6809 | 19.4k | if (RAW == 0) { |
6810 | 545 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6811 | 545 | return; |
6812 | 545 | } |
6813 | 18.9k | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6814 | 409 | SKIP(3); |
6815 | 409 | ignoreDepth++; |
6816 | | /* Check for integer overflow */ |
6817 | 409 | if (ignoreDepth == 0) { |
6818 | 0 | xmlErrMemory(ctxt); |
6819 | 0 | return; |
6820 | 0 | } |
6821 | 18.5k | } else if ((RAW == ']') && (NXT(1) == ']') && |
6822 | 18.5k | (NXT(2) == '>')) { |
6823 | 1.16k | SKIP(3); |
6824 | 1.16k | if (ignoreDepth == 0) |
6825 | 943 | break; |
6826 | 226 | ignoreDepth--; |
6827 | 17.3k | } else { |
6828 | 17.3k | NEXT; |
6829 | 17.3k | } |
6830 | 18.9k | } |
6831 | | |
6832 | 956 | #ifdef LIBXML_VALID_ENABLED |
6833 | 956 | if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) { |
6834 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6835 | 0 | "All markup of the conditional section is" |
6836 | 0 | " not in the same entity\n", |
6837 | 0 | NULL, NULL); |
6838 | 0 | } |
6839 | 956 | #endif |
6840 | 2.52k | } else { |
6841 | 2.52k | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6842 | 2.52k | return; |
6843 | 2.52k | } |
6844 | 22.7k | } else if ((depth > 0) && |
6845 | 22.7k | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6846 | 2.73k | if (isFreshPE) { |
6847 | 5 | xmlFatalErrMsg(ctxt, XML_ERR_CONDSEC_INVALID, |
6848 | 5 | "Parameter entity must match " |
6849 | 5 | "extSubsetDecl\n"); |
6850 | 5 | return; |
6851 | 5 | } |
6852 | | |
6853 | 2.72k | depth--; |
6854 | 2.72k | #ifdef LIBXML_VALID_ENABLED |
6855 | 2.72k | if ((ctxt->validate) && (ctxt->inputNr > declInputNr)) { |
6856 | 0 | xmlValidityError(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6857 | 0 | "All markup of the conditional section is not" |
6858 | 0 | " in the same entity\n", |
6859 | 0 | NULL, NULL); |
6860 | 0 | } |
6861 | 2.72k | #endif |
6862 | 2.72k | SKIP(3); |
6863 | 19.9k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
6864 | 18.4k | isFreshPE = 0; |
6865 | 18.4k | xmlParseMarkupDecl(ctxt); |
6866 | 18.4k | } else if (RAW == '%') { |
6867 | 1.47k | xmlParsePERefInternal(ctxt, 1); |
6868 | 1.47k | if (ctxt->inputNr > declInputNr) { |
6869 | 738 | isFreshPE = 1; |
6870 | 738 | declInputNr = ctxt->inputNr; |
6871 | 738 | } |
6872 | 1.47k | } else { |
6873 | 48 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6874 | 48 | return; |
6875 | 48 | } |
6876 | | |
6877 | 28.4k | if (depth == 0) |
6878 | 3.66k | break; |
6879 | | |
6880 | 24.7k | SKIP_BLANKS; |
6881 | 24.7k | SHRINK; |
6882 | 24.7k | GROW; |
6883 | 24.7k | } |
6884 | 8.14k | } |
6885 | | |
6886 | | /** |
6887 | | * Parse markup declarations. Always consumes '<!' or '<?'. |
6888 | | * |
6889 | | * @deprecated Internal function, don't use. |
6890 | | * |
6891 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6892 | | * NotationDecl | PI | Comment |
6893 | | * |
6894 | | * [ VC: Proper Declaration/PE Nesting ] |
6895 | | * Parameter-entity replacement text must be properly nested with |
6896 | | * markup declarations. That is to say, if either the first character |
6897 | | * or the last character of a markup declaration (markupdecl above) is |
6898 | | * contained in the replacement text for a parameter-entity reference, |
6899 | | * both must be contained in the same replacement text. |
6900 | | * |
6901 | | * [ WFC: PEs in Internal Subset ] |
6902 | | * In the internal DTD subset, parameter-entity references can occur |
6903 | | * only where markup declarations can occur, not within markup declarations. |
6904 | | * (This does not apply to references that occur in external parameter |
6905 | | * entities or to the external subset.) |
6906 | | * |
6907 | | * @param ctxt an XML parser context |
6908 | | */ |
6909 | | void |
6910 | 524k | xmlParseMarkupDecl(xmlParserCtxt *ctxt) { |
6911 | 524k | GROW; |
6912 | 524k | if (CUR == '<') { |
6913 | 524k | if (NXT(1) == '!') { |
6914 | 502k | switch (NXT(2)) { |
6915 | 100k | case 'E': |
6916 | 100k | if (NXT(3) == 'L') |
6917 | 33.5k | xmlParseElementDecl(ctxt); |
6918 | 67.3k | else if (NXT(3) == 'N') |
6919 | 67.3k | xmlParseEntityDecl(ctxt); |
6920 | 26 | else |
6921 | 26 | SKIP(2); |
6922 | 100k | break; |
6923 | 70.0k | case 'A': |
6924 | 70.0k | xmlParseAttributeListDecl(ctxt); |
6925 | 70.0k | break; |
6926 | 10.0k | case 'N': |
6927 | 10.0k | xmlParseNotationDecl(ctxt); |
6928 | 10.0k | break; |
6929 | 308k | case '-': |
6930 | 308k | xmlParseComment(ctxt); |
6931 | 308k | break; |
6932 | 12.4k | default: |
6933 | 12.4k | xmlFatalErr(ctxt, |
6934 | 12.4k | ctxt->inSubset == 2 ? |
6935 | 1.27k | XML_ERR_EXT_SUBSET_NOT_FINISHED : |
6936 | 12.4k | XML_ERR_INT_SUBSET_NOT_FINISHED, |
6937 | 12.4k | NULL); |
6938 | 12.4k | SKIP(2); |
6939 | 12.4k | break; |
6940 | 502k | } |
6941 | 502k | } else if (NXT(1) == '?') { |
6942 | 21.6k | xmlParsePI(ctxt); |
6943 | 21.6k | } |
6944 | 524k | } |
6945 | 524k | } |
6946 | | |
6947 | | /** |
6948 | | * Parse an XML declaration header for external entities |
6949 | | * |
6950 | | * @deprecated Internal function, don't use. |
6951 | | * |
6952 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
6953 | | * @param ctxt an XML parser context |
6954 | | */ |
6955 | | |
6956 | | void |
6957 | 24.1k | xmlParseTextDecl(xmlParserCtxt *ctxt) { |
6958 | 24.1k | xmlChar *version; |
6959 | | |
6960 | | /* |
6961 | | * We know that '<?xml' is here. |
6962 | | */ |
6963 | 24.1k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
6964 | 24.1k | SKIP(5); |
6965 | 24.1k | } else { |
6966 | 6 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
6967 | 6 | return; |
6968 | 6 | } |
6969 | | |
6970 | 24.1k | if (SKIP_BLANKS == 0) { |
6971 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6972 | 0 | "Space needed after '<?xml'\n"); |
6973 | 0 | } |
6974 | | |
6975 | | /* |
6976 | | * We may have the VersionInfo here. |
6977 | | */ |
6978 | 24.1k | version = xmlParseVersionInfo(ctxt); |
6979 | 24.1k | if (version == NULL) { |
6980 | 19.6k | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
6981 | 19.6k | if (version == NULL) { |
6982 | 41 | xmlErrMemory(ctxt); |
6983 | 41 | return; |
6984 | 41 | } |
6985 | 19.6k | } else { |
6986 | 4.52k | if (SKIP_BLANKS == 0) { |
6987 | 840 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6988 | 840 | "Space needed here\n"); |
6989 | 840 | } |
6990 | 4.52k | } |
6991 | 24.1k | ctxt->input->version = version; |
6992 | | |
6993 | | /* |
6994 | | * We must have the encoding declaration |
6995 | | */ |
6996 | 24.1k | xmlParseEncodingDecl(ctxt); |
6997 | | |
6998 | 24.1k | SKIP_BLANKS; |
6999 | 24.1k | if ((RAW == '?') && (NXT(1) == '>')) { |
7000 | 1.86k | SKIP(2); |
7001 | 22.2k | } else if (RAW == '>') { |
7002 | | /* Deprecated old WD ... */ |
7003 | 1.04k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7004 | 1.04k | NEXT; |
7005 | 21.2k | } else { |
7006 | 21.2k | int c; |
7007 | | |
7008 | 21.2k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7009 | 369M | while ((PARSER_STOPPED(ctxt) == 0) && ((c = CUR) != 0)) { |
7010 | 369M | NEXT; |
7011 | 369M | if (c == '>') |
7012 | 6.00k | break; |
7013 | 369M | } |
7014 | 21.2k | } |
7015 | 24.1k | } |
7016 | | |
7017 | | /** |
7018 | | * Parse Markup declarations from an external subset |
7019 | | * |
7020 | | * @deprecated Internal function, don't use. |
7021 | | * |
7022 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7023 | | * |
7024 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | |
7025 | | * PEReference | S) * |
7026 | | * @param ctxt an XML parser context |
7027 | | * @param publicId the public identifier |
7028 | | * @param systemId the system identifier (URL) |
7029 | | */ |
7030 | | void |
7031 | | xmlParseExternalSubset(xmlParserCtxt *ctxt, const xmlChar *publicId, |
7032 | 1.51k | const xmlChar *systemId) { |
7033 | 1.51k | int oldInputNr; |
7034 | | |
7035 | 1.51k | xmlCtxtInitializeLate(ctxt); |
7036 | | |
7037 | 1.51k | xmlDetectEncoding(ctxt); |
7038 | | |
7039 | 1.51k | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7040 | 104 | xmlParseTextDecl(ctxt); |
7041 | 104 | } |
7042 | 1.51k | if (ctxt->myDoc == NULL) { |
7043 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7044 | 0 | if (ctxt->myDoc == NULL) { |
7045 | 0 | xmlErrMemory(ctxt); |
7046 | 0 | return; |
7047 | 0 | } |
7048 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7049 | 0 | } |
7050 | 1.51k | if ((ctxt->myDoc->intSubset == NULL) && |
7051 | 1.51k | (xmlCreateIntSubset(ctxt->myDoc, NULL, publicId, systemId) == NULL)) { |
7052 | 0 | xmlErrMemory(ctxt); |
7053 | 0 | } |
7054 | | |
7055 | 1.51k | ctxt->inSubset = 2; |
7056 | 1.51k | oldInputNr = ctxt->inputNr; |
7057 | | |
7058 | 1.51k | SKIP_BLANKS; |
7059 | 308k | while (!PARSER_STOPPED(ctxt)) { |
7060 | 307k | if (ctxt->input->cur >= ctxt->input->end) { |
7061 | 1.27k | if (ctxt->inputNr <= oldInputNr) { |
7062 | 371 | xmlParserCheckEOF(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED); |
7063 | 371 | break; |
7064 | 371 | } |
7065 | | |
7066 | 899 | xmlPopPE(ctxt); |
7067 | 306k | } else if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7068 | 4.72k | xmlParseConditionalSections(ctxt); |
7069 | 301k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7070 | 298k | xmlParseMarkupDecl(ctxt); |
7071 | 298k | } else if (RAW == '%') { |
7072 | 2.11k | xmlParsePERefInternal(ctxt, 1); |
7073 | 2.11k | } else { |
7074 | 761 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7075 | | |
7076 | 1.01k | while (ctxt->inputNr > oldInputNr) |
7077 | 256 | xmlPopPE(ctxt); |
7078 | 761 | break; |
7079 | 761 | } |
7080 | 306k | SKIP_BLANKS; |
7081 | 306k | SHRINK; |
7082 | 306k | GROW; |
7083 | 306k | } |
7084 | 1.51k | } |
7085 | | |
7086 | | /** |
7087 | | * Parse and handle entity references in content, depending on the SAX |
7088 | | * interface, this may end-up in a call to character() if this is a |
7089 | | * CharRef, a predefined entity, if there is no reference() callback. |
7090 | | * or if the parser was asked to switch to that mode. |
7091 | | * |
7092 | | * @deprecated Internal function, don't use. |
7093 | | * |
7094 | | * Always consumes '&'. |
7095 | | * |
7096 | | * [67] Reference ::= EntityRef | CharRef |
7097 | | * @param ctxt an XML parser context |
7098 | | */ |
7099 | | void |
7100 | 244k | xmlParseReference(xmlParserCtxt *ctxt) { |
7101 | 244k | xmlEntityPtr ent = NULL; |
7102 | 244k | const xmlChar *name; |
7103 | 244k | xmlChar *val; |
7104 | | |
7105 | 244k | if (RAW != '&') |
7106 | 0 | return; |
7107 | | |
7108 | | /* |
7109 | | * Simple case of a CharRef |
7110 | | */ |
7111 | 244k | if (NXT(1) == '#') { |
7112 | 38.2k | int i = 0; |
7113 | 38.2k | xmlChar out[16]; |
7114 | 38.2k | int value = xmlParseCharRef(ctxt); |
7115 | | |
7116 | 38.2k | if (value == 0) |
7117 | 13.0k | return; |
7118 | | |
7119 | | /* |
7120 | | * Just encode the value in UTF-8 |
7121 | | */ |
7122 | 25.1k | COPY_BUF(out, i, value); |
7123 | 25.1k | out[i] = 0; |
7124 | 25.1k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7125 | 25.1k | (!ctxt->disableSAX)) |
7126 | 24.7k | ctxt->sax->characters(ctxt->userData, out, i); |
7127 | 25.1k | return; |
7128 | 38.2k | } |
7129 | | |
7130 | | /* |
7131 | | * We are seeing an entity reference |
7132 | | */ |
7133 | 206k | name = xmlParseEntityRefInternal(ctxt); |
7134 | 206k | if (name == NULL) |
7135 | 34.0k | return; |
7136 | 172k | ent = xmlLookupGeneralEntity(ctxt, name, /* isAttr */ 0); |
7137 | 172k | if (ent == NULL) { |
7138 | | /* |
7139 | | * Create a reference for undeclared entities. |
7140 | | */ |
7141 | 64.2k | if ((ctxt->replaceEntities == 0) && |
7142 | 64.2k | (ctxt->sax != NULL) && |
7143 | 64.2k | (ctxt->disableSAX == 0) && |
7144 | 64.2k | (ctxt->sax->reference != NULL)) { |
7145 | 15.4k | ctxt->sax->reference(ctxt->userData, name); |
7146 | 15.4k | } |
7147 | 64.2k | return; |
7148 | 64.2k | } |
7149 | 107k | if (!ctxt->wellFormed) |
7150 | 24.5k | return; |
7151 | | |
7152 | | /* special case of predefined entities */ |
7153 | 83.2k | if ((ent->name == NULL) || |
7154 | 83.2k | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7155 | 329 | val = ent->content; |
7156 | 329 | if (val == NULL) return; |
7157 | | /* |
7158 | | * inline the entity. |
7159 | | */ |
7160 | 329 | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7161 | 329 | (!ctxt->disableSAX)) |
7162 | 329 | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7163 | 329 | return; |
7164 | 329 | } |
7165 | | |
7166 | | /* |
7167 | | * Some users try to parse entities on their own and used to set |
7168 | | * the renamed "checked" member. Fix the flags to cover this |
7169 | | * case. |
7170 | | */ |
7171 | 82.9k | if (((ent->flags & XML_ENT_PARSED) == 0) && (ent->children != NULL)) |
7172 | 0 | ent->flags |= XML_ENT_PARSED; |
7173 | | |
7174 | | /* |
7175 | | * The first reference to the entity trigger a parsing phase |
7176 | | * where the ent->children is filled with the result from |
7177 | | * the parsing. |
7178 | | * Note: external parsed entities will not be loaded, it is not |
7179 | | * required for a non-validating parser, unless the parsing option |
7180 | | * of validating, or substituting entities were given. Doing so is |
7181 | | * far more secure as the parser will only process data coming from |
7182 | | * the document entity by default. |
7183 | | * |
7184 | | * FIXME: This doesn't work correctly since entities can be |
7185 | | * expanded with different namespace declarations in scope. |
7186 | | * For example: |
7187 | | * |
7188 | | * <!DOCTYPE doc [ |
7189 | | * <!ENTITY ent "<ns:elem/>"> |
7190 | | * ]> |
7191 | | * <doc> |
7192 | | * <decl1 xmlns:ns="urn:ns1"> |
7193 | | * &ent; |
7194 | | * </decl1> |
7195 | | * <decl2 xmlns:ns="urn:ns2"> |
7196 | | * &ent; |
7197 | | * </decl2> |
7198 | | * </doc> |
7199 | | * |
7200 | | * Proposed fix: |
7201 | | * |
7202 | | * - Ignore current namespace declarations when parsing the |
7203 | | * entity. If a prefix can't be resolved, don't report an error |
7204 | | * but mark it as unresolved. |
7205 | | * - Try to resolve these prefixes when expanding the entity. |
7206 | | * This will require a specialized version of xmlStaticCopyNode |
7207 | | * which can also make use of the namespace hash table to avoid |
7208 | | * quadratic behavior. |
7209 | | * |
7210 | | * Alternatively, we could simply reparse the entity on each |
7211 | | * expansion like we already do with custom SAX callbacks. |
7212 | | * External entity content should be cached in this case. |
7213 | | */ |
7214 | 82.9k | if ((ent->etype == XML_INTERNAL_GENERAL_ENTITY) || |
7215 | 82.9k | (((ctxt->options & XML_PARSE_NO_XXE) == 0) && |
7216 | 9.83k | ((ctxt->replaceEntities) || |
7217 | 82.4k | (ctxt->validate)))) { |
7218 | 82.4k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
7219 | 4.52k | xmlCtxtParseEntity(ctxt, ent); |
7220 | 77.9k | } else if (ent->children == NULL) { |
7221 | | /* |
7222 | | * Probably running in SAX mode and the callbacks don't |
7223 | | * build the entity content. Parse the entity again. |
7224 | | * |
7225 | | * This will also be triggered in normal tree builder mode |
7226 | | * if an entity happens to be empty, causing unnecessary |
7227 | | * reloads. It's hard to come up with a reliable check in |
7228 | | * which mode we're running. |
7229 | | */ |
7230 | 69.5k | xmlCtxtParseEntity(ctxt, ent); |
7231 | 69.5k | } |
7232 | 82.4k | } |
7233 | | |
7234 | | /* |
7235 | | * We also check for amplification if entities aren't substituted. |
7236 | | * They might be expanded later. |
7237 | | */ |
7238 | 82.9k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) |
7239 | 105 | return; |
7240 | | |
7241 | 82.8k | if ((ctxt->sax == NULL) || (ctxt->disableSAX)) |
7242 | 886 | return; |
7243 | | |
7244 | 81.9k | if (ctxt->replaceEntities == 0) { |
7245 | | /* |
7246 | | * Create a reference |
7247 | | */ |
7248 | 791 | if (ctxt->sax->reference != NULL) |
7249 | 791 | ctxt->sax->reference(ctxt->userData, ent->name); |
7250 | 81.1k | } else if ((ent->children != NULL) && (ctxt->node != NULL)) { |
7251 | 11.2k | xmlNodePtr copy, cur; |
7252 | | |
7253 | | /* |
7254 | | * Seems we are generating the DOM content, copy the tree |
7255 | | */ |
7256 | 11.2k | cur = ent->children; |
7257 | | |
7258 | | /* |
7259 | | * Handle first text node with SAX to coalesce text efficiently |
7260 | | */ |
7261 | 11.2k | if ((cur->type == XML_TEXT_NODE) || |
7262 | 11.2k | (cur->type == XML_CDATA_SECTION_NODE)) { |
7263 | 10.7k | int len = xmlStrlen(cur->content); |
7264 | | |
7265 | 10.7k | if ((cur->type == XML_TEXT_NODE) || |
7266 | 10.7k | (ctxt->options & XML_PARSE_NOCDATA)) { |
7267 | 10.2k | if (ctxt->sax->characters != NULL) |
7268 | 10.2k | ctxt->sax->characters(ctxt, cur->content, len); |
7269 | 10.2k | } else { |
7270 | 532 | if (ctxt->sax->cdataBlock != NULL) |
7271 | 532 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7272 | 532 | } |
7273 | | |
7274 | 10.7k | cur = cur->next; |
7275 | 10.7k | } |
7276 | | |
7277 | 72.3k | while (cur != NULL) { |
7278 | 63.5k | xmlNodePtr last; |
7279 | | |
7280 | | /* |
7281 | | * Handle last text node with SAX to coalesce text efficiently |
7282 | | */ |
7283 | 63.5k | if ((cur->next == NULL) && |
7284 | 63.5k | ((cur->type == XML_TEXT_NODE) || |
7285 | 4.60k | (cur->type == XML_CDATA_SECTION_NODE))) { |
7286 | 2.18k | int len = xmlStrlen(cur->content); |
7287 | | |
7288 | 2.18k | if ((cur->type == XML_TEXT_NODE) || |
7289 | 2.18k | (ctxt->options & XML_PARSE_NOCDATA)) { |
7290 | 2.11k | if (ctxt->sax->characters != NULL) |
7291 | 2.11k | ctxt->sax->characters(ctxt, cur->content, len); |
7292 | 2.11k | } else { |
7293 | 76 | if (ctxt->sax->cdataBlock != NULL) |
7294 | 76 | ctxt->sax->cdataBlock(ctxt, cur->content, len); |
7295 | 76 | } |
7296 | | |
7297 | 2.18k | break; |
7298 | 2.18k | } |
7299 | | |
7300 | | /* |
7301 | | * Reset coalesce buffer stats only for non-text nodes. |
7302 | | */ |
7303 | 61.3k | ctxt->nodemem = 0; |
7304 | 61.3k | ctxt->nodelen = 0; |
7305 | | |
7306 | 61.3k | copy = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7307 | | |
7308 | 61.3k | if (copy == NULL) { |
7309 | 241 | xmlErrMemory(ctxt); |
7310 | 241 | break; |
7311 | 241 | } |
7312 | | |
7313 | 61.0k | if (ctxt->parseMode == XML_PARSE_READER) { |
7314 | | /* Needed for reader */ |
7315 | 0 | copy->extra = cur->extra; |
7316 | | /* Maybe needed for reader */ |
7317 | 0 | copy->_private = cur->_private; |
7318 | 0 | } |
7319 | | |
7320 | 61.0k | copy->parent = ctxt->node; |
7321 | 61.0k | last = ctxt->node->last; |
7322 | 61.0k | if (last == NULL) { |
7323 | 138 | ctxt->node->children = copy; |
7324 | 60.9k | } else { |
7325 | 60.9k | last->next = copy; |
7326 | 60.9k | copy->prev = last; |
7327 | 60.9k | } |
7328 | 61.0k | ctxt->node->last = copy; |
7329 | | |
7330 | 61.0k | cur = cur->next; |
7331 | 61.0k | } |
7332 | 11.2k | } |
7333 | 81.9k | } |
7334 | | |
7335 | | static void |
7336 | 359k | xmlHandleUndeclaredEntity(xmlParserCtxtPtr ctxt, const xmlChar *name) { |
7337 | | /* |
7338 | | * [ WFC: Entity Declared ] |
7339 | | * In a document without any DTD, a document with only an |
7340 | | * internal DTD subset which contains no parameter entity |
7341 | | * references, or a document with "standalone='yes'", the |
7342 | | * Name given in the entity reference must match that in an |
7343 | | * entity declaration, except that well-formed documents |
7344 | | * need not declare any of the following entities: amp, lt, |
7345 | | * gt, apos, quot. |
7346 | | * The declaration of a parameter entity must precede any |
7347 | | * reference to it. |
7348 | | * Similarly, the declaration of a general entity must |
7349 | | * precede any reference to it which appears in a default |
7350 | | * value in an attribute-list declaration. Note that if |
7351 | | * entities are declared in the external subset or in |
7352 | | * external parameter entities, a non-validating processor |
7353 | | * is not obligated to read and process their declarations; |
7354 | | * for such documents, the rule that an entity must be |
7355 | | * declared is a well-formedness constraint only if |
7356 | | * standalone='yes'. |
7357 | | */ |
7358 | 359k | if ((ctxt->standalone == 1) || |
7359 | 359k | ((ctxt->hasExternalSubset == 0) && |
7360 | 358k | (ctxt->hasPErefs == 0))) { |
7361 | 280k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7362 | 280k | "Entity '%s' not defined\n", name); |
7363 | 280k | #ifdef LIBXML_VALID_ENABLED |
7364 | 280k | } else if (ctxt->validate) { |
7365 | | /* |
7366 | | * [ VC: Entity Declared ] |
7367 | | * In a document with an external subset or external |
7368 | | * parameter entities with "standalone='no'", ... |
7369 | | * ... The declaration of a parameter entity must |
7370 | | * precede any reference to it... |
7371 | | */ |
7372 | 0 | xmlValidityError(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7373 | 0 | "Entity '%s' not defined\n", name, NULL); |
7374 | 0 | #endif |
7375 | 79.0k | } else if ((ctxt->loadsubset & ~XML_SKIP_IDS) || |
7376 | 79.0k | ((ctxt->replaceEntities) && |
7377 | 73.9k | ((ctxt->options & XML_PARSE_NO_XXE) == 0))) { |
7378 | | /* |
7379 | | * Also raise a non-fatal error |
7380 | | * |
7381 | | * - if the external subset is loaded and all entity declarations |
7382 | | * should be available, or |
7383 | | * - entity substition was requested without restricting |
7384 | | * external entity access. |
7385 | | */ |
7386 | 73.9k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7387 | 73.9k | "Entity '%s' not defined\n", name); |
7388 | 73.9k | } else { |
7389 | 5.03k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7390 | 5.03k | "Entity '%s' not defined\n", name, NULL); |
7391 | 5.03k | } |
7392 | | |
7393 | 359k | ctxt->valid = 0; |
7394 | 359k | } |
7395 | | |
7396 | | static xmlEntityPtr |
7397 | 1.96M | xmlLookupGeneralEntity(xmlParserCtxtPtr ctxt, const xmlChar *name, int inAttr) { |
7398 | 1.96M | xmlEntityPtr ent = NULL; |
7399 | | |
7400 | | /* |
7401 | | * Predefined entities override any extra definition |
7402 | | */ |
7403 | 1.96M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7404 | 1.21M | ent = xmlGetPredefinedEntity(name); |
7405 | 1.21M | if (ent != NULL) |
7406 | 131k | return(ent); |
7407 | 1.21M | } |
7408 | | |
7409 | | /* |
7410 | | * Ask first SAX for entity resolution, otherwise try the |
7411 | | * entities which may have stored in the parser context. |
7412 | | */ |
7413 | 1.83M | if (ctxt->sax != NULL) { |
7414 | 1.83M | if (ctxt->sax->getEntity != NULL) |
7415 | 1.83M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7416 | 1.83M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7417 | 1.83M | (ctxt->options & XML_PARSE_OLDSAX)) |
7418 | 271 | ent = xmlGetPredefinedEntity(name); |
7419 | 1.83M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7420 | 1.83M | (ctxt->userData==ctxt)) { |
7421 | 17.0k | ent = xmlSAX2GetEntity(ctxt, name); |
7422 | 17.0k | } |
7423 | 1.83M | } |
7424 | | |
7425 | 1.83M | if (ent == NULL) { |
7426 | 303k | xmlHandleUndeclaredEntity(ctxt, name); |
7427 | 303k | } |
7428 | | |
7429 | | /* |
7430 | | * [ WFC: Parsed Entity ] |
7431 | | * An entity reference must not contain the name of an |
7432 | | * unparsed entity |
7433 | | */ |
7434 | 1.53M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7435 | 216 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7436 | 216 | "Entity reference to unparsed entity %s\n", name); |
7437 | 216 | ent = NULL; |
7438 | 216 | } |
7439 | | |
7440 | | /* |
7441 | | * [ WFC: No External Entity References ] |
7442 | | * Attribute values cannot contain direct or indirect |
7443 | | * entity references to external entities. |
7444 | | */ |
7445 | 1.53M | else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7446 | 21.4k | if (inAttr) { |
7447 | 1.91k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7448 | 1.91k | "Attribute references external entity '%s'\n", name); |
7449 | 1.91k | ent = NULL; |
7450 | 1.91k | } |
7451 | 21.4k | } |
7452 | | |
7453 | 1.83M | return(ent); |
7454 | 1.96M | } |
7455 | | |
7456 | | /** |
7457 | | * Parse an entity reference. Always consumes '&'. |
7458 | | * |
7459 | | * [68] EntityRef ::= '&' Name ';' |
7460 | | * |
7461 | | * @param ctxt an XML parser context |
7462 | | * @returns the name, or NULL in case of error. |
7463 | | */ |
7464 | | static const xmlChar * |
7465 | 1.32M | xmlParseEntityRefInternal(xmlParserCtxtPtr ctxt) { |
7466 | 1.32M | const xmlChar *name; |
7467 | | |
7468 | 1.32M | GROW; |
7469 | | |
7470 | 1.32M | if (RAW != '&') |
7471 | 0 | return(NULL); |
7472 | 1.32M | NEXT; |
7473 | 1.32M | name = xmlParseName(ctxt); |
7474 | 1.32M | if (name == NULL) { |
7475 | 54.5k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7476 | 54.5k | "xmlParseEntityRef: no name\n"); |
7477 | 54.5k | return(NULL); |
7478 | 54.5k | } |
7479 | 1.27M | if (RAW != ';') { |
7480 | 30.1k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7481 | 30.1k | return(NULL); |
7482 | 30.1k | } |
7483 | 1.24M | NEXT; |
7484 | | |
7485 | 1.24M | return(name); |
7486 | 1.27M | } |
7487 | | |
7488 | | /** |
7489 | | * @deprecated Internal function, don't use. |
7490 | | * |
7491 | | * @param ctxt an XML parser context |
7492 | | * @returns the xmlEntity if found, or NULL otherwise. |
7493 | | */ |
7494 | | xmlEntity * |
7495 | 0 | xmlParseEntityRef(xmlParserCtxt *ctxt) { |
7496 | 0 | const xmlChar *name; |
7497 | |
|
7498 | 0 | if (ctxt == NULL) |
7499 | 0 | return(NULL); |
7500 | | |
7501 | 0 | name = xmlParseEntityRefInternal(ctxt); |
7502 | 0 | if (name == NULL) |
7503 | 0 | return(NULL); |
7504 | | |
7505 | 0 | return(xmlLookupGeneralEntity(ctxt, name, /* inAttr */ 0)); |
7506 | 0 | } |
7507 | | |
7508 | | /** |
7509 | | * Parse ENTITY references declarations, but this version parses it from |
7510 | | * a string value. |
7511 | | * |
7512 | | * [68] EntityRef ::= '&' Name ';' |
7513 | | * |
7514 | | * [ WFC: Entity Declared ] |
7515 | | * In a document without any DTD, a document with only an internal DTD |
7516 | | * subset which contains no parameter entity references, or a document |
7517 | | * with "standalone='yes'", the Name given in the entity reference |
7518 | | * must match that in an entity declaration, except that well-formed |
7519 | | * documents need not declare any of the following entities: amp, lt, |
7520 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7521 | | * any reference to it. Similarly, the declaration of a general entity |
7522 | | * must precede any reference to it which appears in a default value in an |
7523 | | * attribute-list declaration. Note that if entities are declared in the |
7524 | | * external subset or in external parameter entities, a non-validating |
7525 | | * processor is not obligated to read and process their declarations; |
7526 | | * for such documents, the rule that an entity must be declared is a |
7527 | | * well-formedness constraint only if standalone='yes'. |
7528 | | * |
7529 | | * [ WFC: Parsed Entity ] |
7530 | | * An entity reference must not contain the name of an unparsed entity |
7531 | | * |
7532 | | * @param ctxt an XML parser context |
7533 | | * @param str a pointer to an index in the string |
7534 | | * @returns the xmlEntity if found, or NULL otherwise. The str pointer |
7535 | | * is updated to the current location in the string. |
7536 | | */ |
7537 | | static xmlChar * |
7538 | 726k | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7539 | 726k | xmlChar *name; |
7540 | 726k | const xmlChar *ptr; |
7541 | 726k | xmlChar cur; |
7542 | | |
7543 | 726k | if ((str == NULL) || (*str == NULL)) |
7544 | 0 | return(NULL); |
7545 | 726k | ptr = *str; |
7546 | 726k | cur = *ptr; |
7547 | 726k | if (cur != '&') |
7548 | 0 | return(NULL); |
7549 | | |
7550 | 726k | ptr++; |
7551 | 726k | name = xmlParseStringName(ctxt, &ptr); |
7552 | 726k | if (name == NULL) { |
7553 | 55 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7554 | 55 | "xmlParseStringEntityRef: no name\n"); |
7555 | 55 | *str = ptr; |
7556 | 55 | return(NULL); |
7557 | 55 | } |
7558 | 726k | if (*ptr != ';') { |
7559 | 19 | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7560 | 19 | xmlFree(name); |
7561 | 19 | *str = ptr; |
7562 | 19 | return(NULL); |
7563 | 19 | } |
7564 | 726k | ptr++; |
7565 | | |
7566 | 726k | *str = ptr; |
7567 | 726k | return(name); |
7568 | 726k | } |
7569 | | |
7570 | | /** |
7571 | | * Parse a parameter entity reference. Always consumes '%'. |
7572 | | * |
7573 | | * The entity content is handled directly by pushing it's content as |
7574 | | * a new input stream. |
7575 | | * |
7576 | | * [69] PEReference ::= '%' Name ';' |
7577 | | * |
7578 | | * [ WFC: No Recursion ] |
7579 | | * A parsed entity must not contain a recursive |
7580 | | * reference to itself, either directly or indirectly. |
7581 | | * |
7582 | | * [ WFC: Entity Declared ] |
7583 | | * In a document without any DTD, a document with only an internal DTD |
7584 | | * subset which contains no parameter entity references, or a document |
7585 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7586 | | * entity must precede any reference to it... |
7587 | | * |
7588 | | * [ VC: Entity Declared ] |
7589 | | * In a document with an external subset or external parameter entities |
7590 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7591 | | * must precede any reference to it... |
7592 | | * |
7593 | | * [ WFC: In DTD ] |
7594 | | * Parameter-entity references may only appear in the DTD. |
7595 | | * NOTE: misleading but this is handled. |
7596 | | * |
7597 | | * @param ctxt an XML parser context |
7598 | | * @param markupDecl whether the PERef starts a markup declaration |
7599 | | */ |
7600 | | static void |
7601 | 162k | xmlParsePERefInternal(xmlParserCtxt *ctxt, int markupDecl) { |
7602 | 162k | const xmlChar *name; |
7603 | 162k | xmlEntityPtr entity = NULL; |
7604 | 162k | xmlParserInputPtr input; |
7605 | | |
7606 | 162k | if (RAW != '%') |
7607 | 0 | return; |
7608 | 162k | NEXT; |
7609 | 162k | name = xmlParseName(ctxt); |
7610 | 162k | if (name == NULL) { |
7611 | 10.4k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
7612 | 10.4k | return; |
7613 | 10.4k | } |
7614 | 151k | if (RAW != ';') { |
7615 | 15.1k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
7616 | 15.1k | return; |
7617 | 15.1k | } |
7618 | | |
7619 | 136k | NEXT; |
7620 | | |
7621 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7622 | 136k | ctxt->hasPErefs = 1; |
7623 | | |
7624 | | /* |
7625 | | * Request the entity from SAX |
7626 | | */ |
7627 | 136k | if ((ctxt->sax != NULL) && |
7628 | 136k | (ctxt->sax->getParameterEntity != NULL)) |
7629 | 136k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7630 | | |
7631 | 136k | if (entity == NULL) { |
7632 | 53.2k | xmlHandleUndeclaredEntity(ctxt, name); |
7633 | 83.2k | } else { |
7634 | | /* |
7635 | | * Internal checking in case the entity quest barfed |
7636 | | */ |
7637 | 83.2k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7638 | 83.2k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7639 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7640 | 0 | "Internal: %%%s; is not a parameter entity\n", |
7641 | 0 | name, NULL); |
7642 | 83.2k | } else { |
7643 | 83.2k | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
7644 | 83.2k | ((ctxt->options & XML_PARSE_NO_XXE) || |
7645 | 58.6k | (((ctxt->loadsubset & ~XML_SKIP_IDS) == 0) && |
7646 | 58.4k | (ctxt->replaceEntities == 0) && |
7647 | 58.4k | (ctxt->validate == 0)))) |
7648 | 388 | return; |
7649 | | |
7650 | 82.9k | if (entity->flags & XML_ENT_EXPANDING) { |
7651 | 28 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7652 | 28 | return; |
7653 | 28 | } |
7654 | | |
7655 | 82.8k | input = xmlNewEntityInputStream(ctxt, entity); |
7656 | 82.8k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
7657 | 4.76k | xmlFreeInputStream(input); |
7658 | 4.76k | return; |
7659 | 4.76k | } |
7660 | | |
7661 | 78.1k | entity->flags |= XML_ENT_EXPANDING; |
7662 | | |
7663 | 78.1k | if (markupDecl) |
7664 | 64.2k | input->flags |= XML_INPUT_MARKUP_DECL; |
7665 | | |
7666 | 78.1k | GROW; |
7667 | | |
7668 | 78.1k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
7669 | 53.5k | xmlDetectEncoding(ctxt); |
7670 | | |
7671 | 53.5k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
7672 | 53.5k | (IS_BLANK_CH(NXT(5)))) { |
7673 | 11.5k | xmlParseTextDecl(ctxt); |
7674 | 11.5k | } |
7675 | 53.5k | } |
7676 | 78.1k | } |
7677 | 83.2k | } |
7678 | 136k | } |
7679 | | |
7680 | | /** |
7681 | | * Parse a parameter entity reference. |
7682 | | * |
7683 | | * @deprecated Internal function, don't use. |
7684 | | * |
7685 | | * @param ctxt an XML parser context |
7686 | | */ |
7687 | | void |
7688 | 0 | xmlParsePEReference(xmlParserCtxt *ctxt) { |
7689 | 0 | xmlParsePERefInternal(ctxt, 0); |
7690 | 0 | } |
7691 | | |
7692 | | /** |
7693 | | * Load the content of an entity. |
7694 | | * |
7695 | | * @param ctxt an XML parser context |
7696 | | * @param entity an unloaded system entity |
7697 | | * @returns 0 in case of success and -1 in case of failure |
7698 | | */ |
7699 | | static int |
7700 | 19.5k | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
7701 | 19.5k | xmlParserInputPtr oldinput, input = NULL; |
7702 | 19.5k | xmlParserInputPtr *oldinputTab; |
7703 | 19.5k | xmlChar *oldencoding; |
7704 | 19.5k | xmlChar *content = NULL; |
7705 | 19.5k | xmlResourceType rtype; |
7706 | 19.5k | size_t length, i; |
7707 | 19.5k | int oldinputNr, oldinputMax; |
7708 | 19.5k | int ret = -1; |
7709 | 19.5k | int res; |
7710 | | |
7711 | 19.5k | if ((ctxt == NULL) || (entity == NULL) || |
7712 | 19.5k | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
7713 | 19.5k | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
7714 | 19.5k | (entity->content != NULL)) { |
7715 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, |
7716 | 0 | "xmlLoadEntityContent parameter error"); |
7717 | 0 | return(-1); |
7718 | 0 | } |
7719 | | |
7720 | 19.5k | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) |
7721 | 19.5k | rtype = XML_RESOURCE_PARAMETER_ENTITY; |
7722 | 0 | else |
7723 | 0 | rtype = XML_RESOURCE_GENERAL_ENTITY; |
7724 | | |
7725 | 19.5k | input = xmlLoadResource(ctxt, (char *) entity->URI, |
7726 | 19.5k | (char *) entity->ExternalID, rtype); |
7727 | 19.5k | if (input == NULL) |
7728 | 544 | return(-1); |
7729 | | |
7730 | 19.0k | oldinput = ctxt->input; |
7731 | 19.0k | oldinputNr = ctxt->inputNr; |
7732 | 19.0k | oldinputMax = ctxt->inputMax; |
7733 | 19.0k | oldinputTab = ctxt->inputTab; |
7734 | 19.0k | oldencoding = ctxt->encoding; |
7735 | | |
7736 | 19.0k | ctxt->input = NULL; |
7737 | 19.0k | ctxt->inputNr = 0; |
7738 | 19.0k | ctxt->inputMax = 1; |
7739 | 19.0k | ctxt->encoding = NULL; |
7740 | 19.0k | ctxt->inputTab = xmlMalloc(sizeof(xmlParserInputPtr)); |
7741 | 19.0k | if (ctxt->inputTab == NULL) { |
7742 | 3 | xmlErrMemory(ctxt); |
7743 | 3 | xmlFreeInputStream(input); |
7744 | 3 | goto error; |
7745 | 3 | } |
7746 | | |
7747 | 19.0k | xmlBufResetInput(input->buf->buffer, input); |
7748 | | |
7749 | 19.0k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
7750 | 19 | xmlFreeInputStream(input); |
7751 | 19 | goto error; |
7752 | 19 | } |
7753 | | |
7754 | 19.0k | xmlDetectEncoding(ctxt); |
7755 | | |
7756 | | /* |
7757 | | * Parse a possible text declaration first |
7758 | | */ |
7759 | 19.0k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7760 | 12.1k | xmlParseTextDecl(ctxt); |
7761 | | /* |
7762 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
7763 | | */ |
7764 | 12.1k | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
7765 | 12.1k | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
7766 | 1.77k | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
7767 | 1.77k | "Version mismatch between document and entity\n"); |
7768 | 1.77k | } |
7769 | 12.1k | } |
7770 | | |
7771 | 19.0k | length = input->cur - input->base; |
7772 | 19.0k | xmlBufShrink(input->buf->buffer, length); |
7773 | 19.0k | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7774 | | |
7775 | 24.5k | while ((res = xmlParserInputBufferGrow(input->buf, 4096)) > 0) |
7776 | 5.54k | ; |
7777 | | |
7778 | 19.0k | xmlBufResetInput(input->buf->buffer, input); |
7779 | | |
7780 | 19.0k | if (res < 0) { |
7781 | 1.19k | xmlCtxtErrIO(ctxt, input->buf->error, NULL); |
7782 | 1.19k | goto error; |
7783 | 1.19k | } |
7784 | | |
7785 | 17.8k | length = xmlBufUse(input->buf->buffer); |
7786 | 17.8k | if (length > INT_MAX) { |
7787 | 0 | xmlErrMemory(ctxt); |
7788 | 0 | goto error; |
7789 | 0 | } |
7790 | | |
7791 | 17.8k | content = xmlStrndup(xmlBufContent(input->buf->buffer), length); |
7792 | 17.8k | if (content == NULL) { |
7793 | 26 | xmlErrMemory(ctxt); |
7794 | 26 | goto error; |
7795 | 26 | } |
7796 | | |
7797 | 4.96M | for (i = 0; i < length; ) { |
7798 | 4.96M | int clen = length - i; |
7799 | 4.96M | int c = xmlGetUTF8Char(content + i, &clen); |
7800 | | |
7801 | 4.96M | if ((c < 0) || (!IS_CHAR(c))) { |
7802 | 17.6k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
7803 | 17.6k | "xmlLoadEntityContent: invalid char value %d\n", |
7804 | 17.6k | content[i]); |
7805 | 17.6k | goto error; |
7806 | 17.6k | } |
7807 | 4.94M | i += clen; |
7808 | 4.94M | } |
7809 | | |
7810 | 175 | xmlSaturatedAdd(&ctxt->sizeentities, length); |
7811 | 175 | entity->content = content; |
7812 | 175 | entity->length = length; |
7813 | 175 | content = NULL; |
7814 | 175 | ret = 0; |
7815 | | |
7816 | 19.0k | error: |
7817 | 38.0k | while (ctxt->inputNr > 0) |
7818 | 19.0k | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
7819 | 19.0k | xmlFree(ctxt->inputTab); |
7820 | 19.0k | xmlFree(ctxt->encoding); |
7821 | | |
7822 | 19.0k | ctxt->input = oldinput; |
7823 | 19.0k | ctxt->inputNr = oldinputNr; |
7824 | 19.0k | ctxt->inputMax = oldinputMax; |
7825 | 19.0k | ctxt->inputTab = oldinputTab; |
7826 | 19.0k | ctxt->encoding = oldencoding; |
7827 | | |
7828 | 19.0k | xmlFree(content); |
7829 | | |
7830 | 19.0k | return(ret); |
7831 | 175 | } |
7832 | | |
7833 | | /** |
7834 | | * Parse PEReference declarations |
7835 | | * |
7836 | | * [69] PEReference ::= '%' Name ';' |
7837 | | * |
7838 | | * [ WFC: No Recursion ] |
7839 | | * A parsed entity must not contain a recursive |
7840 | | * reference to itself, either directly or indirectly. |
7841 | | * |
7842 | | * [ WFC: Entity Declared ] |
7843 | | * In a document without any DTD, a document with only an internal DTD |
7844 | | * subset which contains no parameter entity references, or a document |
7845 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7846 | | * entity must precede any reference to it... |
7847 | | * |
7848 | | * [ VC: Entity Declared ] |
7849 | | * In a document with an external subset or external parameter entities |
7850 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7851 | | * must precede any reference to it... |
7852 | | * |
7853 | | * [ WFC: In DTD ] |
7854 | | * Parameter-entity references may only appear in the DTD. |
7855 | | * NOTE: misleading but this is handled. |
7856 | | * |
7857 | | * @param ctxt an XML parser context |
7858 | | * @param str a pointer to an index in the string |
7859 | | * @returns the string of the entity content. |
7860 | | * str is updated to the current value of the index |
7861 | | */ |
7862 | | static xmlEntityPtr |
7863 | 46.4k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
7864 | 46.4k | const xmlChar *ptr; |
7865 | 46.4k | xmlChar cur; |
7866 | 46.4k | xmlChar *name; |
7867 | 46.4k | xmlEntityPtr entity = NULL; |
7868 | | |
7869 | 46.4k | if ((str == NULL) || (*str == NULL)) return(NULL); |
7870 | 46.4k | ptr = *str; |
7871 | 46.4k | cur = *ptr; |
7872 | 46.4k | if (cur != '%') |
7873 | 0 | return(NULL); |
7874 | 46.4k | ptr++; |
7875 | 46.4k | name = xmlParseStringName(ctxt, &ptr); |
7876 | 46.4k | if (name == NULL) { |
7877 | 4.67k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7878 | 4.67k | "xmlParseStringPEReference: no name\n"); |
7879 | 4.67k | *str = ptr; |
7880 | 4.67k | return(NULL); |
7881 | 4.67k | } |
7882 | 41.7k | cur = *ptr; |
7883 | 41.7k | if (cur != ';') { |
7884 | 4.30k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7885 | 4.30k | xmlFree(name); |
7886 | 4.30k | *str = ptr; |
7887 | 4.30k | return(NULL); |
7888 | 4.30k | } |
7889 | 37.4k | ptr++; |
7890 | | |
7891 | | /* Must be set before xmlHandleUndeclaredEntity */ |
7892 | 37.4k | ctxt->hasPErefs = 1; |
7893 | | |
7894 | | /* |
7895 | | * Request the entity from SAX |
7896 | | */ |
7897 | 37.4k | if ((ctxt->sax != NULL) && |
7898 | 37.4k | (ctxt->sax->getParameterEntity != NULL)) |
7899 | 37.4k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
7900 | | |
7901 | 37.4k | if (entity == NULL) { |
7902 | 2.59k | xmlHandleUndeclaredEntity(ctxt, name); |
7903 | 34.8k | } else { |
7904 | | /* |
7905 | | * Internal checking in case the entity quest barfed |
7906 | | */ |
7907 | 34.8k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
7908 | 34.8k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
7909 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7910 | 0 | "%%%s; is not a parameter entity\n", |
7911 | 0 | name, NULL); |
7912 | 0 | } |
7913 | 34.8k | } |
7914 | | |
7915 | 37.4k | xmlFree(name); |
7916 | 37.4k | *str = ptr; |
7917 | 37.4k | return(entity); |
7918 | 41.7k | } |
7919 | | |
7920 | | /** |
7921 | | * Parse a DOCTYPE declaration |
7922 | | * |
7923 | | * @deprecated Internal function, don't use. |
7924 | | * |
7925 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
7926 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
7927 | | * |
7928 | | * [ VC: Root Element Type ] |
7929 | | * The Name in the document type declaration must match the element |
7930 | | * type of the root element. |
7931 | | * |
7932 | | * @param ctxt an XML parser context |
7933 | | */ |
7934 | | |
7935 | | void |
7936 | 27.3k | xmlParseDocTypeDecl(xmlParserCtxt *ctxt) { |
7937 | 27.3k | const xmlChar *name = NULL; |
7938 | 27.3k | xmlChar *publicId = NULL; |
7939 | 27.3k | xmlChar *URI = NULL; |
7940 | | |
7941 | | /* |
7942 | | * We know that '<!DOCTYPE' has been detected. |
7943 | | */ |
7944 | 27.3k | SKIP(9); |
7945 | | |
7946 | 27.3k | if (SKIP_BLANKS == 0) { |
7947 | 7.61k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7948 | 7.61k | "Space required after 'DOCTYPE'\n"); |
7949 | 7.61k | } |
7950 | | |
7951 | | /* |
7952 | | * Parse the DOCTYPE name. |
7953 | | */ |
7954 | 27.3k | name = xmlParseName(ctxt); |
7955 | 27.3k | if (name == NULL) { |
7956 | 6.36k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7957 | 6.36k | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
7958 | 6.36k | } |
7959 | 27.3k | ctxt->intSubName = name; |
7960 | | |
7961 | 27.3k | SKIP_BLANKS; |
7962 | | |
7963 | | /* |
7964 | | * Check for public and system identifier (URI) |
7965 | | */ |
7966 | 27.3k | URI = xmlParseExternalID(ctxt, &publicId, 1); |
7967 | | |
7968 | 27.3k | if ((URI != NULL) || (publicId != NULL)) { |
7969 | 5.43k | ctxt->hasExternalSubset = 1; |
7970 | 5.43k | } |
7971 | 27.3k | ctxt->extSubURI = URI; |
7972 | 27.3k | ctxt->extSubSystem = publicId; |
7973 | | |
7974 | 27.3k | SKIP_BLANKS; |
7975 | | |
7976 | | /* |
7977 | | * Create and update the internal subset. |
7978 | | */ |
7979 | 27.3k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
7980 | 27.3k | (!ctxt->disableSAX)) |
7981 | 25.1k | ctxt->sax->internalSubset(ctxt->userData, name, publicId, URI); |
7982 | | |
7983 | 27.3k | if ((RAW != '[') && (RAW != '>')) { |
7984 | 1.18k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
7985 | 1.18k | } |
7986 | 27.3k | } |
7987 | | |
7988 | | /** |
7989 | | * Parse the internal subset declaration |
7990 | | * |
7991 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
7992 | | * @param ctxt an XML parser context |
7993 | | */ |
7994 | | |
7995 | | static void |
7996 | 20.8k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
7997 | | /* |
7998 | | * Is there any DTD definition ? |
7999 | | */ |
8000 | 20.8k | if (RAW == '[') { |
8001 | 20.8k | int oldInputNr = ctxt->inputNr; |
8002 | | |
8003 | 20.8k | NEXT; |
8004 | | /* |
8005 | | * Parse the succession of Markup declarations and |
8006 | | * PEReferences. |
8007 | | * Subsequence (markupdecl | PEReference | S)* |
8008 | | */ |
8009 | 20.8k | SKIP_BLANKS; |
8010 | 430k | while (1) { |
8011 | 430k | if (PARSER_STOPPED(ctxt)) { |
8012 | 2.70k | return; |
8013 | 427k | } else if (ctxt->input->cur >= ctxt->input->end) { |
8014 | 62.9k | if (ctxt->inputNr <= oldInputNr) { |
8015 | 3.15k | xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL); |
8016 | 3.15k | return; |
8017 | 3.15k | } |
8018 | 59.7k | xmlPopPE(ctxt); |
8019 | 364k | } else if ((RAW == ']') && (ctxt->inputNr <= oldInputNr)) { |
8020 | 8.83k | NEXT; |
8021 | 8.83k | SKIP_BLANKS; |
8022 | 8.83k | break; |
8023 | 355k | } else if ((PARSER_EXTERNAL(ctxt)) && |
8024 | 355k | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8025 | | /* |
8026 | | * Conditional sections are allowed in external entities |
8027 | | * included by PE References in the internal subset. |
8028 | | */ |
8029 | 3.42k | xmlParseConditionalSections(ctxt); |
8030 | 352k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
8031 | 206k | xmlParseMarkupDecl(ctxt); |
8032 | 206k | } else if (RAW == '%') { |
8033 | 139k | xmlParsePERefInternal(ctxt, 1); |
8034 | 139k | } else { |
8035 | 6.20k | xmlFatalErr(ctxt, XML_ERR_INT_SUBSET_NOT_FINISHED, NULL); |
8036 | | |
8037 | 6.93k | while (ctxt->inputNr > oldInputNr) |
8038 | 730 | xmlPopPE(ctxt); |
8039 | 6.20k | return; |
8040 | 6.20k | } |
8041 | 409k | SKIP_BLANKS; |
8042 | 409k | SHRINK; |
8043 | 409k | GROW; |
8044 | 409k | } |
8045 | 20.8k | } |
8046 | | |
8047 | | /* |
8048 | | * We should be at the end of the DOCTYPE declaration. |
8049 | | */ |
8050 | 8.83k | if (RAW != '>') { |
8051 | 322 | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8052 | 322 | return; |
8053 | 322 | } |
8054 | 8.51k | NEXT; |
8055 | 8.51k | } |
8056 | | |
8057 | | #ifdef LIBXML_SAX1_ENABLED |
8058 | | /** |
8059 | | * Parse an attribute |
8060 | | * |
8061 | | * @deprecated Internal function, don't use. |
8062 | | * |
8063 | | * [41] Attribute ::= Name Eq AttValue |
8064 | | * |
8065 | | * [ WFC: No External Entity References ] |
8066 | | * Attribute values cannot contain direct or indirect entity references |
8067 | | * to external entities. |
8068 | | * |
8069 | | * [ WFC: No < in Attribute Values ] |
8070 | | * The replacement text of any entity referred to directly or indirectly in |
8071 | | * an attribute value (other than "<") must not contain a <. |
8072 | | * |
8073 | | * [ VC: Attribute Value Type ] |
8074 | | * The attribute must have been declared; the value must be of the type |
8075 | | * declared for it. |
8076 | | * |
8077 | | * [25] Eq ::= S? '=' S? |
8078 | | * |
8079 | | * With namespace: |
8080 | | * |
8081 | | * [NS 11] Attribute ::= QName Eq AttValue |
8082 | | * |
8083 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8084 | | * definition. |
8085 | | * |
8086 | | * @param ctxt an XML parser context |
8087 | | * @param value a xmlChar ** used to store the value of the attribute |
8088 | | * @returns the attribute name, and the value in *value. |
8089 | | */ |
8090 | | |
8091 | | const xmlChar * |
8092 | 0 | xmlParseAttribute(xmlParserCtxt *ctxt, xmlChar **value) { |
8093 | 0 | const xmlChar *name; |
8094 | 0 | xmlChar *val; |
8095 | |
|
8096 | 0 | *value = NULL; |
8097 | 0 | GROW; |
8098 | 0 | name = xmlParseName(ctxt); |
8099 | 0 | if (name == NULL) { |
8100 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8101 | 0 | "error parsing attribute name\n"); |
8102 | 0 | return(NULL); |
8103 | 0 | } |
8104 | | |
8105 | | /* |
8106 | | * read the value |
8107 | | */ |
8108 | 0 | SKIP_BLANKS; |
8109 | 0 | if (RAW == '=') { |
8110 | 0 | NEXT; |
8111 | 0 | SKIP_BLANKS; |
8112 | 0 | val = xmlParseAttValue(ctxt); |
8113 | 0 | } else { |
8114 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8115 | 0 | "Specification mandates value for attribute %s\n", name); |
8116 | 0 | return(name); |
8117 | 0 | } |
8118 | | |
8119 | | /* |
8120 | | * Check that xml:lang conforms to the specification |
8121 | | * No more registered as an error, just generate a warning now |
8122 | | * since this was deprecated in XML second edition |
8123 | | */ |
8124 | 0 | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8125 | 0 | if (!xmlCheckLanguageID(val)) { |
8126 | 0 | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8127 | 0 | "Malformed value for xml:lang : %s\n", |
8128 | 0 | val, NULL); |
8129 | 0 | } |
8130 | 0 | } |
8131 | | |
8132 | | /* |
8133 | | * Check that xml:space conforms to the specification |
8134 | | */ |
8135 | 0 | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8136 | 0 | if (xmlStrEqual(val, BAD_CAST "default")) |
8137 | 0 | *(ctxt->space) = 0; |
8138 | 0 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8139 | 0 | *(ctxt->space) = 1; |
8140 | 0 | else { |
8141 | 0 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8142 | 0 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8143 | 0 | val, NULL); |
8144 | 0 | } |
8145 | 0 | } |
8146 | |
|
8147 | 0 | *value = val; |
8148 | 0 | return(name); |
8149 | 0 | } |
8150 | | |
8151 | | /** |
8152 | | * Parse a start tag. Always consumes '<'. |
8153 | | * |
8154 | | * @deprecated Internal function, don't use. |
8155 | | * |
8156 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8157 | | * |
8158 | | * [ WFC: Unique Att Spec ] |
8159 | | * No attribute name may appear more than once in the same start-tag or |
8160 | | * empty-element tag. |
8161 | | * |
8162 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8163 | | * |
8164 | | * [ WFC: Unique Att Spec ] |
8165 | | * No attribute name may appear more than once in the same start-tag or |
8166 | | * empty-element tag. |
8167 | | * |
8168 | | * With namespace: |
8169 | | * |
8170 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8171 | | * |
8172 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8173 | | * |
8174 | | * @param ctxt an XML parser context |
8175 | | * @returns the element name parsed |
8176 | | */ |
8177 | | |
8178 | | const xmlChar * |
8179 | 0 | xmlParseStartTag(xmlParserCtxt *ctxt) { |
8180 | 0 | const xmlChar *name; |
8181 | 0 | const xmlChar *attname; |
8182 | 0 | xmlChar *attvalue; |
8183 | 0 | const xmlChar **atts = ctxt->atts; |
8184 | 0 | int nbatts = 0; |
8185 | 0 | int maxatts = ctxt->maxatts; |
8186 | 0 | int i; |
8187 | |
|
8188 | 0 | if (RAW != '<') return(NULL); |
8189 | 0 | NEXT1; |
8190 | |
|
8191 | 0 | name = xmlParseName(ctxt); |
8192 | 0 | if (name == NULL) { |
8193 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8194 | 0 | "xmlParseStartTag: invalid element name\n"); |
8195 | 0 | return(NULL); |
8196 | 0 | } |
8197 | | |
8198 | | /* |
8199 | | * Now parse the attributes, it ends up with the ending |
8200 | | * |
8201 | | * (S Attribute)* S? |
8202 | | */ |
8203 | 0 | SKIP_BLANKS; |
8204 | 0 | GROW; |
8205 | |
|
8206 | 0 | while (((RAW != '>') && |
8207 | 0 | ((RAW != '/') || (NXT(1) != '>')) && |
8208 | 0 | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8209 | 0 | attname = xmlParseAttribute(ctxt, &attvalue); |
8210 | 0 | if (attname == NULL) |
8211 | 0 | break; |
8212 | 0 | if (attvalue != NULL) { |
8213 | | /* |
8214 | | * [ WFC: Unique Att Spec ] |
8215 | | * No attribute name may appear more than once in the same |
8216 | | * start-tag or empty-element tag. |
8217 | | */ |
8218 | 0 | for (i = 0; i < nbatts;i += 2) { |
8219 | 0 | if (xmlStrEqual(atts[i], attname)) { |
8220 | 0 | xmlErrAttributeDup(ctxt, NULL, attname); |
8221 | 0 | goto failed; |
8222 | 0 | } |
8223 | 0 | } |
8224 | | /* |
8225 | | * Add the pair to atts |
8226 | | */ |
8227 | 0 | if (nbatts + 4 > maxatts) { |
8228 | 0 | const xmlChar **n; |
8229 | 0 | int newSize; |
8230 | |
|
8231 | 0 | newSize = xmlGrowCapacity(maxatts, sizeof(n[0]) * 2, |
8232 | 0 | 11, XML_MAX_ATTRS); |
8233 | 0 | if (newSize < 0) { |
8234 | 0 | xmlErrMemory(ctxt); |
8235 | 0 | goto failed; |
8236 | 0 | } |
8237 | 0 | #ifdef FUZZING_BUILD_MODE_UNSAFE_FOR_PRODUCTION |
8238 | 0 | if (newSize < 2) |
8239 | 0 | newSize = 2; |
8240 | 0 | #endif |
8241 | 0 | n = xmlRealloc(atts, newSize * sizeof(n[0]) * 2); |
8242 | 0 | if (n == NULL) { |
8243 | 0 | xmlErrMemory(ctxt); |
8244 | 0 | goto failed; |
8245 | 0 | } |
8246 | 0 | atts = n; |
8247 | 0 | maxatts = newSize * 2; |
8248 | 0 | ctxt->atts = atts; |
8249 | 0 | ctxt->maxatts = maxatts; |
8250 | 0 | } |
8251 | | |
8252 | 0 | atts[nbatts++] = attname; |
8253 | 0 | atts[nbatts++] = attvalue; |
8254 | 0 | atts[nbatts] = NULL; |
8255 | 0 | atts[nbatts + 1] = NULL; |
8256 | |
|
8257 | 0 | attvalue = NULL; |
8258 | 0 | } |
8259 | | |
8260 | 0 | failed: |
8261 | |
|
8262 | 0 | if (attvalue != NULL) |
8263 | 0 | xmlFree(attvalue); |
8264 | |
|
8265 | 0 | GROW |
8266 | 0 | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8267 | 0 | break; |
8268 | 0 | if (SKIP_BLANKS == 0) { |
8269 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8270 | 0 | "attributes construct error\n"); |
8271 | 0 | } |
8272 | 0 | SHRINK; |
8273 | 0 | GROW; |
8274 | 0 | } |
8275 | | |
8276 | | /* |
8277 | | * SAX: Start of Element ! |
8278 | | */ |
8279 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8280 | 0 | (!ctxt->disableSAX)) { |
8281 | 0 | if (nbatts > 0) |
8282 | 0 | ctxt->sax->startElement(ctxt->userData, name, atts); |
8283 | 0 | else |
8284 | 0 | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8285 | 0 | } |
8286 | |
|
8287 | 0 | if (atts != NULL) { |
8288 | | /* Free only the content strings */ |
8289 | 0 | for (i = 1;i < nbatts;i+=2) |
8290 | 0 | if (atts[i] != NULL) |
8291 | 0 | xmlFree((xmlChar *) atts[i]); |
8292 | 0 | } |
8293 | 0 | return(name); |
8294 | 0 | } |
8295 | | |
8296 | | /** |
8297 | | * Parse an end tag. Always consumes '</'. |
8298 | | * |
8299 | | * [42] ETag ::= '</' Name S? '>' |
8300 | | * |
8301 | | * With namespace |
8302 | | * |
8303 | | * [NS 9] ETag ::= '</' QName S? '>' |
8304 | | * @param ctxt an XML parser context |
8305 | | * @param line line of the start tag |
8306 | | */ |
8307 | | |
8308 | | static void |
8309 | 0 | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8310 | 0 | const xmlChar *name; |
8311 | |
|
8312 | 0 | GROW; |
8313 | 0 | if ((RAW != '<') || (NXT(1) != '/')) { |
8314 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8315 | 0 | "xmlParseEndTag: '</' not found\n"); |
8316 | 0 | return; |
8317 | 0 | } |
8318 | 0 | SKIP(2); |
8319 | |
|
8320 | 0 | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8321 | | |
8322 | | /* |
8323 | | * We should definitely be at the ending "S? '>'" part |
8324 | | */ |
8325 | 0 | GROW; |
8326 | 0 | SKIP_BLANKS; |
8327 | 0 | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8328 | 0 | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8329 | 0 | } else |
8330 | 0 | NEXT1; |
8331 | | |
8332 | | /* |
8333 | | * [ WFC: Element Type Match ] |
8334 | | * The Name in an element's end-tag must match the element type in the |
8335 | | * start-tag. |
8336 | | * |
8337 | | */ |
8338 | 0 | if (name != (xmlChar*)1) { |
8339 | 0 | if (name == NULL) name = BAD_CAST "unparsable"; |
8340 | 0 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8341 | 0 | "Opening and ending tag mismatch: %s line %d and %s\n", |
8342 | 0 | ctxt->name, line, name); |
8343 | 0 | } |
8344 | | |
8345 | | /* |
8346 | | * SAX: End of Tag |
8347 | | */ |
8348 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8349 | 0 | (!ctxt->disableSAX)) |
8350 | 0 | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8351 | |
|
8352 | 0 | namePop(ctxt); |
8353 | 0 | spacePop(ctxt); |
8354 | 0 | } |
8355 | | |
8356 | | /** |
8357 | | * Parse an end of tag |
8358 | | * |
8359 | | * @deprecated Internal function, don't use. |
8360 | | * |
8361 | | * [42] ETag ::= '</' Name S? '>' |
8362 | | * |
8363 | | * With namespace |
8364 | | * |
8365 | | * [NS 9] ETag ::= '</' QName S? '>' |
8366 | | * @param ctxt an XML parser context |
8367 | | */ |
8368 | | |
8369 | | void |
8370 | 0 | xmlParseEndTag(xmlParserCtxt *ctxt) { |
8371 | 0 | xmlParseEndTag1(ctxt, 0); |
8372 | 0 | } |
8373 | | #endif /* LIBXML_SAX1_ENABLED */ |
8374 | | |
8375 | | /************************************************************************ |
8376 | | * * |
8377 | | * SAX 2 specific operations * |
8378 | | * * |
8379 | | ************************************************************************/ |
8380 | | |
8381 | | /** |
8382 | | * Parse an XML Namespace QName |
8383 | | * |
8384 | | * [6] QName ::= (Prefix ':')? LocalPart |
8385 | | * [7] Prefix ::= NCName |
8386 | | * [8] LocalPart ::= NCName |
8387 | | * |
8388 | | * @param ctxt an XML parser context |
8389 | | * @param prefix pointer to store the prefix part |
8390 | | * @returns the Name parsed or NULL |
8391 | | */ |
8392 | | |
8393 | | static xmlHashedString |
8394 | 1.45M | xmlParseQNameHashed(xmlParserCtxtPtr ctxt, xmlHashedString *prefix) { |
8395 | 1.45M | xmlHashedString l, p; |
8396 | 1.45M | int start, isNCName = 0; |
8397 | | |
8398 | 1.45M | l.name = NULL; |
8399 | 1.45M | p.name = NULL; |
8400 | | |
8401 | 1.45M | GROW; |
8402 | 1.45M | start = CUR_PTR - BASE_PTR; |
8403 | | |
8404 | 1.45M | l = xmlParseNCName(ctxt); |
8405 | 1.45M | if (l.name != NULL) { |
8406 | 1.00M | isNCName = 1; |
8407 | 1.00M | if (CUR == ':') { |
8408 | 155k | NEXT; |
8409 | 155k | p = l; |
8410 | 155k | l = xmlParseNCName(ctxt); |
8411 | 155k | } |
8412 | 1.00M | } |
8413 | 1.45M | if ((l.name == NULL) || (CUR == ':')) { |
8414 | 462k | xmlChar *tmp; |
8415 | | |
8416 | 462k | l.name = NULL; |
8417 | 462k | p.name = NULL; |
8418 | 462k | if ((isNCName == 0) && (CUR != ':')) |
8419 | 445k | return(l); |
8420 | 17.1k | tmp = xmlParseNmtoken(ctxt); |
8421 | 17.1k | if (tmp != NULL) |
8422 | 9.40k | xmlFree(tmp); |
8423 | 17.1k | l = xmlDictLookupHashed(ctxt->dict, BASE_PTR + start, |
8424 | 17.1k | CUR_PTR - (BASE_PTR + start)); |
8425 | 17.1k | if (l.name == NULL) { |
8426 | 3 | xmlErrMemory(ctxt); |
8427 | 3 | return(l); |
8428 | 3 | } |
8429 | 17.1k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8430 | 17.1k | "Failed to parse QName '%s'\n", l.name, NULL, NULL); |
8431 | 17.1k | } |
8432 | | |
8433 | 1.00M | *prefix = p; |
8434 | 1.00M | return(l); |
8435 | 1.45M | } |
8436 | | |
8437 | | /** |
8438 | | * Parse an XML Namespace QName |
8439 | | * |
8440 | | * [6] QName ::= (Prefix ':')? LocalPart |
8441 | | * [7] Prefix ::= NCName |
8442 | | * [8] LocalPart ::= NCName |
8443 | | * |
8444 | | * @param ctxt an XML parser context |
8445 | | * @param prefix pointer to store the prefix part |
8446 | | * @returns the Name parsed or NULL |
8447 | | */ |
8448 | | |
8449 | | static const xmlChar * |
8450 | 29.0k | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8451 | 29.0k | xmlHashedString n, p; |
8452 | | |
8453 | 29.0k | n = xmlParseQNameHashed(ctxt, &p); |
8454 | 29.0k | if (n.name == NULL) |
8455 | 2.69k | return(NULL); |
8456 | 26.3k | *prefix = p.name; |
8457 | 26.3k | return(n.name); |
8458 | 29.0k | } |
8459 | | |
8460 | | /** |
8461 | | * Parse an XML name and compares for match |
8462 | | * (specialized for endtag parsing) |
8463 | | * |
8464 | | * @param ctxt an XML parser context |
8465 | | * @param name the localname |
8466 | | * @param prefix the prefix, if any. |
8467 | | * @returns NULL for an illegal name, (xmlChar*) 1 for success |
8468 | | * and the name for mismatch |
8469 | | */ |
8470 | | |
8471 | | static const xmlChar * |
8472 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8473 | 31.5k | xmlChar const *prefix) { |
8474 | 31.5k | const xmlChar *cmp; |
8475 | 31.5k | const xmlChar *in; |
8476 | 31.5k | const xmlChar *ret; |
8477 | 31.5k | const xmlChar *prefix2; |
8478 | | |
8479 | 31.5k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8480 | | |
8481 | 31.5k | GROW; |
8482 | 31.5k | in = ctxt->input->cur; |
8483 | | |
8484 | 31.5k | cmp = prefix; |
8485 | 61.9k | while (*in != 0 && *in == *cmp) { |
8486 | 30.3k | ++in; |
8487 | 30.3k | ++cmp; |
8488 | 30.3k | } |
8489 | 31.5k | if ((*cmp == 0) && (*in == ':')) { |
8490 | 4.43k | in++; |
8491 | 4.43k | cmp = name; |
8492 | 19.3k | while (*in != 0 && *in == *cmp) { |
8493 | 14.9k | ++in; |
8494 | 14.9k | ++cmp; |
8495 | 14.9k | } |
8496 | 4.43k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8497 | | /* success */ |
8498 | 2.54k | ctxt->input->col += in - ctxt->input->cur; |
8499 | 2.54k | ctxt->input->cur = in; |
8500 | 2.54k | return((const xmlChar*) 1); |
8501 | 2.54k | } |
8502 | 4.43k | } |
8503 | | /* |
8504 | | * all strings coms from the dictionary, equality can be done directly |
8505 | | */ |
8506 | 29.0k | ret = xmlParseQName (ctxt, &prefix2); |
8507 | 29.0k | if (ret == NULL) |
8508 | 2.69k | return(NULL); |
8509 | 26.3k | if ((ret == name) && (prefix == prefix2)) |
8510 | 872 | return((const xmlChar*) 1); |
8511 | 25.4k | return ret; |
8512 | 26.3k | } |
8513 | | |
8514 | | /** |
8515 | | * Parse an attribute in the new SAX2 framework. |
8516 | | * |
8517 | | * @param ctxt an XML parser context |
8518 | | * @param pref the element prefix |
8519 | | * @param elem the element name |
8520 | | * @param hprefix resulting attribute prefix |
8521 | | * @param value resulting value of the attribute |
8522 | | * @param len resulting length of the attribute |
8523 | | * @param alloc resulting indicator if the attribute was allocated |
8524 | | * @returns the attribute name, and the value in *value, . |
8525 | | */ |
8526 | | |
8527 | | static xmlHashedString |
8528 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
8529 | | const xmlChar * pref, const xmlChar * elem, |
8530 | | xmlHashedString * hprefix, xmlChar ** value, |
8531 | | int *len, int *alloc) |
8532 | 399k | { |
8533 | 399k | xmlHashedString hname; |
8534 | 399k | const xmlChar *prefix, *name; |
8535 | 399k | xmlChar *val = NULL, *internal_val = NULL; |
8536 | 399k | int special = 0; |
8537 | 399k | int isNamespace; |
8538 | 399k | int flags; |
8539 | | |
8540 | 399k | *value = NULL; |
8541 | 399k | GROW; |
8542 | 399k | hname = xmlParseQNameHashed(ctxt, hprefix); |
8543 | 399k | if (hname.name == NULL) { |
8544 | 92.8k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8545 | 92.8k | "error parsing attribute name\n"); |
8546 | 92.8k | return(hname); |
8547 | 92.8k | } |
8548 | 306k | name = hname.name; |
8549 | 306k | prefix = hprefix->name; |
8550 | | |
8551 | | /* |
8552 | | * get the type if needed |
8553 | | */ |
8554 | 306k | if (ctxt->attsSpecial != NULL) { |
8555 | 100k | special = XML_PTR_TO_INT(xmlHashQLookup2(ctxt->attsSpecial, pref, elem, |
8556 | 100k | prefix, name)); |
8557 | 100k | } |
8558 | | |
8559 | | /* |
8560 | | * read the value |
8561 | | */ |
8562 | 306k | SKIP_BLANKS; |
8563 | 306k | if (RAW != '=') { |
8564 | 31.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8565 | 31.4k | "Specification mandates value for attribute %s\n", |
8566 | 31.4k | name); |
8567 | 31.4k | goto error; |
8568 | 31.4k | } |
8569 | | |
8570 | | |
8571 | 275k | NEXT; |
8572 | 275k | SKIP_BLANKS; |
8573 | 275k | flags = 0; |
8574 | 275k | isNamespace = (((prefix == NULL) && (name == ctxt->str_xmlns)) || |
8575 | 275k | (prefix == ctxt->str_xmlns)); |
8576 | 275k | val = xmlParseAttValueInternal(ctxt, len, &flags, special, |
8577 | 275k | isNamespace); |
8578 | 275k | if (val == NULL) |
8579 | 11.5k | goto error; |
8580 | | |
8581 | 263k | *alloc = (flags & XML_ATTVAL_ALLOC) != 0; |
8582 | | |
8583 | 263k | #ifdef LIBXML_VALID_ENABLED |
8584 | 263k | if ((ctxt->validate) && |
8585 | 263k | (ctxt->standalone) && |
8586 | 263k | (special & XML_SPECIAL_EXTERNAL) && |
8587 | 263k | (flags & XML_ATTVAL_NORM_CHANGE)) { |
8588 | 0 | xmlValidityError(ctxt, XML_DTD_NOT_STANDALONE, |
8589 | 0 | "standalone: normalization of attribute %s on %s " |
8590 | 0 | "by external subset declaration\n", |
8591 | 0 | name, elem); |
8592 | 0 | } |
8593 | 263k | #endif |
8594 | | |
8595 | 263k | if (prefix == ctxt->str_xml) { |
8596 | | /* |
8597 | | * Check that xml:lang conforms to the specification |
8598 | | * No more registered as an error, just generate a warning now |
8599 | | * since this was deprecated in XML second edition |
8600 | | */ |
8601 | 20.8k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
8602 | 11.5k | internal_val = xmlStrndup(val, *len); |
8603 | 11.5k | if (internal_val == NULL) |
8604 | 5 | goto mem_error; |
8605 | 11.5k | if (!xmlCheckLanguageID(internal_val)) { |
8606 | 9.04k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8607 | 9.04k | "Malformed value for xml:lang : %s\n", |
8608 | 9.04k | internal_val, NULL); |
8609 | 9.04k | } |
8610 | 11.5k | } |
8611 | | |
8612 | | /* |
8613 | | * Check that xml:space conforms to the specification |
8614 | | */ |
8615 | 20.8k | if (xmlStrEqual(name, BAD_CAST "space")) { |
8616 | 711 | internal_val = xmlStrndup(val, *len); |
8617 | 711 | if (internal_val == NULL) |
8618 | 3 | goto mem_error; |
8619 | 708 | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
8620 | 234 | *(ctxt->space) = 0; |
8621 | 474 | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
8622 | 212 | *(ctxt->space) = 1; |
8623 | 262 | else { |
8624 | 262 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8625 | 262 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8626 | 262 | internal_val, NULL); |
8627 | 262 | } |
8628 | 708 | } |
8629 | 20.8k | if (internal_val) { |
8630 | 12.2k | xmlFree(internal_val); |
8631 | 12.2k | } |
8632 | 20.8k | } |
8633 | | |
8634 | 263k | *value = val; |
8635 | 263k | return (hname); |
8636 | | |
8637 | 8 | mem_error: |
8638 | 8 | xmlErrMemory(ctxt); |
8639 | 42.9k | error: |
8640 | 42.9k | if ((val != NULL) && (*alloc != 0)) |
8641 | 3 | xmlFree(val); |
8642 | 42.9k | return(hname); |
8643 | 8 | } |
8644 | | |
8645 | | /** |
8646 | | * Inserts a new attribute into the hash table. |
8647 | | * |
8648 | | * @param ctxt parser context |
8649 | | * @param size size of the hash table |
8650 | | * @param name attribute name |
8651 | | * @param uri namespace uri |
8652 | | * @param hashValue combined hash value of name and uri |
8653 | | * @param aindex attribute index (this is a multiple of 5) |
8654 | | * @returns INT_MAX if no existing attribute was found, the attribute |
8655 | | * index if an attribute was found, -1 if a memory allocation failed. |
8656 | | */ |
8657 | | static int |
8658 | | xmlAttrHashInsert(xmlParserCtxtPtr ctxt, unsigned size, const xmlChar *name, |
8659 | 200k | const xmlChar *uri, unsigned hashValue, int aindex) { |
8660 | 200k | xmlAttrHashBucket *table = ctxt->attrHash; |
8661 | 200k | xmlAttrHashBucket *bucket; |
8662 | 200k | unsigned hindex; |
8663 | | |
8664 | 200k | hindex = hashValue & (size - 1); |
8665 | 200k | bucket = &table[hindex]; |
8666 | | |
8667 | 236k | while (bucket->index >= 0) { |
8668 | 57.8k | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8669 | | |
8670 | 57.8k | if (name == atts[0]) { |
8671 | 23.8k | int nsIndex = XML_PTR_TO_INT(atts[2]); |
8672 | | |
8673 | 23.8k | if ((nsIndex == NS_INDEX_EMPTY) ? (uri == NULL) : |
8674 | 23.8k | (nsIndex == NS_INDEX_XML) ? (uri == ctxt->str_xml_ns) : |
8675 | 8.31k | (uri == ctxt->nsTab[nsIndex * 2 + 1])) |
8676 | 21.7k | return(bucket->index); |
8677 | 23.8k | } |
8678 | | |
8679 | 36.1k | hindex++; |
8680 | 36.1k | bucket++; |
8681 | 36.1k | if (hindex >= size) { |
8682 | 4.39k | hindex = 0; |
8683 | 4.39k | bucket = table; |
8684 | 4.39k | } |
8685 | 36.1k | } |
8686 | | |
8687 | 178k | bucket->index = aindex; |
8688 | | |
8689 | 178k | return(INT_MAX); |
8690 | 200k | } |
8691 | | |
8692 | | static int |
8693 | | xmlAttrHashInsertQName(xmlParserCtxtPtr ctxt, unsigned size, |
8694 | | const xmlChar *name, const xmlChar *prefix, |
8695 | 3.91k | unsigned hashValue, int aindex) { |
8696 | 3.91k | xmlAttrHashBucket *table = ctxt->attrHash; |
8697 | 3.91k | xmlAttrHashBucket *bucket; |
8698 | 3.91k | unsigned hindex; |
8699 | | |
8700 | 3.91k | hindex = hashValue & (size - 1); |
8701 | 3.91k | bucket = &table[hindex]; |
8702 | | |
8703 | 6.02k | while (bucket->index >= 0) { |
8704 | 3.27k | const xmlChar **atts = &ctxt->atts[bucket->index]; |
8705 | | |
8706 | 3.27k | if ((name == atts[0]) && (prefix == atts[1])) |
8707 | 1.16k | return(bucket->index); |
8708 | | |
8709 | 2.11k | hindex++; |
8710 | 2.11k | bucket++; |
8711 | 2.11k | if (hindex >= size) { |
8712 | 194 | hindex = 0; |
8713 | 194 | bucket = table; |
8714 | 194 | } |
8715 | 2.11k | } |
8716 | | |
8717 | 2.75k | bucket->index = aindex; |
8718 | | |
8719 | 2.75k | return(INT_MAX); |
8720 | 3.91k | } |
8721 | | /** |
8722 | | * Parse a start tag. Always consumes '<'. |
8723 | | * |
8724 | | * This routine is called when running SAX2 parsing |
8725 | | * |
8726 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8727 | | * |
8728 | | * [ WFC: Unique Att Spec ] |
8729 | | * No attribute name may appear more than once in the same start-tag or |
8730 | | * empty-element tag. |
8731 | | * |
8732 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8733 | | * |
8734 | | * [ WFC: Unique Att Spec ] |
8735 | | * No attribute name may appear more than once in the same start-tag or |
8736 | | * empty-element tag. |
8737 | | * |
8738 | | * With namespace: |
8739 | | * |
8740 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8741 | | * |
8742 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8743 | | * |
8744 | | * @param ctxt an XML parser context |
8745 | | * @param pref resulting namespace prefix |
8746 | | * @param URI resulting namespace URI |
8747 | | * @param nbNsPtr resulting number of namespace declarations |
8748 | | * @returns the element name parsed |
8749 | | */ |
8750 | | |
8751 | | static const xmlChar * |
8752 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
8753 | 1.02M | const xmlChar **URI, int *nbNsPtr) { |
8754 | 1.02M | xmlHashedString hlocalname; |
8755 | 1.02M | xmlHashedString hprefix; |
8756 | 1.02M | xmlHashedString hattname; |
8757 | 1.02M | xmlHashedString haprefix; |
8758 | 1.02M | const xmlChar *localname; |
8759 | 1.02M | const xmlChar *prefix; |
8760 | 1.02M | const xmlChar *attname; |
8761 | 1.02M | const xmlChar *aprefix; |
8762 | 1.02M | const xmlChar *uri; |
8763 | 1.02M | xmlChar *attvalue = NULL; |
8764 | 1.02M | const xmlChar **atts = ctxt->atts; |
8765 | 1.02M | unsigned attrHashSize = 0; |
8766 | 1.02M | int maxatts = ctxt->maxatts; |
8767 | 1.02M | int nratts, nbatts, nbdef; |
8768 | 1.02M | int i, j, nbNs, nbTotalDef, attval, nsIndex, maxAtts; |
8769 | 1.02M | int alloc = 0; |
8770 | 1.02M | int numNsErr = 0; |
8771 | 1.02M | int numDupErr = 0; |
8772 | | |
8773 | 1.02M | if (RAW != '<') return(NULL); |
8774 | 1.02M | NEXT1; |
8775 | | |
8776 | 1.02M | nbatts = 0; |
8777 | 1.02M | nratts = 0; |
8778 | 1.02M | nbdef = 0; |
8779 | 1.02M | nbNs = 0; |
8780 | 1.02M | nbTotalDef = 0; |
8781 | 1.02M | attval = 0; |
8782 | | |
8783 | 1.02M | if (xmlParserNsStartElement(ctxt->nsdb) < 0) { |
8784 | 0 | xmlErrMemory(ctxt); |
8785 | 0 | return(NULL); |
8786 | 0 | } |
8787 | | |
8788 | 1.02M | hlocalname = xmlParseQNameHashed(ctxt, &hprefix); |
8789 | 1.02M | if (hlocalname.name == NULL) { |
8790 | 350k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8791 | 350k | "StartTag: invalid element name\n"); |
8792 | 350k | return(NULL); |
8793 | 350k | } |
8794 | 673k | localname = hlocalname.name; |
8795 | 673k | prefix = hprefix.name; |
8796 | | |
8797 | | /* |
8798 | | * Now parse the attributes, it ends up with the ending |
8799 | | * |
8800 | | * (S Attribute)* S? |
8801 | | */ |
8802 | 673k | SKIP_BLANKS; |
8803 | 673k | GROW; |
8804 | | |
8805 | | /* |
8806 | | * The ctxt->atts array will be ultimately passed to the SAX callback |
8807 | | * containing five xmlChar pointers for each attribute: |
8808 | | * |
8809 | | * [0] attribute name |
8810 | | * [1] attribute prefix |
8811 | | * [2] namespace URI |
8812 | | * [3] attribute value |
8813 | | * [4] end of attribute value |
8814 | | * |
8815 | | * To save memory, we reuse this array temporarily and store integers |
8816 | | * in these pointer variables. |
8817 | | * |
8818 | | * [0] attribute name |
8819 | | * [1] attribute prefix |
8820 | | * [2] hash value of attribute prefix, and later namespace index |
8821 | | * [3] for non-allocated values: ptrdiff_t offset into input buffer |
8822 | | * [4] for non-allocated values: ptrdiff_t offset into input buffer |
8823 | | * |
8824 | | * The ctxt->attallocs array contains an additional unsigned int for |
8825 | | * each attribute, containing the hash value of the attribute name |
8826 | | * and the alloc flag in bit 31. |
8827 | | */ |
8828 | | |
8829 | 796k | while (((RAW != '>') && |
8830 | 796k | ((RAW != '/') || (NXT(1) != '>')) && |
8831 | 796k | (IS_BYTE_CHAR(RAW))) && (PARSER_STOPPED(ctxt) == 0)) { |
8832 | 399k | int len = -1; |
8833 | | |
8834 | 399k | hattname = xmlParseAttribute2(ctxt, prefix, localname, |
8835 | 399k | &haprefix, &attvalue, &len, |
8836 | 399k | &alloc); |
8837 | 399k | if (hattname.name == NULL) |
8838 | 92.8k | break; |
8839 | 306k | if (attvalue == NULL) |
8840 | 42.9k | goto next_attr; |
8841 | 263k | attname = hattname.name; |
8842 | 263k | aprefix = haprefix.name; |
8843 | 263k | if (len < 0) len = xmlStrlen(attvalue); |
8844 | | |
8845 | 263k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
8846 | 55.3k | xmlHashedString huri; |
8847 | 55.3k | xmlURIPtr parsedUri; |
8848 | | |
8849 | 55.3k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
8850 | 55.3k | uri = huri.name; |
8851 | 55.3k | if (uri == NULL) { |
8852 | 7 | xmlErrMemory(ctxt); |
8853 | 7 | goto next_attr; |
8854 | 7 | } |
8855 | 55.3k | if (*uri != 0) { |
8856 | 54.0k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
8857 | 81 | xmlErrMemory(ctxt); |
8858 | 81 | goto next_attr; |
8859 | 81 | } |
8860 | 53.9k | if (parsedUri == NULL) { |
8861 | 40.0k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
8862 | 40.0k | "xmlns: '%s' is not a valid URI\n", |
8863 | 40.0k | uri, NULL, NULL); |
8864 | 40.0k | } else { |
8865 | 13.9k | if (parsedUri->scheme == NULL) { |
8866 | 9.59k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
8867 | 9.59k | "xmlns: URI %s is not absolute\n", |
8868 | 9.59k | uri, NULL, NULL); |
8869 | 9.59k | } |
8870 | 13.9k | xmlFreeURI(parsedUri); |
8871 | 13.9k | } |
8872 | 53.9k | if (uri == ctxt->str_xml_ns) { |
8873 | 200 | if (attname != ctxt->str_xml) { |
8874 | 200 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8875 | 200 | "xml namespace URI cannot be the default namespace\n", |
8876 | 200 | NULL, NULL, NULL); |
8877 | 200 | } |
8878 | 200 | goto next_attr; |
8879 | 200 | } |
8880 | 53.7k | if ((len == 29) && |
8881 | 53.7k | (xmlStrEqual(uri, |
8882 | 775 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
8883 | 406 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8884 | 406 | "reuse of the xmlns namespace name is forbidden\n", |
8885 | 406 | NULL, NULL, NULL); |
8886 | 406 | goto next_attr; |
8887 | 406 | } |
8888 | 53.7k | } |
8889 | | |
8890 | 54.6k | if (xmlParserNsPush(ctxt, NULL, &huri, NULL, 0) > 0) |
8891 | 44.3k | nbNs++; |
8892 | 208k | } else if (aprefix == ctxt->str_xmlns) { |
8893 | 19.2k | xmlHashedString huri; |
8894 | 19.2k | xmlURIPtr parsedUri; |
8895 | | |
8896 | 19.2k | huri = xmlDictLookupHashed(ctxt->dict, attvalue, len); |
8897 | 19.2k | uri = huri.name; |
8898 | 19.2k | if (uri == NULL) { |
8899 | 3 | xmlErrMemory(ctxt); |
8900 | 3 | goto next_attr; |
8901 | 3 | } |
8902 | | |
8903 | 19.2k | if (attname == ctxt->str_xml) { |
8904 | 198 | if (uri != ctxt->str_xml_ns) { |
8905 | 198 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8906 | 198 | "xml namespace prefix mapped to wrong URI\n", |
8907 | 198 | NULL, NULL, NULL); |
8908 | 198 | } |
8909 | | /* |
8910 | | * Do not keep a namespace definition node |
8911 | | */ |
8912 | 198 | goto next_attr; |
8913 | 198 | } |
8914 | 19.0k | if (uri == ctxt->str_xml_ns) { |
8915 | 194 | if (attname != ctxt->str_xml) { |
8916 | 194 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8917 | 194 | "xml namespace URI mapped to wrong prefix\n", |
8918 | 194 | NULL, NULL, NULL); |
8919 | 194 | } |
8920 | 194 | goto next_attr; |
8921 | 194 | } |
8922 | 18.8k | if (attname == ctxt->str_xmlns) { |
8923 | 195 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8924 | 195 | "redefinition of the xmlns prefix is forbidden\n", |
8925 | 195 | NULL, NULL, NULL); |
8926 | 195 | goto next_attr; |
8927 | 195 | } |
8928 | 18.6k | if ((len == 29) && |
8929 | 18.6k | (xmlStrEqual(uri, |
8930 | 951 | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
8931 | 358 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8932 | 358 | "reuse of the xmlns namespace name is forbidden\n", |
8933 | 358 | NULL, NULL, NULL); |
8934 | 358 | goto next_attr; |
8935 | 358 | } |
8936 | 18.3k | if ((uri == NULL) || (uri[0] == 0)) { |
8937 | 246 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
8938 | 246 | "xmlns:%s: Empty XML namespace is not allowed\n", |
8939 | 246 | attname, NULL, NULL); |
8940 | 246 | goto next_attr; |
8941 | 18.0k | } else { |
8942 | 18.0k | if (xmlParseURISafe((const char *) uri, &parsedUri) < 0) { |
8943 | 5 | xmlErrMemory(ctxt); |
8944 | 5 | goto next_attr; |
8945 | 5 | } |
8946 | 18.0k | if (parsedUri == NULL) { |
8947 | 6.68k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
8948 | 6.68k | "xmlns:%s: '%s' is not a valid URI\n", |
8949 | 6.68k | attname, uri, NULL); |
8950 | 11.3k | } else { |
8951 | 11.3k | if ((ctxt->pedantic) && (parsedUri->scheme == NULL)) { |
8952 | 1.54k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
8953 | 1.54k | "xmlns:%s: URI %s is not absolute\n", |
8954 | 1.54k | attname, uri, NULL); |
8955 | 1.54k | } |
8956 | 11.3k | xmlFreeURI(parsedUri); |
8957 | 11.3k | } |
8958 | 18.0k | } |
8959 | | |
8960 | 18.0k | if (xmlParserNsPush(ctxt, &hattname, &huri, NULL, 0) > 0) |
8961 | 16.2k | nbNs++; |
8962 | 189k | } else { |
8963 | | /* |
8964 | | * Populate attributes array, see above for repurposing |
8965 | | * of xmlChar pointers. |
8966 | | */ |
8967 | 189k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
8968 | 11.0k | int res = xmlCtxtGrowAttrs(ctxt); |
8969 | | |
8970 | 11.0k | maxatts = ctxt->maxatts; |
8971 | 11.0k | atts = ctxt->atts; |
8972 | | |
8973 | 11.0k | if (res < 0) |
8974 | 15 | goto next_attr; |
8975 | 11.0k | } |
8976 | 189k | ctxt->attallocs[nratts++] = (hattname.hashValue & 0x7FFFFFFF) | |
8977 | 189k | ((unsigned) alloc << 31); |
8978 | 189k | atts[nbatts++] = attname; |
8979 | 189k | atts[nbatts++] = aprefix; |
8980 | 189k | atts[nbatts++] = XML_INT_TO_PTR(haprefix.hashValue); |
8981 | 189k | if (alloc) { |
8982 | 34.3k | atts[nbatts++] = attvalue; |
8983 | 34.3k | attvalue += len; |
8984 | 34.3k | atts[nbatts++] = attvalue; |
8985 | 155k | } else { |
8986 | | /* |
8987 | | * attvalue points into the input buffer which can be |
8988 | | * reallocated. Store differences to input->base instead. |
8989 | | * The pointers will be reconstructed later. |
8990 | | */ |
8991 | 155k | atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR); |
8992 | 155k | attvalue += len; |
8993 | 155k | atts[nbatts++] = XML_INT_TO_PTR(attvalue - BASE_PTR); |
8994 | 155k | } |
8995 | | /* |
8996 | | * tag if some deallocation is needed |
8997 | | */ |
8998 | 189k | if (alloc != 0) attval = 1; |
8999 | 189k | attvalue = NULL; /* moved into atts */ |
9000 | 189k | } |
9001 | | |
9002 | 306k | next_attr: |
9003 | 306k | if ((attvalue != NULL) && (alloc != 0)) { |
9004 | 27.8k | xmlFree(attvalue); |
9005 | 27.8k | attvalue = NULL; |
9006 | 27.8k | } |
9007 | | |
9008 | 306k | GROW |
9009 | 306k | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9010 | 94.6k | break; |
9011 | 212k | if (SKIP_BLANKS == 0) { |
9012 | 89.1k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9013 | 89.1k | "attributes construct error\n"); |
9014 | 89.1k | break; |
9015 | 89.1k | } |
9016 | 123k | GROW; |
9017 | 123k | } |
9018 | | |
9019 | | /* |
9020 | | * Namespaces from default attributes |
9021 | | */ |
9022 | 673k | if (ctxt->attsDefault != NULL) { |
9023 | 135k | xmlDefAttrsPtr defaults; |
9024 | | |
9025 | 135k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9026 | 135k | if (defaults != NULL) { |
9027 | 576k | for (i = 0; i < defaults->nbAttrs; i++) { |
9028 | 466k | xmlDefAttr *attr = &defaults->attrs[i]; |
9029 | | |
9030 | 466k | attname = attr->name.name; |
9031 | 466k | aprefix = attr->prefix.name; |
9032 | | |
9033 | 466k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9034 | 8.43k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9035 | | |
9036 | 8.43k | if (xmlParserNsPush(ctxt, NULL, &attr->value, NULL, 1) > 0) |
9037 | 7.61k | nbNs++; |
9038 | 458k | } else if (aprefix == ctxt->str_xmlns) { |
9039 | 274k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9040 | | |
9041 | 274k | if (xmlParserNsPush(ctxt, &attr->name, &attr->value, |
9042 | 274k | NULL, 1) > 0) |
9043 | 262k | nbNs++; |
9044 | 274k | } else { |
9045 | 184k | if (nratts + nbTotalDef >= XML_MAX_ATTRS) { |
9046 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
9047 | 0 | "Maximum number of attributes exceeded"); |
9048 | 0 | break; |
9049 | 0 | } |
9050 | 184k | nbTotalDef += 1; |
9051 | 184k | } |
9052 | 466k | } |
9053 | 109k | } |
9054 | 135k | } |
9055 | | |
9056 | | /* |
9057 | | * Resolve attribute namespaces |
9058 | | */ |
9059 | 862k | for (i = 0; i < nbatts; i += 5) { |
9060 | 189k | attname = atts[i]; |
9061 | 189k | aprefix = atts[i+1]; |
9062 | | |
9063 | | /* |
9064 | | * The default namespace does not apply to attribute names. |
9065 | | */ |
9066 | 189k | if (aprefix == NULL) { |
9067 | 135k | nsIndex = NS_INDEX_EMPTY; |
9068 | 135k | } else if (aprefix == ctxt->str_xml) { |
9069 | 20.8k | nsIndex = NS_INDEX_XML; |
9070 | 32.8k | } else { |
9071 | 32.8k | haprefix.name = aprefix; |
9072 | 32.8k | haprefix.hashValue = (size_t) atts[i+2]; |
9073 | 32.8k | nsIndex = xmlParserNsLookup(ctxt, &haprefix, NULL); |
9074 | | |
9075 | 32.8k | if ((nsIndex == INT_MAX) || (nsIndex < ctxt->nsdb->minNsIndex)) { |
9076 | 17.4k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9077 | 17.4k | "Namespace prefix %s for %s on %s is not defined\n", |
9078 | 17.4k | aprefix, attname, localname); |
9079 | 17.4k | nsIndex = NS_INDEX_EMPTY; |
9080 | 17.4k | } |
9081 | 32.8k | } |
9082 | | |
9083 | 189k | atts[i+2] = XML_INT_TO_PTR(nsIndex); |
9084 | 189k | } |
9085 | | |
9086 | | /* |
9087 | | * Maximum number of attributes including default attributes. |
9088 | | */ |
9089 | 673k | maxAtts = nratts + nbTotalDef; |
9090 | | |
9091 | | /* |
9092 | | * Verify that attribute names are unique. |
9093 | | */ |
9094 | 673k | if (maxAtts > 1) { |
9095 | 47.6k | attrHashSize = 4; |
9096 | 63.7k | while (attrHashSize / 2 < (unsigned) maxAtts) |
9097 | 16.1k | attrHashSize *= 2; |
9098 | | |
9099 | 47.6k | if (attrHashSize > ctxt->attrHashMax) { |
9100 | 2.29k | xmlAttrHashBucket *tmp; |
9101 | | |
9102 | 2.29k | tmp = xmlRealloc(ctxt->attrHash, attrHashSize * sizeof(tmp[0])); |
9103 | 2.29k | if (tmp == NULL) { |
9104 | 3 | xmlErrMemory(ctxt); |
9105 | 3 | goto done; |
9106 | 3 | } |
9107 | | |
9108 | 2.29k | ctxt->attrHash = tmp; |
9109 | 2.29k | ctxt->attrHashMax = attrHashSize; |
9110 | 2.29k | } |
9111 | | |
9112 | 47.6k | memset(ctxt->attrHash, -1, attrHashSize * sizeof(ctxt->attrHash[0])); |
9113 | | |
9114 | 117k | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9115 | 70.2k | const xmlChar *nsuri; |
9116 | 70.2k | unsigned hashValue, nameHashValue, uriHashValue; |
9117 | 70.2k | int res; |
9118 | | |
9119 | 70.2k | attname = atts[i]; |
9120 | 70.2k | aprefix = atts[i+1]; |
9121 | 70.2k | nsIndex = XML_PTR_TO_INT(atts[i+2]); |
9122 | | /* Hash values always have bit 31 set, see dict.c */ |
9123 | 70.2k | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9124 | | |
9125 | 70.2k | if (nsIndex == NS_INDEX_EMPTY) { |
9126 | | /* |
9127 | | * Prefix with empty namespace means an undeclared |
9128 | | * prefix which was already reported above. |
9129 | | */ |
9130 | 53.6k | if (aprefix != NULL) |
9131 | 16.0k | continue; |
9132 | 37.6k | nsuri = NULL; |
9133 | 37.6k | uriHashValue = URI_HASH_EMPTY; |
9134 | 37.6k | } else if (nsIndex == NS_INDEX_XML) { |
9135 | 3.55k | nsuri = ctxt->str_xml_ns; |
9136 | 3.55k | uriHashValue = URI_HASH_XML; |
9137 | 12.9k | } else { |
9138 | 12.9k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9139 | 12.9k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9140 | 12.9k | } |
9141 | | |
9142 | 54.2k | hashValue = xmlDictCombineHash(nameHashValue, uriHashValue); |
9143 | 54.2k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9144 | 54.2k | hashValue, i); |
9145 | 54.2k | if (res < 0) |
9146 | 0 | continue; |
9147 | | |
9148 | | /* |
9149 | | * [ WFC: Unique Att Spec ] |
9150 | | * No attribute name may appear more than once in the same |
9151 | | * start-tag or empty-element tag. |
9152 | | * As extended by the Namespace in XML REC. |
9153 | | */ |
9154 | 54.2k | if (res < INT_MAX) { |
9155 | 17.3k | if (aprefix == atts[res+1]) { |
9156 | 15.3k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9157 | 15.3k | numDupErr += 1; |
9158 | 15.3k | } else { |
9159 | 2.08k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9160 | 2.08k | "Namespaced Attribute %s in '%s' redefined\n", |
9161 | 2.08k | attname, nsuri, NULL); |
9162 | 2.08k | numNsErr += 1; |
9163 | 2.08k | } |
9164 | 17.3k | } |
9165 | 54.2k | } |
9166 | 47.6k | } |
9167 | | |
9168 | | /* |
9169 | | * Default attributes |
9170 | | */ |
9171 | 673k | if (ctxt->attsDefault != NULL) { |
9172 | 135k | xmlDefAttrsPtr defaults; |
9173 | | |
9174 | 135k | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9175 | 135k | if (defaults != NULL) { |
9176 | 576k | for (i = 0; i < defaults->nbAttrs; i++) { |
9177 | 466k | xmlDefAttr *attr = &defaults->attrs[i]; |
9178 | 466k | const xmlChar *nsuri = NULL; |
9179 | 466k | unsigned hashValue, uriHashValue = 0; |
9180 | 466k | int res; |
9181 | | |
9182 | 466k | attname = attr->name.name; |
9183 | 466k | aprefix = attr->prefix.name; |
9184 | | |
9185 | 466k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) |
9186 | 8.43k | continue; |
9187 | 458k | if (aprefix == ctxt->str_xmlns) |
9188 | 274k | continue; |
9189 | | |
9190 | 184k | if (aprefix == NULL) { |
9191 | 111k | nsIndex = NS_INDEX_EMPTY; |
9192 | 111k | nsuri = NULL; |
9193 | 111k | uriHashValue = URI_HASH_EMPTY; |
9194 | 111k | } else if (aprefix == ctxt->str_xml) { |
9195 | 21.1k | nsIndex = NS_INDEX_XML; |
9196 | 21.1k | nsuri = ctxt->str_xml_ns; |
9197 | 21.1k | uriHashValue = URI_HASH_XML; |
9198 | 52.1k | } else { |
9199 | 52.1k | nsIndex = xmlParserNsLookup(ctxt, &attr->prefix, NULL); |
9200 | 52.1k | if ((nsIndex == INT_MAX) || |
9201 | 52.1k | (nsIndex < ctxt->nsdb->minNsIndex)) { |
9202 | 46.0k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9203 | 46.0k | "Namespace prefix %s for %s on %s is not " |
9204 | 46.0k | "defined\n", |
9205 | 46.0k | aprefix, attname, localname); |
9206 | 46.0k | nsIndex = NS_INDEX_EMPTY; |
9207 | 46.0k | nsuri = NULL; |
9208 | 46.0k | uriHashValue = URI_HASH_EMPTY; |
9209 | 46.0k | } else { |
9210 | 6.12k | nsuri = ctxt->nsTab[nsIndex * 2 + 1]; |
9211 | 6.12k | uriHashValue = ctxt->nsdb->extra[nsIndex].uriHashValue; |
9212 | 6.12k | } |
9213 | 52.1k | } |
9214 | | |
9215 | | /* |
9216 | | * Check whether the attribute exists |
9217 | | */ |
9218 | 184k | if (maxAtts > 1) { |
9219 | 145k | hashValue = xmlDictCombineHash(attr->name.hashValue, |
9220 | 145k | uriHashValue); |
9221 | 145k | res = xmlAttrHashInsert(ctxt, attrHashSize, attname, nsuri, |
9222 | 145k | hashValue, nbatts); |
9223 | 145k | if (res < 0) |
9224 | 0 | continue; |
9225 | 145k | if (res < INT_MAX) { |
9226 | 4.33k | if (aprefix == atts[res+1]) |
9227 | 3.15k | continue; |
9228 | 1.18k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9229 | 1.18k | "Namespaced Attribute %s in '%s' redefined\n", |
9230 | 1.18k | attname, nsuri, NULL); |
9231 | 1.18k | } |
9232 | 145k | } |
9233 | | |
9234 | 181k | xmlParserEntityCheck(ctxt, attr->expandedSize); |
9235 | | |
9236 | 181k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9237 | 2.58k | res = xmlCtxtGrowAttrs(ctxt); |
9238 | | |
9239 | 2.58k | maxatts = ctxt->maxatts; |
9240 | 2.58k | atts = ctxt->atts; |
9241 | | |
9242 | 2.58k | if (res < 0) { |
9243 | 6 | localname = NULL; |
9244 | 6 | goto done; |
9245 | 6 | } |
9246 | 2.58k | } |
9247 | | |
9248 | 181k | atts[nbatts++] = attname; |
9249 | 181k | atts[nbatts++] = aprefix; |
9250 | 181k | atts[nbatts++] = XML_INT_TO_PTR(nsIndex); |
9251 | 181k | atts[nbatts++] = attr->value.name; |
9252 | 181k | atts[nbatts++] = attr->valueEnd; |
9253 | | |
9254 | 181k | #ifdef LIBXML_VALID_ENABLED |
9255 | | /* |
9256 | | * This should be moved to valid.c, but we don't keep track |
9257 | | * whether an attribute was defaulted. |
9258 | | */ |
9259 | 181k | if ((ctxt->validate) && |
9260 | 181k | (ctxt->standalone == 1) && |
9261 | 181k | (attr->external != 0)) { |
9262 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9263 | 0 | "standalone: attribute %s on %s defaulted " |
9264 | 0 | "from external subset\n", |
9265 | 0 | attname, localname); |
9266 | 0 | } |
9267 | 181k | #endif |
9268 | 181k | nbdef++; |
9269 | 181k | } |
9270 | 109k | } |
9271 | 135k | } |
9272 | | |
9273 | | /* |
9274 | | * Using a single hash table for nsUri/localName pairs cannot |
9275 | | * detect duplicate QNames reliably. The following example will |
9276 | | * only result in two namespace errors. |
9277 | | * |
9278 | | * <doc xmlns:a="a" xmlns:b="a"> |
9279 | | * <elem a:a="" b:a="" b:a=""/> |
9280 | | * </doc> |
9281 | | * |
9282 | | * If we saw more than one namespace error but no duplicate QNames |
9283 | | * were found, we have to scan for duplicate QNames. |
9284 | | */ |
9285 | 673k | if ((numDupErr == 0) && (numNsErr > 1)) { |
9286 | 725 | memset(ctxt->attrHash, -1, |
9287 | 725 | attrHashSize * sizeof(ctxt->attrHash[0])); |
9288 | | |
9289 | 4.92k | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9290 | 4.19k | unsigned hashValue, nameHashValue, prefixHashValue; |
9291 | 4.19k | int res; |
9292 | | |
9293 | 4.19k | aprefix = atts[i+1]; |
9294 | 4.19k | if (aprefix == NULL) |
9295 | 284 | continue; |
9296 | | |
9297 | 3.91k | attname = atts[i]; |
9298 | | /* Hash values always have bit 31 set, see dict.c */ |
9299 | 3.91k | nameHashValue = ctxt->attallocs[j] | 0x80000000; |
9300 | 3.91k | prefixHashValue = xmlDictComputeHash(ctxt->dict, aprefix); |
9301 | | |
9302 | 3.91k | hashValue = xmlDictCombineHash(nameHashValue, prefixHashValue); |
9303 | 3.91k | res = xmlAttrHashInsertQName(ctxt, attrHashSize, attname, |
9304 | 3.91k | aprefix, hashValue, i); |
9305 | 3.91k | if (res < INT_MAX) |
9306 | 1.16k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9307 | 3.91k | } |
9308 | 725 | } |
9309 | | |
9310 | | /* |
9311 | | * Reconstruct attribute pointers |
9312 | | */ |
9313 | 1.04M | for (i = 0, j = 0; i < nbatts; i += 5, j++) { |
9314 | | /* namespace URI */ |
9315 | 370k | nsIndex = XML_PTR_TO_INT(atts[i+2]); |
9316 | 370k | if (nsIndex == INT_MAX) |
9317 | 307k | atts[i+2] = NULL; |
9318 | 63.0k | else if (nsIndex == INT_MAX - 1) |
9319 | 41.7k | atts[i+2] = ctxt->str_xml_ns; |
9320 | 21.3k | else |
9321 | 21.3k | atts[i+2] = ctxt->nsTab[nsIndex * 2 + 1]; |
9322 | | |
9323 | 370k | if ((j < nratts) && (ctxt->attallocs[j] & 0x80000000) == 0) { |
9324 | 155k | atts[i+3] = BASE_PTR + XML_PTR_TO_INT(atts[i+3]); /* value */ |
9325 | 155k | atts[i+4] = BASE_PTR + XML_PTR_TO_INT(atts[i+4]); /* valuend */ |
9326 | 155k | } |
9327 | 370k | } |
9328 | | |
9329 | 673k | uri = xmlParserNsLookupUri(ctxt, &hprefix); |
9330 | 673k | if ((prefix != NULL) && (uri == NULL)) { |
9331 | 39.6k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9332 | 39.6k | "Namespace prefix %s on %s is not defined\n", |
9333 | 39.6k | prefix, localname, NULL); |
9334 | 39.6k | } |
9335 | 673k | *pref = prefix; |
9336 | 673k | *URI = uri; |
9337 | | |
9338 | | /* |
9339 | | * SAX callback |
9340 | | */ |
9341 | 673k | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9342 | 673k | (!ctxt->disableSAX)) { |
9343 | 601k | if (nbNs > 0) |
9344 | 85.8k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9345 | 85.8k | nbNs, ctxt->nsTab + 2 * (ctxt->nsNr - nbNs), |
9346 | 85.8k | nbatts / 5, nbdef, atts); |
9347 | 515k | else |
9348 | 515k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, uri, |
9349 | 515k | 0, NULL, nbatts / 5, nbdef, atts); |
9350 | 601k | } |
9351 | | |
9352 | 673k | done: |
9353 | | /* |
9354 | | * Free allocated attribute values |
9355 | | */ |
9356 | 673k | if (attval != 0) { |
9357 | 77.9k | for (i = 0, j = 0; j < nratts; i += 5, j++) |
9358 | 46.2k | if (ctxt->attallocs[j] & 0x80000000) |
9359 | 34.3k | xmlFree((xmlChar *) atts[i+3]); |
9360 | 31.7k | } |
9361 | | |
9362 | 673k | *nbNsPtr = nbNs; |
9363 | 673k | return(localname); |
9364 | 673k | } |
9365 | | |
9366 | | /** |
9367 | | * Parse an end tag. Always consumes '</'. |
9368 | | * |
9369 | | * [42] ETag ::= '</' Name S? '>' |
9370 | | * |
9371 | | * With namespace |
9372 | | * |
9373 | | * [NS 9] ETag ::= '</' QName S? '>' |
9374 | | * @param ctxt an XML parser context |
9375 | | * @param tag the corresponding start tag |
9376 | | */ |
9377 | | |
9378 | | static void |
9379 | 101k | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9380 | 101k | const xmlChar *name; |
9381 | | |
9382 | 101k | GROW; |
9383 | 101k | if ((RAW != '<') || (NXT(1) != '/')) { |
9384 | 1.13k | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9385 | 1.13k | return; |
9386 | 1.13k | } |
9387 | 99.9k | SKIP(2); |
9388 | | |
9389 | 99.9k | if (tag->prefix == NULL) |
9390 | 68.3k | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9391 | 31.5k | else |
9392 | 31.5k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9393 | | |
9394 | | /* |
9395 | | * We should definitely be at the ending "S? '>'" part |
9396 | | */ |
9397 | 99.9k | GROW; |
9398 | 99.9k | SKIP_BLANKS; |
9399 | 99.9k | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9400 | 46.4k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9401 | 46.4k | } else |
9402 | 53.5k | NEXT1; |
9403 | | |
9404 | | /* |
9405 | | * [ WFC: Element Type Match ] |
9406 | | * The Name in an element's end-tag must match the element type in the |
9407 | | * start-tag. |
9408 | | * |
9409 | | */ |
9410 | 99.9k | if (name != (xmlChar*)1) { |
9411 | 57.2k | if (name == NULL) name = BAD_CAST "unparsable"; |
9412 | 57.2k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9413 | 57.2k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9414 | 57.2k | ctxt->name, tag->line, name); |
9415 | 57.2k | } |
9416 | | |
9417 | | /* |
9418 | | * SAX: End of Tag |
9419 | | */ |
9420 | 99.9k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9421 | 99.9k | (!ctxt->disableSAX)) |
9422 | 92.6k | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9423 | 92.6k | tag->URI); |
9424 | | |
9425 | 99.9k | spacePop(ctxt); |
9426 | 99.9k | if (tag->nsNr != 0) |
9427 | 4.48k | xmlParserNsPop(ctxt, tag->nsNr); |
9428 | 99.9k | } |
9429 | | |
9430 | | /** |
9431 | | * Parse escaped pure raw content. Always consumes '<!['. |
9432 | | * |
9433 | | * @deprecated Internal function, don't use. |
9434 | | * |
9435 | | * [18] CDSect ::= CDStart CData CDEnd |
9436 | | * |
9437 | | * [19] CDStart ::= '<![CDATA[' |
9438 | | * |
9439 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9440 | | * |
9441 | | * [21] CDEnd ::= ']]>' |
9442 | | * @param ctxt an XML parser context |
9443 | | */ |
9444 | | void |
9445 | 26.2k | xmlParseCDSect(xmlParserCtxt *ctxt) { |
9446 | 26.2k | xmlChar *buf = NULL; |
9447 | 26.2k | int len = 0; |
9448 | 26.2k | int size = XML_PARSER_BUFFER_SIZE; |
9449 | 26.2k | int r, rl; |
9450 | 26.2k | int s, sl; |
9451 | 26.2k | int cur, l; |
9452 | 26.2k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9453 | 11.8k | XML_MAX_HUGE_LENGTH : |
9454 | 26.2k | XML_MAX_TEXT_LENGTH; |
9455 | | |
9456 | 26.2k | if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '[')) |
9457 | 0 | return; |
9458 | 26.2k | SKIP(3); |
9459 | | |
9460 | 26.2k | if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '[')) |
9461 | 0 | return; |
9462 | 26.2k | SKIP(6); |
9463 | | |
9464 | 26.2k | r = xmlCurrentCharRecover(ctxt, &rl); |
9465 | 26.2k | if (!IS_CHAR(r)) { |
9466 | 575 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9467 | 575 | goto out; |
9468 | 575 | } |
9469 | 25.6k | NEXTL(rl); |
9470 | 25.6k | s = xmlCurrentCharRecover(ctxt, &sl); |
9471 | 25.6k | if (!IS_CHAR(s)) { |
9472 | 644 | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9473 | 644 | goto out; |
9474 | 644 | } |
9475 | 25.0k | NEXTL(sl); |
9476 | 25.0k | cur = xmlCurrentCharRecover(ctxt, &l); |
9477 | 25.0k | buf = xmlMalloc(size); |
9478 | 25.0k | if (buf == NULL) { |
9479 | 4 | xmlErrMemory(ctxt); |
9480 | 4 | goto out; |
9481 | 4 | } |
9482 | 4.74M | while (IS_CHAR(cur) && |
9483 | 4.74M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9484 | 4.72M | if (len + 5 >= size) { |
9485 | 5.24k | xmlChar *tmp; |
9486 | 5.24k | int newSize; |
9487 | | |
9488 | 5.24k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
9489 | 5.24k | if (newSize < 0) { |
9490 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9491 | 0 | "CData section too big found\n"); |
9492 | 0 | goto out; |
9493 | 0 | } |
9494 | 5.24k | tmp = xmlRealloc(buf, newSize); |
9495 | 5.24k | if (tmp == NULL) { |
9496 | 8 | xmlErrMemory(ctxt); |
9497 | 8 | goto out; |
9498 | 8 | } |
9499 | 5.23k | buf = tmp; |
9500 | 5.23k | size = newSize; |
9501 | 5.23k | } |
9502 | 4.72M | COPY_BUF(buf, len, r); |
9503 | 4.72M | r = s; |
9504 | 4.72M | rl = sl; |
9505 | 4.72M | s = cur; |
9506 | 4.72M | sl = l; |
9507 | 4.72M | NEXTL(l); |
9508 | 4.72M | cur = xmlCurrentCharRecover(ctxt, &l); |
9509 | 4.72M | } |
9510 | 25.0k | buf[len] = 0; |
9511 | 25.0k | if (cur != '>') { |
9512 | 5.56k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9513 | 5.56k | "CData section not finished\n%.50s\n", buf); |
9514 | 5.56k | goto out; |
9515 | 5.56k | } |
9516 | 19.4k | NEXTL(l); |
9517 | | |
9518 | | /* |
9519 | | * OK the buffer is to be consumed as cdata. |
9520 | | */ |
9521 | 19.4k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9522 | 16.6k | if ((ctxt->sax->cdataBlock != NULL) && |
9523 | 16.6k | ((ctxt->options & XML_PARSE_NOCDATA) == 0)) { |
9524 | 14.3k | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9525 | 14.3k | } else if (ctxt->sax->characters != NULL) { |
9526 | 2.29k | ctxt->sax->characters(ctxt->userData, buf, len); |
9527 | 2.29k | } |
9528 | 16.6k | } |
9529 | | |
9530 | 26.2k | out: |
9531 | 26.2k | xmlFree(buf); |
9532 | 26.2k | } |
9533 | | |
9534 | | /** |
9535 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9536 | | * unexpected EOF to the caller. |
9537 | | * |
9538 | | * @param ctxt an XML parser context |
9539 | | */ |
9540 | | |
9541 | | static void |
9542 | 82.7k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9543 | 82.7k | int oldNameNr = ctxt->nameNr; |
9544 | 82.7k | int oldSpaceNr = ctxt->spaceNr; |
9545 | 82.7k | int oldNodeNr = ctxt->nodeNr; |
9546 | | |
9547 | 82.7k | GROW; |
9548 | 4.61M | while ((ctxt->input->cur < ctxt->input->end) && |
9549 | 4.61M | (PARSER_STOPPED(ctxt) == 0)) { |
9550 | 4.53M | const xmlChar *cur = ctxt->input->cur; |
9551 | | |
9552 | | /* |
9553 | | * First case : a Processing Instruction. |
9554 | | */ |
9555 | 4.53M | if ((*cur == '<') && (cur[1] == '?')) { |
9556 | 25.3k | xmlParsePI(ctxt); |
9557 | 25.3k | } |
9558 | | |
9559 | | /* |
9560 | | * Second case : a CDSection |
9561 | | */ |
9562 | | /* 2.6.0 test was *cur not RAW */ |
9563 | 4.50M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9564 | 24.1k | xmlParseCDSect(ctxt); |
9565 | 24.1k | } |
9566 | | |
9567 | | /* |
9568 | | * Third case : a comment |
9569 | | */ |
9570 | 4.48M | else if ((*cur == '<') && (NXT(1) == '!') && |
9571 | 4.48M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9572 | 38.0k | xmlParseComment(ctxt); |
9573 | 38.0k | } |
9574 | | |
9575 | | /* |
9576 | | * Fourth case : a sub-element. |
9577 | | */ |
9578 | 4.44M | else if (*cur == '<') { |
9579 | 985k | if (NXT(1) == '/') { |
9580 | 93.7k | if (ctxt->nameNr <= oldNameNr) |
9581 | 386 | break; |
9582 | 93.3k | xmlParseElementEnd(ctxt); |
9583 | 891k | } else { |
9584 | 891k | xmlParseElementStart(ctxt); |
9585 | 891k | } |
9586 | 985k | } |
9587 | | |
9588 | | /* |
9589 | | * Fifth case : a reference. If if has not been resolved, |
9590 | | * parsing returns it's Name, create the node |
9591 | | */ |
9592 | | |
9593 | 3.45M | else if (*cur == '&') { |
9594 | 225k | xmlParseReference(ctxt); |
9595 | 225k | } |
9596 | | |
9597 | | /* |
9598 | | * Last case, text. Note that References are handled directly. |
9599 | | */ |
9600 | 3.23M | else { |
9601 | 3.23M | xmlParseCharDataInternal(ctxt, 0); |
9602 | 3.23M | } |
9603 | | |
9604 | 4.53M | SHRINK; |
9605 | 4.53M | GROW; |
9606 | 4.53M | } |
9607 | | |
9608 | 82.7k | if ((ctxt->nameNr > oldNameNr) && |
9609 | 82.7k | (ctxt->input->cur >= ctxt->input->end) && |
9610 | 82.7k | (ctxt->wellFormed)) { |
9611 | 123 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9612 | 123 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9613 | 123 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9614 | 123 | "Premature end of data in tag %s line %d\n", |
9615 | 123 | name, line, NULL); |
9616 | 123 | } |
9617 | | |
9618 | | /* |
9619 | | * Clean up in error case |
9620 | | */ |
9621 | | |
9622 | 265k | while (ctxt->nodeNr > oldNodeNr) |
9623 | 182k | nodePop(ctxt); |
9624 | | |
9625 | 304k | while (ctxt->nameNr > oldNameNr) { |
9626 | 221k | xmlStartTag *tag = &ctxt->pushTab[ctxt->nameNr - 1]; |
9627 | | |
9628 | 221k | if (tag->nsNr != 0) |
9629 | 60.2k | xmlParserNsPop(ctxt, tag->nsNr); |
9630 | | |
9631 | 221k | namePop(ctxt); |
9632 | 221k | } |
9633 | | |
9634 | 304k | while (ctxt->spaceNr > oldSpaceNr) |
9635 | 222k | spacePop(ctxt); |
9636 | 82.7k | } |
9637 | | |
9638 | | /** |
9639 | | * Parse XML element content. This is useful if you're only interested |
9640 | | * in custom SAX callbacks. If you want a node list, use |
9641 | | * #xmlCtxtParseContent. |
9642 | | * |
9643 | | * @param ctxt an XML parser context |
9644 | | */ |
9645 | | void |
9646 | 0 | xmlParseContent(xmlParserCtxt *ctxt) { |
9647 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
9648 | 0 | return; |
9649 | | |
9650 | 0 | xmlCtxtInitializeLate(ctxt); |
9651 | |
|
9652 | 0 | xmlParseContentInternal(ctxt); |
9653 | |
|
9654 | 0 | xmlParserCheckEOF(ctxt, XML_ERR_NOT_WELL_BALANCED); |
9655 | 0 | } |
9656 | | |
9657 | | /** |
9658 | | * Parse an XML element |
9659 | | * |
9660 | | * @deprecated Internal function, don't use. |
9661 | | * |
9662 | | * [39] element ::= EmptyElemTag | STag content ETag |
9663 | | * |
9664 | | * [ WFC: Element Type Match ] |
9665 | | * The Name in an element's end-tag must match the element type in the |
9666 | | * start-tag. |
9667 | | * |
9668 | | * @param ctxt an XML parser context |
9669 | | */ |
9670 | | |
9671 | | void |
9672 | 15.3k | xmlParseElement(xmlParserCtxt *ctxt) { |
9673 | 15.3k | if (xmlParseElementStart(ctxt) != 0) |
9674 | 5.71k | return; |
9675 | | |
9676 | 9.67k | xmlParseContentInternal(ctxt); |
9677 | | |
9678 | 9.67k | if (ctxt->input->cur >= ctxt->input->end) { |
9679 | 8.11k | if (ctxt->wellFormed) { |
9680 | 349 | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
9681 | 349 | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
9682 | 349 | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
9683 | 349 | "Premature end of data in tag %s line %d\n", |
9684 | 349 | name, line, NULL); |
9685 | 349 | } |
9686 | 8.11k | return; |
9687 | 8.11k | } |
9688 | | |
9689 | 1.56k | xmlParseElementEnd(ctxt); |
9690 | 1.56k | } |
9691 | | |
9692 | | /** |
9693 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
9694 | | * opening tag was parsed, 1 if an empty element was parsed. |
9695 | | * |
9696 | | * Always consumes '<'. |
9697 | | * |
9698 | | * @param ctxt an XML parser context |
9699 | | */ |
9700 | | static int |
9701 | 906k | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
9702 | 906k | int maxDepth = (ctxt->options & XML_PARSE_HUGE) ? 2048 : 256; |
9703 | 906k | const xmlChar *name; |
9704 | 906k | const xmlChar *prefix = NULL; |
9705 | 906k | const xmlChar *URI = NULL; |
9706 | 906k | xmlParserNodeInfo node_info; |
9707 | 906k | int line; |
9708 | 906k | xmlNodePtr cur; |
9709 | 906k | int nbNs = 0; |
9710 | | |
9711 | 906k | if (ctxt->nameNr > maxDepth) { |
9712 | 12 | xmlFatalErrMsgInt(ctxt, XML_ERR_RESOURCE_LIMIT, |
9713 | 12 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
9714 | 12 | ctxt->nameNr); |
9715 | 12 | return(-1); |
9716 | 12 | } |
9717 | | |
9718 | | /* Capture start position */ |
9719 | 906k | if (ctxt->record_info) { |
9720 | 0 | node_info.begin_pos = ctxt->input->consumed + |
9721 | 0 | (CUR_PTR - ctxt->input->base); |
9722 | 0 | node_info.begin_line = ctxt->input->line; |
9723 | 0 | } |
9724 | | |
9725 | 906k | if (ctxt->spaceNr == 0) |
9726 | 15.3k | spacePush(ctxt, -1); |
9727 | 891k | else if (*ctxt->space == -2) |
9728 | 187k | spacePush(ctxt, -1); |
9729 | 703k | else |
9730 | 703k | spacePush(ctxt, *ctxt->space); |
9731 | | |
9732 | 906k | line = ctxt->input->line; |
9733 | 906k | #ifdef LIBXML_SAX1_ENABLED |
9734 | 906k | if (ctxt->sax2) |
9735 | 906k | #endif /* LIBXML_SAX1_ENABLED */ |
9736 | 906k | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
9737 | 0 | #ifdef LIBXML_SAX1_ENABLED |
9738 | 0 | else |
9739 | 0 | name = xmlParseStartTag(ctxt); |
9740 | 906k | #endif /* LIBXML_SAX1_ENABLED */ |
9741 | 906k | if (name == NULL) { |
9742 | 346k | spacePop(ctxt); |
9743 | 346k | return(-1); |
9744 | 346k | } |
9745 | 559k | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
9746 | 559k | cur = ctxt->node; |
9747 | | |
9748 | 559k | #ifdef LIBXML_VALID_ENABLED |
9749 | | /* |
9750 | | * [ VC: Root Element Type ] |
9751 | | * The Name in the document type declaration must match the element |
9752 | | * type of the root element. |
9753 | | */ |
9754 | 559k | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
9755 | 559k | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
9756 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
9757 | 559k | #endif /* LIBXML_VALID_ENABLED */ |
9758 | | |
9759 | | /* |
9760 | | * Check for an Empty Element. |
9761 | | */ |
9762 | 559k | if ((RAW == '/') && (NXT(1) == '>')) { |
9763 | 68.2k | SKIP(2); |
9764 | 68.2k | if (ctxt->sax2) { |
9765 | 68.2k | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9766 | 68.2k | (!ctxt->disableSAX)) |
9767 | 60.2k | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
9768 | 68.2k | #ifdef LIBXML_SAX1_ENABLED |
9769 | 68.2k | } else { |
9770 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
9771 | 0 | (!ctxt->disableSAX)) |
9772 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
9773 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
9774 | 0 | } |
9775 | 68.2k | namePop(ctxt); |
9776 | 68.2k | spacePop(ctxt); |
9777 | 68.2k | if (nbNs > 0) |
9778 | 3.59k | xmlParserNsPop(ctxt, nbNs); |
9779 | 68.2k | if (cur != NULL && ctxt->record_info) { |
9780 | 0 | node_info.node = cur; |
9781 | 0 | node_info.end_pos = ctxt->input->consumed + |
9782 | 0 | (CUR_PTR - ctxt->input->base); |
9783 | 0 | node_info.end_line = ctxt->input->line; |
9784 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9785 | 0 | } |
9786 | 68.2k | return(1); |
9787 | 68.2k | } |
9788 | 491k | if (RAW == '>') { |
9789 | 325k | NEXT1; |
9790 | 325k | if (cur != NULL && ctxt->record_info) { |
9791 | 0 | node_info.node = cur; |
9792 | 0 | node_info.end_pos = 0; |
9793 | 0 | node_info.end_line = 0; |
9794 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
9795 | 0 | } |
9796 | 325k | } else { |
9797 | 166k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
9798 | 166k | "Couldn't find end of Start Tag %s line %d\n", |
9799 | 166k | name, line, NULL); |
9800 | | |
9801 | | /* |
9802 | | * end of parsing of this node. |
9803 | | */ |
9804 | 166k | nodePop(ctxt); |
9805 | 166k | namePop(ctxt); |
9806 | 166k | spacePop(ctxt); |
9807 | 166k | if (nbNs > 0) |
9808 | 20.7k | xmlParserNsPop(ctxt, nbNs); |
9809 | 166k | return(-1); |
9810 | 166k | } |
9811 | | |
9812 | 325k | return(0); |
9813 | 491k | } |
9814 | | |
9815 | | /** |
9816 | | * Parse the end of an XML element. Always consumes '</'. |
9817 | | * |
9818 | | * @param ctxt an XML parser context |
9819 | | */ |
9820 | | static void |
9821 | 94.9k | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
9822 | 94.9k | xmlNodePtr cur = ctxt->node; |
9823 | | |
9824 | 94.9k | if (ctxt->nameNr <= 0) { |
9825 | 18 | if ((RAW == '<') && (NXT(1) == '/')) |
9826 | 2 | SKIP(2); |
9827 | 18 | return; |
9828 | 18 | } |
9829 | | |
9830 | | /* |
9831 | | * parse the end of tag: '</' should be here. |
9832 | | */ |
9833 | 94.8k | if (ctxt->sax2) { |
9834 | 94.8k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
9835 | 94.8k | namePop(ctxt); |
9836 | 94.8k | } |
9837 | 0 | #ifdef LIBXML_SAX1_ENABLED |
9838 | 0 | else |
9839 | 0 | xmlParseEndTag1(ctxt, 0); |
9840 | 94.8k | #endif /* LIBXML_SAX1_ENABLED */ |
9841 | | |
9842 | | /* |
9843 | | * Capture end position |
9844 | | */ |
9845 | 94.8k | if (cur != NULL && ctxt->record_info) { |
9846 | 0 | xmlParserNodeInfoPtr node_info; |
9847 | |
|
9848 | 0 | node_info = (xmlParserNodeInfoPtr) xmlParserFindNodeInfo(ctxt, cur); |
9849 | 0 | if (node_info != NULL) { |
9850 | 0 | node_info->end_pos = ctxt->input->consumed + |
9851 | 0 | (CUR_PTR - ctxt->input->base); |
9852 | 0 | node_info->end_line = ctxt->input->line; |
9853 | 0 | } |
9854 | 0 | } |
9855 | 94.8k | } |
9856 | | |
9857 | | /** |
9858 | | * Parse the XML version value. |
9859 | | * |
9860 | | * @deprecated Internal function, don't use. |
9861 | | * |
9862 | | * [26] VersionNum ::= '1.' [0-9]+ |
9863 | | * |
9864 | | * In practice allow [0-9].[0-9]+ at that level |
9865 | | * |
9866 | | * @param ctxt an XML parser context |
9867 | | * @returns the string giving the XML version number, or NULL |
9868 | | */ |
9869 | | xmlChar * |
9870 | 9.31k | xmlParseVersionNum(xmlParserCtxt *ctxt) { |
9871 | 9.31k | xmlChar *buf = NULL; |
9872 | 9.31k | int len = 0; |
9873 | 9.31k | int size = 10; |
9874 | 9.31k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9875 | 4.07k | XML_MAX_TEXT_LENGTH : |
9876 | 9.31k | XML_MAX_NAME_LENGTH; |
9877 | 9.31k | xmlChar cur; |
9878 | | |
9879 | 9.31k | buf = xmlMalloc(size); |
9880 | 9.31k | if (buf == NULL) { |
9881 | 28 | xmlErrMemory(ctxt); |
9882 | 28 | return(NULL); |
9883 | 28 | } |
9884 | 9.28k | cur = CUR; |
9885 | 9.28k | if (!((cur >= '0') && (cur <= '9'))) { |
9886 | 917 | xmlFree(buf); |
9887 | 917 | return(NULL); |
9888 | 917 | } |
9889 | 8.37k | buf[len++] = cur; |
9890 | 8.37k | NEXT; |
9891 | 8.37k | cur=CUR; |
9892 | 8.37k | if (cur != '.') { |
9893 | 448 | xmlFree(buf); |
9894 | 448 | return(NULL); |
9895 | 448 | } |
9896 | 7.92k | buf[len++] = cur; |
9897 | 7.92k | NEXT; |
9898 | 7.92k | cur=CUR; |
9899 | 281k | while ((cur >= '0') && (cur <= '9')) { |
9900 | 273k | if (len + 1 >= size) { |
9901 | 475 | xmlChar *tmp; |
9902 | 475 | int newSize; |
9903 | | |
9904 | 475 | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
9905 | 475 | if (newSize < 0) { |
9906 | 3 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "VersionNum"); |
9907 | 3 | xmlFree(buf); |
9908 | 3 | return(NULL); |
9909 | 3 | } |
9910 | 472 | tmp = xmlRealloc(buf, newSize); |
9911 | 472 | if (tmp == NULL) { |
9912 | 4 | xmlErrMemory(ctxt); |
9913 | 4 | xmlFree(buf); |
9914 | 4 | return(NULL); |
9915 | 4 | } |
9916 | 468 | buf = tmp; |
9917 | 468 | size = newSize; |
9918 | 468 | } |
9919 | 273k | buf[len++] = cur; |
9920 | 273k | NEXT; |
9921 | 273k | cur=CUR; |
9922 | 273k | } |
9923 | 7.91k | buf[len] = 0; |
9924 | 7.91k | return(buf); |
9925 | 7.92k | } |
9926 | | |
9927 | | /** |
9928 | | * Parse the XML version. |
9929 | | * |
9930 | | * @deprecated Internal function, don't use. |
9931 | | * |
9932 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
9933 | | * |
9934 | | * [25] Eq ::= S? '=' S? |
9935 | | * |
9936 | | * @param ctxt an XML parser context |
9937 | | * @returns the version string, e.g. "1.0" |
9938 | | */ |
9939 | | |
9940 | | xmlChar * |
9941 | 31.0k | xmlParseVersionInfo(xmlParserCtxt *ctxt) { |
9942 | 31.0k | xmlChar *version = NULL; |
9943 | | |
9944 | 31.0k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
9945 | 10.4k | SKIP(7); |
9946 | 10.4k | SKIP_BLANKS; |
9947 | 10.4k | if (RAW != '=') { |
9948 | 674 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
9949 | 674 | return(NULL); |
9950 | 674 | } |
9951 | 9.78k | NEXT; |
9952 | 9.78k | SKIP_BLANKS; |
9953 | 9.78k | if (RAW == '"') { |
9954 | 5.44k | NEXT; |
9955 | 5.44k | version = xmlParseVersionNum(ctxt); |
9956 | 5.44k | if (RAW != '"') { |
9957 | 1.08k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
9958 | 1.08k | } else |
9959 | 4.36k | NEXT; |
9960 | 5.44k | } else if (RAW == '\''){ |
9961 | 3.86k | NEXT; |
9962 | 3.86k | version = xmlParseVersionNum(ctxt); |
9963 | 3.86k | if (RAW != '\'') { |
9964 | 829 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
9965 | 829 | } else |
9966 | 3.03k | NEXT; |
9967 | 3.86k | } else { |
9968 | 468 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
9969 | 468 | } |
9970 | 9.78k | } |
9971 | 30.3k | return(version); |
9972 | 31.0k | } |
9973 | | |
9974 | | /** |
9975 | | * Parse the XML encoding name |
9976 | | * |
9977 | | * @deprecated Internal function, don't use. |
9978 | | * |
9979 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
9980 | | * |
9981 | | * @param ctxt an XML parser context |
9982 | | * @returns the encoding name value or NULL |
9983 | | */ |
9984 | | xmlChar * |
9985 | 10.0k | xmlParseEncName(xmlParserCtxt *ctxt) { |
9986 | 10.0k | xmlChar *buf = NULL; |
9987 | 10.0k | int len = 0; |
9988 | 10.0k | int size = 10; |
9989 | 10.0k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9990 | 1.99k | XML_MAX_TEXT_LENGTH : |
9991 | 10.0k | XML_MAX_NAME_LENGTH; |
9992 | 10.0k | xmlChar cur; |
9993 | | |
9994 | 10.0k | cur = CUR; |
9995 | 10.0k | if (((cur >= 'a') && (cur <= 'z')) || |
9996 | 10.0k | ((cur >= 'A') && (cur <= 'Z'))) { |
9997 | 9.65k | buf = xmlMalloc(size); |
9998 | 9.65k | if (buf == NULL) { |
9999 | 30 | xmlErrMemory(ctxt); |
10000 | 30 | return(NULL); |
10001 | 30 | } |
10002 | | |
10003 | 9.62k | buf[len++] = cur; |
10004 | 9.62k | NEXT; |
10005 | 9.62k | cur = CUR; |
10006 | 377k | while (((cur >= 'a') && (cur <= 'z')) || |
10007 | 377k | ((cur >= 'A') && (cur <= 'Z')) || |
10008 | 377k | ((cur >= '0') && (cur <= '9')) || |
10009 | 377k | (cur == '.') || (cur == '_') || |
10010 | 377k | (cur == '-')) { |
10011 | 367k | if (len + 1 >= size) { |
10012 | 5.69k | xmlChar *tmp; |
10013 | 5.69k | int newSize; |
10014 | | |
10015 | 5.69k | newSize = xmlGrowCapacity(size, 1, 1, maxLength); |
10016 | 5.69k | if (newSize < 0) { |
10017 | 3 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "EncName"); |
10018 | 3 | xmlFree(buf); |
10019 | 3 | return(NULL); |
10020 | 3 | } |
10021 | 5.69k | tmp = xmlRealloc(buf, newSize); |
10022 | 5.69k | if (tmp == NULL) { |
10023 | 6 | xmlErrMemory(ctxt); |
10024 | 6 | xmlFree(buf); |
10025 | 6 | return(NULL); |
10026 | 6 | } |
10027 | 5.68k | buf = tmp; |
10028 | 5.68k | size = newSize; |
10029 | 5.68k | } |
10030 | 367k | buf[len++] = cur; |
10031 | 367k | NEXT; |
10032 | 367k | cur = CUR; |
10033 | 367k | } |
10034 | 9.61k | buf[len] = 0; |
10035 | 9.61k | } else { |
10036 | 414 | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10037 | 414 | } |
10038 | 10.0k | return(buf); |
10039 | 10.0k | } |
10040 | | |
10041 | | /** |
10042 | | * Parse the XML encoding declaration |
10043 | | * |
10044 | | * @deprecated Internal function, don't use. |
10045 | | * |
10046 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | |
10047 | | * "'" EncName "'") |
10048 | | * |
10049 | | * this setups the conversion filters. |
10050 | | * |
10051 | | * @param ctxt an XML parser context |
10052 | | * @returns the encoding value or NULL |
10053 | | */ |
10054 | | |
10055 | | const xmlChar * |
10056 | 30.1k | xmlParseEncodingDecl(xmlParserCtxt *ctxt) { |
10057 | 30.1k | xmlChar *encoding = NULL; |
10058 | | |
10059 | 30.1k | SKIP_BLANKS; |
10060 | 30.1k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g') == 0) |
10061 | 19.2k | return(NULL); |
10062 | | |
10063 | 10.9k | SKIP(8); |
10064 | 10.9k | SKIP_BLANKS; |
10065 | 10.9k | if (RAW != '=') { |
10066 | 403 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10067 | 403 | return(NULL); |
10068 | 403 | } |
10069 | 10.5k | NEXT; |
10070 | 10.5k | SKIP_BLANKS; |
10071 | 10.5k | if (RAW == '"') { |
10072 | 5.61k | NEXT; |
10073 | 5.61k | encoding = xmlParseEncName(ctxt); |
10074 | 5.61k | if (RAW != '"') { |
10075 | 531 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10076 | 531 | xmlFree(encoding); |
10077 | 531 | return(NULL); |
10078 | 531 | } else |
10079 | 5.08k | NEXT; |
10080 | 5.61k | } else if (RAW == '\''){ |
10081 | 4.46k | NEXT; |
10082 | 4.46k | encoding = xmlParseEncName(ctxt); |
10083 | 4.46k | if (RAW != '\'') { |
10084 | 790 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10085 | 790 | xmlFree(encoding); |
10086 | 790 | return(NULL); |
10087 | 790 | } else |
10088 | 3.67k | NEXT; |
10089 | 4.46k | } else { |
10090 | 432 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10091 | 432 | } |
10092 | | |
10093 | 9.18k | if (encoding == NULL) |
10094 | 435 | return(NULL); |
10095 | | |
10096 | 8.74k | xmlSetDeclaredEncoding(ctxt, encoding); |
10097 | | |
10098 | 8.74k | return(ctxt->encoding); |
10099 | 9.18k | } |
10100 | | |
10101 | | /** |
10102 | | * Parse the XML standalone declaration |
10103 | | * |
10104 | | * @deprecated Internal function, don't use. |
10105 | | * |
10106 | | * [32] SDDecl ::= S 'standalone' Eq |
10107 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10108 | | * |
10109 | | * [ VC: Standalone Document Declaration ] |
10110 | | * TODO The standalone document declaration must have the value "no" |
10111 | | * if any external markup declarations contain declarations of: |
10112 | | * - attributes with default values, if elements to which these |
10113 | | * attributes apply appear in the document without specifications |
10114 | | * of values for these attributes, or |
10115 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10116 | | * to those entities appear in the document, or |
10117 | | * - attributes with values subject to normalization, where the |
10118 | | * attribute appears in the document with a value which will change |
10119 | | * as a result of normalization, or |
10120 | | * - element types with element content, if white space occurs directly |
10121 | | * within any instance of those types. |
10122 | | * |
10123 | | * @param ctxt an XML parser context |
10124 | | * @returns |
10125 | | * 1 if standalone="yes" |
10126 | | * 0 if standalone="no" |
10127 | | * -2 if standalone attribute is missing or invalid |
10128 | | * (A standalone value of -2 means that the XML declaration was found, |
10129 | | * but no value was specified for the standalone attribute). |
10130 | | */ |
10131 | | |
10132 | | int |
10133 | 3.45k | xmlParseSDDecl(xmlParserCtxt *ctxt) { |
10134 | 3.45k | int standalone = -2; |
10135 | | |
10136 | 3.45k | SKIP_BLANKS; |
10137 | 3.45k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10138 | 176 | SKIP(10); |
10139 | 176 | SKIP_BLANKS; |
10140 | 176 | if (RAW != '=') { |
10141 | 3 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10142 | 3 | return(standalone); |
10143 | 3 | } |
10144 | 173 | NEXT; |
10145 | 173 | SKIP_BLANKS; |
10146 | 173 | if (RAW == '\''){ |
10147 | 83 | NEXT; |
10148 | 83 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10149 | 60 | standalone = 0; |
10150 | 60 | SKIP(2); |
10151 | 60 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10152 | 23 | (NXT(2) == 's')) { |
10153 | 6 | standalone = 1; |
10154 | 6 | SKIP(3); |
10155 | 17 | } else { |
10156 | 17 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10157 | 17 | } |
10158 | 83 | if (RAW != '\'') { |
10159 | 25 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10160 | 25 | } else |
10161 | 58 | NEXT; |
10162 | 90 | } else if (RAW == '"'){ |
10163 | 84 | NEXT; |
10164 | 84 | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10165 | 8 | standalone = 0; |
10166 | 8 | SKIP(2); |
10167 | 76 | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10168 | 76 | (NXT(2) == 's')) { |
10169 | 62 | standalone = 1; |
10170 | 62 | SKIP(3); |
10171 | 62 | } else { |
10172 | 14 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10173 | 14 | } |
10174 | 84 | if (RAW != '"') { |
10175 | 44 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10176 | 44 | } else |
10177 | 40 | NEXT; |
10178 | 84 | } else { |
10179 | 6 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10180 | 6 | } |
10181 | 173 | } |
10182 | 3.45k | return(standalone); |
10183 | 3.45k | } |
10184 | | |
10185 | | /** |
10186 | | * Parse an XML declaration header |
10187 | | * |
10188 | | * @deprecated Internal function, don't use. |
10189 | | * |
10190 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10191 | | * @param ctxt an XML parser context |
10192 | | */ |
10193 | | |
10194 | | void |
10195 | 6.88k | xmlParseXMLDecl(xmlParserCtxt *ctxt) { |
10196 | 6.88k | xmlChar *version; |
10197 | | |
10198 | | /* |
10199 | | * This value for standalone indicates that the document has an |
10200 | | * XML declaration but it does not have a standalone attribute. |
10201 | | * It will be overwritten later if a standalone attribute is found. |
10202 | | */ |
10203 | | |
10204 | 6.88k | ctxt->standalone = -2; |
10205 | | |
10206 | | /* |
10207 | | * We know that '<?xml' is here. |
10208 | | */ |
10209 | 6.88k | SKIP(5); |
10210 | | |
10211 | 6.88k | if (!IS_BLANK_CH(RAW)) { |
10212 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10213 | 0 | "Blank needed after '<?xml'\n"); |
10214 | 0 | } |
10215 | 6.88k | SKIP_BLANKS; |
10216 | | |
10217 | | /* |
10218 | | * We must have the VersionInfo here. |
10219 | | */ |
10220 | 6.88k | version = xmlParseVersionInfo(ctxt); |
10221 | 6.88k | if (version == NULL) { |
10222 | 3.49k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10223 | 3.49k | } else { |
10224 | 3.38k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10225 | | /* |
10226 | | * Changed here for XML-1.0 5th edition |
10227 | | */ |
10228 | 1.49k | if (ctxt->options & XML_PARSE_OLD10) { |
10229 | 230 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10230 | 230 | "Unsupported version '%s'\n", |
10231 | 230 | version); |
10232 | 1.26k | } else { |
10233 | 1.26k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10234 | 693 | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10235 | 693 | "Unsupported version '%s'\n", |
10236 | 693 | version, NULL); |
10237 | 693 | } else { |
10238 | 570 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10239 | 570 | "Unsupported version '%s'\n", |
10240 | 570 | version); |
10241 | 570 | } |
10242 | 1.26k | } |
10243 | 1.49k | } |
10244 | 3.38k | if (ctxt->version != NULL) |
10245 | 0 | xmlFree(ctxt->version); |
10246 | 3.38k | ctxt->version = version; |
10247 | 3.38k | } |
10248 | | |
10249 | | /* |
10250 | | * We may have the encoding declaration |
10251 | | */ |
10252 | 6.88k | if (!IS_BLANK_CH(RAW)) { |
10253 | 4.49k | if ((RAW == '?') && (NXT(1) == '>')) { |
10254 | 868 | SKIP(2); |
10255 | 868 | return; |
10256 | 868 | } |
10257 | 3.62k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10258 | 3.62k | } |
10259 | 6.01k | xmlParseEncodingDecl(ctxt); |
10260 | | |
10261 | | /* |
10262 | | * We may have the standalone status. |
10263 | | */ |
10264 | 6.01k | if ((ctxt->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10265 | 3.48k | if ((RAW == '?') && (NXT(1) == '>')) { |
10266 | 2.56k | SKIP(2); |
10267 | 2.56k | return; |
10268 | 2.56k | } |
10269 | 921 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10270 | 921 | } |
10271 | | |
10272 | | /* |
10273 | | * We can grow the input buffer freely at that point |
10274 | | */ |
10275 | 3.45k | GROW; |
10276 | | |
10277 | 3.45k | SKIP_BLANKS; |
10278 | 3.45k | ctxt->standalone = xmlParseSDDecl(ctxt); |
10279 | | |
10280 | 3.45k | SKIP_BLANKS; |
10281 | 3.45k | if ((RAW == '?') && (NXT(1) == '>')) { |
10282 | 152 | SKIP(2); |
10283 | 3.30k | } else if (RAW == '>') { |
10284 | | /* Deprecated old WD ... */ |
10285 | 224 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10286 | 224 | NEXT; |
10287 | 3.07k | } else { |
10288 | 3.07k | int c; |
10289 | | |
10290 | 3.07k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10291 | 319k | while ((PARSER_STOPPED(ctxt) == 0) && |
10292 | 319k | ((c = CUR) != 0)) { |
10293 | 318k | NEXT; |
10294 | 318k | if (c == '>') |
10295 | 2.16k | break; |
10296 | 318k | } |
10297 | 3.07k | } |
10298 | 3.45k | } |
10299 | | |
10300 | | /** |
10301 | | * @since 2.14.0 |
10302 | | * |
10303 | | * @param ctxt parser context |
10304 | | * @returns the version from the XML declaration. |
10305 | | */ |
10306 | | const xmlChar * |
10307 | 0 | xmlCtxtGetVersion(xmlParserCtxt *ctxt) { |
10308 | 0 | if (ctxt == NULL) |
10309 | 0 | return(NULL); |
10310 | | |
10311 | 0 | return(ctxt->version); |
10312 | 0 | } |
10313 | | |
10314 | | /** |
10315 | | * @since 2.14.0 |
10316 | | * |
10317 | | * @param ctxt parser context |
10318 | | * @returns the value from the standalone document declaration. |
10319 | | */ |
10320 | | int |
10321 | 0 | xmlCtxtGetStandalone(xmlParserCtxt *ctxt) { |
10322 | 0 | if (ctxt == NULL) |
10323 | 0 | return(0); |
10324 | | |
10325 | 0 | return(ctxt->standalone); |
10326 | 0 | } |
10327 | | |
10328 | | /** |
10329 | | * Parse an XML Misc* optional field. |
10330 | | * |
10331 | | * @deprecated Internal function, don't use. |
10332 | | * |
10333 | | * [27] Misc ::= Comment | PI | S |
10334 | | * @param ctxt an XML parser context |
10335 | | */ |
10336 | | |
10337 | | void |
10338 | 53.6k | xmlParseMisc(xmlParserCtxt *ctxt) { |
10339 | 59.8k | while (PARSER_STOPPED(ctxt) == 0) { |
10340 | 54.0k | SKIP_BLANKS; |
10341 | 54.0k | GROW; |
10342 | 54.0k | if ((RAW == '<') && (NXT(1) == '?')) { |
10343 | 4.44k | xmlParsePI(ctxt); |
10344 | 49.6k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10345 | 1.74k | xmlParseComment(ctxt); |
10346 | 47.8k | } else { |
10347 | 47.8k | break; |
10348 | 47.8k | } |
10349 | 54.0k | } |
10350 | 53.6k | } |
10351 | | |
10352 | | static void |
10353 | 39.0k | xmlFinishDocument(xmlParserCtxtPtr ctxt) { |
10354 | 39.0k | xmlDocPtr doc; |
10355 | | |
10356 | | /* |
10357 | | * SAX: end of the document processing. |
10358 | | */ |
10359 | 39.0k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10360 | 39.0k | ctxt->sax->endDocument(ctxt->userData); |
10361 | | |
10362 | | /* |
10363 | | * Remove locally kept entity definitions if the tree was not built |
10364 | | */ |
10365 | 39.0k | doc = ctxt->myDoc; |
10366 | 39.0k | if ((doc != NULL) && |
10367 | 39.0k | (xmlStrEqual(doc->version, SAX_COMPAT_MODE))) { |
10368 | 224 | xmlFreeDoc(doc); |
10369 | 224 | ctxt->myDoc = NULL; |
10370 | 224 | } |
10371 | 39.0k | } |
10372 | | |
10373 | | /** |
10374 | | * Parse an XML document and invoke the SAX handlers. This is useful |
10375 | | * if you're only interested in custom SAX callbacks. If you want a |
10376 | | * document tree, use #xmlCtxtParseDocument. |
10377 | | * |
10378 | | * @param ctxt an XML parser context |
10379 | | * @returns 0, -1 in case of error. |
10380 | | */ |
10381 | | |
10382 | | int |
10383 | 24.7k | xmlParseDocument(xmlParserCtxt *ctxt) { |
10384 | 24.7k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10385 | 0 | return(-1); |
10386 | | |
10387 | 24.7k | GROW; |
10388 | | |
10389 | | /* |
10390 | | * SAX: detecting the level. |
10391 | | */ |
10392 | 24.7k | xmlCtxtInitializeLate(ctxt); |
10393 | | |
10394 | 24.7k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10395 | 24.7k | ctxt->sax->setDocumentLocator(ctxt->userData, |
10396 | 24.7k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10397 | 24.7k | } |
10398 | | |
10399 | 24.7k | xmlDetectEncoding(ctxt); |
10400 | | |
10401 | 24.7k | if (CUR == 0) { |
10402 | 278 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10403 | 278 | return(-1); |
10404 | 278 | } |
10405 | | |
10406 | 24.4k | GROW; |
10407 | 24.4k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10408 | | |
10409 | | /* |
10410 | | * Note that we will switch encoding on the fly. |
10411 | | */ |
10412 | 3.49k | xmlParseXMLDecl(ctxt); |
10413 | 3.49k | SKIP_BLANKS; |
10414 | 21.0k | } else { |
10415 | 21.0k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10416 | 21.0k | if (ctxt->version == NULL) { |
10417 | 12 | xmlErrMemory(ctxt); |
10418 | 12 | return(-1); |
10419 | 12 | } |
10420 | 21.0k | } |
10421 | 24.4k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10422 | 23.5k | ctxt->sax->startDocument(ctxt->userData); |
10423 | 24.4k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10424 | 24.4k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10425 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10426 | 0 | } |
10427 | | |
10428 | | /* |
10429 | | * The Misc part of the Prolog |
10430 | | */ |
10431 | 24.4k | xmlParseMisc(ctxt); |
10432 | | |
10433 | | /* |
10434 | | * Then possibly doc type declaration(s) and more Misc |
10435 | | * (doctypedecl Misc*)? |
10436 | | */ |
10437 | 24.4k | GROW; |
10438 | 24.4k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10439 | | |
10440 | 13.8k | ctxt->inSubset = 1; |
10441 | 13.8k | xmlParseDocTypeDecl(ctxt); |
10442 | 13.8k | if (RAW == '[') { |
10443 | 10.9k | xmlParseInternalSubset(ctxt); |
10444 | 10.9k | } else if (RAW == '>') { |
10445 | 2.21k | NEXT; |
10446 | 2.21k | } |
10447 | | |
10448 | | /* |
10449 | | * Create and update the external subset. |
10450 | | */ |
10451 | 13.8k | ctxt->inSubset = 2; |
10452 | 13.8k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10453 | 13.8k | (!ctxt->disableSAX)) |
10454 | 10.5k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10455 | 10.5k | ctxt->extSubSystem, ctxt->extSubURI); |
10456 | 13.8k | ctxt->inSubset = 0; |
10457 | | |
10458 | 13.8k | xmlCleanSpecialAttr(ctxt); |
10459 | | |
10460 | 13.8k | xmlParseMisc(ctxt); |
10461 | 13.8k | } |
10462 | | |
10463 | | /* |
10464 | | * Time to start parsing the tree itself |
10465 | | */ |
10466 | 24.4k | GROW; |
10467 | 24.4k | if (RAW != '<') { |
10468 | 9.09k | if (ctxt->wellFormed) |
10469 | 931 | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10470 | 931 | "Start tag expected, '<' not found\n"); |
10471 | 15.3k | } else { |
10472 | 15.3k | xmlParseElement(ctxt); |
10473 | | |
10474 | | /* |
10475 | | * The Misc part at the end |
10476 | | */ |
10477 | 15.3k | xmlParseMisc(ctxt); |
10478 | | |
10479 | 15.3k | xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END); |
10480 | 15.3k | } |
10481 | | |
10482 | 24.4k | ctxt->instate = XML_PARSER_EOF; |
10483 | 24.4k | xmlFinishDocument(ctxt); |
10484 | | |
10485 | 24.4k | if (! ctxt->wellFormed) { |
10486 | 24.3k | ctxt->valid = 0; |
10487 | 24.3k | return(-1); |
10488 | 24.3k | } |
10489 | | |
10490 | 156 | return(0); |
10491 | 24.4k | } |
10492 | | |
10493 | | /** |
10494 | | * Parse a general parsed entity |
10495 | | * An external general parsed entity is well-formed if it matches the |
10496 | | * production labeled extParsedEnt. |
10497 | | * |
10498 | | * @deprecated Internal function, don't use. |
10499 | | * |
10500 | | * [78] extParsedEnt ::= TextDecl? content |
10501 | | * |
10502 | | * @param ctxt an XML parser context |
10503 | | * @returns 0, -1 in case of error. the parser context is augmented |
10504 | | * as a result of the parsing. |
10505 | | */ |
10506 | | |
10507 | | int |
10508 | 0 | xmlParseExtParsedEnt(xmlParserCtxt *ctxt) { |
10509 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10510 | 0 | return(-1); |
10511 | | |
10512 | 0 | xmlCtxtInitializeLate(ctxt); |
10513 | |
|
10514 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10515 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, |
10516 | 0 | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10517 | 0 | } |
10518 | |
|
10519 | 0 | xmlDetectEncoding(ctxt); |
10520 | |
|
10521 | 0 | if (CUR == 0) { |
10522 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10523 | 0 | } |
10524 | | |
10525 | | /* |
10526 | | * Check for the XMLDecl in the Prolog. |
10527 | | */ |
10528 | 0 | GROW; |
10529 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10530 | | |
10531 | | /* |
10532 | | * Note that we will switch encoding on the fly. |
10533 | | */ |
10534 | 0 | xmlParseXMLDecl(ctxt); |
10535 | 0 | SKIP_BLANKS; |
10536 | 0 | } else { |
10537 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10538 | 0 | } |
10539 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10540 | 0 | ctxt->sax->startDocument(ctxt->userData); |
10541 | | |
10542 | | /* |
10543 | | * Doing validity checking on chunk doesn't make sense |
10544 | | */ |
10545 | 0 | ctxt->options &= ~XML_PARSE_DTDVALID; |
10546 | 0 | ctxt->validate = 0; |
10547 | 0 | ctxt->depth = 0; |
10548 | |
|
10549 | 0 | xmlParseContentInternal(ctxt); |
10550 | |
|
10551 | 0 | if (ctxt->input->cur < ctxt->input->end) |
10552 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
10553 | | |
10554 | | /* |
10555 | | * SAX: end of the document processing. |
10556 | | */ |
10557 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10558 | 0 | ctxt->sax->endDocument(ctxt->userData); |
10559 | |
|
10560 | 0 | if (! ctxt->wellFormed) return(-1); |
10561 | 0 | return(0); |
10562 | 0 | } |
10563 | | |
10564 | | #ifdef LIBXML_PUSH_ENABLED |
10565 | | /************************************************************************ |
10566 | | * * |
10567 | | * Progressive parsing interfaces * |
10568 | | * * |
10569 | | ************************************************************************/ |
10570 | | |
10571 | | /** |
10572 | | * Check whether the input buffer contains a character. |
10573 | | * |
10574 | | * @param ctxt an XML parser context |
10575 | | * @param c character |
10576 | | */ |
10577 | | static int |
10578 | 13.8k | xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) { |
10579 | 13.8k | const xmlChar *cur; |
10580 | | |
10581 | 13.8k | if (ctxt->checkIndex == 0) { |
10582 | 10.7k | cur = ctxt->input->cur + 1; |
10583 | 10.7k | } else { |
10584 | 3.08k | cur = ctxt->input->cur + ctxt->checkIndex; |
10585 | 3.08k | } |
10586 | | |
10587 | 13.8k | if (memchr(cur, c, ctxt->input->end - cur) == NULL) { |
10588 | 3.69k | size_t index = ctxt->input->end - ctxt->input->cur; |
10589 | | |
10590 | 3.69k | if (index > LONG_MAX) { |
10591 | 0 | ctxt->checkIndex = 0; |
10592 | 0 | return(1); |
10593 | 0 | } |
10594 | 3.69k | ctxt->checkIndex = index; |
10595 | 3.69k | return(0); |
10596 | 10.1k | } else { |
10597 | 10.1k | ctxt->checkIndex = 0; |
10598 | 10.1k | return(1); |
10599 | 10.1k | } |
10600 | 13.8k | } |
10601 | | |
10602 | | /** |
10603 | | * Check whether the input buffer contains a string. |
10604 | | * |
10605 | | * @param ctxt an XML parser context |
10606 | | * @param startDelta delta to apply at the start |
10607 | | * @param str string |
10608 | | * @param strLen length of string |
10609 | | */ |
10610 | | static const xmlChar * |
10611 | | xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta, |
10612 | 95.0k | const char *str, size_t strLen) { |
10613 | 95.0k | const xmlChar *cur, *term; |
10614 | | |
10615 | 95.0k | if (ctxt->checkIndex == 0) { |
10616 | 9.00k | cur = ctxt->input->cur + startDelta; |
10617 | 86.0k | } else { |
10618 | 86.0k | cur = ctxt->input->cur + ctxt->checkIndex; |
10619 | 86.0k | } |
10620 | | |
10621 | 95.0k | term = BAD_CAST strstr((const char *) cur, str); |
10622 | 95.0k | if (term == NULL) { |
10623 | 88.0k | const xmlChar *end = ctxt->input->end; |
10624 | 88.0k | size_t index; |
10625 | | |
10626 | | /* Rescan (strLen - 1) characters. */ |
10627 | 88.0k | if ((size_t) (end - cur) < strLen) |
10628 | 3.54k | end = cur; |
10629 | 84.4k | else |
10630 | 84.4k | end -= strLen - 1; |
10631 | 88.0k | index = end - ctxt->input->cur; |
10632 | 88.0k | if (index > LONG_MAX) { |
10633 | 0 | ctxt->checkIndex = 0; |
10634 | 0 | return(ctxt->input->end - strLen); |
10635 | 0 | } |
10636 | 88.0k | ctxt->checkIndex = index; |
10637 | 88.0k | } else { |
10638 | 7.01k | ctxt->checkIndex = 0; |
10639 | 7.01k | } |
10640 | | |
10641 | 95.0k | return(term); |
10642 | 95.0k | } |
10643 | | |
10644 | | /** |
10645 | | * Check whether the input buffer contains terminated char data. |
10646 | | * |
10647 | | * @param ctxt an XML parser context |
10648 | | */ |
10649 | | static int |
10650 | 30.6k | xmlParseLookupCharData(xmlParserCtxtPtr ctxt) { |
10651 | 30.6k | const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex; |
10652 | 30.6k | const xmlChar *end = ctxt->input->end; |
10653 | 30.6k | size_t index; |
10654 | | |
10655 | 1.13M | while (cur < end) { |
10656 | 1.11M | if ((*cur == '<') || (*cur == '&')) { |
10657 | 14.9k | ctxt->checkIndex = 0; |
10658 | 14.9k | return(1); |
10659 | 14.9k | } |
10660 | 1.10M | cur++; |
10661 | 1.10M | } |
10662 | | |
10663 | 15.7k | index = cur - ctxt->input->cur; |
10664 | 15.7k | if (index > LONG_MAX) { |
10665 | 0 | ctxt->checkIndex = 0; |
10666 | 0 | return(1); |
10667 | 0 | } |
10668 | 15.7k | ctxt->checkIndex = index; |
10669 | 15.7k | return(0); |
10670 | 15.7k | } |
10671 | | |
10672 | | /** |
10673 | | * Check whether there's enough data in the input buffer to finish parsing |
10674 | | * a start tag. This has to take quotes into account. |
10675 | | * |
10676 | | * @param ctxt an XML parser context |
10677 | | */ |
10678 | | static int |
10679 | 133k | xmlParseLookupGt(xmlParserCtxtPtr ctxt) { |
10680 | 133k | const xmlChar *cur; |
10681 | 133k | const xmlChar *end = ctxt->input->end; |
10682 | 133k | int state = ctxt->endCheckState; |
10683 | 133k | size_t index; |
10684 | | |
10685 | 133k | if (ctxt->checkIndex == 0) |
10686 | 61.0k | cur = ctxt->input->cur + 1; |
10687 | 72.6k | else |
10688 | 72.6k | cur = ctxt->input->cur + ctxt->checkIndex; |
10689 | | |
10690 | 58.1M | while (cur < end) { |
10691 | 58.0M | if (state) { |
10692 | 51.8M | if (*cur == state) |
10693 | 60.5k | state = 0; |
10694 | 51.8M | } else if (*cur == '\'' || *cur == '"') { |
10695 | 64.1k | state = *cur; |
10696 | 6.17M | } else if (*cur == '>') { |
10697 | 53.8k | ctxt->checkIndex = 0; |
10698 | 53.8k | ctxt->endCheckState = 0; |
10699 | 53.8k | return(1); |
10700 | 53.8k | } |
10701 | 58.0M | cur++; |
10702 | 58.0M | } |
10703 | | |
10704 | 79.8k | index = cur - ctxt->input->cur; |
10705 | 79.8k | if (index > LONG_MAX) { |
10706 | 0 | ctxt->checkIndex = 0; |
10707 | 0 | ctxt->endCheckState = 0; |
10708 | 0 | return(1); |
10709 | 0 | } |
10710 | 79.8k | ctxt->checkIndex = index; |
10711 | 79.8k | ctxt->endCheckState = state; |
10712 | 79.8k | return(0); |
10713 | 79.8k | } |
10714 | | |
10715 | | /** |
10716 | | * Check whether there's enough data in the input buffer to finish parsing |
10717 | | * the internal subset. |
10718 | | * |
10719 | | * @param ctxt an XML parser context |
10720 | | */ |
10721 | | static int |
10722 | 294k | xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) { |
10723 | | /* |
10724 | | * Sorry, but progressive parsing of the internal subset is not |
10725 | | * supported. We first check that the full content of the internal |
10726 | | * subset is available and parsing is launched only at that point. |
10727 | | * Internal subset ends with "']' S? '>'" in an unescaped section and |
10728 | | * not in a ']]>' sequence which are conditional sections. |
10729 | | */ |
10730 | 294k | const xmlChar *cur, *start; |
10731 | 294k | const xmlChar *end = ctxt->input->end; |
10732 | 294k | int state = ctxt->endCheckState; |
10733 | 294k | size_t index; |
10734 | | |
10735 | 294k | if (ctxt->checkIndex == 0) { |
10736 | 5.13k | cur = ctxt->input->cur + 1; |
10737 | 289k | } else { |
10738 | 289k | cur = ctxt->input->cur + ctxt->checkIndex; |
10739 | 289k | } |
10740 | 294k | start = cur; |
10741 | | |
10742 | 268M | while (cur < end) { |
10743 | 268M | if (state == '-') { |
10744 | 23.9k | if ((*cur == '-') && |
10745 | 23.9k | (cur[1] == '-') && |
10746 | 23.9k | (cur[2] == '>')) { |
10747 | 1.76k | state = 0; |
10748 | 1.76k | cur += 3; |
10749 | 1.76k | start = cur; |
10750 | 1.76k | continue; |
10751 | 1.76k | } |
10752 | 23.9k | } |
10753 | 268M | else if (state == ']') { |
10754 | 4.67k | if (*cur == '>') { |
10755 | 2.61k | ctxt->checkIndex = 0; |
10756 | 2.61k | ctxt->endCheckState = 0; |
10757 | 2.61k | return(1); |
10758 | 2.61k | } |
10759 | 2.05k | if (IS_BLANK_CH(*cur)) { |
10760 | 1.05k | state = ' '; |
10761 | 1.05k | } else if (*cur != ']') { |
10762 | 554 | state = 0; |
10763 | 554 | start = cur; |
10764 | 554 | continue; |
10765 | 554 | } |
10766 | 2.05k | } |
10767 | 268M | else if (state == ' ') { |
10768 | 2.28k | if (*cur == '>') { |
10769 | 45 | ctxt->checkIndex = 0; |
10770 | 45 | ctxt->endCheckState = 0; |
10771 | 45 | return(1); |
10772 | 45 | } |
10773 | 2.24k | if (!IS_BLANK_CH(*cur)) { |
10774 | 993 | state = 0; |
10775 | 993 | start = cur; |
10776 | 993 | continue; |
10777 | 993 | } |
10778 | 2.24k | } |
10779 | 268M | else if (state != 0) { |
10780 | 267M | if (*cur == state) { |
10781 | 15.1k | state = 0; |
10782 | 15.1k | start = cur + 1; |
10783 | 15.1k | } |
10784 | 267M | } |
10785 | 949k | else if (*cur == '<') { |
10786 | 22.7k | if ((cur[1] == '!') && |
10787 | 22.7k | (cur[2] == '-') && |
10788 | 22.7k | (cur[3] == '-')) { |
10789 | 1.79k | state = '-'; |
10790 | 1.79k | cur += 4; |
10791 | | /* Don't treat <!--> as comment */ |
10792 | 1.79k | start = cur; |
10793 | 1.79k | continue; |
10794 | 1.79k | } |
10795 | 22.7k | } |
10796 | 926k | else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) { |
10797 | 19.9k | state = *cur; |
10798 | 19.9k | } |
10799 | | |
10800 | 268M | cur++; |
10801 | 268M | } |
10802 | | |
10803 | | /* |
10804 | | * Rescan the three last characters to detect "<!--" and "-->" |
10805 | | * split across chunks. |
10806 | | */ |
10807 | 292k | if ((state == 0) || (state == '-')) { |
10808 | 10.7k | if (cur - start < 3) |
10809 | 1.72k | cur = start; |
10810 | 9.02k | else |
10811 | 9.02k | cur -= 3; |
10812 | 10.7k | } |
10813 | 292k | index = cur - ctxt->input->cur; |
10814 | 292k | if (index > LONG_MAX) { |
10815 | 0 | ctxt->checkIndex = 0; |
10816 | 0 | ctxt->endCheckState = 0; |
10817 | 0 | return(1); |
10818 | 0 | } |
10819 | 292k | ctxt->checkIndex = index; |
10820 | 292k | ctxt->endCheckState = state; |
10821 | 292k | return(0); |
10822 | 292k | } |
10823 | | |
10824 | | /** |
10825 | | * Try to progress on parsing |
10826 | | * |
10827 | | * @param ctxt an XML parser context |
10828 | | * @param terminate last chunk indicator |
10829 | | * @returns zero if no parsing was possible |
10830 | | */ |
10831 | | static int |
10832 | 539k | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
10833 | 539k | int ret = 0; |
10834 | 539k | size_t avail; |
10835 | 539k | xmlChar cur, next; |
10836 | | |
10837 | 539k | if (ctxt->input == NULL) |
10838 | 0 | return(0); |
10839 | | |
10840 | 539k | if ((ctxt->input != NULL) && |
10841 | 539k | (ctxt->input->cur - ctxt->input->base > 4096)) { |
10842 | 825 | xmlParserShrink(ctxt); |
10843 | 825 | } |
10844 | | |
10845 | 1.09M | while (ctxt->disableSAX == 0) { |
10846 | 1.08M | avail = ctxt->input->end - ctxt->input->cur; |
10847 | 1.08M | if (avail < 1) |
10848 | 13.3k | goto done; |
10849 | 1.07M | switch (ctxt->instate) { |
10850 | 11.1k | case XML_PARSER_EOF: |
10851 | | /* |
10852 | | * Document parsing is done ! |
10853 | | */ |
10854 | 11.1k | goto done; |
10855 | 33.3k | case XML_PARSER_START: |
10856 | | /* |
10857 | | * Very first chars read from the document flow. |
10858 | | */ |
10859 | 33.3k | if ((!terminate) && (avail < 4)) |
10860 | 7.56k | goto done; |
10861 | | |
10862 | | /* |
10863 | | * We need more bytes to detect EBCDIC code pages. |
10864 | | * See xmlDetectEBCDIC. |
10865 | | */ |
10866 | 25.7k | if ((CMP4(CUR_PTR, 0x4C, 0x6F, 0xA7, 0x94)) && |
10867 | 25.7k | (!terminate) && (avail < 200)) |
10868 | 1.36k | goto done; |
10869 | | |
10870 | 24.3k | xmlDetectEncoding(ctxt); |
10871 | 24.3k | ctxt->instate = XML_PARSER_XML_DECL; |
10872 | 24.3k | break; |
10873 | | |
10874 | 43.3k | case XML_PARSER_XML_DECL: |
10875 | 43.3k | if ((!terminate) && (avail < 2)) |
10876 | 207 | goto done; |
10877 | 43.1k | cur = ctxt->input->cur[0]; |
10878 | 43.1k | next = ctxt->input->cur[1]; |
10879 | 43.1k | if ((cur == '<') && (next == '?')) { |
10880 | | /* PI or XML decl */ |
10881 | 23.9k | if ((!terminate) && |
10882 | 23.9k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
10883 | 18.8k | goto done; |
10884 | 5.13k | if ((ctxt->input->cur[2] == 'x') && |
10885 | 5.13k | (ctxt->input->cur[3] == 'm') && |
10886 | 5.13k | (ctxt->input->cur[4] == 'l') && |
10887 | 5.13k | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
10888 | 3.39k | ret += 5; |
10889 | 3.39k | xmlParseXMLDecl(ctxt); |
10890 | 3.39k | } else { |
10891 | 1.73k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10892 | 1.73k | if (ctxt->version == NULL) { |
10893 | 4 | xmlErrMemory(ctxt); |
10894 | 4 | break; |
10895 | 4 | } |
10896 | 1.73k | } |
10897 | 19.1k | } else { |
10898 | 19.1k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10899 | 19.1k | if (ctxt->version == NULL) { |
10900 | 36 | xmlErrMemory(ctxt); |
10901 | 36 | break; |
10902 | 36 | } |
10903 | 19.1k | } |
10904 | 24.2k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) { |
10905 | 24.2k | ctxt->sax->setDocumentLocator(ctxt->userData, |
10906 | 24.2k | (xmlSAXLocator *) &xmlDefaultSAXLocator); |
10907 | 24.2k | } |
10908 | 24.2k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
10909 | 24.2k | (!ctxt->disableSAX)) |
10910 | 23.6k | ctxt->sax->startDocument(ctxt->userData); |
10911 | 24.2k | ctxt->instate = XML_PARSER_MISC; |
10912 | 24.2k | break; |
10913 | 160k | case XML_PARSER_START_TAG: { |
10914 | 160k | const xmlChar *name; |
10915 | 160k | const xmlChar *prefix = NULL; |
10916 | 160k | const xmlChar *URI = NULL; |
10917 | 160k | int line = ctxt->input->line; |
10918 | 160k | int nbNs = 0; |
10919 | | |
10920 | 160k | if ((!terminate) && (avail < 2)) |
10921 | 26 | goto done; |
10922 | 160k | cur = ctxt->input->cur[0]; |
10923 | 160k | if (cur != '<') { |
10924 | 2.57k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10925 | 2.57k | "Start tag expected, '<' not found"); |
10926 | 2.57k | ctxt->instate = XML_PARSER_EOF; |
10927 | 2.57k | xmlFinishDocument(ctxt); |
10928 | 2.57k | goto done; |
10929 | 2.57k | } |
10930 | 157k | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
10931 | 41.3k | goto done; |
10932 | 116k | if (ctxt->spaceNr == 0) |
10933 | 0 | spacePush(ctxt, -1); |
10934 | 116k | else if (*ctxt->space == -2) |
10935 | 23.6k | spacePush(ctxt, -1); |
10936 | 92.7k | else |
10937 | 92.7k | spacePush(ctxt, *ctxt->space); |
10938 | 116k | #ifdef LIBXML_SAX1_ENABLED |
10939 | 116k | if (ctxt->sax2) |
10940 | 116k | #endif /* LIBXML_SAX1_ENABLED */ |
10941 | 116k | name = xmlParseStartTag2(ctxt, &prefix, &URI, &nbNs); |
10942 | 0 | #ifdef LIBXML_SAX1_ENABLED |
10943 | 0 | else |
10944 | 0 | name = xmlParseStartTag(ctxt); |
10945 | 116k | #endif /* LIBXML_SAX1_ENABLED */ |
10946 | 116k | if (name == NULL) { |
10947 | 3.14k | spacePop(ctxt); |
10948 | 3.14k | ctxt->instate = XML_PARSER_EOF; |
10949 | 3.14k | xmlFinishDocument(ctxt); |
10950 | 3.14k | goto done; |
10951 | 3.14k | } |
10952 | 113k | #ifdef LIBXML_VALID_ENABLED |
10953 | | /* |
10954 | | * [ VC: Root Element Type ] |
10955 | | * The Name in the document type declaration must match |
10956 | | * the element type of the root element. |
10957 | | */ |
10958 | 113k | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10959 | 113k | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10960 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10961 | 113k | #endif /* LIBXML_VALID_ENABLED */ |
10962 | | |
10963 | | /* |
10964 | | * Check for an Empty Element. |
10965 | | */ |
10966 | 113k | if ((RAW == '/') && (NXT(1) == '>')) { |
10967 | 4.20k | SKIP(2); |
10968 | | |
10969 | 4.20k | if (ctxt->sax2) { |
10970 | 4.20k | if ((ctxt->sax != NULL) && |
10971 | 4.20k | (ctxt->sax->endElementNs != NULL) && |
10972 | 4.20k | (!ctxt->disableSAX)) |
10973 | 4.18k | ctxt->sax->endElementNs(ctxt->userData, name, |
10974 | 4.18k | prefix, URI); |
10975 | 4.20k | if (nbNs > 0) |
10976 | 1.74k | xmlParserNsPop(ctxt, nbNs); |
10977 | 4.20k | #ifdef LIBXML_SAX1_ENABLED |
10978 | 4.20k | } else { |
10979 | 0 | if ((ctxt->sax != NULL) && |
10980 | 0 | (ctxt->sax->endElement != NULL) && |
10981 | 0 | (!ctxt->disableSAX)) |
10982 | 0 | ctxt->sax->endElement(ctxt->userData, name); |
10983 | 0 | #endif /* LIBXML_SAX1_ENABLED */ |
10984 | 0 | } |
10985 | 4.20k | spacePop(ctxt); |
10986 | 109k | } else if (RAW == '>') { |
10987 | 75.2k | NEXT; |
10988 | 75.2k | nameNsPush(ctxt, name, prefix, URI, line, nbNs); |
10989 | 75.2k | } else { |
10990 | 33.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
10991 | 33.7k | "Couldn't find end of Start Tag %s\n", |
10992 | 33.7k | name); |
10993 | 33.7k | nodePop(ctxt); |
10994 | 33.7k | spacePop(ctxt); |
10995 | 33.7k | if (nbNs > 0) |
10996 | 6.45k | xmlParserNsPop(ctxt, nbNs); |
10997 | 33.7k | } |
10998 | | |
10999 | 113k | if (ctxt->nameNr == 0) |
11000 | 4.09k | ctxt->instate = XML_PARSER_EPILOG; |
11001 | 109k | else |
11002 | 109k | ctxt->instate = XML_PARSER_CONTENT; |
11003 | 113k | break; |
11004 | 116k | } |
11005 | 427k | case XML_PARSER_CONTENT: { |
11006 | 427k | cur = ctxt->input->cur[0]; |
11007 | | |
11008 | 427k | if (cur == '<') { |
11009 | 184k | if ((!terminate) && (avail < 2)) |
11010 | 1.85k | goto done; |
11011 | 182k | next = ctxt->input->cur[1]; |
11012 | | |
11013 | 182k | if (next == '/') { |
11014 | 6.18k | ctxt->instate = XML_PARSER_END_TAG; |
11015 | 6.18k | break; |
11016 | 175k | } else if (next == '?') { |
11017 | 19.7k | if ((!terminate) && |
11018 | 19.7k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11019 | 14.6k | goto done; |
11020 | 5.06k | xmlParsePI(ctxt); |
11021 | 5.06k | ctxt->instate = XML_PARSER_CONTENT; |
11022 | 5.06k | break; |
11023 | 156k | } else if (next == '!') { |
11024 | 54.4k | if ((!terminate) && (avail < 3)) |
11025 | 566 | goto done; |
11026 | 53.8k | next = ctxt->input->cur[2]; |
11027 | | |
11028 | 53.8k | if (next == '-') { |
11029 | 5.23k | if ((!terminate) && (avail < 4)) |
11030 | 437 | goto done; |
11031 | 4.79k | if (ctxt->input->cur[3] == '-') { |
11032 | 4.77k | if ((!terminate) && |
11033 | 4.77k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11034 | 2.42k | goto done; |
11035 | 2.34k | xmlParseComment(ctxt); |
11036 | 2.34k | ctxt->instate = XML_PARSER_CONTENT; |
11037 | 2.34k | break; |
11038 | 4.77k | } |
11039 | 48.6k | } else if (next == '[') { |
11040 | 48.0k | if ((!terminate) && (avail < 9)) |
11041 | 620 | goto done; |
11042 | 47.4k | if ((ctxt->input->cur[2] == '[') && |
11043 | 47.4k | (ctxt->input->cur[3] == 'C') && |
11044 | 47.4k | (ctxt->input->cur[4] == 'D') && |
11045 | 47.4k | (ctxt->input->cur[5] == 'A') && |
11046 | 47.4k | (ctxt->input->cur[6] == 'T') && |
11047 | 47.4k | (ctxt->input->cur[7] == 'A') && |
11048 | 47.4k | (ctxt->input->cur[8] == '[')) { |
11049 | 47.3k | if ((!terminate) && |
11050 | 47.3k | (!xmlParseLookupString(ctxt, 9, "]]>", 3))) |
11051 | 45.2k | goto done; |
11052 | 2.04k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11053 | 2.04k | xmlParseCDSect(ctxt); |
11054 | 2.04k | ctxt->instate = XML_PARSER_CONTENT; |
11055 | 2.04k | break; |
11056 | 47.3k | } |
11057 | 47.4k | } |
11058 | 53.8k | } |
11059 | 243k | } else if (cur == '&') { |
11060 | 21.6k | if ((!terminate) && (!xmlParseLookupChar(ctxt, ';'))) |
11061 | 2.75k | goto done; |
11062 | 18.8k | xmlParseReference(ctxt); |
11063 | 18.8k | break; |
11064 | 221k | } else { |
11065 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11066 | | /* |
11067 | | * Goal of the following test is: |
11068 | | * - minimize calls to the SAX 'character' callback |
11069 | | * when they are mergeable |
11070 | | * - handle an problem for isBlank when we only parse |
11071 | | * a sequence of blank chars and the next one is |
11072 | | * not available to check against '<' presence. |
11073 | | * - tries to homogenize the differences in SAX |
11074 | | * callbacks between the push and pull versions |
11075 | | * of the parser. |
11076 | | */ |
11077 | 221k | if (avail < XML_PARSER_BIG_BUFFER_SIZE) { |
11078 | 52.6k | if ((!terminate) && (!xmlParseLookupCharData(ctxt))) |
11079 | 15.7k | goto done; |
11080 | 52.6k | } |
11081 | 205k | ctxt->checkIndex = 0; |
11082 | 205k | xmlParseCharDataInternal(ctxt, !terminate); |
11083 | 205k | break; |
11084 | 221k | } |
11085 | | |
11086 | 102k | ctxt->instate = XML_PARSER_START_TAG; |
11087 | 102k | break; |
11088 | 427k | } |
11089 | 7.12k | case XML_PARSER_END_TAG: |
11090 | 7.12k | if ((!terminate) && (!xmlParseLookupChar(ctxt, '>'))) |
11091 | 939 | goto done; |
11092 | 6.18k | if (ctxt->sax2) { |
11093 | 6.18k | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11094 | 6.18k | nameNsPop(ctxt); |
11095 | 6.18k | } |
11096 | 0 | #ifdef LIBXML_SAX1_ENABLED |
11097 | 0 | else |
11098 | 0 | xmlParseEndTag1(ctxt, 0); |
11099 | 6.18k | #endif /* LIBXML_SAX1_ENABLED */ |
11100 | 6.18k | if (ctxt->nameNr == 0) { |
11101 | 282 | ctxt->instate = XML_PARSER_EPILOG; |
11102 | 5.90k | } else { |
11103 | 5.90k | ctxt->instate = XML_PARSER_CONTENT; |
11104 | 5.90k | } |
11105 | 6.18k | break; |
11106 | 75.8k | case XML_PARSER_MISC: |
11107 | 85.5k | case XML_PARSER_PROLOG: |
11108 | 89.6k | case XML_PARSER_EPILOG: |
11109 | 89.6k | SKIP_BLANKS; |
11110 | 89.6k | avail = ctxt->input->end - ctxt->input->cur; |
11111 | 89.6k | if (avail < 1) |
11112 | 866 | goto done; |
11113 | 88.7k | if (ctxt->input->cur[0] == '<') { |
11114 | 85.3k | if ((!terminate) && (avail < 2)) |
11115 | 727 | goto done; |
11116 | 84.6k | next = ctxt->input->cur[1]; |
11117 | 84.6k | if (next == '?') { |
11118 | 6.75k | if ((!terminate) && |
11119 | 6.75k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11120 | 2.66k | goto done; |
11121 | 4.09k | xmlParsePI(ctxt); |
11122 | 4.09k | break; |
11123 | 77.9k | } else if (next == '!') { |
11124 | 64.0k | if ((!terminate) && (avail < 3)) |
11125 | 290 | goto done; |
11126 | | |
11127 | 63.7k | if (ctxt->input->cur[2] == '-') { |
11128 | 6.02k | if ((!terminate) && (avail < 4)) |
11129 | 147 | goto done; |
11130 | 5.88k | if (ctxt->input->cur[3] == '-') { |
11131 | 5.86k | if ((!terminate) && |
11132 | 5.86k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11133 | 4.17k | goto done; |
11134 | 1.69k | xmlParseComment(ctxt); |
11135 | 1.69k | break; |
11136 | 5.86k | } |
11137 | 57.7k | } else if (ctxt->instate == XML_PARSER_MISC) { |
11138 | 57.7k | if ((!terminate) && (avail < 9)) |
11139 | 5.58k | goto done; |
11140 | 52.1k | if ((ctxt->input->cur[2] == 'D') && |
11141 | 52.1k | (ctxt->input->cur[3] == 'O') && |
11142 | 52.1k | (ctxt->input->cur[4] == 'C') && |
11143 | 52.1k | (ctxt->input->cur[5] == 'T') && |
11144 | 52.1k | (ctxt->input->cur[6] == 'Y') && |
11145 | 52.1k | (ctxt->input->cur[7] == 'P') && |
11146 | 52.1k | (ctxt->input->cur[8] == 'E')) { |
11147 | 52.0k | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11148 | 38.5k | goto done; |
11149 | 13.5k | ctxt->inSubset = 1; |
11150 | 13.5k | xmlParseDocTypeDecl(ctxt); |
11151 | 13.5k | if (RAW == '[') { |
11152 | 10.7k | ctxt->instate = XML_PARSER_DTD; |
11153 | 10.7k | } else { |
11154 | 2.77k | if (RAW == '>') |
11155 | 2.20k | NEXT; |
11156 | | /* |
11157 | | * Create and update the external subset. |
11158 | | */ |
11159 | 2.77k | ctxt->inSubset = 2; |
11160 | 2.77k | if ((ctxt->sax != NULL) && |
11161 | 2.77k | (!ctxt->disableSAX) && |
11162 | 2.77k | (ctxt->sax->externalSubset != NULL)) |
11163 | 2.63k | ctxt->sax->externalSubset( |
11164 | 2.63k | ctxt->userData, |
11165 | 2.63k | ctxt->intSubName, |
11166 | 2.63k | ctxt->extSubSystem, |
11167 | 2.63k | ctxt->extSubURI); |
11168 | 2.77k | ctxt->inSubset = 0; |
11169 | 2.77k | xmlCleanSpecialAttr(ctxt); |
11170 | 2.77k | ctxt->instate = XML_PARSER_PROLOG; |
11171 | 2.77k | } |
11172 | 13.5k | break; |
11173 | 52.0k | } |
11174 | 52.1k | } |
11175 | 63.7k | } |
11176 | 84.6k | } |
11177 | | |
11178 | 17.3k | if (ctxt->instate == XML_PARSER_EPILOG) { |
11179 | 870 | if (ctxt->errNo == XML_ERR_OK) |
11180 | 20 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11181 | 870 | ctxt->instate = XML_PARSER_EOF; |
11182 | 870 | xmlFinishDocument(ctxt); |
11183 | 16.4k | } else { |
11184 | 16.4k | ctxt->instate = XML_PARSER_START_TAG; |
11185 | 16.4k | } |
11186 | 17.3k | break; |
11187 | 302k | case XML_PARSER_DTD: { |
11188 | 302k | if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt))) |
11189 | 292k | goto done; |
11190 | 9.90k | xmlParseInternalSubset(ctxt); |
11191 | 9.90k | ctxt->inSubset = 2; |
11192 | 9.90k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11193 | 9.90k | (ctxt->sax->externalSubset != NULL)) |
11194 | 7.90k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
11195 | 7.90k | ctxt->extSubSystem, ctxt->extSubURI); |
11196 | 9.90k | ctxt->inSubset = 0; |
11197 | 9.90k | xmlCleanSpecialAttr(ctxt); |
11198 | 9.90k | ctxt->instate = XML_PARSER_PROLOG; |
11199 | 9.90k | break; |
11200 | 302k | } |
11201 | 0 | default: |
11202 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
11203 | 0 | "PP: internal error\n"); |
11204 | 0 | ctxt->instate = XML_PARSER_EOF; |
11205 | 0 | break; |
11206 | 1.07M | } |
11207 | 1.07M | } |
11208 | 539k | done: |
11209 | 539k | return(ret); |
11210 | 539k | } |
11211 | | |
11212 | | /** |
11213 | | * Parse a chunk of memory in push parser mode. |
11214 | | * |
11215 | | * Assumes that the parser context was initialized with |
11216 | | * #xmlCreatePushParserCtxt. |
11217 | | * |
11218 | | * The last chunk, which will often be empty, must be marked with |
11219 | | * the `terminate` flag. With the default SAX callbacks, the resulting |
11220 | | * document will be available in ctxt->myDoc. This pointer will not |
11221 | | * be freed when calling #xmlFreeParserCtxt and must be freed by the |
11222 | | * caller. If the document isn't well-formed, it will still be returned |
11223 | | * in ctxt->myDoc. |
11224 | | * |
11225 | | * As an exception, #xmlCtxtResetPush will free the document in |
11226 | | * ctxt->myDoc. So ctxt->myDoc should be set to NULL after extracting |
11227 | | * the document. |
11228 | | * |
11229 | | * Since 2.14.0, #xmlCtxtGetDocument can be used to retrieve the |
11230 | | * result document. |
11231 | | * |
11232 | | * @param ctxt an XML parser context |
11233 | | * @param chunk chunk of memory |
11234 | | * @param size size of chunk in bytes |
11235 | | * @param terminate last chunk indicator |
11236 | | * @returns an xmlParserErrors code (0 on success). |
11237 | | */ |
11238 | | int |
11239 | | xmlParseChunk(xmlParserCtxt *ctxt, const char *chunk, int size, |
11240 | 728k | int terminate) { |
11241 | 728k | size_t curBase; |
11242 | 728k | size_t maxLength; |
11243 | 728k | size_t pos; |
11244 | 728k | int end_in_lf = 0; |
11245 | 728k | int res; |
11246 | | |
11247 | 728k | if ((ctxt == NULL) || (size < 0)) |
11248 | 0 | return(XML_ERR_ARGUMENT); |
11249 | 728k | if ((chunk == NULL) && (size > 0)) |
11250 | 0 | return(XML_ERR_ARGUMENT); |
11251 | 728k | if ((ctxt->input == NULL) || (ctxt->input->buf == NULL)) |
11252 | 0 | return(XML_ERR_ARGUMENT); |
11253 | 728k | if (ctxt->disableSAX != 0) |
11254 | 187k | return(ctxt->errNo); |
11255 | | |
11256 | 540k | ctxt->input->flags |= XML_INPUT_PROGRESSIVE; |
11257 | 540k | if (ctxt->instate == XML_PARSER_START) |
11258 | 33.7k | xmlCtxtInitializeLate(ctxt); |
11259 | 540k | if ((size > 0) && (chunk != NULL) && (!terminate) && |
11260 | 540k | (chunk[size - 1] == '\r')) { |
11261 | 1.54k | end_in_lf = 1; |
11262 | 1.54k | size--; |
11263 | 1.54k | } |
11264 | | |
11265 | | /* |
11266 | | * Also push an empty chunk to make sure that the raw buffer |
11267 | | * will be flushed if there is an encoder. |
11268 | | */ |
11269 | 540k | pos = ctxt->input->cur - ctxt->input->base; |
11270 | 540k | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
11271 | 540k | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11272 | 540k | if (res < 0) { |
11273 | 298 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11274 | 298 | return(ctxt->errNo); |
11275 | 298 | } |
11276 | | |
11277 | 539k | xmlParseTryOrFinish(ctxt, terminate); |
11278 | | |
11279 | 539k | curBase = ctxt->input->cur - ctxt->input->base; |
11280 | 539k | maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
11281 | 145k | XML_MAX_HUGE_LENGTH : |
11282 | 539k | XML_MAX_LOOKUP_LIMIT; |
11283 | 539k | if (curBase > maxLength) { |
11284 | 0 | xmlFatalErr(ctxt, XML_ERR_RESOURCE_LIMIT, |
11285 | 0 | "Buffer size limit exceeded, try XML_PARSE_HUGE\n"); |
11286 | 0 | } |
11287 | | |
11288 | 539k | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX != 0)) |
11289 | 10.7k | return(ctxt->errNo); |
11290 | | |
11291 | 529k | if (end_in_lf == 1) { |
11292 | 1.53k | pos = ctxt->input->cur - ctxt->input->base; |
11293 | 1.53k | res = xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
11294 | 1.53k | xmlBufUpdateInput(ctxt->input->buf->buffer, ctxt->input, pos); |
11295 | 1.53k | if (res < 0) { |
11296 | 1 | xmlCtxtErrIO(ctxt, ctxt->input->buf->error, NULL); |
11297 | 1 | return(ctxt->errNo); |
11298 | 1 | } |
11299 | 1.53k | } |
11300 | 529k | if (terminate) { |
11301 | | /* |
11302 | | * Check for termination |
11303 | | */ |
11304 | 13.9k | if ((ctxt->instate != XML_PARSER_EOF) && |
11305 | 13.9k | (ctxt->instate != XML_PARSER_EPILOG)) { |
11306 | 6.52k | if (ctxt->nameNr > 0) { |
11307 | 3.97k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
11308 | 3.97k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
11309 | 3.97k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
11310 | 3.97k | "Premature end of data in tag %s line %d\n", |
11311 | 3.97k | name, line, NULL); |
11312 | 3.97k | } else if (ctxt->instate == XML_PARSER_START) { |
11313 | 171 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11314 | 2.38k | } else { |
11315 | 2.38k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
11316 | 2.38k | "Start tag expected, '<' not found\n"); |
11317 | 2.38k | } |
11318 | 7.40k | } else { |
11319 | 7.40k | xmlParserCheckEOF(ctxt, XML_ERR_DOCUMENT_END); |
11320 | 7.40k | } |
11321 | 13.9k | if (ctxt->instate != XML_PARSER_EOF) { |
11322 | 7.99k | ctxt->instate = XML_PARSER_EOF; |
11323 | 7.99k | xmlFinishDocument(ctxt); |
11324 | 7.99k | } |
11325 | 13.9k | } |
11326 | 529k | if (ctxt->wellFormed == 0) |
11327 | 340k | return((xmlParserErrors) ctxt->errNo); |
11328 | 189k | else |
11329 | 189k | return(0); |
11330 | 529k | } |
11331 | | |
11332 | | /************************************************************************ |
11333 | | * * |
11334 | | * I/O front end functions to the parser * |
11335 | | * * |
11336 | | ************************************************************************/ |
11337 | | |
11338 | | /** |
11339 | | * Create a parser context for using the XML parser in push mode. |
11340 | | * See #xmlParseChunk. |
11341 | | * |
11342 | | * Passing an initial chunk is useless and deprecated. |
11343 | | * |
11344 | | * The push parser doesn't support recovery mode or the |
11345 | | * XML_PARSE_NOBLANKS option. |
11346 | | * |
11347 | | * `filename` is used as base URI to fetch external entities and for |
11348 | | * error reports. |
11349 | | * |
11350 | | * @param sax a SAX handler (optional) |
11351 | | * @param user_data user data for SAX callbacks (optional) |
11352 | | * @param chunk initial chunk (optional, deprecated) |
11353 | | * @param size size of initial chunk in bytes |
11354 | | * @param filename file name or URI (optional) |
11355 | | * @returns the new parser context or NULL if a memory allocation |
11356 | | * failed. |
11357 | | */ |
11358 | | |
11359 | | xmlParserCtxt * |
11360 | | xmlCreatePushParserCtxt(xmlSAXHandler *sax, void *user_data, |
11361 | 24.8k | const char *chunk, int size, const char *filename) { |
11362 | 24.8k | xmlParserCtxtPtr ctxt; |
11363 | 24.8k | xmlParserInputPtr input; |
11364 | | |
11365 | 24.8k | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11366 | 24.8k | if (ctxt == NULL) |
11367 | 30 | return(NULL); |
11368 | | |
11369 | 24.8k | ctxt->options &= ~XML_PARSE_NODICT; |
11370 | 24.8k | ctxt->dictNames = 1; |
11371 | | |
11372 | 24.8k | input = xmlNewPushInput(filename, chunk, size); |
11373 | 24.8k | if (input == NULL) { |
11374 | 15 | xmlFreeParserCtxt(ctxt); |
11375 | 15 | return(NULL); |
11376 | 15 | } |
11377 | 24.7k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11378 | 7 | xmlFreeInputStream(input); |
11379 | 7 | xmlFreeParserCtxt(ctxt); |
11380 | 7 | return(NULL); |
11381 | 7 | } |
11382 | | |
11383 | 24.7k | return(ctxt); |
11384 | 24.7k | } |
11385 | | #endif /* LIBXML_PUSH_ENABLED */ |
11386 | | |
11387 | | /** |
11388 | | * Blocks further parser processing |
11389 | | * |
11390 | | * @param ctxt an XML parser context |
11391 | | */ |
11392 | | void |
11393 | 0 | xmlStopParser(xmlParserCtxt *ctxt) { |
11394 | 0 | if (ctxt == NULL) |
11395 | 0 | return; |
11396 | | |
11397 | | /* This stops the parser */ |
11398 | 0 | ctxt->disableSAX = 2; |
11399 | | |
11400 | | /* |
11401 | | * xmlStopParser is often called from error handlers, |
11402 | | * so we can't raise an error here to avoid infinite |
11403 | | * loops. Just make sure that an error condition is |
11404 | | * reported. |
11405 | | */ |
11406 | 0 | if (ctxt->errNo == XML_ERR_OK) { |
11407 | 0 | ctxt->errNo = XML_ERR_USER_STOP; |
11408 | 0 | ctxt->lastError.code = XML_ERR_USER_STOP; |
11409 | 0 | ctxt->wellFormed = 0; |
11410 | 0 | } |
11411 | 0 | } |
11412 | | |
11413 | | /** |
11414 | | * Create a parser context for using the XML parser with an existing |
11415 | | * I/O stream |
11416 | | * |
11417 | | * @param sax a SAX handler (optional) |
11418 | | * @param user_data user data for SAX callbacks (optional) |
11419 | | * @param ioread an I/O read function |
11420 | | * @param ioclose an I/O close function (optional) |
11421 | | * @param ioctx an I/O handler |
11422 | | * @param enc the charset encoding if known (deprecated) |
11423 | | * @returns the new parser context or NULL |
11424 | | */ |
11425 | | xmlParserCtxt * |
11426 | | xmlCreateIOParserCtxt(xmlSAXHandler *sax, void *user_data, |
11427 | | xmlInputReadCallback ioread, |
11428 | | xmlInputCloseCallback ioclose, |
11429 | 0 | void *ioctx, xmlCharEncoding enc) { |
11430 | 0 | xmlParserCtxtPtr ctxt; |
11431 | 0 | xmlParserInputPtr input; |
11432 | 0 | const char *encoding; |
11433 | |
|
11434 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11435 | 0 | if (ctxt == NULL) |
11436 | 0 | return(NULL); |
11437 | | |
11438 | 0 | encoding = xmlGetCharEncodingName(enc); |
11439 | 0 | input = xmlCtxtNewInputFromIO(ctxt, NULL, ioread, ioclose, ioctx, |
11440 | 0 | encoding, 0); |
11441 | 0 | if (input == NULL) { |
11442 | 0 | xmlFreeParserCtxt(ctxt); |
11443 | 0 | return (NULL); |
11444 | 0 | } |
11445 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11446 | 0 | xmlFreeInputStream(input); |
11447 | 0 | xmlFreeParserCtxt(ctxt); |
11448 | 0 | return(NULL); |
11449 | 0 | } |
11450 | | |
11451 | 0 | return(ctxt); |
11452 | 0 | } |
11453 | | |
11454 | | #ifdef LIBXML_VALID_ENABLED |
11455 | | /************************************************************************ |
11456 | | * * |
11457 | | * Front ends when parsing a DTD * |
11458 | | * * |
11459 | | ************************************************************************/ |
11460 | | |
11461 | | /** |
11462 | | * Parse a DTD. |
11463 | | * |
11464 | | * Option XML_PARSE_DTDLOAD should be enabled in the parser context |
11465 | | * to make external entities work. |
11466 | | * |
11467 | | * @since 2.14.0 |
11468 | | * |
11469 | | * @param ctxt a parser context |
11470 | | * @param input a parser input |
11471 | | * @param publicId public ID of the DTD (optional) |
11472 | | * @param systemId system ID of the DTD (optional) |
11473 | | * @returns the resulting xmlDtd or NULL in case of error. |
11474 | | * `input` will be freed by the function in any case. |
11475 | | */ |
11476 | | xmlDtd * |
11477 | | xmlCtxtParseDtd(xmlParserCtxt *ctxt, xmlParserInput *input, |
11478 | 0 | const xmlChar *publicId, const xmlChar *systemId) { |
11479 | 0 | xmlDtdPtr ret = NULL; |
11480 | |
|
11481 | 0 | if ((ctxt == NULL) || (input == NULL)) { |
11482 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
11483 | 0 | xmlFreeInputStream(input); |
11484 | 0 | return(NULL); |
11485 | 0 | } |
11486 | | |
11487 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
11488 | 0 | xmlFreeInputStream(input); |
11489 | 0 | return(NULL); |
11490 | 0 | } |
11491 | | |
11492 | 0 | if (publicId == NULL) |
11493 | 0 | publicId = BAD_CAST "none"; |
11494 | 0 | if (systemId == NULL) |
11495 | 0 | systemId = BAD_CAST "none"; |
11496 | |
|
11497 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
11498 | 0 | if (ctxt->myDoc == NULL) { |
11499 | 0 | xmlErrMemory(ctxt); |
11500 | 0 | goto error; |
11501 | 0 | } |
11502 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
11503 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
11504 | 0 | publicId, systemId); |
11505 | 0 | if (ctxt->myDoc->extSubset == NULL) { |
11506 | 0 | xmlErrMemory(ctxt); |
11507 | 0 | xmlFreeDoc(ctxt->myDoc); |
11508 | 0 | goto error; |
11509 | 0 | } |
11510 | | |
11511 | 0 | xmlParseExternalSubset(ctxt, publicId, systemId); |
11512 | |
|
11513 | 0 | if (ctxt->wellFormed) { |
11514 | 0 | ret = ctxt->myDoc->extSubset; |
11515 | 0 | ctxt->myDoc->extSubset = NULL; |
11516 | 0 | if (ret != NULL) { |
11517 | 0 | xmlNodePtr tmp; |
11518 | |
|
11519 | 0 | ret->doc = NULL; |
11520 | 0 | tmp = ret->children; |
11521 | 0 | while (tmp != NULL) { |
11522 | 0 | tmp->doc = NULL; |
11523 | 0 | tmp = tmp->next; |
11524 | 0 | } |
11525 | 0 | } |
11526 | 0 | } else { |
11527 | 0 | ret = NULL; |
11528 | 0 | } |
11529 | 0 | xmlFreeDoc(ctxt->myDoc); |
11530 | 0 | ctxt->myDoc = NULL; |
11531 | |
|
11532 | 0 | error: |
11533 | 0 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
11534 | |
|
11535 | 0 | return(ret); |
11536 | 0 | } |
11537 | | |
11538 | | /** |
11539 | | * Load and parse a DTD |
11540 | | * |
11541 | | * @deprecated Use #xmlCtxtParseDtd. |
11542 | | * |
11543 | | * @param sax the SAX handler block or NULL |
11544 | | * @param input an Input Buffer |
11545 | | * @param enc the charset encoding if known |
11546 | | * @returns the resulting xmlDtd or NULL in case of error. |
11547 | | * `input` will be freed by the function in any case. |
11548 | | */ |
11549 | | |
11550 | | xmlDtd * |
11551 | | xmlIOParseDTD(xmlSAXHandler *sax, xmlParserInputBuffer *input, |
11552 | 0 | xmlCharEncoding enc) { |
11553 | 0 | xmlDtdPtr ret = NULL; |
11554 | 0 | xmlParserCtxtPtr ctxt; |
11555 | 0 | xmlParserInputPtr pinput = NULL; |
11556 | |
|
11557 | 0 | if (input == NULL) |
11558 | 0 | return(NULL); |
11559 | | |
11560 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11561 | 0 | if (ctxt == NULL) { |
11562 | 0 | xmlFreeParserInputBuffer(input); |
11563 | 0 | return(NULL); |
11564 | 0 | } |
11565 | 0 | xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD); |
11566 | | |
11567 | | /* |
11568 | | * generate a parser input from the I/O handler |
11569 | | */ |
11570 | |
|
11571 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
11572 | 0 | if (pinput == NULL) { |
11573 | 0 | xmlFreeParserInputBuffer(input); |
11574 | 0 | xmlFreeParserCtxt(ctxt); |
11575 | 0 | return(NULL); |
11576 | 0 | } |
11577 | | |
11578 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11579 | 0 | xmlSwitchEncoding(ctxt, enc); |
11580 | 0 | } |
11581 | |
|
11582 | 0 | ret = xmlCtxtParseDtd(ctxt, pinput, NULL, NULL); |
11583 | |
|
11584 | 0 | xmlFreeParserCtxt(ctxt); |
11585 | 0 | return(ret); |
11586 | 0 | } |
11587 | | |
11588 | | /** |
11589 | | * Load and parse an external subset. |
11590 | | * |
11591 | | * @deprecated Use #xmlCtxtParseDtd. |
11592 | | * |
11593 | | * @param sax the SAX handler block |
11594 | | * @param publicId public identifier of the DTD (optional) |
11595 | | * @param systemId system identifier (URL) of the DTD |
11596 | | * @returns the resulting xmlDtd or NULL in case of error. |
11597 | | */ |
11598 | | |
11599 | | xmlDtd * |
11600 | | xmlSAXParseDTD(xmlSAXHandler *sax, const xmlChar *publicId, |
11601 | 0 | const xmlChar *systemId) { |
11602 | 0 | xmlDtdPtr ret = NULL; |
11603 | 0 | xmlParserCtxtPtr ctxt; |
11604 | 0 | xmlParserInputPtr input = NULL; |
11605 | 0 | xmlChar* systemIdCanonic; |
11606 | |
|
11607 | 0 | if ((publicId == NULL) && (systemId == NULL)) return(NULL); |
11608 | | |
11609 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
11610 | 0 | if (ctxt == NULL) { |
11611 | 0 | return(NULL); |
11612 | 0 | } |
11613 | 0 | xmlCtxtSetOptions(ctxt, XML_PARSE_DTDLOAD); |
11614 | | |
11615 | | /* |
11616 | | * Canonicalise the system ID |
11617 | | */ |
11618 | 0 | systemIdCanonic = xmlCanonicPath(systemId); |
11619 | 0 | if ((systemId != NULL) && (systemIdCanonic == NULL)) { |
11620 | 0 | xmlFreeParserCtxt(ctxt); |
11621 | 0 | return(NULL); |
11622 | 0 | } |
11623 | | |
11624 | | /* |
11625 | | * Ask the Entity resolver to load the damn thing |
11626 | | */ |
11627 | | |
11628 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
11629 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, publicId, |
11630 | 0 | systemIdCanonic); |
11631 | 0 | if (input == NULL) { |
11632 | 0 | xmlFreeParserCtxt(ctxt); |
11633 | 0 | if (systemIdCanonic != NULL) |
11634 | 0 | xmlFree(systemIdCanonic); |
11635 | 0 | return(NULL); |
11636 | 0 | } |
11637 | | |
11638 | 0 | if (input->filename == NULL) |
11639 | 0 | input->filename = (char *) systemIdCanonic; |
11640 | 0 | else |
11641 | 0 | xmlFree(systemIdCanonic); |
11642 | |
|
11643 | 0 | ret = xmlCtxtParseDtd(ctxt, input, publicId, systemId); |
11644 | |
|
11645 | 0 | xmlFreeParserCtxt(ctxt); |
11646 | 0 | return(ret); |
11647 | 0 | } |
11648 | | |
11649 | | |
11650 | | /** |
11651 | | * Load and parse an external subset. |
11652 | | * |
11653 | | * @param publicId public identifier of the DTD (optional) |
11654 | | * @param systemId system identifier (URL) of the DTD |
11655 | | * @returns the resulting xmlDtd or NULL in case of error. |
11656 | | */ |
11657 | | |
11658 | | xmlDtd * |
11659 | 0 | xmlParseDTD(const xmlChar *publicId, const xmlChar *systemId) { |
11660 | 0 | return(xmlSAXParseDTD(NULL, publicId, systemId)); |
11661 | 0 | } |
11662 | | #endif /* LIBXML_VALID_ENABLED */ |
11663 | | |
11664 | | /************************************************************************ |
11665 | | * * |
11666 | | * Front ends when parsing an Entity * |
11667 | | * * |
11668 | | ************************************************************************/ |
11669 | | |
11670 | | static xmlNodePtr |
11671 | | xmlCtxtParseContentInternal(xmlParserCtxtPtr ctxt, xmlParserInputPtr input, |
11672 | 73.1k | int hasTextDecl, int buildTree) { |
11673 | 73.1k | xmlNodePtr root = NULL; |
11674 | 73.1k | xmlNodePtr list = NULL; |
11675 | 73.1k | xmlChar *rootName = BAD_CAST "#root"; |
11676 | 73.1k | int result; |
11677 | | |
11678 | 73.1k | if (buildTree) { |
11679 | 73.1k | root = xmlNewDocNode(ctxt->myDoc, NULL, rootName, NULL); |
11680 | 73.1k | if (root == NULL) { |
11681 | 27 | xmlErrMemory(ctxt); |
11682 | 27 | goto error; |
11683 | 27 | } |
11684 | 73.1k | } |
11685 | | |
11686 | 73.1k | if (xmlCtxtPushInput(ctxt, input) < 0) |
11687 | 20 | goto error; |
11688 | | |
11689 | 73.0k | nameNsPush(ctxt, rootName, NULL, NULL, 0, 0); |
11690 | 73.0k | spacePush(ctxt, -1); |
11691 | | |
11692 | 73.0k | if (buildTree) |
11693 | 73.0k | nodePush(ctxt, root); |
11694 | | |
11695 | 73.0k | if (hasTextDecl) { |
11696 | 3.02k | xmlDetectEncoding(ctxt); |
11697 | | |
11698 | | /* |
11699 | | * Parse a possible text declaration first |
11700 | | */ |
11701 | 3.02k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
11702 | 3.02k | (IS_BLANK_CH(NXT(5)))) { |
11703 | 347 | xmlParseTextDecl(ctxt); |
11704 | | /* |
11705 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
11706 | | */ |
11707 | 347 | if ((xmlStrEqual(ctxt->version, BAD_CAST "1.0")) && |
11708 | 347 | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
11709 | 5 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
11710 | 5 | "Version mismatch between document and " |
11711 | 5 | "entity\n"); |
11712 | 5 | } |
11713 | 347 | } |
11714 | 3.02k | } |
11715 | | |
11716 | 73.0k | xmlParseContentInternal(ctxt); |
11717 | | |
11718 | 73.0k | if (ctxt->input->cur < ctxt->input->end) |
11719 | 443 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
11720 | | |
11721 | 73.0k | if ((ctxt->wellFormed) || |
11722 | 73.0k | ((ctxt->recovery) && (!xmlCtxtIsCatastrophicError(ctxt)))) { |
11723 | 72.2k | if (root != NULL) { |
11724 | 72.2k | xmlNodePtr cur; |
11725 | | |
11726 | | /* |
11727 | | * Unlink newly created node list. |
11728 | | */ |
11729 | 72.2k | list = root->children; |
11730 | 72.2k | root->children = NULL; |
11731 | 72.2k | root->last = NULL; |
11732 | 103k | for (cur = list; cur != NULL; cur = cur->next) |
11733 | 31.1k | cur->parent = NULL; |
11734 | 72.2k | } |
11735 | 72.2k | } |
11736 | | |
11737 | | /* |
11738 | | * Read the rest of the stream in case of errors. We want |
11739 | | * to account for the whole entity size. |
11740 | | */ |
11741 | 73.4k | do { |
11742 | 73.4k | ctxt->input->cur = ctxt->input->end; |
11743 | 73.4k | xmlParserShrink(ctxt); |
11744 | 73.4k | result = xmlParserGrow(ctxt); |
11745 | 73.4k | } while (result > 0); |
11746 | | |
11747 | 73.0k | if (buildTree) |
11748 | 73.0k | nodePop(ctxt); |
11749 | | |
11750 | 73.0k | namePop(ctxt); |
11751 | 73.0k | spacePop(ctxt); |
11752 | | |
11753 | 73.0k | xmlCtxtPopInput(ctxt); |
11754 | | |
11755 | 73.1k | error: |
11756 | 73.1k | xmlFreeNode(root); |
11757 | | |
11758 | 73.1k | return(list); |
11759 | 73.0k | } |
11760 | | |
11761 | | static void |
11762 | 74.0k | xmlCtxtParseEntity(xmlParserCtxtPtr ctxt, xmlEntityPtr ent) { |
11763 | 74.0k | xmlParserInputPtr input; |
11764 | 74.0k | xmlNodePtr list; |
11765 | 74.0k | unsigned long consumed; |
11766 | 74.0k | int isExternal; |
11767 | 74.0k | int buildTree; |
11768 | 74.0k | int oldMinNsIndex; |
11769 | 74.0k | int oldNodelen, oldNodemem; |
11770 | | |
11771 | 74.0k | isExternal = (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY); |
11772 | 74.0k | buildTree = (ctxt->node != NULL); |
11773 | | |
11774 | | /* |
11775 | | * Recursion check |
11776 | | */ |
11777 | 74.0k | if (ent->flags & XML_ENT_EXPANDING) { |
11778 | 14 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
11779 | 14 | goto error; |
11780 | 14 | } |
11781 | | |
11782 | | /* |
11783 | | * Load entity |
11784 | | */ |
11785 | 74.0k | input = xmlNewEntityInputStream(ctxt, ent); |
11786 | 74.0k | if (input == NULL) |
11787 | 941 | goto error; |
11788 | | |
11789 | | /* |
11790 | | * When building a tree, we need to limit the scope of namespace |
11791 | | * declarations, so that entities don't reference xmlNs structs |
11792 | | * from the parent of a reference. |
11793 | | */ |
11794 | 73.1k | oldMinNsIndex = ctxt->nsdb->minNsIndex; |
11795 | 73.1k | if (buildTree) |
11796 | 73.1k | ctxt->nsdb->minNsIndex = ctxt->nsNr; |
11797 | | |
11798 | 73.1k | oldNodelen = ctxt->nodelen; |
11799 | 73.1k | oldNodemem = ctxt->nodemem; |
11800 | 73.1k | ctxt->nodelen = 0; |
11801 | 73.1k | ctxt->nodemem = 0; |
11802 | | |
11803 | | /* |
11804 | | * Parse content |
11805 | | * |
11806 | | * This initiates a recursive call chain: |
11807 | | * |
11808 | | * - xmlCtxtParseContentInternal |
11809 | | * - xmlParseContentInternal |
11810 | | * - xmlParseReference |
11811 | | * - xmlCtxtParseEntity |
11812 | | * |
11813 | | * The nesting depth is limited by the maximum number of inputs, |
11814 | | * see xmlCtxtPushInput. |
11815 | | * |
11816 | | * It's possible to make this non-recursive (minNsIndex must be |
11817 | | * stored in the input struct) at the expense of code readability. |
11818 | | */ |
11819 | | |
11820 | 73.1k | ent->flags |= XML_ENT_EXPANDING; |
11821 | | |
11822 | 73.1k | list = xmlCtxtParseContentInternal(ctxt, input, isExternal, buildTree); |
11823 | | |
11824 | 73.1k | ent->flags &= ~XML_ENT_EXPANDING; |
11825 | | |
11826 | 73.1k | ctxt->nsdb->minNsIndex = oldMinNsIndex; |
11827 | 73.1k | ctxt->nodelen = oldNodelen; |
11828 | 73.1k | ctxt->nodemem = oldNodemem; |
11829 | | |
11830 | | /* |
11831 | | * Entity size accounting |
11832 | | */ |
11833 | 73.1k | consumed = input->consumed; |
11834 | 73.1k | xmlSaturatedAddSizeT(&consumed, input->end - input->base); |
11835 | | |
11836 | 73.1k | if ((ent->flags & XML_ENT_CHECKED) == 0) |
11837 | 4.37k | xmlSaturatedAdd(&ent->expandedSize, consumed); |
11838 | | |
11839 | 73.1k | if ((ent->flags & XML_ENT_PARSED) == 0) { |
11840 | 4.37k | if (isExternal) |
11841 | 2.82k | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
11842 | | |
11843 | 4.37k | ent->children = list; |
11844 | | |
11845 | 35.5k | while (list != NULL) { |
11846 | 31.1k | list->parent = (xmlNodePtr) ent; |
11847 | | |
11848 | | /* |
11849 | | * Downstream code like the nginx xslt module can set |
11850 | | * ctxt->myDoc->extSubset to a separate DTD, so the entity |
11851 | | * might have a different or a NULL document. |
11852 | | */ |
11853 | 31.1k | if (list->doc != ent->doc) |
11854 | 0 | xmlSetTreeDoc(list, ent->doc); |
11855 | | |
11856 | 31.1k | if (list->next == NULL) |
11857 | 3.34k | ent->last = list; |
11858 | 31.1k | list = list->next; |
11859 | 31.1k | } |
11860 | 68.7k | } else { |
11861 | 68.7k | xmlFreeNodeList(list); |
11862 | 68.7k | } |
11863 | | |
11864 | 73.1k | xmlFreeInputStream(input); |
11865 | | |
11866 | 74.0k | error: |
11867 | 74.0k | ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED; |
11868 | 74.0k | } |
11869 | | |
11870 | | /** |
11871 | | * Parse an external general entity within an existing parsing context |
11872 | | * An external general parsed entity is well-formed if it matches the |
11873 | | * production labeled extParsedEnt. |
11874 | | * |
11875 | | * [78] extParsedEnt ::= TextDecl? content |
11876 | | * |
11877 | | * @param ctxt the existing parsing context |
11878 | | * @param URL the URL for the entity to load |
11879 | | * @param ID the System ID for the entity to load |
11880 | | * @param listOut the return value for the set of parsed nodes |
11881 | | * @returns 0 if the entity is well formed, -1 in case of args problem and |
11882 | | * the parser error code otherwise |
11883 | | */ |
11884 | | |
11885 | | int |
11886 | | xmlParseCtxtExternalEntity(xmlParserCtxt *ctxt, const xmlChar *URL, |
11887 | 0 | const xmlChar *ID, xmlNode **listOut) { |
11888 | 0 | xmlParserInputPtr input; |
11889 | 0 | xmlNodePtr list; |
11890 | |
|
11891 | 0 | if (listOut != NULL) |
11892 | 0 | *listOut = NULL; |
11893 | |
|
11894 | 0 | if (ctxt == NULL) |
11895 | 0 | return(XML_ERR_ARGUMENT); |
11896 | | |
11897 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
11898 | 0 | XML_RESOURCE_GENERAL_ENTITY); |
11899 | 0 | if (input == NULL) |
11900 | 0 | return(ctxt->errNo); |
11901 | | |
11902 | 0 | xmlCtxtInitializeLate(ctxt); |
11903 | |
|
11904 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 1, 1); |
11905 | 0 | if (listOut != NULL) |
11906 | 0 | *listOut = list; |
11907 | 0 | else |
11908 | 0 | xmlFreeNodeList(list); |
11909 | |
|
11910 | 0 | xmlFreeInputStream(input); |
11911 | 0 | return(ctxt->errNo); |
11912 | 0 | } |
11913 | | |
11914 | | #ifdef LIBXML_SAX1_ENABLED |
11915 | | /** |
11916 | | * Parse an external general entity |
11917 | | * An external general parsed entity is well-formed if it matches the |
11918 | | * production labeled extParsedEnt. |
11919 | | * |
11920 | | * This function uses deprecated global variables to set parser options |
11921 | | * which default to XML_PARSE_NODICT. |
11922 | | * |
11923 | | * @deprecated Use #xmlParseCtxtExternalEntity. |
11924 | | * |
11925 | | * [78] extParsedEnt ::= TextDecl? content |
11926 | | * |
11927 | | * @param doc the document the chunk pertains to |
11928 | | * @param sax the SAX handler block (possibly NULL) |
11929 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
11930 | | * @param depth Used for loop detection, use 0 |
11931 | | * @param URL the URL for the entity to load |
11932 | | * @param ID the System ID for the entity to load |
11933 | | * @param list the return value for the set of parsed nodes |
11934 | | * @returns 0 if the entity is well formed, -1 in case of args problem and |
11935 | | * the parser error code otherwise |
11936 | | */ |
11937 | | |
11938 | | int |
11939 | | xmlParseExternalEntity(xmlDoc *doc, xmlSAXHandler *sax, void *user_data, |
11940 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNode **list) { |
11941 | 0 | xmlParserCtxtPtr ctxt; |
11942 | 0 | int ret; |
11943 | |
|
11944 | 0 | if (list != NULL) |
11945 | 0 | *list = NULL; |
11946 | |
|
11947 | 0 | if (doc == NULL) |
11948 | 0 | return(XML_ERR_ARGUMENT); |
11949 | | |
11950 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
11951 | 0 | if (ctxt == NULL) |
11952 | 0 | return(XML_ERR_NO_MEMORY); |
11953 | | |
11954 | 0 | ctxt->depth = depth; |
11955 | 0 | ctxt->myDoc = doc; |
11956 | 0 | ret = xmlParseCtxtExternalEntity(ctxt, URL, ID, list); |
11957 | |
|
11958 | 0 | xmlFreeParserCtxt(ctxt); |
11959 | 0 | return(ret); |
11960 | 0 | } |
11961 | | |
11962 | | /** |
11963 | | * Parse a well-balanced chunk of an XML document |
11964 | | * called by the parser |
11965 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
11966 | | * the content production in the XML grammar: |
11967 | | * |
11968 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
11969 | | * Comment)* |
11970 | | * |
11971 | | * This function uses deprecated global variables to set parser options |
11972 | | * which default to XML_PARSE_NODICT. |
11973 | | * |
11974 | | * @param doc the document the chunk pertains to (must not be NULL) |
11975 | | * @param sax the SAX handler block (possibly NULL) |
11976 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
11977 | | * @param depth Used for loop detection, use 0 |
11978 | | * @param string the input string in UTF8 or ISO-Latin (zero terminated) |
11979 | | * @param lst the return value for the set of parsed nodes |
11980 | | * @returns 0 if the chunk is well balanced, -1 in case of args problem and |
11981 | | * the parser error code otherwise |
11982 | | */ |
11983 | | |
11984 | | int |
11985 | | xmlParseBalancedChunkMemory(xmlDoc *doc, xmlSAXHandler *sax, |
11986 | 0 | void *user_data, int depth, const xmlChar *string, xmlNode **lst) { |
11987 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
11988 | 0 | depth, string, lst, 0 ); |
11989 | 0 | } |
11990 | | #endif /* LIBXML_SAX1_ENABLED */ |
11991 | | |
11992 | | /** |
11993 | | * Parse a well-balanced chunk of XML matching the 'content' production. |
11994 | | * |
11995 | | * Namespaces in scope of `node` and entities of `node`'s document are |
11996 | | * recognized. When validating, the DTD of `node`'s document is used. |
11997 | | * |
11998 | | * Always consumes `input` even in error case. |
11999 | | * |
12000 | | * @since 2.14.0 |
12001 | | * |
12002 | | * @param ctxt parser context |
12003 | | * @param input parser input |
12004 | | * @param node target node or document |
12005 | | * @param hasTextDecl whether to parse text declaration |
12006 | | * @returns a node list or NULL in case of error. |
12007 | | */ |
12008 | | xmlNode * |
12009 | | xmlCtxtParseContent(xmlParserCtxt *ctxt, xmlParserInput *input, |
12010 | 0 | xmlNode *node, int hasTextDecl) { |
12011 | 0 | xmlDocPtr doc; |
12012 | 0 | xmlNodePtr cur, list = NULL; |
12013 | 0 | int nsnr = 0; |
12014 | 0 | xmlDictPtr oldDict; |
12015 | 0 | int oldOptions, oldDictNames, oldLoadSubset; |
12016 | |
|
12017 | 0 | if ((ctxt == NULL) || (input == NULL) || (node == NULL)) { |
12018 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12019 | 0 | goto exit; |
12020 | 0 | } |
12021 | | |
12022 | 0 | doc = node->doc; |
12023 | 0 | if (doc == NULL) { |
12024 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12025 | 0 | goto exit; |
12026 | 0 | } |
12027 | | |
12028 | 0 | switch (node->type) { |
12029 | 0 | case XML_ELEMENT_NODE: |
12030 | 0 | case XML_DOCUMENT_NODE: |
12031 | 0 | case XML_HTML_DOCUMENT_NODE: |
12032 | 0 | break; |
12033 | | |
12034 | 0 | case XML_ATTRIBUTE_NODE: |
12035 | 0 | case XML_TEXT_NODE: |
12036 | 0 | case XML_CDATA_SECTION_NODE: |
12037 | 0 | case XML_ENTITY_REF_NODE: |
12038 | 0 | case XML_PI_NODE: |
12039 | 0 | case XML_COMMENT_NODE: |
12040 | 0 | for (cur = node->parent; cur != NULL; cur = node->parent) { |
12041 | 0 | if ((cur->type == XML_ELEMENT_NODE) || |
12042 | 0 | (cur->type == XML_DOCUMENT_NODE) || |
12043 | 0 | (cur->type == XML_HTML_DOCUMENT_NODE)) { |
12044 | 0 | node = cur; |
12045 | 0 | break; |
12046 | 0 | } |
12047 | 0 | } |
12048 | 0 | break; |
12049 | | |
12050 | 0 | default: |
12051 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
12052 | 0 | goto exit; |
12053 | 0 | } |
12054 | | |
12055 | 0 | xmlCtxtReset(ctxt); |
12056 | |
|
12057 | 0 | oldDict = ctxt->dict; |
12058 | 0 | oldOptions = ctxt->options; |
12059 | 0 | oldDictNames = ctxt->dictNames; |
12060 | 0 | oldLoadSubset = ctxt->loadsubset; |
12061 | | |
12062 | | /* |
12063 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
12064 | | */ |
12065 | 0 | if (doc->dict != NULL) { |
12066 | 0 | ctxt->dict = doc->dict; |
12067 | 0 | } else { |
12068 | 0 | ctxt->options |= XML_PARSE_NODICT; |
12069 | 0 | ctxt->dictNames = 0; |
12070 | 0 | } |
12071 | | |
12072 | | /* |
12073 | | * Disable IDs |
12074 | | */ |
12075 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
12076 | 0 | ctxt->options |= XML_PARSE_SKIP_IDS; |
12077 | |
|
12078 | 0 | ctxt->myDoc = doc; |
12079 | |
|
12080 | 0 | #ifdef LIBXML_HTML_ENABLED |
12081 | 0 | if (ctxt->html) { |
12082 | | /* |
12083 | | * When parsing in context, it makes no sense to add implied |
12084 | | * elements like html/body/etc... |
12085 | | */ |
12086 | 0 | ctxt->options |= HTML_PARSE_NOIMPLIED; |
12087 | |
|
12088 | 0 | list = htmlCtxtParseContentInternal(ctxt, input); |
12089 | 0 | } else |
12090 | 0 | #endif |
12091 | 0 | { |
12092 | 0 | xmlCtxtInitializeLate(ctxt); |
12093 | | |
12094 | | /* |
12095 | | * initialize the SAX2 namespaces stack |
12096 | | */ |
12097 | 0 | cur = node; |
12098 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
12099 | 0 | xmlNsPtr ns = cur->nsDef; |
12100 | 0 | xmlHashedString hprefix, huri; |
12101 | |
|
12102 | 0 | while (ns != NULL) { |
12103 | 0 | hprefix = xmlDictLookupHashed(ctxt->dict, ns->prefix, -1); |
12104 | 0 | huri = xmlDictLookupHashed(ctxt->dict, ns->href, -1); |
12105 | 0 | if (xmlParserNsPush(ctxt, &hprefix, &huri, ns, 1) > 0) |
12106 | 0 | nsnr++; |
12107 | 0 | ns = ns->next; |
12108 | 0 | } |
12109 | 0 | cur = cur->parent; |
12110 | 0 | } |
12111 | |
|
12112 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, hasTextDecl, 1); |
12113 | |
|
12114 | 0 | if (nsnr > 0) |
12115 | 0 | xmlParserNsPop(ctxt, nsnr); |
12116 | 0 | } |
12117 | |
|
12118 | 0 | ctxt->dict = oldDict; |
12119 | 0 | ctxt->options = oldOptions; |
12120 | 0 | ctxt->dictNames = oldDictNames; |
12121 | 0 | ctxt->loadsubset = oldLoadSubset; |
12122 | 0 | ctxt->myDoc = NULL; |
12123 | 0 | ctxt->node = NULL; |
12124 | |
|
12125 | 0 | exit: |
12126 | 0 | xmlFreeInputStream(input); |
12127 | 0 | return(list); |
12128 | 0 | } |
12129 | | |
12130 | | /** |
12131 | | * Parse a well-balanced chunk of an XML document |
12132 | | * within the context (DTD, namespaces, etc ...) of the given node. |
12133 | | * |
12134 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
12135 | | * the content production in the XML grammar: |
12136 | | * |
12137 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
12138 | | * Comment)* |
12139 | | * |
12140 | | * This function assumes the encoding of `node`'s document which is |
12141 | | * typically not what you want. A better alternative is |
12142 | | * #xmlCtxtParseContent. |
12143 | | * |
12144 | | * @param node the context node |
12145 | | * @param data the input string |
12146 | | * @param datalen the input string length in bytes |
12147 | | * @param options a combination of xmlParserOption |
12148 | | * @param listOut the return value for the set of parsed nodes |
12149 | | * @returns XML_ERR_OK if the chunk is well balanced, and the parser |
12150 | | * error code otherwise |
12151 | | */ |
12152 | | xmlParserErrors |
12153 | | xmlParseInNodeContext(xmlNode *node, const char *data, int datalen, |
12154 | 0 | int options, xmlNode **listOut) { |
12155 | 0 | xmlParserCtxtPtr ctxt; |
12156 | 0 | xmlParserInputPtr input; |
12157 | 0 | xmlDocPtr doc; |
12158 | 0 | xmlNodePtr list; |
12159 | 0 | xmlParserErrors ret; |
12160 | |
|
12161 | 0 | if (listOut == NULL) |
12162 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12163 | 0 | *listOut = NULL; |
12164 | |
|
12165 | 0 | if ((node == NULL) || (data == NULL) || (datalen < 0)) |
12166 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12167 | | |
12168 | 0 | doc = node->doc; |
12169 | 0 | if (doc == NULL) |
12170 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12171 | | |
12172 | 0 | #ifdef LIBXML_HTML_ENABLED |
12173 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) { |
12174 | 0 | ctxt = htmlNewParserCtxt(); |
12175 | 0 | } |
12176 | 0 | else |
12177 | 0 | #endif |
12178 | 0 | ctxt = xmlNewParserCtxt(); |
12179 | |
|
12180 | 0 | if (ctxt == NULL) |
12181 | 0 | return(XML_ERR_NO_MEMORY); |
12182 | | |
12183 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, data, datalen, |
12184 | 0 | (const char *) doc->encoding, |
12185 | 0 | XML_INPUT_BUF_STATIC); |
12186 | 0 | if (input == NULL) { |
12187 | 0 | xmlFreeParserCtxt(ctxt); |
12188 | 0 | return(XML_ERR_NO_MEMORY); |
12189 | 0 | } |
12190 | | |
12191 | 0 | xmlCtxtUseOptions(ctxt, options); |
12192 | |
|
12193 | 0 | list = xmlCtxtParseContent(ctxt, input, node, /* hasTextDecl */ 0); |
12194 | |
|
12195 | 0 | if (list == NULL) { |
12196 | 0 | ret = ctxt->errNo; |
12197 | 0 | if (ret == XML_ERR_ARGUMENT) |
12198 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
12199 | 0 | } else { |
12200 | 0 | ret = XML_ERR_OK; |
12201 | 0 | *listOut = list; |
12202 | 0 | } |
12203 | |
|
12204 | 0 | xmlFreeParserCtxt(ctxt); |
12205 | |
|
12206 | 0 | return(ret); |
12207 | 0 | } |
12208 | | |
12209 | | #ifdef LIBXML_SAX1_ENABLED |
12210 | | /** |
12211 | | * Parse a well-balanced chunk of an XML document |
12212 | | * |
12213 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
12214 | | * the content production in the XML grammar: |
12215 | | * |
12216 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | |
12217 | | * Comment)* |
12218 | | * |
12219 | | * In case recover is set to 1, the nodelist will not be empty even if |
12220 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
12221 | | * some extent. |
12222 | | * |
12223 | | * This function uses deprecated global variables to set parser options |
12224 | | * which default to XML_PARSE_NODICT. |
12225 | | * |
12226 | | * @param doc the document the chunk pertains to (must not be NULL) |
12227 | | * @param sax the SAX handler block (possibly NULL) |
12228 | | * @param user_data The user data returned on SAX callbacks (possibly NULL) |
12229 | | * @param depth Used for loop detection, use 0 |
12230 | | * @param string the input string in UTF8 or ISO-Latin (zero terminated) |
12231 | | * @param listOut the return value for the set of parsed nodes |
12232 | | * @param recover return nodes even if the data is broken (use 0) |
12233 | | * @returns 0 if the chunk is well balanced, or thehe parser error code |
12234 | | * otherwise. |
12235 | | */ |
12236 | | int |
12237 | | xmlParseBalancedChunkMemoryRecover(xmlDoc *doc, xmlSAXHandler *sax, |
12238 | | void *user_data, int depth, const xmlChar *string, xmlNode **listOut, |
12239 | 0 | int recover) { |
12240 | 0 | xmlParserCtxtPtr ctxt; |
12241 | 0 | xmlParserInputPtr input; |
12242 | 0 | xmlNodePtr list; |
12243 | 0 | int ret; |
12244 | |
|
12245 | 0 | if (listOut != NULL) |
12246 | 0 | *listOut = NULL; |
12247 | |
|
12248 | 0 | if (string == NULL) |
12249 | 0 | return(XML_ERR_ARGUMENT); |
12250 | | |
12251 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12252 | 0 | if (ctxt == NULL) |
12253 | 0 | return(XML_ERR_NO_MEMORY); |
12254 | | |
12255 | 0 | xmlCtxtInitializeLate(ctxt); |
12256 | |
|
12257 | 0 | ctxt->depth = depth; |
12258 | 0 | ctxt->myDoc = doc; |
12259 | 0 | if (recover) { |
12260 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12261 | 0 | ctxt->recovery = 1; |
12262 | 0 | } |
12263 | |
|
12264 | 0 | input = xmlNewStringInputStream(ctxt, string); |
12265 | 0 | if (input == NULL) { |
12266 | 0 | ret = ctxt->errNo; |
12267 | 0 | goto error; |
12268 | 0 | } |
12269 | | |
12270 | 0 | list = xmlCtxtParseContentInternal(ctxt, input, /* hasTextDecl */ 0, 1); |
12271 | 0 | if (listOut != NULL) |
12272 | 0 | *listOut = list; |
12273 | 0 | else |
12274 | 0 | xmlFreeNodeList(list); |
12275 | |
|
12276 | 0 | if (!ctxt->wellFormed) |
12277 | 0 | ret = ctxt->errNo; |
12278 | 0 | else |
12279 | 0 | ret = XML_ERR_OK; |
12280 | |
|
12281 | 0 | error: |
12282 | 0 | xmlFreeInputStream(input); |
12283 | 0 | xmlFreeParserCtxt(ctxt); |
12284 | 0 | return(ret); |
12285 | 0 | } |
12286 | | |
12287 | | /** |
12288 | | * Parse an XML external entity out of context and build a tree. |
12289 | | * It use the given SAX function block to handle the parsing callback. |
12290 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12291 | | * |
12292 | | * @deprecated Don't use. |
12293 | | * |
12294 | | * [78] extParsedEnt ::= TextDecl? content |
12295 | | * |
12296 | | * This correspond to a "Well Balanced" chunk |
12297 | | * |
12298 | | * This function uses deprecated global variables to set parser options |
12299 | | * which default to XML_PARSE_NODICT. |
12300 | | * |
12301 | | * @param sax the SAX handler block |
12302 | | * @param filename the filename |
12303 | | * @returns the resulting document tree |
12304 | | */ |
12305 | | |
12306 | | xmlDoc * |
12307 | 0 | xmlSAXParseEntity(xmlSAXHandler *sax, const char *filename) { |
12308 | 0 | xmlDocPtr ret; |
12309 | 0 | xmlParserCtxtPtr ctxt; |
12310 | |
|
12311 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12312 | 0 | if (ctxt == NULL) { |
12313 | 0 | return(NULL); |
12314 | 0 | } |
12315 | 0 | if (sax != NULL) { |
12316 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12317 | 0 | *ctxt->sax = *sax; |
12318 | 0 | } else { |
12319 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12320 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12321 | 0 | } |
12322 | 0 | ctxt->userData = NULL; |
12323 | 0 | } |
12324 | |
|
12325 | 0 | xmlParseExtParsedEnt(ctxt); |
12326 | |
|
12327 | 0 | if (ctxt->wellFormed) { |
12328 | 0 | ret = ctxt->myDoc; |
12329 | 0 | } else { |
12330 | 0 | ret = NULL; |
12331 | 0 | xmlFreeDoc(ctxt->myDoc); |
12332 | 0 | } |
12333 | |
|
12334 | 0 | xmlFreeParserCtxt(ctxt); |
12335 | |
|
12336 | 0 | return(ret); |
12337 | 0 | } |
12338 | | |
12339 | | /** |
12340 | | * Parse an XML external entity out of context and build a tree. |
12341 | | * |
12342 | | * [78] extParsedEnt ::= TextDecl? content |
12343 | | * |
12344 | | * This correspond to a "Well Balanced" chunk |
12345 | | * |
12346 | | * This function uses deprecated global variables to set parser options |
12347 | | * which default to XML_PARSE_NODICT. |
12348 | | * |
12349 | | * @deprecated Don't use. |
12350 | | * |
12351 | | * @param filename the filename |
12352 | | * @returns the resulting document tree |
12353 | | */ |
12354 | | |
12355 | | xmlDoc * |
12356 | 0 | xmlParseEntity(const char *filename) { |
12357 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
12358 | 0 | } |
12359 | | #endif /* LIBXML_SAX1_ENABLED */ |
12360 | | |
12361 | | /** |
12362 | | * Create a parser context for an external entity |
12363 | | * Automatic support for ZLIB/Compress compressed document is provided |
12364 | | * by default if found at compile-time. |
12365 | | * |
12366 | | * @deprecated Don't use. |
12367 | | * |
12368 | | * @param URL the entity URL |
12369 | | * @param ID the entity PUBLIC ID |
12370 | | * @param base a possible base for the target URI |
12371 | | * @returns the new parser context or NULL |
12372 | | */ |
12373 | | xmlParserCtxt * |
12374 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
12375 | 0 | const xmlChar *base) { |
12376 | 0 | xmlParserCtxtPtr ctxt; |
12377 | 0 | xmlParserInputPtr input; |
12378 | 0 | xmlChar *uri = NULL; |
12379 | |
|
12380 | 0 | ctxt = xmlNewParserCtxt(); |
12381 | 0 | if (ctxt == NULL) |
12382 | 0 | return(NULL); |
12383 | | |
12384 | 0 | if (base != NULL) { |
12385 | 0 | if (xmlBuildURISafe(URL, base, &uri) < 0) |
12386 | 0 | goto error; |
12387 | 0 | if (uri != NULL) |
12388 | 0 | URL = uri; |
12389 | 0 | } |
12390 | | |
12391 | 0 | input = xmlLoadResource(ctxt, (char *) URL, (char *) ID, |
12392 | 0 | XML_RESOURCE_UNKNOWN); |
12393 | 0 | if (input == NULL) |
12394 | 0 | goto error; |
12395 | | |
12396 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12397 | 0 | xmlFreeInputStream(input); |
12398 | 0 | goto error; |
12399 | 0 | } |
12400 | | |
12401 | 0 | xmlFree(uri); |
12402 | 0 | return(ctxt); |
12403 | | |
12404 | 0 | error: |
12405 | 0 | xmlFree(uri); |
12406 | 0 | xmlFreeParserCtxt(ctxt); |
12407 | 0 | return(NULL); |
12408 | 0 | } |
12409 | | |
12410 | | /************************************************************************ |
12411 | | * * |
12412 | | * Front ends when parsing from a file * |
12413 | | * * |
12414 | | ************************************************************************/ |
12415 | | |
12416 | | /** |
12417 | | * Create a parser context for a file or URL content. |
12418 | | * Automatic support for ZLIB/Compress compressed document is provided |
12419 | | * by default if found at compile-time and for file accesses |
12420 | | * |
12421 | | * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile. |
12422 | | * |
12423 | | * @param filename the filename or URL |
12424 | | * @param options a combination of xmlParserOption |
12425 | | * @returns the new parser context or NULL |
12426 | | */ |
12427 | | xmlParserCtxt * |
12428 | | xmlCreateURLParserCtxt(const char *filename, int options) |
12429 | 0 | { |
12430 | 0 | xmlParserCtxtPtr ctxt; |
12431 | 0 | xmlParserInputPtr input; |
12432 | |
|
12433 | 0 | ctxt = xmlNewParserCtxt(); |
12434 | 0 | if (ctxt == NULL) |
12435 | 0 | return(NULL); |
12436 | | |
12437 | 0 | options |= XML_PARSE_UNZIP; |
12438 | |
|
12439 | 0 | xmlCtxtUseOptions(ctxt, options); |
12440 | |
|
12441 | 0 | input = xmlLoadResource(ctxt, filename, NULL, XML_RESOURCE_MAIN_DOCUMENT); |
12442 | 0 | if (input == NULL) { |
12443 | 0 | xmlFreeParserCtxt(ctxt); |
12444 | 0 | return(NULL); |
12445 | 0 | } |
12446 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12447 | 0 | xmlFreeInputStream(input); |
12448 | 0 | xmlFreeParserCtxt(ctxt); |
12449 | 0 | return(NULL); |
12450 | 0 | } |
12451 | | |
12452 | 0 | return(ctxt); |
12453 | 0 | } |
12454 | | |
12455 | | /** |
12456 | | * Create a parser context for a file content. |
12457 | | * Automatic support for ZLIB/Compress compressed document is provided |
12458 | | * by default if found at compile-time. |
12459 | | * |
12460 | | * @deprecated Use #xmlNewParserCtxt and #xmlCtxtReadFile. |
12461 | | * |
12462 | | * @param filename the filename |
12463 | | * @returns the new parser context or NULL |
12464 | | */ |
12465 | | xmlParserCtxt * |
12466 | | xmlCreateFileParserCtxt(const char *filename) |
12467 | 0 | { |
12468 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
12469 | 0 | } |
12470 | | |
12471 | | #ifdef LIBXML_SAX1_ENABLED |
12472 | | /** |
12473 | | * Parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12474 | | * compressed document is provided by default if found at compile-time. |
12475 | | * It use the given SAX function block to handle the parsing callback. |
12476 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12477 | | * |
12478 | | * This function uses deprecated global variables to set parser options |
12479 | | * which default to XML_PARSE_NODICT. |
12480 | | * |
12481 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile. |
12482 | | * |
12483 | | * User data (void *) is stored within the parser context in the |
12484 | | * context's _private member, so it is available nearly everywhere in libxml |
12485 | | * |
12486 | | * @param sax the SAX handler block |
12487 | | * @param filename the filename |
12488 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12489 | | * documents |
12490 | | * @param data the userdata |
12491 | | * @returns the resulting document tree |
12492 | | */ |
12493 | | |
12494 | | xmlDoc * |
12495 | | xmlSAXParseFileWithData(xmlSAXHandler *sax, const char *filename, |
12496 | 0 | int recovery, void *data) { |
12497 | 0 | xmlDocPtr ret = NULL; |
12498 | 0 | xmlParserCtxtPtr ctxt; |
12499 | 0 | xmlParserInputPtr input; |
12500 | |
|
12501 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12502 | 0 | if (ctxt == NULL) |
12503 | 0 | return(NULL); |
12504 | | |
12505 | 0 | if (data != NULL) |
12506 | 0 | ctxt->_private = data; |
12507 | |
|
12508 | 0 | if (recovery) { |
12509 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12510 | 0 | ctxt->recovery = 1; |
12511 | 0 | } |
12512 | |
|
12513 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
12514 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, NULL, 0); |
12515 | 0 | else |
12516 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, NULL, 0); |
12517 | |
|
12518 | 0 | if (input != NULL) |
12519 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12520 | |
|
12521 | 0 | xmlFreeParserCtxt(ctxt); |
12522 | 0 | return(ret); |
12523 | 0 | } |
12524 | | |
12525 | | /** |
12526 | | * Parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12527 | | * compressed document is provided by default if found at compile-time. |
12528 | | * It use the given SAX function block to handle the parsing callback. |
12529 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12530 | | * |
12531 | | * This function uses deprecated global variables to set parser options |
12532 | | * which default to XML_PARSE_NODICT. |
12533 | | * |
12534 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile. |
12535 | | * |
12536 | | * @param sax the SAX handler block |
12537 | | * @param filename the filename |
12538 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12539 | | * documents |
12540 | | * @returns the resulting document tree |
12541 | | */ |
12542 | | |
12543 | | xmlDoc * |
12544 | | xmlSAXParseFile(xmlSAXHandler *sax, const char *filename, |
12545 | 0 | int recovery) { |
12546 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
12547 | 0 | } |
12548 | | |
12549 | | /** |
12550 | | * Parse an XML in-memory document and build a tree. |
12551 | | * In the case the document is not Well Formed, a attempt to build a |
12552 | | * tree is tried anyway |
12553 | | * |
12554 | | * This function uses deprecated global variables to set parser options |
12555 | | * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER. |
12556 | | * |
12557 | | * @deprecated Use #xmlReadDoc with XML_PARSE_RECOVER. |
12558 | | * |
12559 | | * @param cur a pointer to an array of xmlChar |
12560 | | * @returns the resulting document tree or NULL in case of failure |
12561 | | */ |
12562 | | |
12563 | | xmlDoc * |
12564 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
12565 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
12566 | 0 | } |
12567 | | |
12568 | | /** |
12569 | | * Parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12570 | | * compressed document is provided by default if found at compile-time. |
12571 | | * |
12572 | | * This function uses deprecated global variables to set parser options |
12573 | | * which default to XML_PARSE_NODICT. |
12574 | | * |
12575 | | * @deprecated Use #xmlReadFile. |
12576 | | * |
12577 | | * @param filename the filename |
12578 | | * @returns the resulting document tree if the file was wellformed, |
12579 | | * NULL otherwise. |
12580 | | */ |
12581 | | |
12582 | | xmlDoc * |
12583 | 0 | xmlParseFile(const char *filename) { |
12584 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
12585 | 0 | } |
12586 | | |
12587 | | /** |
12588 | | * Parse an XML file and build a tree. Automatic support for ZLIB/Compress |
12589 | | * compressed document is provided by default if found at compile-time. |
12590 | | * In the case the document is not Well Formed, it attempts to build |
12591 | | * a tree anyway |
12592 | | * |
12593 | | * This function uses deprecated global variables to set parser options |
12594 | | * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER. |
12595 | | * |
12596 | | * @deprecated Use #xmlReadFile with XML_PARSE_RECOVER. |
12597 | | * |
12598 | | * @param filename the filename |
12599 | | * @returns the resulting document tree or NULL in case of failure |
12600 | | */ |
12601 | | |
12602 | | xmlDoc * |
12603 | 0 | xmlRecoverFile(const char *filename) { |
12604 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
12605 | 0 | } |
12606 | | |
12607 | | |
12608 | | /** |
12609 | | * Setup the parser context to parse a new buffer; Clears any prior |
12610 | | * contents from the parser context. The buffer parameter must not be |
12611 | | * NULL, but the filename parameter can be |
12612 | | * |
12613 | | * @deprecated Don't use. |
12614 | | * |
12615 | | * @param ctxt an XML parser context |
12616 | | * @param buffer a xmlChar * buffer |
12617 | | * @param filename a file name |
12618 | | */ |
12619 | | void |
12620 | | xmlSetupParserForBuffer(xmlParserCtxt *ctxt, const xmlChar* buffer, |
12621 | | const char* filename) |
12622 | 0 | { |
12623 | 0 | xmlParserInputPtr input; |
12624 | |
|
12625 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
12626 | 0 | return; |
12627 | | |
12628 | 0 | xmlCtxtReset(ctxt); |
12629 | |
|
12630 | 0 | input = xmlCtxtNewInputFromString(ctxt, filename, (const char *) buffer, |
12631 | 0 | NULL, 0); |
12632 | 0 | if (input == NULL) |
12633 | 0 | return; |
12634 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) |
12635 | 0 | xmlFreeInputStream(input); |
12636 | 0 | } |
12637 | | |
12638 | | /** |
12639 | | * Parse an XML file and call the given SAX handler routines. |
12640 | | * Automatic support for ZLIB/Compress compressed document is provided |
12641 | | * |
12642 | | * This function uses deprecated global variables to set parser options |
12643 | | * which default to XML_PARSE_NODICT. |
12644 | | * |
12645 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadFile. |
12646 | | * |
12647 | | * @param sax a SAX handler |
12648 | | * @param user_data The user data returned on SAX callbacks |
12649 | | * @param filename a file name |
12650 | | * @returns 0 in case of success or a error number otherwise |
12651 | | */ |
12652 | | int |
12653 | | xmlSAXUserParseFile(xmlSAXHandler *sax, void *user_data, |
12654 | 0 | const char *filename) { |
12655 | 0 | int ret = 0; |
12656 | 0 | xmlParserCtxtPtr ctxt; |
12657 | |
|
12658 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
12659 | 0 | if (ctxt == NULL) return -1; |
12660 | 0 | if (sax != NULL) { |
12661 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12662 | 0 | *ctxt->sax = *sax; |
12663 | 0 | } else { |
12664 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12665 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12666 | 0 | } |
12667 | 0 | ctxt->userData = user_data; |
12668 | 0 | } |
12669 | |
|
12670 | 0 | xmlParseDocument(ctxt); |
12671 | |
|
12672 | 0 | if (ctxt->wellFormed) |
12673 | 0 | ret = 0; |
12674 | 0 | else { |
12675 | 0 | if (ctxt->errNo != 0) |
12676 | 0 | ret = ctxt->errNo; |
12677 | 0 | else |
12678 | 0 | ret = -1; |
12679 | 0 | } |
12680 | 0 | if (ctxt->myDoc != NULL) { |
12681 | 0 | xmlFreeDoc(ctxt->myDoc); |
12682 | 0 | ctxt->myDoc = NULL; |
12683 | 0 | } |
12684 | 0 | xmlFreeParserCtxt(ctxt); |
12685 | |
|
12686 | 0 | return ret; |
12687 | 0 | } |
12688 | | #endif /* LIBXML_SAX1_ENABLED */ |
12689 | | |
12690 | | /************************************************************************ |
12691 | | * * |
12692 | | * Front ends when parsing from memory * |
12693 | | * * |
12694 | | ************************************************************************/ |
12695 | | |
12696 | | /** |
12697 | | * Create a parser context for an XML in-memory document. The input buffer |
12698 | | * must not contain a terminating null byte. |
12699 | | * |
12700 | | * @param buffer a pointer to a char array |
12701 | | * @param size the size of the array |
12702 | | * @returns the new parser context or NULL |
12703 | | */ |
12704 | | xmlParserCtxt * |
12705 | 0 | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
12706 | 0 | xmlParserCtxtPtr ctxt; |
12707 | 0 | xmlParserInputPtr input; |
12708 | |
|
12709 | 0 | if (size < 0) |
12710 | 0 | return(NULL); |
12711 | | |
12712 | 0 | ctxt = xmlNewParserCtxt(); |
12713 | 0 | if (ctxt == NULL) |
12714 | 0 | return(NULL); |
12715 | | |
12716 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, 0); |
12717 | 0 | if (input == NULL) { |
12718 | 0 | xmlFreeParserCtxt(ctxt); |
12719 | 0 | return(NULL); |
12720 | 0 | } |
12721 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12722 | 0 | xmlFreeInputStream(input); |
12723 | 0 | xmlFreeParserCtxt(ctxt); |
12724 | 0 | return(NULL); |
12725 | 0 | } |
12726 | | |
12727 | 0 | return(ctxt); |
12728 | 0 | } |
12729 | | |
12730 | | #ifdef LIBXML_SAX1_ENABLED |
12731 | | /** |
12732 | | * Parse an XML in-memory block and use the given SAX function block |
12733 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12734 | | * DOM tree building routines. |
12735 | | * |
12736 | | * This function uses deprecated global variables to set parser options |
12737 | | * which default to XML_PARSE_NODICT. |
12738 | | * |
12739 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory. |
12740 | | * |
12741 | | * User data (void *) is stored within the parser context in the |
12742 | | * context's _private member, so it is available nearly everywhere in libxml |
12743 | | * |
12744 | | * @param sax the SAX handler block |
12745 | | * @param buffer an pointer to a char array |
12746 | | * @param size the size of the array |
12747 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12748 | | * documents |
12749 | | * @param data the userdata |
12750 | | * @returns the resulting document tree |
12751 | | */ |
12752 | | |
12753 | | xmlDoc * |
12754 | | xmlSAXParseMemoryWithData(xmlSAXHandler *sax, const char *buffer, |
12755 | 0 | int size, int recovery, void *data) { |
12756 | 0 | xmlDocPtr ret = NULL; |
12757 | 0 | xmlParserCtxtPtr ctxt; |
12758 | 0 | xmlParserInputPtr input; |
12759 | |
|
12760 | 0 | if (size < 0) |
12761 | 0 | return(NULL); |
12762 | | |
12763 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12764 | 0 | if (ctxt == NULL) |
12765 | 0 | return(NULL); |
12766 | | |
12767 | 0 | if (data != NULL) |
12768 | 0 | ctxt->_private=data; |
12769 | |
|
12770 | 0 | if (recovery) { |
12771 | 0 | ctxt->options |= XML_PARSE_RECOVER; |
12772 | 0 | ctxt->recovery = 1; |
12773 | 0 | } |
12774 | |
|
12775 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, NULL, buffer, size, NULL, |
12776 | 0 | XML_INPUT_BUF_STATIC); |
12777 | |
|
12778 | 0 | if (input != NULL) |
12779 | 0 | ret = xmlCtxtParseDocument(ctxt, input); |
12780 | |
|
12781 | 0 | xmlFreeParserCtxt(ctxt); |
12782 | 0 | return(ret); |
12783 | 0 | } |
12784 | | |
12785 | | /** |
12786 | | * Parse an XML in-memory block and use the given SAX function block |
12787 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
12788 | | * DOM tree building routines. |
12789 | | * |
12790 | | * This function uses deprecated global variables to set parser options |
12791 | | * which default to XML_PARSE_NODICT. |
12792 | | * |
12793 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory. |
12794 | | * |
12795 | | * @param sax the SAX handler block |
12796 | | * @param buffer an pointer to a char array |
12797 | | * @param size the size of the array |
12798 | | * @param recovery work in recovery mode, i.e. tries to read not Well Formed |
12799 | | * documents |
12800 | | * @returns the resulting document tree |
12801 | | */ |
12802 | | xmlDoc * |
12803 | | xmlSAXParseMemory(xmlSAXHandler *sax, const char *buffer, |
12804 | 0 | int size, int recovery) { |
12805 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
12806 | 0 | } |
12807 | | |
12808 | | /** |
12809 | | * Parse an XML in-memory block and build a tree. |
12810 | | * |
12811 | | * This function uses deprecated global variables to set parser options |
12812 | | * which default to XML_PARSE_NODICT. |
12813 | | * |
12814 | | * @deprecated Use #xmlReadMemory. |
12815 | | * |
12816 | | * @param buffer an pointer to a char array |
12817 | | * @param size the size of the array |
12818 | | * @returns the resulting document tree |
12819 | | */ |
12820 | | |
12821 | 0 | xmlDoc *xmlParseMemory(const char *buffer, int size) { |
12822 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
12823 | 0 | } |
12824 | | |
12825 | | /** |
12826 | | * Parse an XML in-memory block and build a tree. |
12827 | | * In the case the document is not Well Formed, an attempt to |
12828 | | * build a tree is tried anyway |
12829 | | * |
12830 | | * This function uses deprecated global variables to set parser options |
12831 | | * which default to XML_PARSE_NODICT | XML_PARSE_RECOVER. |
12832 | | * |
12833 | | * @deprecated Use #xmlReadMemory with XML_PARSE_RECOVER. |
12834 | | * |
12835 | | * @param buffer an pointer to a char array |
12836 | | * @param size the size of the array |
12837 | | * @returns the resulting document tree or NULL in case of error |
12838 | | */ |
12839 | | |
12840 | 0 | xmlDoc *xmlRecoverMemory(const char *buffer, int size) { |
12841 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
12842 | 0 | } |
12843 | | |
12844 | | /** |
12845 | | * Parse an XML in-memory buffer and call the given SAX handler routines. |
12846 | | * |
12847 | | * This function uses deprecated global variables to set parser options |
12848 | | * which default to XML_PARSE_NODICT. |
12849 | | * |
12850 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadMemory. |
12851 | | * |
12852 | | * @param sax a SAX handler |
12853 | | * @param user_data The user data returned on SAX callbacks |
12854 | | * @param buffer an in-memory XML document input |
12855 | | * @param size the length of the XML document in bytes |
12856 | | * @returns 0 in case of success or a error number otherwise |
12857 | | */ |
12858 | | int xmlSAXUserParseMemory(xmlSAXHandler *sax, void *user_data, |
12859 | 0 | const char *buffer, int size) { |
12860 | 0 | int ret = 0; |
12861 | 0 | xmlParserCtxtPtr ctxt; |
12862 | |
|
12863 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
12864 | 0 | if (ctxt == NULL) return -1; |
12865 | 0 | if (sax != NULL) { |
12866 | 0 | if (sax->initialized == XML_SAX2_MAGIC) { |
12867 | 0 | *ctxt->sax = *sax; |
12868 | 0 | } else { |
12869 | 0 | memset(ctxt->sax, 0, sizeof(*ctxt->sax)); |
12870 | 0 | memcpy(ctxt->sax, sax, sizeof(xmlSAXHandlerV1)); |
12871 | 0 | } |
12872 | 0 | ctxt->userData = user_data; |
12873 | 0 | } |
12874 | |
|
12875 | 0 | xmlParseDocument(ctxt); |
12876 | |
|
12877 | 0 | if (ctxt->wellFormed) |
12878 | 0 | ret = 0; |
12879 | 0 | else { |
12880 | 0 | if (ctxt->errNo != 0) |
12881 | 0 | ret = ctxt->errNo; |
12882 | 0 | else |
12883 | 0 | ret = -1; |
12884 | 0 | } |
12885 | 0 | if (ctxt->myDoc != NULL) { |
12886 | 0 | xmlFreeDoc(ctxt->myDoc); |
12887 | 0 | ctxt->myDoc = NULL; |
12888 | 0 | } |
12889 | 0 | xmlFreeParserCtxt(ctxt); |
12890 | |
|
12891 | 0 | return ret; |
12892 | 0 | } |
12893 | | #endif /* LIBXML_SAX1_ENABLED */ |
12894 | | |
12895 | | /** |
12896 | | * Creates a parser context for an XML in-memory document. |
12897 | | * |
12898 | | * @param str a pointer to an array of xmlChar |
12899 | | * @returns the new parser context or NULL |
12900 | | */ |
12901 | | xmlParserCtxt * |
12902 | 0 | xmlCreateDocParserCtxt(const xmlChar *str) { |
12903 | 0 | xmlParserCtxtPtr ctxt; |
12904 | 0 | xmlParserInputPtr input; |
12905 | |
|
12906 | 0 | ctxt = xmlNewParserCtxt(); |
12907 | 0 | if (ctxt == NULL) |
12908 | 0 | return(NULL); |
12909 | | |
12910 | 0 | input = xmlCtxtNewInputFromString(ctxt, NULL, (const char *) str, NULL, 0); |
12911 | 0 | if (input == NULL) { |
12912 | 0 | xmlFreeParserCtxt(ctxt); |
12913 | 0 | return(NULL); |
12914 | 0 | } |
12915 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
12916 | 0 | xmlFreeInputStream(input); |
12917 | 0 | xmlFreeParserCtxt(ctxt); |
12918 | 0 | return(NULL); |
12919 | 0 | } |
12920 | | |
12921 | 0 | return(ctxt); |
12922 | 0 | } |
12923 | | |
12924 | | #ifdef LIBXML_SAX1_ENABLED |
12925 | | /** |
12926 | | * Parse an XML in-memory document and build a tree. |
12927 | | * It use the given SAX function block to handle the parsing callback. |
12928 | | * If sax is NULL, fallback to the default DOM tree building routines. |
12929 | | * |
12930 | | * This function uses deprecated global variables to set parser options |
12931 | | * which default to XML_PARSE_NODICT. |
12932 | | * |
12933 | | * @deprecated Use #xmlNewSAXParserCtxt and #xmlCtxtReadDoc. |
12934 | | * |
12935 | | * @param sax the SAX handler block |
12936 | | * @param cur a pointer to an array of xmlChar |
12937 | | * @param recovery work in recovery mode, i.e. tries to read no Well Formed |
12938 | | * documents |
12939 | | * @returns the resulting document tree |
12940 | | */ |
12941 | | |
12942 | | xmlDoc * |
12943 | 0 | xmlSAXParseDoc(xmlSAXHandler *sax, const xmlChar *cur, int recovery) { |
12944 | 0 | xmlDocPtr ret; |
12945 | 0 | xmlParserCtxtPtr ctxt; |
12946 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
12947 | |
|
12948 | 0 | if (cur == NULL) return(NULL); |
12949 | | |
12950 | | |
12951 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
12952 | 0 | if (ctxt == NULL) return(NULL); |
12953 | 0 | if (sax != NULL) { |
12954 | 0 | oldsax = ctxt->sax; |
12955 | 0 | ctxt->sax = sax; |
12956 | 0 | ctxt->userData = NULL; |
12957 | 0 | } |
12958 | |
|
12959 | 0 | xmlParseDocument(ctxt); |
12960 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
12961 | 0 | else { |
12962 | 0 | ret = NULL; |
12963 | 0 | xmlFreeDoc(ctxt->myDoc); |
12964 | 0 | ctxt->myDoc = NULL; |
12965 | 0 | } |
12966 | 0 | if (sax != NULL) |
12967 | 0 | ctxt->sax = oldsax; |
12968 | 0 | xmlFreeParserCtxt(ctxt); |
12969 | |
|
12970 | 0 | return(ret); |
12971 | 0 | } |
12972 | | |
12973 | | /** |
12974 | | * Parse an XML in-memory document and build a tree. |
12975 | | * |
12976 | | * This function uses deprecated global variables to set parser options |
12977 | | * which default to XML_PARSE_NODICT. |
12978 | | * |
12979 | | * @deprecated Use #xmlReadDoc. |
12980 | | * |
12981 | | * @param cur a pointer to an array of xmlChar |
12982 | | * @returns the resulting document tree |
12983 | | */ |
12984 | | |
12985 | | xmlDoc * |
12986 | 0 | xmlParseDoc(const xmlChar *cur) { |
12987 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
12988 | 0 | } |
12989 | | #endif /* LIBXML_SAX1_ENABLED */ |
12990 | | |
12991 | | /************************************************************************ |
12992 | | * * |
12993 | | * New set (2.6.0) of simpler and more flexible APIs * |
12994 | | * * |
12995 | | ************************************************************************/ |
12996 | | |
12997 | | /** |
12998 | | * Reset a parser context |
12999 | | * |
13000 | | * @param ctxt an XML parser context |
13001 | | */ |
13002 | | void |
13003 | | xmlCtxtReset(xmlParserCtxt *ctxt) |
13004 | 24.8k | { |
13005 | 24.8k | xmlParserInputPtr input; |
13006 | | |
13007 | 24.8k | if (ctxt == NULL) |
13008 | 0 | return; |
13009 | | |
13010 | 24.8k | while ((input = xmlCtxtPopInput(ctxt)) != NULL) { /* Non consuming */ |
13011 | 0 | xmlFreeInputStream(input); |
13012 | 0 | } |
13013 | 24.8k | ctxt->inputNr = 0; |
13014 | 24.8k | ctxt->input = NULL; |
13015 | | |
13016 | 24.8k | ctxt->spaceNr = 0; |
13017 | 24.8k | if (ctxt->spaceTab != NULL) { |
13018 | 24.8k | ctxt->spaceTab[0] = -1; |
13019 | 24.8k | ctxt->space = &ctxt->spaceTab[0]; |
13020 | 24.8k | } else { |
13021 | 0 | ctxt->space = NULL; |
13022 | 0 | } |
13023 | | |
13024 | | |
13025 | 24.8k | ctxt->nodeNr = 0; |
13026 | 24.8k | ctxt->node = NULL; |
13027 | | |
13028 | 24.8k | ctxt->nameNr = 0; |
13029 | 24.8k | ctxt->name = NULL; |
13030 | | |
13031 | 24.8k | ctxt->nsNr = 0; |
13032 | 24.8k | xmlParserNsReset(ctxt->nsdb); |
13033 | | |
13034 | 24.8k | if (ctxt->version != NULL) { |
13035 | 0 | xmlFree(ctxt->version); |
13036 | 0 | ctxt->version = NULL; |
13037 | 0 | } |
13038 | 24.8k | if (ctxt->encoding != NULL) { |
13039 | 0 | xmlFree(ctxt->encoding); |
13040 | 0 | ctxt->encoding = NULL; |
13041 | 0 | } |
13042 | 24.8k | if (ctxt->extSubURI != NULL) { |
13043 | 0 | xmlFree(ctxt->extSubURI); |
13044 | 0 | ctxt->extSubURI = NULL; |
13045 | 0 | } |
13046 | 24.8k | if (ctxt->extSubSystem != NULL) { |
13047 | 0 | xmlFree(ctxt->extSubSystem); |
13048 | 0 | ctxt->extSubSystem = NULL; |
13049 | 0 | } |
13050 | 24.8k | if (ctxt->directory != NULL) { |
13051 | 0 | xmlFree(ctxt->directory); |
13052 | 0 | ctxt->directory = NULL; |
13053 | 0 | } |
13054 | | |
13055 | 24.8k | if (ctxt->myDoc != NULL) |
13056 | 0 | xmlFreeDoc(ctxt->myDoc); |
13057 | 24.8k | ctxt->myDoc = NULL; |
13058 | | |
13059 | 24.8k | ctxt->standalone = -1; |
13060 | 24.8k | ctxt->hasExternalSubset = 0; |
13061 | 24.8k | ctxt->hasPErefs = 0; |
13062 | 24.8k | ctxt->html = ctxt->html ? 1 : 0; |
13063 | 24.8k | ctxt->instate = XML_PARSER_START; |
13064 | | |
13065 | 24.8k | ctxt->wellFormed = 1; |
13066 | 24.8k | ctxt->nsWellFormed = 1; |
13067 | 24.8k | ctxt->disableSAX = 0; |
13068 | 24.8k | ctxt->valid = 1; |
13069 | 24.8k | ctxt->record_info = 0; |
13070 | 24.8k | ctxt->checkIndex = 0; |
13071 | 24.8k | ctxt->endCheckState = 0; |
13072 | 24.8k | ctxt->inSubset = 0; |
13073 | 24.8k | ctxt->errNo = XML_ERR_OK; |
13074 | 24.8k | ctxt->depth = 0; |
13075 | 24.8k | ctxt->catalogs = NULL; |
13076 | 24.8k | ctxt->sizeentities = 0; |
13077 | 24.8k | ctxt->sizeentcopy = 0; |
13078 | 24.8k | xmlInitNodeInfoSeq(&ctxt->node_seq); |
13079 | | |
13080 | 24.8k | if (ctxt->attsDefault != NULL) { |
13081 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
13082 | 0 | ctxt->attsDefault = NULL; |
13083 | 0 | } |
13084 | 24.8k | if (ctxt->attsSpecial != NULL) { |
13085 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
13086 | 0 | ctxt->attsSpecial = NULL; |
13087 | 0 | } |
13088 | | |
13089 | 24.8k | #ifdef LIBXML_CATALOG_ENABLED |
13090 | 24.8k | if (ctxt->catalogs != NULL) |
13091 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
13092 | 24.8k | #endif |
13093 | 24.8k | ctxt->nbErrors = 0; |
13094 | 24.8k | ctxt->nbWarnings = 0; |
13095 | 24.8k | if (ctxt->lastError.code != XML_ERR_OK) |
13096 | 0 | xmlResetError(&ctxt->lastError); |
13097 | 24.8k | } |
13098 | | |
13099 | | /** |
13100 | | * Reset a push parser context |
13101 | | * |
13102 | | * @param ctxt an XML parser context |
13103 | | * @param chunk a pointer to an array of chars |
13104 | | * @param size number of chars in the array |
13105 | | * @param filename an optional file name or URI |
13106 | | * @param encoding the document encoding, or NULL |
13107 | | * @returns 0 in case of success and 1 in case of error |
13108 | | */ |
13109 | | int |
13110 | | xmlCtxtResetPush(xmlParserCtxt *ctxt, const char *chunk, |
13111 | | int size, const char *filename, const char *encoding) |
13112 | 0 | { |
13113 | 0 | xmlParserInputPtr input; |
13114 | |
|
13115 | 0 | if (ctxt == NULL) |
13116 | 0 | return(1); |
13117 | | |
13118 | 0 | xmlCtxtReset(ctxt); |
13119 | |
|
13120 | 0 | input = xmlNewPushInput(filename, chunk, size); |
13121 | 0 | if (input == NULL) |
13122 | 0 | return(1); |
13123 | | |
13124 | 0 | if (xmlCtxtPushInput(ctxt, input) < 0) { |
13125 | 0 | xmlFreeInputStream(input); |
13126 | 0 | return(1); |
13127 | 0 | } |
13128 | | |
13129 | 0 | if (encoding != NULL) |
13130 | 0 | xmlSwitchEncodingName(ctxt, encoding); |
13131 | |
|
13132 | 0 | return(0); |
13133 | 0 | } |
13134 | | |
13135 | | static int |
13136 | | xmlCtxtSetOptionsInternal(xmlParserCtxtPtr ctxt, int options, int keepMask) |
13137 | 49.5k | { |
13138 | 49.5k | int allMask; |
13139 | | |
13140 | 49.5k | if (ctxt == NULL) |
13141 | 0 | return(-1); |
13142 | | |
13143 | | /* |
13144 | | * XInclude options aren't handled by the parser. |
13145 | | * |
13146 | | * XML_PARSE_XINCLUDE |
13147 | | * XML_PARSE_NOXINCNODE |
13148 | | * XML_PARSE_NOBASEFIX |
13149 | | */ |
13150 | 49.5k | allMask = XML_PARSE_RECOVER | |
13151 | 49.5k | XML_PARSE_NOENT | |
13152 | 49.5k | XML_PARSE_DTDLOAD | |
13153 | 49.5k | XML_PARSE_DTDATTR | |
13154 | 49.5k | XML_PARSE_DTDVALID | |
13155 | 49.5k | XML_PARSE_NOERROR | |
13156 | 49.5k | XML_PARSE_NOWARNING | |
13157 | 49.5k | XML_PARSE_PEDANTIC | |
13158 | 49.5k | XML_PARSE_NOBLANKS | |
13159 | 49.5k | #ifdef LIBXML_SAX1_ENABLED |
13160 | 49.5k | XML_PARSE_SAX1 | |
13161 | 49.5k | #endif |
13162 | 49.5k | XML_PARSE_NONET | |
13163 | 49.5k | XML_PARSE_NODICT | |
13164 | 49.5k | XML_PARSE_NSCLEAN | |
13165 | 49.5k | XML_PARSE_NOCDATA | |
13166 | 49.5k | XML_PARSE_COMPACT | |
13167 | 49.5k | XML_PARSE_OLD10 | |
13168 | 49.5k | XML_PARSE_HUGE | |
13169 | 49.5k | XML_PARSE_OLDSAX | |
13170 | 49.5k | XML_PARSE_IGNORE_ENC | |
13171 | 49.5k | XML_PARSE_BIG_LINES | |
13172 | 49.5k | XML_PARSE_NO_XXE | |
13173 | 49.5k | XML_PARSE_UNZIP | |
13174 | 49.5k | XML_PARSE_NO_SYS_CATALOG | |
13175 | 49.5k | XML_PARSE_CATALOG_PI; |
13176 | | |
13177 | 49.5k | ctxt->options = (ctxt->options & keepMask) | (options & allMask); |
13178 | | |
13179 | | /* |
13180 | | * For some options, struct members are historically the source |
13181 | | * of truth. The values are initalized from global variables and |
13182 | | * old code could also modify them directly. Several older API |
13183 | | * functions that don't take an options argument rely on these |
13184 | | * deprecated mechanisms. |
13185 | | * |
13186 | | * Once public access to struct members and the globals are |
13187 | | * disabled, we can use the options bitmask as source of |
13188 | | * truth, making all these struct members obsolete. |
13189 | | * |
13190 | | * The XML_DETECT_IDS flags is misnamed. It simply enables |
13191 | | * loading of the external subset. |
13192 | | */ |
13193 | 49.5k | ctxt->recovery = (options & XML_PARSE_RECOVER) ? 1 : 0; |
13194 | 49.5k | ctxt->replaceEntities = (options & XML_PARSE_NOENT) ? 1 : 0; |
13195 | 49.5k | ctxt->loadsubset = (options & XML_PARSE_DTDLOAD) ? XML_DETECT_IDS : 0; |
13196 | 49.5k | ctxt->loadsubset |= (options & XML_PARSE_DTDATTR) ? XML_COMPLETE_ATTRS : 0; |
13197 | 49.5k | ctxt->loadsubset |= (options & XML_PARSE_SKIP_IDS) ? XML_SKIP_IDS : 0; |
13198 | 49.5k | ctxt->validate = (options & XML_PARSE_DTDVALID) ? 1 : 0; |
13199 | 49.5k | ctxt->pedantic = (options & XML_PARSE_PEDANTIC) ? 1 : 0; |
13200 | 49.5k | ctxt->keepBlanks = (options & XML_PARSE_NOBLANKS) ? 0 : 1; |
13201 | 49.5k | ctxt->dictNames = (options & XML_PARSE_NODICT) ? 0 : 1; |
13202 | | |
13203 | 49.5k | return(options & ~allMask); |
13204 | 49.5k | } |
13205 | | |
13206 | | /** |
13207 | | * Applies the options to the parser context. Unset options are |
13208 | | * cleared. |
13209 | | * |
13210 | | * @since 2.13.0 |
13211 | | * |
13212 | | * With older versions, you can use #xmlCtxtUseOptions. |
13213 | | * |
13214 | | * @param ctxt an XML parser context |
13215 | | * @param options a bitmask of xmlParserOption values |
13216 | | * @returns 0 in case of success, the set of unknown or unimplemented options |
13217 | | * in case of error. |
13218 | | */ |
13219 | | int |
13220 | | xmlCtxtSetOptions(xmlParserCtxt *ctxt, int options) |
13221 | 0 | { |
13222 | 0 | #ifdef LIBXML_HTML_ENABLED |
13223 | 0 | if ((ctxt != NULL) && (ctxt->html)) |
13224 | 0 | return(htmlCtxtSetOptions(ctxt, options)); |
13225 | 0 | #endif |
13226 | | |
13227 | 0 | return(xmlCtxtSetOptionsInternal(ctxt, options, 0)); |
13228 | 0 | } |
13229 | | |
13230 | | /** |
13231 | | * Get the current options of the parser context. |
13232 | | * |
13233 | | * @since 2.14.0 |
13234 | | * |
13235 | | * @param ctxt an XML parser context |
13236 | | * @returns the current options set in the parser context, or -1 if ctxt is NULL. |
13237 | | */ |
13238 | | int |
13239 | | xmlCtxtGetOptions(xmlParserCtxt *ctxt) |
13240 | 0 | { |
13241 | 0 | if (ctxt == NULL) |
13242 | 0 | return(-1); |
13243 | | |
13244 | 0 | return(ctxt->options); |
13245 | 0 | } |
13246 | | |
13247 | | /** |
13248 | | * Applies the options to the parser context. The following options |
13249 | | * are never cleared and can only be enabled: |
13250 | | * |
13251 | | * - XML_PARSE_NOERROR |
13252 | | * - XML_PARSE_NOWARNING |
13253 | | * - XML_PARSE_NONET |
13254 | | * - XML_PARSE_NSCLEAN |
13255 | | * - XML_PARSE_NOCDATA |
13256 | | * - XML_PARSE_COMPACT |
13257 | | * - XML_PARSE_OLD10 |
13258 | | * - XML_PARSE_HUGE |
13259 | | * - XML_PARSE_OLDSAX |
13260 | | * - XML_PARSE_IGNORE_ENC |
13261 | | * - XML_PARSE_BIG_LINES |
13262 | | * |
13263 | | * @deprecated Use #xmlCtxtSetOptions. |
13264 | | * |
13265 | | * @param ctxt an XML parser context |
13266 | | * @param options a combination of xmlParserOption |
13267 | | * @returns 0 in case of success, the set of unknown or unimplemented options |
13268 | | * in case of error. |
13269 | | */ |
13270 | | int |
13271 | | xmlCtxtUseOptions(xmlParserCtxt *ctxt, int options) |
13272 | 49.5k | { |
13273 | 49.5k | int keepMask; |
13274 | | |
13275 | 49.5k | #ifdef LIBXML_HTML_ENABLED |
13276 | 49.5k | if ((ctxt != NULL) && (ctxt->html)) |
13277 | 0 | return(htmlCtxtUseOptions(ctxt, options)); |
13278 | 49.5k | #endif |
13279 | | |
13280 | | /* |
13281 | | * For historic reasons, some options can only be enabled. |
13282 | | */ |
13283 | 49.5k | keepMask = XML_PARSE_NOERROR | |
13284 | 49.5k | XML_PARSE_NOWARNING | |
13285 | 49.5k | XML_PARSE_NONET | |
13286 | 49.5k | XML_PARSE_NSCLEAN | |
13287 | 49.5k | XML_PARSE_NOCDATA | |
13288 | 49.5k | XML_PARSE_COMPACT | |
13289 | 49.5k | XML_PARSE_OLD10 | |
13290 | 49.5k | XML_PARSE_HUGE | |
13291 | 49.5k | XML_PARSE_OLDSAX | |
13292 | 49.5k | XML_PARSE_IGNORE_ENC | |
13293 | 49.5k | XML_PARSE_BIG_LINES; |
13294 | | |
13295 | 49.5k | return(xmlCtxtSetOptionsInternal(ctxt, options, keepMask)); |
13296 | 49.5k | } |
13297 | | |
13298 | | /** |
13299 | | * To protect against exponential entity expansion ("billion laughs"), the |
13300 | | * size of serialized output is (roughly) limited to the input size |
13301 | | * multiplied by this factor. The default value is 5. |
13302 | | * |
13303 | | * When working with documents making heavy use of entity expansion, it can |
13304 | | * be necessary to increase the value. For security reasons, this should only |
13305 | | * be considered when processing trusted input. |
13306 | | * |
13307 | | * @param ctxt an XML parser context |
13308 | | * @param maxAmpl maximum amplification factor |
13309 | | */ |
13310 | | void |
13311 | | xmlCtxtSetMaxAmplification(xmlParserCtxt *ctxt, unsigned maxAmpl) |
13312 | 0 | { |
13313 | 0 | if (ctxt == NULL) |
13314 | 0 | return; |
13315 | 0 | ctxt->maxAmpl = maxAmpl; |
13316 | 0 | } |
13317 | | |
13318 | | /** |
13319 | | * Parse an XML document and return the resulting document tree. |
13320 | | * Takes ownership of the input object. |
13321 | | * |
13322 | | * @since 2.13.0 |
13323 | | * |
13324 | | * @param ctxt an XML parser context |
13325 | | * @param input parser input |
13326 | | * @returns the resulting document tree or NULL |
13327 | | */ |
13328 | | xmlDoc * |
13329 | | xmlCtxtParseDocument(xmlParserCtxt *ctxt, xmlParserInput *input) |
13330 | 24.7k | { |
13331 | 24.7k | xmlDocPtr ret = NULL; |
13332 | | |
13333 | 24.7k | if ((ctxt == NULL) || (input == NULL)) { |
13334 | 0 | xmlFatalErr(ctxt, XML_ERR_ARGUMENT, NULL); |
13335 | 0 | xmlFreeInputStream(input); |
13336 | 0 | return(NULL); |
13337 | 0 | } |
13338 | | |
13339 | | /* assert(ctxt->inputNr == 0); */ |
13340 | 24.7k | while (ctxt->inputNr > 0) |
13341 | 0 | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
13342 | | |
13343 | 24.7k | if (xmlCtxtPushInput(ctxt, input) < 0) { |
13344 | 8 | xmlFreeInputStream(input); |
13345 | 8 | return(NULL); |
13346 | 8 | } |
13347 | | |
13348 | 24.7k | xmlParseDocument(ctxt); |
13349 | | |
13350 | 24.7k | ret = xmlCtxtGetDocument(ctxt); |
13351 | | |
13352 | | /* assert(ctxt->inputNr == 1); */ |
13353 | 50.2k | while (ctxt->inputNr > 0) |
13354 | 25.4k | xmlFreeInputStream(xmlCtxtPopInput(ctxt)); |
13355 | | |
13356 | 24.7k | return(ret); |
13357 | 24.7k | } |
13358 | | |
13359 | | /** |
13360 | | * Convenience function to parse an XML document from a |
13361 | | * zero-terminated string. |
13362 | | * |
13363 | | * See #xmlCtxtReadDoc for details. |
13364 | | * |
13365 | | * @param cur a pointer to a zero terminated string |
13366 | | * @param URL base URL (optional) |
13367 | | * @param encoding the document encoding (optional) |
13368 | | * @param options a combination of xmlParserOption |
13369 | | * @returns the resulting document tree |
13370 | | */ |
13371 | | xmlDoc * |
13372 | | xmlReadDoc(const xmlChar *cur, const char *URL, const char *encoding, |
13373 | | int options) |
13374 | 0 | { |
13375 | 0 | xmlParserCtxtPtr ctxt; |
13376 | 0 | xmlParserInputPtr input; |
13377 | 0 | xmlDocPtr doc = NULL; |
13378 | |
|
13379 | 0 | ctxt = xmlNewParserCtxt(); |
13380 | 0 | if (ctxt == NULL) |
13381 | 0 | return(NULL); |
13382 | | |
13383 | 0 | xmlCtxtUseOptions(ctxt, options); |
13384 | |
|
13385 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) cur, encoding, |
13386 | 0 | XML_INPUT_BUF_STATIC); |
13387 | |
|
13388 | 0 | if (input != NULL) |
13389 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13390 | |
|
13391 | 0 | xmlFreeParserCtxt(ctxt); |
13392 | 0 | return(doc); |
13393 | 0 | } |
13394 | | |
13395 | | /** |
13396 | | * Convenience function to parse an XML file from the filesystem |
13397 | | * or a global, user-defined resource loader. |
13398 | | * |
13399 | | * This function always enables the XML_PARSE_UNZIP option for |
13400 | | * backward compatibility. If a "-" filename is passed, it will |
13401 | | * read from stdin. Both of these features are potentially |
13402 | | * insecure and might be removed from later versions. |
13403 | | * |
13404 | | * See #xmlCtxtReadFile for details. |
13405 | | * |
13406 | | * @param filename a file or URL |
13407 | | * @param encoding the document encoding (optional) |
13408 | | * @param options a combination of xmlParserOption |
13409 | | * @returns the resulting document tree |
13410 | | */ |
13411 | | xmlDoc * |
13412 | | xmlReadFile(const char *filename, const char *encoding, int options) |
13413 | 0 | { |
13414 | 0 | xmlParserCtxtPtr ctxt; |
13415 | 0 | xmlParserInputPtr input; |
13416 | 0 | xmlDocPtr doc = NULL; |
13417 | |
|
13418 | 0 | ctxt = xmlNewParserCtxt(); |
13419 | 0 | if (ctxt == NULL) |
13420 | 0 | return(NULL); |
13421 | | |
13422 | 0 | options |= XML_PARSE_UNZIP; |
13423 | |
|
13424 | 0 | xmlCtxtUseOptions(ctxt, options); |
13425 | | |
13426 | | /* |
13427 | | * Backward compatibility for users of command line utilities like |
13428 | | * xmlstarlet expecting "-" to mean stdin. This is dangerous and |
13429 | | * should be removed at some point. |
13430 | | */ |
13431 | 0 | if ((filename != NULL) && (filename[0] == '-') && (filename[1] == 0)) |
13432 | 0 | input = xmlCtxtNewInputFromFd(ctxt, filename, STDIN_FILENO, |
13433 | 0 | encoding, 0); |
13434 | 0 | else |
13435 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13436 | |
|
13437 | 0 | if (input != NULL) |
13438 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13439 | |
|
13440 | 0 | xmlFreeParserCtxt(ctxt); |
13441 | 0 | return(doc); |
13442 | 0 | } |
13443 | | |
13444 | | /** |
13445 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13446 | | * not contain a terminating null byte. |
13447 | | * |
13448 | | * See #xmlCtxtReadMemory for details. |
13449 | | * |
13450 | | * @param buffer a pointer to a char array |
13451 | | * @param size the size of the array |
13452 | | * @param url base URL (optional) |
13453 | | * @param encoding the document encoding (optional) |
13454 | | * @param options a combination of xmlParserOption |
13455 | | * @returns the resulting document tree |
13456 | | */ |
13457 | | xmlDoc * |
13458 | | xmlReadMemory(const char *buffer, int size, const char *url, |
13459 | | const char *encoding, int options) |
13460 | 0 | { |
13461 | 0 | xmlParserCtxtPtr ctxt; |
13462 | 0 | xmlParserInputPtr input; |
13463 | 0 | xmlDocPtr doc = NULL; |
13464 | |
|
13465 | 0 | if (size < 0) |
13466 | 0 | return(NULL); |
13467 | | |
13468 | 0 | ctxt = xmlNewParserCtxt(); |
13469 | 0 | if (ctxt == NULL) |
13470 | 0 | return(NULL); |
13471 | | |
13472 | 0 | xmlCtxtUseOptions(ctxt, options); |
13473 | |
|
13474 | 0 | input = xmlCtxtNewInputFromMemory(ctxt, url, buffer, size, encoding, |
13475 | 0 | XML_INPUT_BUF_STATIC); |
13476 | |
|
13477 | 0 | if (input != NULL) |
13478 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13479 | |
|
13480 | 0 | xmlFreeParserCtxt(ctxt); |
13481 | 0 | return(doc); |
13482 | 0 | } |
13483 | | |
13484 | | /** |
13485 | | * Parse an XML from a file descriptor and build a tree. |
13486 | | * |
13487 | | * See #xmlCtxtReadFd for details. |
13488 | | * |
13489 | | * NOTE that the file descriptor will not be closed when the |
13490 | | * context is freed or reset. |
13491 | | * |
13492 | | * @param fd an open file descriptor |
13493 | | * @param URL base URL (optional) |
13494 | | * @param encoding the document encoding (optional) |
13495 | | * @param options a combination of xmlParserOption |
13496 | | * @returns the resulting document tree |
13497 | | */ |
13498 | | xmlDoc * |
13499 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
13500 | 0 | { |
13501 | 0 | xmlParserCtxtPtr ctxt; |
13502 | 0 | xmlParserInputPtr input; |
13503 | 0 | xmlDocPtr doc = NULL; |
13504 | |
|
13505 | 0 | ctxt = xmlNewParserCtxt(); |
13506 | 0 | if (ctxt == NULL) |
13507 | 0 | return(NULL); |
13508 | | |
13509 | 0 | xmlCtxtUseOptions(ctxt, options); |
13510 | |
|
13511 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
13512 | |
|
13513 | 0 | if (input != NULL) |
13514 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13515 | |
|
13516 | 0 | xmlFreeParserCtxt(ctxt); |
13517 | 0 | return(doc); |
13518 | 0 | } |
13519 | | |
13520 | | /** |
13521 | | * Parse an XML document from I/O functions and context and build a tree. |
13522 | | * |
13523 | | * See #xmlCtxtReadIO for details. |
13524 | | * |
13525 | | * @param ioread an I/O read function |
13526 | | * @param ioclose an I/O close function (optional) |
13527 | | * @param ioctx an I/O handler |
13528 | | * @param URL base URL (optional) |
13529 | | * @param encoding the document encoding (optional) |
13530 | | * @param options a combination of xmlParserOption |
13531 | | * @returns the resulting document tree |
13532 | | */ |
13533 | | xmlDoc * |
13534 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
13535 | | void *ioctx, const char *URL, const char *encoding, int options) |
13536 | 0 | { |
13537 | 0 | xmlParserCtxtPtr ctxt; |
13538 | 0 | xmlParserInputPtr input; |
13539 | 0 | xmlDocPtr doc = NULL; |
13540 | |
|
13541 | 0 | ctxt = xmlNewParserCtxt(); |
13542 | 0 | if (ctxt == NULL) |
13543 | 0 | return(NULL); |
13544 | | |
13545 | 0 | xmlCtxtUseOptions(ctxt, options); |
13546 | |
|
13547 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
13548 | 0 | encoding, 0); |
13549 | |
|
13550 | 0 | if (input != NULL) |
13551 | 0 | doc = xmlCtxtParseDocument(ctxt, input); |
13552 | |
|
13553 | 0 | xmlFreeParserCtxt(ctxt); |
13554 | 0 | return(doc); |
13555 | 0 | } |
13556 | | |
13557 | | /** |
13558 | | * Parse an XML in-memory document and build a tree. |
13559 | | * |
13560 | | * `URL` is used as base to resolve external entities and for error |
13561 | | * reporting. |
13562 | | * |
13563 | | * @param ctxt an XML parser context |
13564 | | * @param str a pointer to a zero terminated string |
13565 | | * @param URL base URL (optional) |
13566 | | * @param encoding the document encoding (optional) |
13567 | | * @param options a combination of xmlParserOption |
13568 | | * @returns the resulting document tree |
13569 | | */ |
13570 | | xmlDoc * |
13571 | | xmlCtxtReadDoc(xmlParserCtxt *ctxt, const xmlChar *str, |
13572 | | const char *URL, const char *encoding, int options) |
13573 | 0 | { |
13574 | 0 | xmlParserInputPtr input; |
13575 | |
|
13576 | 0 | if (ctxt == NULL) |
13577 | 0 | return(NULL); |
13578 | | |
13579 | 0 | xmlCtxtReset(ctxt); |
13580 | 0 | xmlCtxtUseOptions(ctxt, options); |
13581 | |
|
13582 | 0 | input = xmlCtxtNewInputFromString(ctxt, URL, (const char *) str, encoding, |
13583 | 0 | XML_INPUT_BUF_STATIC); |
13584 | 0 | if (input == NULL) |
13585 | 0 | return(NULL); |
13586 | | |
13587 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13588 | 0 | } |
13589 | | |
13590 | | /** |
13591 | | * Parse an XML file from the filesystem or a global, user-defined |
13592 | | * resource loader. |
13593 | | * |
13594 | | * This function always enables the XML_PARSE_UNZIP option for |
13595 | | * backward compatibility. This feature is potentially insecure |
13596 | | * and might be removed from later versions. |
13597 | | * |
13598 | | * @param ctxt an XML parser context |
13599 | | * @param filename a file or URL |
13600 | | * @param encoding the document encoding (optional) |
13601 | | * @param options a combination of xmlParserOption |
13602 | | * @returns the resulting document tree |
13603 | | */ |
13604 | | xmlDoc * |
13605 | | xmlCtxtReadFile(xmlParserCtxt *ctxt, const char *filename, |
13606 | | const char *encoding, int options) |
13607 | 0 | { |
13608 | 0 | xmlParserInputPtr input; |
13609 | |
|
13610 | 0 | if (ctxt == NULL) |
13611 | 0 | return(NULL); |
13612 | | |
13613 | 0 | options |= XML_PARSE_UNZIP; |
13614 | |
|
13615 | 0 | xmlCtxtReset(ctxt); |
13616 | 0 | xmlCtxtUseOptions(ctxt, options); |
13617 | |
|
13618 | 0 | input = xmlCtxtNewInputFromUrl(ctxt, filename, NULL, encoding, 0); |
13619 | 0 | if (input == NULL) |
13620 | 0 | return(NULL); |
13621 | | |
13622 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13623 | 0 | } |
13624 | | |
13625 | | /** |
13626 | | * Parse an XML in-memory document and build a tree. The input buffer must |
13627 | | * not contain a terminating null byte. |
13628 | | * |
13629 | | * `URL` is used as base to resolve external entities and for error |
13630 | | * reporting. |
13631 | | * |
13632 | | * @param ctxt an XML parser context |
13633 | | * @param buffer a pointer to a char array |
13634 | | * @param size the size of the array |
13635 | | * @param URL base URL (optional) |
13636 | | * @param encoding the document encoding (optional) |
13637 | | * @param options a combination of xmlParserOption |
13638 | | * @returns the resulting document tree |
13639 | | */ |
13640 | | xmlDoc * |
13641 | | xmlCtxtReadMemory(xmlParserCtxt *ctxt, const char *buffer, int size, |
13642 | | const char *URL, const char *encoding, int options) |
13643 | 24.8k | { |
13644 | 24.8k | xmlParserInputPtr input; |
13645 | | |
13646 | 24.8k | if ((ctxt == NULL) || (size < 0)) |
13647 | 0 | return(NULL); |
13648 | | |
13649 | 24.8k | xmlCtxtReset(ctxt); |
13650 | 24.8k | xmlCtxtUseOptions(ctxt, options); |
13651 | | |
13652 | 24.8k | input = xmlCtxtNewInputFromMemory(ctxt, URL, buffer, size, encoding, |
13653 | 24.8k | XML_INPUT_BUF_STATIC); |
13654 | 24.8k | if (input == NULL) |
13655 | 22 | return(NULL); |
13656 | | |
13657 | 24.7k | return(xmlCtxtParseDocument(ctxt, input)); |
13658 | 24.8k | } |
13659 | | |
13660 | | /** |
13661 | | * Parse an XML document from a file descriptor and build a tree. |
13662 | | * |
13663 | | * NOTE that the file descriptor will not be closed when the |
13664 | | * context is freed or reset. |
13665 | | * |
13666 | | * `URL` is used as base to resolve external entities and for error |
13667 | | * reporting. |
13668 | | * |
13669 | | * @param ctxt an XML parser context |
13670 | | * @param fd an open file descriptor |
13671 | | * @param URL base URL (optional) |
13672 | | * @param encoding the document encoding (optional) |
13673 | | * @param options a combination of xmlParserOption |
13674 | | * @returns the resulting document tree |
13675 | | */ |
13676 | | xmlDoc * |
13677 | | xmlCtxtReadFd(xmlParserCtxt *ctxt, int fd, |
13678 | | const char *URL, const char *encoding, int options) |
13679 | 0 | { |
13680 | 0 | xmlParserInputPtr input; |
13681 | |
|
13682 | 0 | if (ctxt == NULL) |
13683 | 0 | return(NULL); |
13684 | | |
13685 | 0 | xmlCtxtReset(ctxt); |
13686 | 0 | xmlCtxtUseOptions(ctxt, options); |
13687 | |
|
13688 | 0 | input = xmlCtxtNewInputFromFd(ctxt, URL, fd, encoding, 0); |
13689 | 0 | if (input == NULL) |
13690 | 0 | return(NULL); |
13691 | | |
13692 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13693 | 0 | } |
13694 | | |
13695 | | /** |
13696 | | * Parse an XML document from I/O functions and source and build a tree. |
13697 | | * This reuses the existing `ctxt` parser context |
13698 | | * |
13699 | | * `URL` is used as base to resolve external entities and for error |
13700 | | * reporting. |
13701 | | * |
13702 | | * @param ctxt an XML parser context |
13703 | | * @param ioread an I/O read function |
13704 | | * @param ioclose an I/O close function |
13705 | | * @param ioctx an I/O handler |
13706 | | * @param URL the base URL to use for the document |
13707 | | * @param encoding the document encoding, or NULL |
13708 | | * @param options a combination of xmlParserOption |
13709 | | * @returns the resulting document tree |
13710 | | */ |
13711 | | xmlDoc * |
13712 | | xmlCtxtReadIO(xmlParserCtxt *ctxt, xmlInputReadCallback ioread, |
13713 | | xmlInputCloseCallback ioclose, void *ioctx, |
13714 | | const char *URL, |
13715 | | const char *encoding, int options) |
13716 | 0 | { |
13717 | 0 | xmlParserInputPtr input; |
13718 | |
|
13719 | 0 | if (ctxt == NULL) |
13720 | 0 | return(NULL); |
13721 | | |
13722 | 0 | xmlCtxtReset(ctxt); |
13723 | 0 | xmlCtxtUseOptions(ctxt, options); |
13724 | |
|
13725 | 0 | input = xmlCtxtNewInputFromIO(ctxt, URL, ioread, ioclose, ioctx, |
13726 | 0 | encoding, 0); |
13727 | 0 | if (input == NULL) |
13728 | 0 | return(NULL); |
13729 | | |
13730 | 0 | return(xmlCtxtParseDocument(ctxt, input)); |
13731 | 0 | } |
13732 | | |