Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * parser.c : an XML 1.0 parser, namespaces and validity support are mostly |
3 | | * implemented on top of the SAX interfaces |
4 | | * |
5 | | * References: |
6 | | * The XML specification: |
7 | | * http://www.w3.org/TR/REC-xml |
8 | | * Original 1.0 version: |
9 | | * http://www.w3.org/TR/1998/REC-xml-19980210 |
10 | | * XML second edition working draft |
11 | | * http://www.w3.org/TR/2000/WD-xml-2e-20000814 |
12 | | * |
13 | | * Okay this is a big file, the parser core is around 7000 lines, then it |
14 | | * is followed by the progressive parser top routines, then the various |
15 | | * high level APIs to call the parser and a few miscellaneous functions. |
16 | | * A number of helper functions and deprecated ones have been moved to |
17 | | * parserInternals.c to reduce this file size. |
18 | | * As much as possible the functions are associated with their relative |
19 | | * production in the XML specification. A few productions defining the |
20 | | * different ranges of character are actually implanted either in |
21 | | * parserInternals.h or parserInternals.c |
22 | | * The DOM tree build is realized from the default SAX callbacks in |
23 | | * the module SAX.c. |
24 | | * The routines doing the validation checks are in valid.c and called either |
25 | | * from the SAX callbacks or as standalone functions using a preparsed |
26 | | * document. |
27 | | * |
28 | | * See Copyright for the status of this software. |
29 | | * |
30 | | * daniel@veillard.com |
31 | | */ |
32 | | |
33 | | /* To avoid EBCDIC trouble when parsing on zOS */ |
34 | | #if defined(__MVS__) |
35 | | #pragma convert("ISO8859-1") |
36 | | #endif |
37 | | |
38 | | #define IN_LIBXML |
39 | | #include "libxml.h" |
40 | | |
41 | | #if defined(_WIN32) |
42 | | #define XML_DIR_SEP '\\' |
43 | | #else |
44 | | #define XML_DIR_SEP '/' |
45 | | #endif |
46 | | |
47 | | #include <stdlib.h> |
48 | | #include <limits.h> |
49 | | #include <string.h> |
50 | | #include <stdarg.h> |
51 | | #include <stddef.h> |
52 | | #include <ctype.h> |
53 | | #include <stdlib.h> |
54 | | #include <libxml/xmlmemory.h> |
55 | | #include <libxml/threads.h> |
56 | | #include <libxml/globals.h> |
57 | | #include <libxml/tree.h> |
58 | | #include <libxml/parser.h> |
59 | | #include <libxml/parserInternals.h> |
60 | | #include <libxml/HTMLparser.h> |
61 | | #include <libxml/valid.h> |
62 | | #include <libxml/entities.h> |
63 | | #include <libxml/xmlerror.h> |
64 | | #include <libxml/encoding.h> |
65 | | #include <libxml/xmlIO.h> |
66 | | #include <libxml/uri.h> |
67 | | #ifdef LIBXML_CATALOG_ENABLED |
68 | | #include <libxml/catalog.h> |
69 | | #endif |
70 | | #ifdef LIBXML_SCHEMAS_ENABLED |
71 | | #include <libxml/xmlschemastypes.h> |
72 | | #include <libxml/relaxng.h> |
73 | | #endif |
74 | | #if defined(LIBXML_XPATH_ENABLED) || defined(LIBXML_SCHEMAS_ENABLED) |
75 | | #include <libxml/xpath.h> |
76 | | #endif |
77 | | |
78 | | #include "private/buf.h" |
79 | | #include "private/dict.h" |
80 | | #include "private/enc.h" |
81 | | #include "private/entities.h" |
82 | | #include "private/error.h" |
83 | | #include "private/globals.h" |
84 | | #include "private/html.h" |
85 | | #include "private/io.h" |
86 | | #include "private/memory.h" |
87 | | #include "private/parser.h" |
88 | | #include "private/threads.h" |
89 | | #include "private/xpath.h" |
90 | | |
91 | | struct _xmlStartTag { |
92 | | const xmlChar *prefix; |
93 | | const xmlChar *URI; |
94 | | int line; |
95 | | int nsNr; |
96 | | }; |
97 | | |
98 | | static xmlParserCtxtPtr |
99 | | xmlCreateEntityParserCtxtInternal(xmlSAXHandlerPtr sax, void *userData, |
100 | | const xmlChar *URL, const xmlChar *ID, const xmlChar *base, |
101 | | xmlParserCtxtPtr pctx); |
102 | | |
103 | | static void xmlHaltParser(xmlParserCtxtPtr ctxt); |
104 | | |
105 | | static int |
106 | | xmlParseElementStart(xmlParserCtxtPtr ctxt); |
107 | | |
108 | | static void |
109 | | xmlParseElementEnd(xmlParserCtxtPtr ctxt); |
110 | | |
111 | | /************************************************************************ |
112 | | * * |
113 | | * Arbitrary limits set in the parser. See XML_PARSE_HUGE * |
114 | | * * |
115 | | ************************************************************************/ |
116 | | |
117 | 10.1M | #define XML_MAX_HUGE_LENGTH 1000000000 |
118 | | |
119 | | #define XML_PARSER_BIG_ENTITY 1000 |
120 | | #define XML_PARSER_LOT_ENTITY 5000 |
121 | | |
122 | | /* |
123 | | * XML_PARSER_NON_LINEAR is the threshold where the ratio of parsed entity |
124 | | * replacement over the size in byte of the input indicates that you have |
125 | | * and exponential behaviour. A value of 10 correspond to at least 3 entity |
126 | | * replacement per byte of input. |
127 | | */ |
128 | 766 | #define XML_PARSER_NON_LINEAR 10 |
129 | | |
130 | 61.1M | #define XML_ENT_FIXED_COST 50 |
131 | | |
132 | | /** |
133 | | * xmlParserMaxDepth: |
134 | | * |
135 | | * arbitrary depth limit for the XML documents that we allow to |
136 | | * process. This is not a limitation of the parser but a safety |
137 | | * boundary feature. It can be disabled with the XML_PARSE_HUGE |
138 | | * parser option. |
139 | | */ |
140 | | unsigned int xmlParserMaxDepth = 256; |
141 | | |
142 | | |
143 | | |
144 | | #define SAX2 1 |
145 | 72.0M | #define XML_PARSER_BIG_BUFFER_SIZE 300 |
146 | 6.88G | #define XML_PARSER_BUFFER_SIZE 100 |
147 | 620k | #define SAX_COMPAT_MODE BAD_CAST "SAX compatibility mode document" |
148 | | |
149 | | /** |
150 | | * XML_PARSER_CHUNK_SIZE |
151 | | * |
152 | | * When calling GROW that's the minimal amount of data |
153 | | * the parser expected to have received. It is not a hard |
154 | | * limit but an optimization when reading strings like Names |
155 | | * It is not strictly needed as long as inputs available characters |
156 | | * are followed by 0, which should be provided by the I/O level |
157 | | */ |
158 | 35.6M | #define XML_PARSER_CHUNK_SIZE 100 |
159 | | |
160 | | /* |
161 | | * List of XML prefixed PI allowed by W3C specs |
162 | | */ |
163 | | |
164 | | static const char* const xmlW3CPIs[] = { |
165 | | "xml-stylesheet", |
166 | | "xml-model", |
167 | | NULL |
168 | | }; |
169 | | |
170 | | |
171 | | /* DEPR void xmlParserHandleReference(xmlParserCtxtPtr ctxt); */ |
172 | | static xmlEntityPtr xmlParseStringPEReference(xmlParserCtxtPtr ctxt, |
173 | | const xmlChar **str); |
174 | | |
175 | | static xmlParserErrors |
176 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
177 | | xmlSAXHandlerPtr sax, |
178 | | void *user_data, int depth, const xmlChar *URL, |
179 | | const xmlChar *ID, xmlNodePtr *list); |
180 | | |
181 | | static int |
182 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, |
183 | | const char *encoding); |
184 | | #ifdef LIBXML_LEGACY_ENABLED |
185 | | static void |
186 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
187 | | xmlNodePtr lastNode); |
188 | | #endif /* LIBXML_LEGACY_ENABLED */ |
189 | | |
190 | | static xmlParserErrors |
191 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
192 | | const xmlChar *string, void *user_data, xmlNodePtr *lst); |
193 | | |
194 | | static int |
195 | | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity); |
196 | | |
197 | | /************************************************************************ |
198 | | * * |
199 | | * Some factorized error routines * |
200 | | * * |
201 | | ************************************************************************/ |
202 | | |
203 | | /** |
204 | | * xmlErrAttributeDup: |
205 | | * @ctxt: an XML parser context |
206 | | * @prefix: the attribute prefix |
207 | | * @localname: the attribute localname |
208 | | * |
209 | | * Handle a redefinition of attribute error |
210 | | */ |
211 | | static void |
212 | | xmlErrAttributeDup(xmlParserCtxtPtr ctxt, const xmlChar * prefix, |
213 | | const xmlChar * localname) |
214 | 31.4k | { |
215 | 31.4k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
216 | 31.4k | (ctxt->instate == XML_PARSER_EOF)) |
217 | 114 | return; |
218 | 31.3k | if (ctxt != NULL) |
219 | 31.3k | ctxt->errNo = XML_ERR_ATTRIBUTE_REDEFINED; |
220 | | |
221 | 31.3k | if (prefix == NULL) |
222 | 22.4k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
223 | 22.4k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
224 | 22.4k | (const char *) localname, NULL, NULL, 0, 0, |
225 | 22.4k | "Attribute %s redefined\n", localname); |
226 | 8.85k | else |
227 | 8.85k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, |
228 | 8.85k | XML_ERR_ATTRIBUTE_REDEFINED, XML_ERR_FATAL, NULL, 0, |
229 | 8.85k | (const char *) prefix, (const char *) localname, |
230 | 8.85k | NULL, 0, 0, "Attribute %s:%s redefined\n", prefix, |
231 | 8.85k | localname); |
232 | 31.3k | if (ctxt != NULL) { |
233 | 31.3k | ctxt->wellFormed = 0; |
234 | 31.3k | if (ctxt->recovery == 0) |
235 | 15.0k | ctxt->disableSAX = 1; |
236 | 31.3k | } |
237 | 31.3k | } |
238 | | |
239 | | /** |
240 | | * xmlFatalErr: |
241 | | * @ctxt: an XML parser context |
242 | | * @error: the error number |
243 | | * @extra: extra information string |
244 | | * |
245 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
246 | | */ |
247 | | static void |
248 | | xmlFatalErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, const char *info) |
249 | 2.05M | { |
250 | 2.05M | const char *errmsg; |
251 | | |
252 | 2.05M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
253 | 2.05M | (ctxt->instate == XML_PARSER_EOF)) |
254 | 13.1k | return; |
255 | 2.04M | switch (error) { |
256 | 21.2k | case XML_ERR_INVALID_HEX_CHARREF: |
257 | 21.2k | errmsg = "CharRef: invalid hexadecimal value"; |
258 | 21.2k | break; |
259 | 37.1k | case XML_ERR_INVALID_DEC_CHARREF: |
260 | 37.1k | errmsg = "CharRef: invalid decimal value"; |
261 | 37.1k | break; |
262 | 0 | case XML_ERR_INVALID_CHARREF: |
263 | 0 | errmsg = "CharRef: invalid value"; |
264 | 0 | break; |
265 | 465k | case XML_ERR_INTERNAL_ERROR: |
266 | 465k | errmsg = "internal error"; |
267 | 465k | break; |
268 | 0 | case XML_ERR_PEREF_AT_EOF: |
269 | 0 | errmsg = "PEReference at end of document"; |
270 | 0 | break; |
271 | 0 | case XML_ERR_PEREF_IN_PROLOG: |
272 | 0 | errmsg = "PEReference in prolog"; |
273 | 0 | break; |
274 | 0 | case XML_ERR_PEREF_IN_EPILOG: |
275 | 0 | errmsg = "PEReference in epilog"; |
276 | 0 | break; |
277 | 0 | case XML_ERR_PEREF_NO_NAME: |
278 | 0 | errmsg = "PEReference: no name"; |
279 | 0 | break; |
280 | 765k | case XML_ERR_PEREF_SEMICOL_MISSING: |
281 | 765k | errmsg = "PEReference: expecting ';'"; |
282 | 765k | break; |
283 | 1.53k | case XML_ERR_ENTITY_LOOP: |
284 | 1.53k | errmsg = "Detected an entity reference loop"; |
285 | 1.53k | break; |
286 | 0 | case XML_ERR_ENTITY_NOT_STARTED: |
287 | 0 | errmsg = "EntityValue: \" or ' expected"; |
288 | 0 | break; |
289 | 6.03k | case XML_ERR_ENTITY_PE_INTERNAL: |
290 | 6.03k | errmsg = "PEReferences forbidden in internal subset"; |
291 | 6.03k | break; |
292 | 1.85k | case XML_ERR_ENTITY_NOT_FINISHED: |
293 | 1.85k | errmsg = "EntityValue: \" or ' expected"; |
294 | 1.85k | break; |
295 | 32.2k | case XML_ERR_ATTRIBUTE_NOT_STARTED: |
296 | 32.2k | errmsg = "AttValue: \" or ' expected"; |
297 | 32.2k | break; |
298 | 94.5k | case XML_ERR_LT_IN_ATTRIBUTE: |
299 | 94.5k | errmsg = "Unescaped '<' not allowed in attributes values"; |
300 | 94.5k | break; |
301 | 12.1k | case XML_ERR_LITERAL_NOT_STARTED: |
302 | 12.1k | errmsg = "SystemLiteral \" or ' expected"; |
303 | 12.1k | break; |
304 | 13.9k | case XML_ERR_LITERAL_NOT_FINISHED: |
305 | 13.9k | errmsg = "Unfinished System or Public ID \" or ' expected"; |
306 | 13.9k | break; |
307 | 14.4k | case XML_ERR_MISPLACED_CDATA_END: |
308 | 14.4k | errmsg = "Sequence ']]>' not allowed in content"; |
309 | 14.4k | break; |
310 | 10.5k | case XML_ERR_URI_REQUIRED: |
311 | 10.5k | errmsg = "SYSTEM or PUBLIC, the URI is missing"; |
312 | 10.5k | break; |
313 | 1.84k | case XML_ERR_PUBID_REQUIRED: |
314 | 1.84k | errmsg = "PUBLIC, the Public Identifier is missing"; |
315 | 1.84k | break; |
316 | 22.4k | case XML_ERR_HYPHEN_IN_COMMENT: |
317 | 22.4k | errmsg = "Comment must not contain '--' (double-hyphen)"; |
318 | 22.4k | break; |
319 | 12.1k | case XML_ERR_PI_NOT_STARTED: |
320 | 12.1k | errmsg = "xmlParsePI : no target name"; |
321 | 12.1k | break; |
322 | 3.16k | case XML_ERR_RESERVED_XML_NAME: |
323 | 3.16k | errmsg = "Invalid PI name"; |
324 | 3.16k | break; |
325 | 2.08k | case XML_ERR_NOTATION_NOT_STARTED: |
326 | 2.08k | errmsg = "NOTATION: Name expected here"; |
327 | 2.08k | break; |
328 | 10.4k | case XML_ERR_NOTATION_NOT_FINISHED: |
329 | 10.4k | errmsg = "'>' required to close NOTATION declaration"; |
330 | 10.4k | break; |
331 | 8.21k | case XML_ERR_VALUE_REQUIRED: |
332 | 8.21k | errmsg = "Entity value required"; |
333 | 8.21k | break; |
334 | 2.36k | case XML_ERR_URI_FRAGMENT: |
335 | 2.36k | errmsg = "Fragment not allowed"; |
336 | 2.36k | break; |
337 | 8.67k | case XML_ERR_ATTLIST_NOT_STARTED: |
338 | 8.67k | errmsg = "'(' required to start ATTLIST enumeration"; |
339 | 8.67k | break; |
340 | 1.66k | case XML_ERR_NMTOKEN_REQUIRED: |
341 | 1.66k | errmsg = "NmToken expected in ATTLIST enumeration"; |
342 | 1.66k | break; |
343 | 3.41k | case XML_ERR_ATTLIST_NOT_FINISHED: |
344 | 3.41k | errmsg = "')' required to finish ATTLIST enumeration"; |
345 | 3.41k | break; |
346 | 2.40k | case XML_ERR_MIXED_NOT_STARTED: |
347 | 2.40k | errmsg = "MixedContentDecl : '|' or ')*' expected"; |
348 | 2.40k | break; |
349 | 0 | case XML_ERR_PCDATA_REQUIRED: |
350 | 0 | errmsg = "MixedContentDecl : '#PCDATA' expected"; |
351 | 0 | break; |
352 | 8.08k | case XML_ERR_ELEMCONTENT_NOT_STARTED: |
353 | 8.08k | errmsg = "ContentDecl : Name or '(' expected"; |
354 | 8.08k | break; |
355 | 9.21k | case XML_ERR_ELEMCONTENT_NOT_FINISHED: |
356 | 9.21k | errmsg = "ContentDecl : ',' '|' or ')' expected"; |
357 | 9.21k | break; |
358 | 0 | case XML_ERR_PEREF_IN_INT_SUBSET: |
359 | 0 | errmsg = |
360 | 0 | "PEReference: forbidden within markup decl in internal subset"; |
361 | 0 | break; |
362 | 91.3k | case XML_ERR_GT_REQUIRED: |
363 | 91.3k | errmsg = "expected '>'"; |
364 | 91.3k | break; |
365 | 126 | case XML_ERR_CONDSEC_INVALID: |
366 | 126 | errmsg = "XML conditional section '[' expected"; |
367 | 126 | break; |
368 | 8.99k | case XML_ERR_EXT_SUBSET_NOT_FINISHED: |
369 | 8.99k | errmsg = "Content error in the external subset"; |
370 | 8.99k | break; |
371 | 828 | case XML_ERR_CONDSEC_INVALID_KEYWORD: |
372 | 828 | errmsg = |
373 | 828 | "conditional section INCLUDE or IGNORE keyword expected"; |
374 | 828 | break; |
375 | 1.35k | case XML_ERR_CONDSEC_NOT_FINISHED: |
376 | 1.35k | errmsg = "XML conditional section not closed"; |
377 | 1.35k | break; |
378 | 196 | case XML_ERR_XMLDECL_NOT_STARTED: |
379 | 196 | errmsg = "Text declaration '<?xml' required"; |
380 | 196 | break; |
381 | 76.5k | case XML_ERR_XMLDECL_NOT_FINISHED: |
382 | 76.5k | errmsg = "parsing XML declaration: '?>' expected"; |
383 | 76.5k | break; |
384 | 0 | case XML_ERR_EXT_ENTITY_STANDALONE: |
385 | 0 | errmsg = "external parsed entities cannot be standalone"; |
386 | 0 | break; |
387 | 89.8k | case XML_ERR_ENTITYREF_SEMICOL_MISSING: |
388 | 89.8k | errmsg = "EntityRef: expecting ';'"; |
389 | 89.8k | break; |
390 | 6.74k | case XML_ERR_DOCTYPE_NOT_FINISHED: |
391 | 6.74k | errmsg = "DOCTYPE improperly terminated"; |
392 | 6.74k | break; |
393 | 0 | case XML_ERR_LTSLASH_REQUIRED: |
394 | 0 | errmsg = "EndTag: '</' not found"; |
395 | 0 | break; |
396 | 5.81k | case XML_ERR_EQUAL_REQUIRED: |
397 | 5.81k | errmsg = "expected '='"; |
398 | 5.81k | break; |
399 | 18.2k | case XML_ERR_STRING_NOT_CLOSED: |
400 | 18.2k | errmsg = "String not closed expecting \" or '"; |
401 | 18.2k | break; |
402 | 5.15k | case XML_ERR_STRING_NOT_STARTED: |
403 | 5.15k | errmsg = "String not started expecting ' or \""; |
404 | 5.15k | break; |
405 | 2.28k | case XML_ERR_ENCODING_NAME: |
406 | 2.28k | errmsg = "Invalid XML encoding name"; |
407 | 2.28k | break; |
408 | 720 | case XML_ERR_STANDALONE_VALUE: |
409 | 720 | errmsg = "standalone accepts only 'yes' or 'no'"; |
410 | 720 | break; |
411 | 16.8k | case XML_ERR_DOCUMENT_EMPTY: |
412 | 16.8k | errmsg = "Document is empty"; |
413 | 16.8k | break; |
414 | 104k | case XML_ERR_DOCUMENT_END: |
415 | 104k | errmsg = "Extra content at the end of the document"; |
416 | 104k | break; |
417 | 17.4k | case XML_ERR_NOT_WELL_BALANCED: |
418 | 17.4k | errmsg = "chunk is not well balanced"; |
419 | 17.4k | break; |
420 | 0 | case XML_ERR_EXTRA_CONTENT: |
421 | 0 | errmsg = "extra content at the end of well balanced chunk"; |
422 | 0 | break; |
423 | 18.8k | case XML_ERR_VERSION_MISSING: |
424 | 18.8k | errmsg = "Malformed declaration expecting version"; |
425 | 18.8k | break; |
426 | 157 | case XML_ERR_NAME_TOO_LONG: |
427 | 157 | errmsg = "Name too long"; |
428 | 157 | break; |
429 | | #if 0 |
430 | | case: |
431 | | errmsg = ""; |
432 | | break; |
433 | | #endif |
434 | 2.22k | default: |
435 | 2.22k | errmsg = "Unregistered error message"; |
436 | 2.04M | } |
437 | 2.04M | if (ctxt != NULL) |
438 | 2.04M | ctxt->errNo = error; |
439 | 2.04M | if (info == NULL) { |
440 | 1.57M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
441 | 1.57M | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s\n", |
442 | 1.57M | errmsg); |
443 | 1.57M | } else { |
444 | 465k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
445 | 465k | XML_ERR_FATAL, NULL, 0, info, NULL, NULL, 0, 0, "%s: %s\n", |
446 | 465k | errmsg, info); |
447 | 465k | } |
448 | 2.04M | if (ctxt != NULL) { |
449 | 2.04M | ctxt->wellFormed = 0; |
450 | 2.04M | if (ctxt->recovery == 0) |
451 | 604k | ctxt->disableSAX = 1; |
452 | 2.04M | } |
453 | 2.04M | } |
454 | | |
455 | | /** |
456 | | * xmlFatalErrMsg: |
457 | | * @ctxt: an XML parser context |
458 | | * @error: the error number |
459 | | * @msg: the error message |
460 | | * |
461 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
462 | | */ |
463 | | static void LIBXML_ATTR_FORMAT(3,0) |
464 | | xmlFatalErrMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
465 | | const char *msg) |
466 | 1.62M | { |
467 | 1.62M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
468 | 1.62M | (ctxt->instate == XML_PARSER_EOF)) |
469 | 485 | return; |
470 | 1.62M | if (ctxt != NULL) |
471 | 1.62M | ctxt->errNo = error; |
472 | 1.62M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_PARSER, error, |
473 | 1.62M | XML_ERR_FATAL, NULL, 0, NULL, NULL, NULL, 0, 0, "%s", msg); |
474 | 1.62M | if (ctxt != NULL) { |
475 | 1.62M | ctxt->wellFormed = 0; |
476 | 1.62M | if (ctxt->recovery == 0) |
477 | 473k | ctxt->disableSAX = 1; |
478 | 1.62M | } |
479 | 1.62M | } |
480 | | |
481 | | /** |
482 | | * xmlWarningMsg: |
483 | | * @ctxt: an XML parser context |
484 | | * @error: the error number |
485 | | * @msg: the error message |
486 | | * @str1: extra data |
487 | | * @str2: extra data |
488 | | * |
489 | | * Handle a warning. |
490 | | */ |
491 | | static void LIBXML_ATTR_FORMAT(3,0) |
492 | | xmlWarningMsg(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
493 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
494 | 5.75M | { |
495 | 5.75M | xmlStructuredErrorFunc schannel = NULL; |
496 | | |
497 | 5.75M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
498 | 5.75M | (ctxt->instate == XML_PARSER_EOF)) |
499 | 0 | return; |
500 | 5.75M | if ((ctxt != NULL) && (ctxt->sax != NULL) && |
501 | 5.75M | (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
502 | 3.97M | schannel = ctxt->sax->serror; |
503 | 5.75M | if (ctxt != NULL) { |
504 | 5.75M | __xmlRaiseError(schannel, |
505 | 5.75M | (ctxt->sax) ? ctxt->sax->warning : NULL, |
506 | 5.75M | ctxt->userData, |
507 | 5.75M | ctxt, NULL, XML_FROM_PARSER, error, |
508 | 5.75M | XML_ERR_WARNING, NULL, 0, |
509 | 5.75M | (const char *) str1, (const char *) str2, NULL, 0, 0, |
510 | 5.75M | msg, (const char *) str1, (const char *) str2); |
511 | 5.75M | } else { |
512 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
513 | 0 | ctxt, NULL, XML_FROM_PARSER, error, |
514 | 0 | XML_ERR_WARNING, NULL, 0, |
515 | 0 | (const char *) str1, (const char *) str2, NULL, 0, 0, |
516 | 0 | msg, (const char *) str1, (const char *) str2); |
517 | 0 | } |
518 | 5.75M | } |
519 | | |
520 | | /** |
521 | | * xmlValidityError: |
522 | | * @ctxt: an XML parser context |
523 | | * @error: the error number |
524 | | * @msg: the error message |
525 | | * @str1: extra data |
526 | | * |
527 | | * Handle a validity error. |
528 | | */ |
529 | | static void LIBXML_ATTR_FORMAT(3,0) |
530 | | xmlValidityError(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
531 | | const char *msg, const xmlChar *str1, const xmlChar *str2) |
532 | 1.48M | { |
533 | 1.48M | xmlStructuredErrorFunc schannel = NULL; |
534 | | |
535 | 1.48M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
536 | 1.48M | (ctxt->instate == XML_PARSER_EOF)) |
537 | 0 | return; |
538 | 1.48M | if (ctxt != NULL) { |
539 | 1.48M | ctxt->errNo = error; |
540 | 1.48M | if ((ctxt->sax != NULL) && (ctxt->sax->initialized == XML_SAX2_MAGIC)) |
541 | 643k | schannel = ctxt->sax->serror; |
542 | 1.48M | } |
543 | 1.48M | if (ctxt != NULL) { |
544 | 1.48M | __xmlRaiseError(schannel, |
545 | 1.48M | ctxt->vctxt.error, ctxt->vctxt.userData, |
546 | 1.48M | ctxt, NULL, XML_FROM_DTD, error, |
547 | 1.48M | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
548 | 1.48M | (const char *) str2, NULL, 0, 0, |
549 | 1.48M | msg, (const char *) str1, (const char *) str2); |
550 | 1.48M | ctxt->valid = 0; |
551 | 1.48M | } else { |
552 | 0 | __xmlRaiseError(schannel, NULL, NULL, |
553 | 0 | ctxt, NULL, XML_FROM_DTD, error, |
554 | 0 | XML_ERR_ERROR, NULL, 0, (const char *) str1, |
555 | 0 | (const char *) str2, NULL, 0, 0, |
556 | 0 | msg, (const char *) str1, (const char *) str2); |
557 | 0 | } |
558 | 1.48M | } |
559 | | |
560 | | /** |
561 | | * xmlFatalErrMsgInt: |
562 | | * @ctxt: an XML parser context |
563 | | * @error: the error number |
564 | | * @msg: the error message |
565 | | * @val: an integer value |
566 | | * |
567 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
568 | | */ |
569 | | static void LIBXML_ATTR_FORMAT(3,0) |
570 | | xmlFatalErrMsgInt(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
571 | | const char *msg, int val) |
572 | 1.37M | { |
573 | 1.37M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
574 | 1.37M | (ctxt->instate == XML_PARSER_EOF)) |
575 | 0 | return; |
576 | 1.37M | if (ctxt != NULL) |
577 | 1.37M | ctxt->errNo = error; |
578 | 1.37M | __xmlRaiseError(NULL, NULL, NULL, |
579 | 1.37M | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
580 | 1.37M | NULL, 0, NULL, NULL, NULL, val, 0, msg, val); |
581 | 1.37M | if (ctxt != NULL) { |
582 | 1.37M | ctxt->wellFormed = 0; |
583 | 1.37M | if (ctxt->recovery == 0) |
584 | 251k | ctxt->disableSAX = 1; |
585 | 1.37M | } |
586 | 1.37M | } |
587 | | |
588 | | /** |
589 | | * xmlFatalErrMsgStrIntStr: |
590 | | * @ctxt: an XML parser context |
591 | | * @error: the error number |
592 | | * @msg: the error message |
593 | | * @str1: an string info |
594 | | * @val: an integer value |
595 | | * @str2: an string info |
596 | | * |
597 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
598 | | */ |
599 | | static void LIBXML_ATTR_FORMAT(3,0) |
600 | | xmlFatalErrMsgStrIntStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
601 | | const char *msg, const xmlChar *str1, int val, |
602 | | const xmlChar *str2) |
603 | 512k | { |
604 | 512k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
605 | 512k | (ctxt->instate == XML_PARSER_EOF)) |
606 | 0 | return; |
607 | 512k | if (ctxt != NULL) |
608 | 512k | ctxt->errNo = error; |
609 | 512k | __xmlRaiseError(NULL, NULL, NULL, |
610 | 512k | ctxt, NULL, XML_FROM_PARSER, error, XML_ERR_FATAL, |
611 | 512k | NULL, 0, (const char *) str1, (const char *) str2, |
612 | 512k | NULL, val, 0, msg, str1, val, str2); |
613 | 512k | if (ctxt != NULL) { |
614 | 512k | ctxt->wellFormed = 0; |
615 | 512k | if (ctxt->recovery == 0) |
616 | 175k | ctxt->disableSAX = 1; |
617 | 512k | } |
618 | 512k | } |
619 | | |
620 | | /** |
621 | | * xmlFatalErrMsgStr: |
622 | | * @ctxt: an XML parser context |
623 | | * @error: the error number |
624 | | * @msg: the error message |
625 | | * @val: a string value |
626 | | * |
627 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
628 | | */ |
629 | | static void LIBXML_ATTR_FORMAT(3,0) |
630 | | xmlFatalErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
631 | | const char *msg, const xmlChar * val) |
632 | 3.43M | { |
633 | 3.43M | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
634 | 3.43M | (ctxt->instate == XML_PARSER_EOF)) |
635 | 46 | return; |
636 | 3.43M | if (ctxt != NULL) |
637 | 3.43M | ctxt->errNo = error; |
638 | 3.43M | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
639 | 3.43M | XML_FROM_PARSER, error, XML_ERR_FATAL, |
640 | 3.43M | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
641 | 3.43M | val); |
642 | 3.43M | if (ctxt != NULL) { |
643 | 3.43M | ctxt->wellFormed = 0; |
644 | 3.43M | if (ctxt->recovery == 0) |
645 | 508k | ctxt->disableSAX = 1; |
646 | 3.43M | } |
647 | 3.43M | } |
648 | | |
649 | | /** |
650 | | * xmlErrMsgStr: |
651 | | * @ctxt: an XML parser context |
652 | | * @error: the error number |
653 | | * @msg: the error message |
654 | | * @val: a string value |
655 | | * |
656 | | * Handle a non fatal parser error |
657 | | */ |
658 | | static void LIBXML_ATTR_FORMAT(3,0) |
659 | | xmlErrMsgStr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
660 | | const char *msg, const xmlChar * val) |
661 | 443k | { |
662 | 443k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
663 | 443k | (ctxt->instate == XML_PARSER_EOF)) |
664 | 0 | return; |
665 | 443k | if (ctxt != NULL) |
666 | 443k | ctxt->errNo = error; |
667 | 443k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, |
668 | 443k | XML_FROM_PARSER, error, XML_ERR_ERROR, |
669 | 443k | NULL, 0, (const char *) val, NULL, NULL, 0, 0, msg, |
670 | 443k | val); |
671 | 443k | } |
672 | | |
673 | | /** |
674 | | * xmlNsErr: |
675 | | * @ctxt: an XML parser context |
676 | | * @error: the error number |
677 | | * @msg: the message |
678 | | * @info1: extra information string |
679 | | * @info2: extra information string |
680 | | * |
681 | | * Handle a fatal parser error, i.e. violating Well-Formedness constraints |
682 | | */ |
683 | | static void LIBXML_ATTR_FORMAT(3,0) |
684 | | xmlNsErr(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
685 | | const char *msg, |
686 | | const xmlChar * info1, const xmlChar * info2, |
687 | | const xmlChar * info3) |
688 | 392k | { |
689 | 392k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
690 | 392k | (ctxt->instate == XML_PARSER_EOF)) |
691 | 133 | return; |
692 | 392k | if (ctxt != NULL) |
693 | 392k | ctxt->errNo = error; |
694 | 392k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
695 | 392k | XML_ERR_ERROR, NULL, 0, (const char *) info1, |
696 | 392k | (const char *) info2, (const char *) info3, 0, 0, msg, |
697 | 392k | info1, info2, info3); |
698 | 392k | if (ctxt != NULL) |
699 | 392k | ctxt->nsWellFormed = 0; |
700 | 392k | } |
701 | | |
702 | | /** |
703 | | * xmlNsWarn |
704 | | * @ctxt: an XML parser context |
705 | | * @error: the error number |
706 | | * @msg: the message |
707 | | * @info1: extra information string |
708 | | * @info2: extra information string |
709 | | * |
710 | | * Handle a namespace warning error |
711 | | */ |
712 | | static void LIBXML_ATTR_FORMAT(3,0) |
713 | | xmlNsWarn(xmlParserCtxtPtr ctxt, xmlParserErrors error, |
714 | | const char *msg, |
715 | | const xmlChar * info1, const xmlChar * info2, |
716 | | const xmlChar * info3) |
717 | 8.98k | { |
718 | 8.98k | if ((ctxt != NULL) && (ctxt->disableSAX != 0) && |
719 | 8.98k | (ctxt->instate == XML_PARSER_EOF)) |
720 | 0 | return; |
721 | 8.98k | __xmlRaiseError(NULL, NULL, NULL, ctxt, NULL, XML_FROM_NAMESPACE, error, |
722 | 8.98k | XML_ERR_WARNING, NULL, 0, (const char *) info1, |
723 | 8.98k | (const char *) info2, (const char *) info3, 0, 0, msg, |
724 | 8.98k | info1, info2, info3); |
725 | 8.98k | } |
726 | | |
727 | | static void |
728 | 217M | xmlSaturatedAdd(unsigned long *dst, unsigned long val) { |
729 | 217M | if (val > ULONG_MAX - *dst) |
730 | 0 | *dst = ULONG_MAX; |
731 | 217M | else |
732 | 217M | *dst += val; |
733 | 217M | } |
734 | | |
735 | | static void |
736 | 62.4M | xmlSaturatedAddSizeT(unsigned long *dst, unsigned long val) { |
737 | 62.4M | if (val > ULONG_MAX - *dst) |
738 | 0 | *dst = ULONG_MAX; |
739 | 62.4M | else |
740 | 62.4M | *dst += val; |
741 | 62.4M | } |
742 | | |
743 | | /** |
744 | | * xmlParserEntityCheck: |
745 | | * @ctxt: parser context |
746 | | * @extra: sum of unexpanded entity sizes |
747 | | * |
748 | | * Check for non-linear entity expansion behaviour. |
749 | | * |
750 | | * In some cases like xmlStringDecodeEntities, this function is called |
751 | | * for each, possibly nested entity and its unexpanded content length. |
752 | | * |
753 | | * In other cases like xmlParseReference, it's only called for each |
754 | | * top-level entity with its unexpanded content length plus the sum of |
755 | | * the unexpanded content lengths (plus fixed cost) of all nested |
756 | | * entities. |
757 | | * |
758 | | * Summing the unexpanded lengths also adds the length of the reference. |
759 | | * This is by design. Taking the length of the entity name into account |
760 | | * discourages attacks that try to waste CPU time with abusively long |
761 | | * entity names. See test/recurse/lol6.xml for example. Each call also |
762 | | * adds some fixed cost XML_ENT_FIXED_COST to discourage attacks with |
763 | | * short entities. |
764 | | * |
765 | | * Returns 1 on error, 0 on success. |
766 | | */ |
767 | | static int |
768 | | xmlParserEntityCheck(xmlParserCtxtPtr ctxt, unsigned long extra) |
769 | 61.1M | { |
770 | 61.1M | unsigned long consumed; |
771 | 61.1M | xmlParserInputPtr input = ctxt->input; |
772 | 61.1M | xmlEntityPtr entity = input->entity; |
773 | | |
774 | | /* |
775 | | * Compute total consumed bytes so far, including input streams of |
776 | | * external entities. |
777 | | */ |
778 | 61.1M | consumed = input->parentConsumed; |
779 | 61.1M | if ((entity == NULL) || |
780 | 61.1M | ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
781 | 31.9M | ((entity->flags & XML_ENT_PARSED) == 0))) { |
782 | 31.9M | xmlSaturatedAdd(&consumed, input->consumed); |
783 | 31.9M | xmlSaturatedAddSizeT(&consumed, input->cur - input->base); |
784 | 31.9M | } |
785 | 61.1M | xmlSaturatedAdd(&consumed, ctxt->sizeentities); |
786 | | |
787 | | /* |
788 | | * Add extra cost and some fixed cost. |
789 | | */ |
790 | 61.1M | xmlSaturatedAdd(&ctxt->sizeentcopy, extra); |
791 | 61.1M | xmlSaturatedAdd(&ctxt->sizeentcopy, XML_ENT_FIXED_COST); |
792 | | |
793 | | /* |
794 | | * It's important to always use saturation arithmetic when tracking |
795 | | * entity sizes to make the size checks reliable. If "sizeentcopy" |
796 | | * overflows, we have to abort. |
797 | | */ |
798 | 61.1M | if ((ctxt->sizeentcopy > XML_MAX_TEXT_LENGTH) && |
799 | 61.1M | ((ctxt->sizeentcopy >= ULONG_MAX) || |
800 | 766 | (ctxt->sizeentcopy / XML_PARSER_NON_LINEAR > consumed))) { |
801 | 766 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_LOOP, |
802 | 766 | "Maximum entity amplification factor exceeded"); |
803 | 766 | xmlHaltParser(ctxt); |
804 | 766 | return(1); |
805 | 766 | } |
806 | | |
807 | 61.1M | return(0); |
808 | 61.1M | } |
809 | | |
810 | | /************************************************************************ |
811 | | * * |
812 | | * Library wide options * |
813 | | * * |
814 | | ************************************************************************/ |
815 | | |
816 | | /** |
817 | | * xmlHasFeature: |
818 | | * @feature: the feature to be examined |
819 | | * |
820 | | * Examines if the library has been compiled with a given feature. |
821 | | * |
822 | | * Returns a non-zero value if the feature exist, otherwise zero. |
823 | | * Returns zero (0) if the feature does not exist or an unknown |
824 | | * unknown feature is requested, non-zero otherwise. |
825 | | */ |
826 | | int |
827 | | xmlHasFeature(xmlFeature feature) |
828 | 0 | { |
829 | 0 | switch (feature) { |
830 | 0 | case XML_WITH_THREAD: |
831 | 0 | #ifdef LIBXML_THREAD_ENABLED |
832 | 0 | return(1); |
833 | | #else |
834 | | return(0); |
835 | | #endif |
836 | 0 | case XML_WITH_TREE: |
837 | 0 | #ifdef LIBXML_TREE_ENABLED |
838 | 0 | return(1); |
839 | | #else |
840 | | return(0); |
841 | | #endif |
842 | 0 | case XML_WITH_OUTPUT: |
843 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
844 | 0 | return(1); |
845 | | #else |
846 | | return(0); |
847 | | #endif |
848 | 0 | case XML_WITH_PUSH: |
849 | 0 | #ifdef LIBXML_PUSH_ENABLED |
850 | 0 | return(1); |
851 | | #else |
852 | | return(0); |
853 | | #endif |
854 | 0 | case XML_WITH_READER: |
855 | 0 | #ifdef LIBXML_READER_ENABLED |
856 | 0 | return(1); |
857 | | #else |
858 | | return(0); |
859 | | #endif |
860 | 0 | case XML_WITH_PATTERN: |
861 | 0 | #ifdef LIBXML_PATTERN_ENABLED |
862 | 0 | return(1); |
863 | | #else |
864 | | return(0); |
865 | | #endif |
866 | 0 | case XML_WITH_WRITER: |
867 | 0 | #ifdef LIBXML_WRITER_ENABLED |
868 | 0 | return(1); |
869 | | #else |
870 | | return(0); |
871 | | #endif |
872 | 0 | case XML_WITH_SAX1: |
873 | 0 | #ifdef LIBXML_SAX1_ENABLED |
874 | 0 | return(1); |
875 | | #else |
876 | | return(0); |
877 | | #endif |
878 | 0 | case XML_WITH_FTP: |
879 | | #ifdef LIBXML_FTP_ENABLED |
880 | | return(1); |
881 | | #else |
882 | 0 | return(0); |
883 | 0 | #endif |
884 | 0 | case XML_WITH_HTTP: |
885 | | #ifdef LIBXML_HTTP_ENABLED |
886 | | return(1); |
887 | | #else |
888 | 0 | return(0); |
889 | 0 | #endif |
890 | 0 | case XML_WITH_VALID: |
891 | 0 | #ifdef LIBXML_VALID_ENABLED |
892 | 0 | return(1); |
893 | | #else |
894 | | return(0); |
895 | | #endif |
896 | 0 | case XML_WITH_HTML: |
897 | 0 | #ifdef LIBXML_HTML_ENABLED |
898 | 0 | return(1); |
899 | | #else |
900 | | return(0); |
901 | | #endif |
902 | 0 | case XML_WITH_LEGACY: |
903 | | #ifdef LIBXML_LEGACY_ENABLED |
904 | | return(1); |
905 | | #else |
906 | 0 | return(0); |
907 | 0 | #endif |
908 | 0 | case XML_WITH_C14N: |
909 | 0 | #ifdef LIBXML_C14N_ENABLED |
910 | 0 | return(1); |
911 | | #else |
912 | | return(0); |
913 | | #endif |
914 | 0 | case XML_WITH_CATALOG: |
915 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
916 | 0 | return(1); |
917 | | #else |
918 | | return(0); |
919 | | #endif |
920 | 0 | case XML_WITH_XPATH: |
921 | 0 | #ifdef LIBXML_XPATH_ENABLED |
922 | 0 | return(1); |
923 | | #else |
924 | | return(0); |
925 | | #endif |
926 | 0 | case XML_WITH_XPTR: |
927 | 0 | #ifdef LIBXML_XPTR_ENABLED |
928 | 0 | return(1); |
929 | | #else |
930 | | return(0); |
931 | | #endif |
932 | 0 | case XML_WITH_XINCLUDE: |
933 | 0 | #ifdef LIBXML_XINCLUDE_ENABLED |
934 | 0 | return(1); |
935 | | #else |
936 | | return(0); |
937 | | #endif |
938 | 0 | case XML_WITH_ICONV: |
939 | 0 | #ifdef LIBXML_ICONV_ENABLED |
940 | 0 | return(1); |
941 | | #else |
942 | | return(0); |
943 | | #endif |
944 | 0 | case XML_WITH_ISO8859X: |
945 | 0 | #ifdef LIBXML_ISO8859X_ENABLED |
946 | 0 | return(1); |
947 | | #else |
948 | | return(0); |
949 | | #endif |
950 | 0 | case XML_WITH_UNICODE: |
951 | 0 | #ifdef LIBXML_UNICODE_ENABLED |
952 | 0 | return(1); |
953 | | #else |
954 | | return(0); |
955 | | #endif |
956 | 0 | case XML_WITH_REGEXP: |
957 | 0 | #ifdef LIBXML_REGEXP_ENABLED |
958 | 0 | return(1); |
959 | | #else |
960 | | return(0); |
961 | | #endif |
962 | 0 | case XML_WITH_AUTOMATA: |
963 | 0 | #ifdef LIBXML_AUTOMATA_ENABLED |
964 | 0 | return(1); |
965 | | #else |
966 | | return(0); |
967 | | #endif |
968 | 0 | case XML_WITH_EXPR: |
969 | | #ifdef LIBXML_EXPR_ENABLED |
970 | | return(1); |
971 | | #else |
972 | 0 | return(0); |
973 | 0 | #endif |
974 | 0 | case XML_WITH_SCHEMAS: |
975 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
976 | 0 | return(1); |
977 | | #else |
978 | | return(0); |
979 | | #endif |
980 | 0 | case XML_WITH_SCHEMATRON: |
981 | 0 | #ifdef LIBXML_SCHEMATRON_ENABLED |
982 | 0 | return(1); |
983 | | #else |
984 | | return(0); |
985 | | #endif |
986 | 0 | case XML_WITH_MODULES: |
987 | 0 | #ifdef LIBXML_MODULES_ENABLED |
988 | 0 | return(1); |
989 | | #else |
990 | | return(0); |
991 | | #endif |
992 | 0 | case XML_WITH_DEBUG: |
993 | | #ifdef LIBXML_DEBUG_ENABLED |
994 | | return(1); |
995 | | #else |
996 | 0 | return(0); |
997 | 0 | #endif |
998 | 0 | case XML_WITH_DEBUG_MEM: |
999 | | #ifdef DEBUG_MEMORY_LOCATION |
1000 | | return(1); |
1001 | | #else |
1002 | 0 | return(0); |
1003 | 0 | #endif |
1004 | 0 | case XML_WITH_DEBUG_RUN: |
1005 | 0 | return(0); |
1006 | 0 | case XML_WITH_ZLIB: |
1007 | 0 | #ifdef LIBXML_ZLIB_ENABLED |
1008 | 0 | return(1); |
1009 | | #else |
1010 | | return(0); |
1011 | | #endif |
1012 | 0 | case XML_WITH_LZMA: |
1013 | 0 | #ifdef LIBXML_LZMA_ENABLED |
1014 | 0 | return(1); |
1015 | | #else |
1016 | | return(0); |
1017 | | #endif |
1018 | 0 | case XML_WITH_ICU: |
1019 | | #ifdef LIBXML_ICU_ENABLED |
1020 | | return(1); |
1021 | | #else |
1022 | 0 | return(0); |
1023 | 0 | #endif |
1024 | 0 | default: |
1025 | 0 | break; |
1026 | 0 | } |
1027 | 0 | return(0); |
1028 | 0 | } |
1029 | | |
1030 | | /************************************************************************ |
1031 | | * * |
1032 | | * SAX2 defaulted attributes handling * |
1033 | | * * |
1034 | | ************************************************************************/ |
1035 | | |
1036 | | /** |
1037 | | * xmlDetectSAX2: |
1038 | | * @ctxt: an XML parser context |
1039 | | * |
1040 | | * Do the SAX2 detection and specific initialization |
1041 | | */ |
1042 | | static void |
1043 | 886k | xmlDetectSAX2(xmlParserCtxtPtr ctxt) { |
1044 | 886k | xmlSAXHandlerPtr sax; |
1045 | | |
1046 | | /* Avoid unused variable warning if features are disabled. */ |
1047 | 886k | (void) sax; |
1048 | | |
1049 | 886k | if (ctxt == NULL) return; |
1050 | 886k | sax = ctxt->sax; |
1051 | 886k | #ifdef LIBXML_SAX1_ENABLED |
1052 | 886k | if ((sax) && (sax->initialized == XML_SAX2_MAGIC) && |
1053 | 886k | ((sax->startElementNs != NULL) || |
1054 | 544k | (sax->endElementNs != NULL) || |
1055 | 544k | ((sax->startElement == NULL) && (sax->endElement == NULL)))) |
1056 | 544k | ctxt->sax2 = 1; |
1057 | | #else |
1058 | | ctxt->sax2 = 1; |
1059 | | #endif /* LIBXML_SAX1_ENABLED */ |
1060 | | |
1061 | 886k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
1062 | 886k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
1063 | 886k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
1064 | 886k | if ((ctxt->str_xml==NULL) || (ctxt->str_xmlns==NULL) || |
1065 | 886k | (ctxt->str_xml_ns == NULL)) { |
1066 | 0 | xmlErrMemory(ctxt, NULL); |
1067 | 0 | } |
1068 | 886k | } |
1069 | | |
1070 | | typedef struct _xmlDefAttrs xmlDefAttrs; |
1071 | | typedef xmlDefAttrs *xmlDefAttrsPtr; |
1072 | | struct _xmlDefAttrs { |
1073 | | int nbAttrs; /* number of defaulted attributes on that element */ |
1074 | | int maxAttrs; /* the size of the array */ |
1075 | | #if __STDC_VERSION__ >= 199901L |
1076 | | /* Using a C99 flexible array member avoids UBSan errors. */ |
1077 | | const xmlChar *values[]; /* array of localname/prefix/values/external */ |
1078 | | #else |
1079 | | const xmlChar *values[5]; |
1080 | | #endif |
1081 | | }; |
1082 | | |
1083 | | /** |
1084 | | * xmlAttrNormalizeSpace: |
1085 | | * @src: the source string |
1086 | | * @dst: the target string |
1087 | | * |
1088 | | * Normalize the space in non CDATA attribute values: |
1089 | | * If the attribute type is not CDATA, then the XML processor MUST further |
1090 | | * process the normalized attribute value by discarding any leading and |
1091 | | * trailing space (#x20) characters, and by replacing sequences of space |
1092 | | * (#x20) characters by a single space (#x20) character. |
1093 | | * Note that the size of dst need to be at least src, and if one doesn't need |
1094 | | * to preserve dst (and it doesn't come from a dictionary or read-only) then |
1095 | | * passing src as dst is just fine. |
1096 | | * |
1097 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1098 | | * is needed. |
1099 | | */ |
1100 | | static xmlChar * |
1101 | | xmlAttrNormalizeSpace(const xmlChar *src, xmlChar *dst) |
1102 | 155k | { |
1103 | 155k | if ((src == NULL) || (dst == NULL)) |
1104 | 0 | return(NULL); |
1105 | | |
1106 | 181k | while (*src == 0x20) src++; |
1107 | 2.50M | while (*src != 0) { |
1108 | 2.35M | if (*src == 0x20) { |
1109 | 519k | while (*src == 0x20) src++; |
1110 | 179k | if (*src != 0) |
1111 | 164k | *dst++ = 0x20; |
1112 | 2.17M | } else { |
1113 | 2.17M | *dst++ = *src++; |
1114 | 2.17M | } |
1115 | 2.35M | } |
1116 | 155k | *dst = 0; |
1117 | 155k | if (dst == src) |
1118 | 133k | return(NULL); |
1119 | 22.1k | return(dst); |
1120 | 155k | } |
1121 | | |
1122 | | /** |
1123 | | * xmlAttrNormalizeSpace2: |
1124 | | * @src: the source string |
1125 | | * |
1126 | | * Normalize the space in non CDATA attribute values, a slightly more complex |
1127 | | * front end to avoid allocation problems when running on attribute values |
1128 | | * coming from the input. |
1129 | | * |
1130 | | * Returns a pointer to the normalized value (dst) or NULL if no conversion |
1131 | | * is needed. |
1132 | | */ |
1133 | | static const xmlChar * |
1134 | | xmlAttrNormalizeSpace2(xmlParserCtxtPtr ctxt, xmlChar *src, int *len) |
1135 | 26.5k | { |
1136 | 26.5k | int i; |
1137 | 26.5k | int remove_head = 0; |
1138 | 26.5k | int need_realloc = 0; |
1139 | 26.5k | const xmlChar *cur; |
1140 | | |
1141 | 26.5k | if ((ctxt == NULL) || (src == NULL) || (len == NULL)) |
1142 | 0 | return(NULL); |
1143 | 26.5k | i = *len; |
1144 | 26.5k | if (i <= 0) |
1145 | 1.43k | return(NULL); |
1146 | | |
1147 | 25.0k | cur = src; |
1148 | 39.3k | while (*cur == 0x20) { |
1149 | 14.2k | cur++; |
1150 | 14.2k | remove_head++; |
1151 | 14.2k | } |
1152 | 790k | while (*cur != 0) { |
1153 | 771k | if (*cur == 0x20) { |
1154 | 48.4k | cur++; |
1155 | 48.4k | if ((*cur == 0x20) || (*cur == 0)) { |
1156 | 5.97k | need_realloc = 1; |
1157 | 5.97k | break; |
1158 | 5.97k | } |
1159 | 48.4k | } else |
1160 | 722k | cur++; |
1161 | 771k | } |
1162 | 25.0k | if (need_realloc) { |
1163 | 5.97k | xmlChar *ret; |
1164 | | |
1165 | 5.97k | ret = xmlStrndup(src + remove_head, i - remove_head + 1); |
1166 | 5.97k | if (ret == NULL) { |
1167 | 0 | xmlErrMemory(ctxt, NULL); |
1168 | 0 | return(NULL); |
1169 | 0 | } |
1170 | 5.97k | xmlAttrNormalizeSpace(ret, ret); |
1171 | 5.97k | *len = strlen((const char *)ret); |
1172 | 5.97k | return(ret); |
1173 | 19.1k | } else if (remove_head) { |
1174 | 1.19k | *len -= remove_head; |
1175 | 1.19k | memmove(src, src + remove_head, 1 + *len); |
1176 | 1.19k | return(src); |
1177 | 1.19k | } |
1178 | 17.9k | return(NULL); |
1179 | 25.0k | } |
1180 | | |
1181 | | /** |
1182 | | * xmlAddDefAttrs: |
1183 | | * @ctxt: an XML parser context |
1184 | | * @fullname: the element fullname |
1185 | | * @fullattr: the attribute fullname |
1186 | | * @value: the attribute value |
1187 | | * |
1188 | | * Add a defaulted attribute for an element |
1189 | | */ |
1190 | | static void |
1191 | | xmlAddDefAttrs(xmlParserCtxtPtr ctxt, |
1192 | | const xmlChar *fullname, |
1193 | | const xmlChar *fullattr, |
1194 | 187k | const xmlChar *value) { |
1195 | 187k | xmlDefAttrsPtr defaults; |
1196 | 187k | int len; |
1197 | 187k | const xmlChar *name; |
1198 | 187k | const xmlChar *prefix; |
1199 | | |
1200 | | /* |
1201 | | * Allows to detect attribute redefinitions |
1202 | | */ |
1203 | 187k | if (ctxt->attsSpecial != NULL) { |
1204 | 165k | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1205 | 47.2k | return; |
1206 | 165k | } |
1207 | | |
1208 | 140k | if (ctxt->attsDefault == NULL) { |
1209 | 29.2k | ctxt->attsDefault = xmlHashCreateDict(10, ctxt->dict); |
1210 | 29.2k | if (ctxt->attsDefault == NULL) |
1211 | 0 | goto mem_error; |
1212 | 29.2k | } |
1213 | | |
1214 | | /* |
1215 | | * split the element name into prefix:localname , the string found |
1216 | | * are within the DTD and then not associated to namespace names. |
1217 | | */ |
1218 | 140k | name = xmlSplitQName3(fullname, &len); |
1219 | 140k | if (name == NULL) { |
1220 | 132k | name = xmlDictLookup(ctxt->dict, fullname, -1); |
1221 | 132k | prefix = NULL; |
1222 | 132k | } else { |
1223 | 8.69k | name = xmlDictLookup(ctxt->dict, name, -1); |
1224 | 8.69k | prefix = xmlDictLookup(ctxt->dict, fullname, len); |
1225 | 8.69k | } |
1226 | | |
1227 | | /* |
1228 | | * make sure there is some storage |
1229 | | */ |
1230 | 140k | defaults = xmlHashLookup2(ctxt->attsDefault, name, prefix); |
1231 | 140k | if (defaults == NULL) { |
1232 | 78.1k | defaults = (xmlDefAttrsPtr) xmlMalloc(sizeof(xmlDefAttrs) + |
1233 | 78.1k | (4 * 5) * sizeof(const xmlChar *)); |
1234 | 78.1k | if (defaults == NULL) |
1235 | 0 | goto mem_error; |
1236 | 78.1k | defaults->nbAttrs = 0; |
1237 | 78.1k | defaults->maxAttrs = 4; |
1238 | 78.1k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1239 | 78.1k | defaults, NULL) < 0) { |
1240 | 0 | xmlFree(defaults); |
1241 | 0 | goto mem_error; |
1242 | 0 | } |
1243 | 78.1k | } else if (defaults->nbAttrs >= defaults->maxAttrs) { |
1244 | 3.83k | xmlDefAttrsPtr temp; |
1245 | | |
1246 | 3.83k | temp = (xmlDefAttrsPtr) xmlRealloc(defaults, sizeof(xmlDefAttrs) + |
1247 | 3.83k | (2 * defaults->maxAttrs * 5) * sizeof(const xmlChar *)); |
1248 | 3.83k | if (temp == NULL) |
1249 | 0 | goto mem_error; |
1250 | 3.83k | defaults = temp; |
1251 | 3.83k | defaults->maxAttrs *= 2; |
1252 | 3.83k | if (xmlHashUpdateEntry2(ctxt->attsDefault, name, prefix, |
1253 | 3.83k | defaults, NULL) < 0) { |
1254 | 0 | xmlFree(defaults); |
1255 | 0 | goto mem_error; |
1256 | 0 | } |
1257 | 3.83k | } |
1258 | | |
1259 | | /* |
1260 | | * Split the element name into prefix:localname , the string found |
1261 | | * are within the DTD and hen not associated to namespace names. |
1262 | | */ |
1263 | 140k | name = xmlSplitQName3(fullattr, &len); |
1264 | 140k | if (name == NULL) { |
1265 | 120k | name = xmlDictLookup(ctxt->dict, fullattr, -1); |
1266 | 120k | prefix = NULL; |
1267 | 120k | } else { |
1268 | 20.3k | name = xmlDictLookup(ctxt->dict, name, -1); |
1269 | 20.3k | prefix = xmlDictLookup(ctxt->dict, fullattr, len); |
1270 | 20.3k | } |
1271 | | |
1272 | 140k | defaults->values[5 * defaults->nbAttrs] = name; |
1273 | 140k | defaults->values[5 * defaults->nbAttrs + 1] = prefix; |
1274 | | /* intern the string and precompute the end */ |
1275 | 140k | len = xmlStrlen(value); |
1276 | 140k | value = xmlDictLookup(ctxt->dict, value, len); |
1277 | 140k | if (value == NULL) |
1278 | 0 | goto mem_error; |
1279 | 140k | defaults->values[5 * defaults->nbAttrs + 2] = value; |
1280 | 140k | defaults->values[5 * defaults->nbAttrs + 3] = value + len; |
1281 | 140k | if (ctxt->external) |
1282 | 30.7k | defaults->values[5 * defaults->nbAttrs + 4] = BAD_CAST "external"; |
1283 | 109k | else |
1284 | 109k | defaults->values[5 * defaults->nbAttrs + 4] = NULL; |
1285 | 140k | defaults->nbAttrs++; |
1286 | | |
1287 | 140k | return; |
1288 | | |
1289 | 0 | mem_error: |
1290 | 0 | xmlErrMemory(ctxt, NULL); |
1291 | 0 | return; |
1292 | 140k | } |
1293 | | |
1294 | | /** |
1295 | | * xmlAddSpecialAttr: |
1296 | | * @ctxt: an XML parser context |
1297 | | * @fullname: the element fullname |
1298 | | * @fullattr: the attribute fullname |
1299 | | * @type: the attribute type |
1300 | | * |
1301 | | * Register this attribute type |
1302 | | */ |
1303 | | static void |
1304 | | xmlAddSpecialAttr(xmlParserCtxtPtr ctxt, |
1305 | | const xmlChar *fullname, |
1306 | | const xmlChar *fullattr, |
1307 | | int type) |
1308 | 2.53M | { |
1309 | 2.53M | if (ctxt->attsSpecial == NULL) { |
1310 | 51.6k | ctxt->attsSpecial = xmlHashCreateDict(10, ctxt->dict); |
1311 | 51.6k | if (ctxt->attsSpecial == NULL) |
1312 | 0 | goto mem_error; |
1313 | 51.6k | } |
1314 | | |
1315 | 2.53M | if (xmlHashLookup2(ctxt->attsSpecial, fullname, fullattr) != NULL) |
1316 | 769k | return; |
1317 | | |
1318 | 1.76M | xmlHashAddEntry2(ctxt->attsSpecial, fullname, fullattr, |
1319 | 1.76M | (void *) (ptrdiff_t) type); |
1320 | 1.76M | return; |
1321 | | |
1322 | 0 | mem_error: |
1323 | 0 | xmlErrMemory(ctxt, NULL); |
1324 | 0 | return; |
1325 | 2.53M | } |
1326 | | |
1327 | | /** |
1328 | | * xmlCleanSpecialAttrCallback: |
1329 | | * |
1330 | | * Removes CDATA attributes from the special attribute table |
1331 | | */ |
1332 | | static void |
1333 | | xmlCleanSpecialAttrCallback(void *payload, void *data, |
1334 | | const xmlChar *fullname, const xmlChar *fullattr, |
1335 | 1.25M | const xmlChar *unused ATTRIBUTE_UNUSED) { |
1336 | 1.25M | xmlParserCtxtPtr ctxt = (xmlParserCtxtPtr) data; |
1337 | | |
1338 | 1.25M | if (((ptrdiff_t) payload) == XML_ATTRIBUTE_CDATA) { |
1339 | 517k | xmlHashRemoveEntry2(ctxt->attsSpecial, fullname, fullattr, NULL); |
1340 | 517k | } |
1341 | 1.25M | } |
1342 | | |
1343 | | /** |
1344 | | * xmlCleanSpecialAttr: |
1345 | | * @ctxt: an XML parser context |
1346 | | * |
1347 | | * Trim the list of attributes defined to remove all those of type |
1348 | | * CDATA as they are not special. This call should be done when finishing |
1349 | | * to parse the DTD and before starting to parse the document root. |
1350 | | */ |
1351 | | static void |
1352 | | xmlCleanSpecialAttr(xmlParserCtxtPtr ctxt) |
1353 | 228k | { |
1354 | 228k | if (ctxt->attsSpecial == NULL) |
1355 | 187k | return; |
1356 | | |
1357 | 40.4k | xmlHashScanFull(ctxt->attsSpecial, xmlCleanSpecialAttrCallback, ctxt); |
1358 | | |
1359 | 40.4k | if (xmlHashSize(ctxt->attsSpecial) == 0) { |
1360 | 11.0k | xmlHashFree(ctxt->attsSpecial, NULL); |
1361 | 11.0k | ctxt->attsSpecial = NULL; |
1362 | 11.0k | } |
1363 | 40.4k | return; |
1364 | 228k | } |
1365 | | |
1366 | | /** |
1367 | | * xmlCheckLanguageID: |
1368 | | * @lang: pointer to the string value |
1369 | | * |
1370 | | * Checks that the value conforms to the LanguageID production: |
1371 | | * |
1372 | | * NOTE: this is somewhat deprecated, those productions were removed from |
1373 | | * the XML Second edition. |
1374 | | * |
1375 | | * [33] LanguageID ::= Langcode ('-' Subcode)* |
1376 | | * [34] Langcode ::= ISO639Code | IanaCode | UserCode |
1377 | | * [35] ISO639Code ::= ([a-z] | [A-Z]) ([a-z] | [A-Z]) |
1378 | | * [36] IanaCode ::= ('i' | 'I') '-' ([a-z] | [A-Z])+ |
1379 | | * [37] UserCode ::= ('x' | 'X') '-' ([a-z] | [A-Z])+ |
1380 | | * [38] Subcode ::= ([a-z] | [A-Z])+ |
1381 | | * |
1382 | | * The current REC reference the successors of RFC 1766, currently 5646 |
1383 | | * |
1384 | | * http://www.rfc-editor.org/rfc/rfc5646.txt |
1385 | | * langtag = language |
1386 | | * ["-" script] |
1387 | | * ["-" region] |
1388 | | * *("-" variant) |
1389 | | * *("-" extension) |
1390 | | * ["-" privateuse] |
1391 | | * language = 2*3ALPHA ; shortest ISO 639 code |
1392 | | * ["-" extlang] ; sometimes followed by |
1393 | | * ; extended language subtags |
1394 | | * / 4ALPHA ; or reserved for future use |
1395 | | * / 5*8ALPHA ; or registered language subtag |
1396 | | * |
1397 | | * extlang = 3ALPHA ; selected ISO 639 codes |
1398 | | * *2("-" 3ALPHA) ; permanently reserved |
1399 | | * |
1400 | | * script = 4ALPHA ; ISO 15924 code |
1401 | | * |
1402 | | * region = 2ALPHA ; ISO 3166-1 code |
1403 | | * / 3DIGIT ; UN M.49 code |
1404 | | * |
1405 | | * variant = 5*8alphanum ; registered variants |
1406 | | * / (DIGIT 3alphanum) |
1407 | | * |
1408 | | * extension = singleton 1*("-" (2*8alphanum)) |
1409 | | * |
1410 | | * ; Single alphanumerics |
1411 | | * ; "x" reserved for private use |
1412 | | * singleton = DIGIT ; 0 - 9 |
1413 | | * / %x41-57 ; A - W |
1414 | | * / %x59-5A ; Y - Z |
1415 | | * / %x61-77 ; a - w |
1416 | | * / %x79-7A ; y - z |
1417 | | * |
1418 | | * it sounds right to still allow Irregular i-xxx IANA and user codes too |
1419 | | * The parser below doesn't try to cope with extension or privateuse |
1420 | | * that could be added but that's not interoperable anyway |
1421 | | * |
1422 | | * Returns 1 if correct 0 otherwise |
1423 | | **/ |
1424 | | int |
1425 | | xmlCheckLanguageID(const xmlChar * lang) |
1426 | 23.8k | { |
1427 | 23.8k | const xmlChar *cur = lang, *nxt; |
1428 | | |
1429 | 23.8k | if (cur == NULL) |
1430 | 481 | return (0); |
1431 | 23.3k | if (((cur[0] == 'i') && (cur[1] == '-')) || |
1432 | 23.3k | ((cur[0] == 'I') && (cur[1] == '-')) || |
1433 | 23.3k | ((cur[0] == 'x') && (cur[1] == '-')) || |
1434 | 23.3k | ((cur[0] == 'X') && (cur[1] == '-'))) { |
1435 | | /* |
1436 | | * Still allow IANA code and user code which were coming |
1437 | | * from the previous version of the XML-1.0 specification |
1438 | | * it's deprecated but we should not fail |
1439 | | */ |
1440 | 1.70k | cur += 2; |
1441 | 10.8k | while (((cur[0] >= 'A') && (cur[0] <= 'Z')) || |
1442 | 10.8k | ((cur[0] >= 'a') && (cur[0] <= 'z'))) |
1443 | 9.10k | cur++; |
1444 | 1.70k | return(cur[0] == 0); |
1445 | 1.70k | } |
1446 | 21.6k | nxt = cur; |
1447 | 79.9k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1448 | 79.9k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1449 | 58.2k | nxt++; |
1450 | 21.6k | if (nxt - cur >= 4) { |
1451 | | /* |
1452 | | * Reserved |
1453 | | */ |
1454 | 933 | if ((nxt - cur > 8) || (nxt[0] != 0)) |
1455 | 716 | return(0); |
1456 | 217 | return(1); |
1457 | 933 | } |
1458 | 20.7k | if (nxt - cur < 2) |
1459 | 1.11k | return(0); |
1460 | | /* we got an ISO 639 code */ |
1461 | 19.6k | if (nxt[0] == 0) |
1462 | 6.42k | return(1); |
1463 | 13.2k | if (nxt[0] != '-') |
1464 | 1.24k | return(0); |
1465 | | |
1466 | 11.9k | nxt++; |
1467 | 11.9k | cur = nxt; |
1468 | | /* now we can have extlang or script or region or variant */ |
1469 | 11.9k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1470 | 1.57k | goto region_m49; |
1471 | | |
1472 | 47.5k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1473 | 47.5k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1474 | 37.1k | nxt++; |
1475 | 10.3k | if (nxt - cur == 4) |
1476 | 3.31k | goto script; |
1477 | 7.07k | if (nxt - cur == 2) |
1478 | 2.13k | goto region; |
1479 | 4.94k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1480 | 934 | goto variant; |
1481 | 4.00k | if (nxt - cur != 3) |
1482 | 1.18k | return(0); |
1483 | | /* we parsed an extlang */ |
1484 | 2.81k | if (nxt[0] == 0) |
1485 | 409 | return(1); |
1486 | 2.41k | if (nxt[0] != '-') |
1487 | 502 | return(0); |
1488 | | |
1489 | 1.90k | nxt++; |
1490 | 1.90k | cur = nxt; |
1491 | | /* now we can have script or region or variant */ |
1492 | 1.90k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1493 | 295 | goto region_m49; |
1494 | | |
1495 | 24.5k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1496 | 24.5k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1497 | 22.9k | nxt++; |
1498 | 1.61k | if (nxt - cur == 2) |
1499 | 444 | goto region; |
1500 | 1.16k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1501 | 263 | goto variant; |
1502 | 906 | if (nxt - cur != 4) |
1503 | 569 | return(0); |
1504 | | /* we parsed a script */ |
1505 | 3.65k | script: |
1506 | 3.65k | if (nxt[0] == 0) |
1507 | 544 | return(1); |
1508 | 3.10k | if (nxt[0] != '-') |
1509 | 885 | return(0); |
1510 | | |
1511 | 2.22k | nxt++; |
1512 | 2.22k | cur = nxt; |
1513 | | /* now we can have region or variant */ |
1514 | 2.22k | if ((nxt[0] >= '0') && (nxt[0] <= '9')) |
1515 | 745 | goto region_m49; |
1516 | | |
1517 | 12.5k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1518 | 12.5k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1519 | 11.0k | nxt++; |
1520 | | |
1521 | 1.47k | if ((nxt - cur >= 5) && (nxt - cur <= 8)) |
1522 | 426 | goto variant; |
1523 | 1.05k | if (nxt - cur != 2) |
1524 | 665 | return(0); |
1525 | | /* we parsed a region */ |
1526 | 4.31k | region: |
1527 | 4.31k | if (nxt[0] == 0) |
1528 | 1.31k | return(1); |
1529 | 2.99k | if (nxt[0] != '-') |
1530 | 1.89k | return(0); |
1531 | | |
1532 | 1.10k | nxt++; |
1533 | 1.10k | cur = nxt; |
1534 | | /* now we can just have a variant */ |
1535 | 11.2k | while (((nxt[0] >= 'A') && (nxt[0] <= 'Z')) || |
1536 | 11.2k | ((nxt[0] >= 'a') && (nxt[0] <= 'z'))) |
1537 | 10.1k | nxt++; |
1538 | | |
1539 | 1.10k | if ((nxt - cur < 5) || (nxt - cur > 8)) |
1540 | 530 | return(0); |
1541 | | |
1542 | | /* we parsed a variant */ |
1543 | 2.19k | variant: |
1544 | 2.19k | if (nxt[0] == 0) |
1545 | 221 | return(1); |
1546 | 1.97k | if (nxt[0] != '-') |
1547 | 1.41k | return(0); |
1548 | | /* extensions and private use subtags not checked */ |
1549 | 559 | return (1); |
1550 | | |
1551 | 2.61k | region_m49: |
1552 | 2.61k | if (((nxt[1] >= '0') && (nxt[1] <= '9')) && |
1553 | 2.61k | ((nxt[2] >= '0') && (nxt[2] <= '9'))) { |
1554 | 1.34k | nxt += 3; |
1555 | 1.34k | goto region; |
1556 | 1.34k | } |
1557 | 1.27k | return(0); |
1558 | 2.61k | } |
1559 | | |
1560 | | /************************************************************************ |
1561 | | * * |
1562 | | * Parser stacks related functions and macros * |
1563 | | * * |
1564 | | ************************************************************************/ |
1565 | | |
1566 | | static xmlEntityPtr xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, |
1567 | | const xmlChar ** str); |
1568 | | |
1569 | | #ifdef SAX2 |
1570 | | /** |
1571 | | * nsPush: |
1572 | | * @ctxt: an XML parser context |
1573 | | * @prefix: the namespace prefix or NULL |
1574 | | * @URL: the namespace name |
1575 | | * |
1576 | | * Pushes a new parser namespace on top of the ns stack |
1577 | | * |
1578 | | * Returns -1 in case of error, -2 if the namespace should be discarded |
1579 | | * and the index in the stack otherwise. |
1580 | | */ |
1581 | | static int |
1582 | | nsPush(xmlParserCtxtPtr ctxt, const xmlChar *prefix, const xmlChar *URL) |
1583 | 153k | { |
1584 | 153k | if (ctxt->options & XML_PARSE_NSCLEAN) { |
1585 | 86.0k | int i; |
1586 | 170k | for (i = ctxt->nsNr - 2;i >= 0;i -= 2) { |
1587 | 116k | if (ctxt->nsTab[i] == prefix) { |
1588 | | /* in scope */ |
1589 | 32.7k | if (ctxt->nsTab[i + 1] == URL) |
1590 | 9.94k | return(-2); |
1591 | | /* out of scope keep it */ |
1592 | 22.8k | break; |
1593 | 32.7k | } |
1594 | 116k | } |
1595 | 86.0k | } |
1596 | 143k | if ((ctxt->nsMax == 0) || (ctxt->nsTab == NULL)) { |
1597 | 47.2k | ctxt->nsMax = 10; |
1598 | 47.2k | ctxt->nsNr = 0; |
1599 | 47.2k | ctxt->nsTab = (const xmlChar **) |
1600 | 47.2k | xmlMalloc(ctxt->nsMax * sizeof(xmlChar *)); |
1601 | 47.2k | if (ctxt->nsTab == NULL) { |
1602 | 0 | xmlErrMemory(ctxt, NULL); |
1603 | 0 | ctxt->nsMax = 0; |
1604 | 0 | return (-1); |
1605 | 0 | } |
1606 | 96.1k | } else if (ctxt->nsNr >= ctxt->nsMax) { |
1607 | 4.86k | const xmlChar ** tmp; |
1608 | 4.86k | ctxt->nsMax *= 2; |
1609 | 4.86k | tmp = (const xmlChar **) xmlRealloc((char *) ctxt->nsTab, |
1610 | 4.86k | ctxt->nsMax * sizeof(ctxt->nsTab[0])); |
1611 | 4.86k | if (tmp == NULL) { |
1612 | 0 | xmlErrMemory(ctxt, NULL); |
1613 | 0 | ctxt->nsMax /= 2; |
1614 | 0 | return (-1); |
1615 | 0 | } |
1616 | 4.86k | ctxt->nsTab = tmp; |
1617 | 4.86k | } |
1618 | 143k | ctxt->nsTab[ctxt->nsNr++] = prefix; |
1619 | 143k | ctxt->nsTab[ctxt->nsNr++] = URL; |
1620 | 143k | return (ctxt->nsNr); |
1621 | 143k | } |
1622 | | /** |
1623 | | * nsPop: |
1624 | | * @ctxt: an XML parser context |
1625 | | * @nr: the number to pop |
1626 | | * |
1627 | | * Pops the top @nr parser prefix/namespace from the ns stack |
1628 | | * |
1629 | | * Returns the number of namespaces removed |
1630 | | */ |
1631 | | static int |
1632 | | nsPop(xmlParserCtxtPtr ctxt, int nr) |
1633 | 49.8k | { |
1634 | 49.8k | int i; |
1635 | | |
1636 | 49.8k | if (ctxt->nsTab == NULL) return(0); |
1637 | 49.8k | if (ctxt->nsNr < nr) { |
1638 | 0 | xmlGenericError(xmlGenericErrorContext, "Pbm popping %d NS\n", nr); |
1639 | 0 | nr = ctxt->nsNr; |
1640 | 0 | } |
1641 | 49.8k | if (ctxt->nsNr <= 0) |
1642 | 0 | return (0); |
1643 | | |
1644 | 190k | for (i = 0;i < nr;i++) { |
1645 | 140k | ctxt->nsNr--; |
1646 | 140k | ctxt->nsTab[ctxt->nsNr] = NULL; |
1647 | 140k | } |
1648 | 49.8k | return(nr); |
1649 | 49.8k | } |
1650 | | #endif |
1651 | | |
1652 | | static int |
1653 | 109k | xmlCtxtGrowAttrs(xmlParserCtxtPtr ctxt, int nr) { |
1654 | 109k | const xmlChar **atts; |
1655 | 109k | int *attallocs; |
1656 | 109k | int maxatts; |
1657 | | |
1658 | 109k | if (nr + 5 > ctxt->maxatts) { |
1659 | 109k | maxatts = ctxt->maxatts == 0 ? 55 : (nr + 5) * 2; |
1660 | 109k | atts = (const xmlChar **) xmlMalloc( |
1661 | 109k | maxatts * sizeof(const xmlChar *)); |
1662 | 109k | if (atts == NULL) goto mem_error; |
1663 | 109k | attallocs = (int *) xmlRealloc((void *) ctxt->attallocs, |
1664 | 109k | (maxatts / 5) * sizeof(int)); |
1665 | 109k | if (attallocs == NULL) { |
1666 | 0 | xmlFree(atts); |
1667 | 0 | goto mem_error; |
1668 | 0 | } |
1669 | 109k | if (ctxt->maxatts > 0) |
1670 | 535 | memcpy(atts, ctxt->atts, ctxt->maxatts * sizeof(const xmlChar *)); |
1671 | 109k | xmlFree(ctxt->atts); |
1672 | 109k | ctxt->atts = atts; |
1673 | 109k | ctxt->attallocs = attallocs; |
1674 | 109k | ctxt->maxatts = maxatts; |
1675 | 109k | } |
1676 | 109k | return(ctxt->maxatts); |
1677 | 0 | mem_error: |
1678 | 0 | xmlErrMemory(ctxt, NULL); |
1679 | 0 | return(-1); |
1680 | 109k | } |
1681 | | |
1682 | | /** |
1683 | | * inputPush: |
1684 | | * @ctxt: an XML parser context |
1685 | | * @value: the parser input |
1686 | | * |
1687 | | * Pushes a new parser input on top of the input stack |
1688 | | * |
1689 | | * Returns -1 in case of error, the index in the stack otherwise |
1690 | | */ |
1691 | | int |
1692 | | inputPush(xmlParserCtxtPtr ctxt, xmlParserInputPtr value) |
1693 | 29.8M | { |
1694 | 29.8M | if ((ctxt == NULL) || (value == NULL)) |
1695 | 0 | return(-1); |
1696 | 29.8M | if (ctxt->inputNr >= ctxt->inputMax) { |
1697 | 1.63k | size_t newSize = ctxt->inputMax * 2; |
1698 | 1.63k | xmlParserInputPtr *tmp; |
1699 | | |
1700 | 1.63k | tmp = (xmlParserInputPtr *) xmlRealloc(ctxt->inputTab, |
1701 | 1.63k | newSize * sizeof(*tmp)); |
1702 | 1.63k | if (tmp == NULL) { |
1703 | 0 | xmlErrMemory(ctxt, NULL); |
1704 | 0 | return (-1); |
1705 | 0 | } |
1706 | 1.63k | ctxt->inputTab = tmp; |
1707 | 1.63k | ctxt->inputMax = newSize; |
1708 | 1.63k | } |
1709 | 29.8M | ctxt->inputTab[ctxt->inputNr] = value; |
1710 | 29.8M | ctxt->input = value; |
1711 | 29.8M | return (ctxt->inputNr++); |
1712 | 29.8M | } |
1713 | | /** |
1714 | | * inputPop: |
1715 | | * @ctxt: an XML parser context |
1716 | | * |
1717 | | * Pops the top parser input from the input stack |
1718 | | * |
1719 | | * Returns the input just removed |
1720 | | */ |
1721 | | xmlParserInputPtr |
1722 | | inputPop(xmlParserCtxtPtr ctxt) |
1723 | 32.2M | { |
1724 | 32.2M | xmlParserInputPtr ret; |
1725 | | |
1726 | 32.2M | if (ctxt == NULL) |
1727 | 0 | return(NULL); |
1728 | 32.2M | if (ctxt->inputNr <= 0) |
1729 | 2.43M | return (NULL); |
1730 | 29.8M | ctxt->inputNr--; |
1731 | 29.8M | if (ctxt->inputNr > 0) |
1732 | 29.2M | ctxt->input = ctxt->inputTab[ctxt->inputNr - 1]; |
1733 | 625k | else |
1734 | 625k | ctxt->input = NULL; |
1735 | 29.8M | ret = ctxt->inputTab[ctxt->inputNr]; |
1736 | 29.8M | ctxt->inputTab[ctxt->inputNr] = NULL; |
1737 | 29.8M | return (ret); |
1738 | 32.2M | } |
1739 | | /** |
1740 | | * nodePush: |
1741 | | * @ctxt: an XML parser context |
1742 | | * @value: the element node |
1743 | | * |
1744 | | * Pushes a new element node on top of the node stack |
1745 | | * |
1746 | | * Returns -1 in case of error, the index in the stack otherwise |
1747 | | */ |
1748 | | int |
1749 | | nodePush(xmlParserCtxtPtr ctxt, xmlNodePtr value) |
1750 | 9.26M | { |
1751 | 9.26M | if (ctxt == NULL) return(0); |
1752 | 9.26M | if (ctxt->nodeNr >= ctxt->nodeMax) { |
1753 | 10.2k | xmlNodePtr *tmp; |
1754 | | |
1755 | 10.2k | tmp = (xmlNodePtr *) xmlRealloc(ctxt->nodeTab, |
1756 | 10.2k | ctxt->nodeMax * 2 * |
1757 | 10.2k | sizeof(ctxt->nodeTab[0])); |
1758 | 10.2k | if (tmp == NULL) { |
1759 | 0 | xmlErrMemory(ctxt, NULL); |
1760 | 0 | return (-1); |
1761 | 0 | } |
1762 | 10.2k | ctxt->nodeTab = tmp; |
1763 | 10.2k | ctxt->nodeMax *= 2; |
1764 | 10.2k | } |
1765 | 9.26M | if ((((unsigned int) ctxt->nodeNr) > xmlParserMaxDepth) && |
1766 | 9.26M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
1767 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
1768 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
1769 | 0 | xmlParserMaxDepth); |
1770 | 0 | xmlHaltParser(ctxt); |
1771 | 0 | return(-1); |
1772 | 0 | } |
1773 | 9.26M | ctxt->nodeTab[ctxt->nodeNr] = value; |
1774 | 9.26M | ctxt->node = value; |
1775 | 9.26M | return (ctxt->nodeNr++); |
1776 | 9.26M | } |
1777 | | |
1778 | | /** |
1779 | | * nodePop: |
1780 | | * @ctxt: an XML parser context |
1781 | | * |
1782 | | * Pops the top element node from the node stack |
1783 | | * |
1784 | | * Returns the node just removed |
1785 | | */ |
1786 | | xmlNodePtr |
1787 | | nodePop(xmlParserCtxtPtr ctxt) |
1788 | 8.74M | { |
1789 | 8.74M | xmlNodePtr ret; |
1790 | | |
1791 | 8.74M | if (ctxt == NULL) return(NULL); |
1792 | 8.74M | if (ctxt->nodeNr <= 0) |
1793 | 115k | return (NULL); |
1794 | 8.62M | ctxt->nodeNr--; |
1795 | 8.62M | if (ctxt->nodeNr > 0) |
1796 | 8.27M | ctxt->node = ctxt->nodeTab[ctxt->nodeNr - 1]; |
1797 | 356k | else |
1798 | 356k | ctxt->node = NULL; |
1799 | 8.62M | ret = ctxt->nodeTab[ctxt->nodeNr]; |
1800 | 8.62M | ctxt->nodeTab[ctxt->nodeNr] = NULL; |
1801 | 8.62M | return (ret); |
1802 | 8.74M | } |
1803 | | |
1804 | | /** |
1805 | | * nameNsPush: |
1806 | | * @ctxt: an XML parser context |
1807 | | * @value: the element name |
1808 | | * @prefix: the element prefix |
1809 | | * @URI: the element namespace name |
1810 | | * @line: the current line number for error messages |
1811 | | * @nsNr: the number of namespaces pushed on the namespace table |
1812 | | * |
1813 | | * Pushes a new element name/prefix/URL on top of the name stack |
1814 | | * |
1815 | | * Returns -1 in case of error, the index in the stack otherwise |
1816 | | */ |
1817 | | static int |
1818 | | nameNsPush(xmlParserCtxtPtr ctxt, const xmlChar * value, |
1819 | | const xmlChar *prefix, const xmlChar *URI, int line, int nsNr) |
1820 | 7.82M | { |
1821 | 7.82M | xmlStartTag *tag; |
1822 | | |
1823 | 7.82M | if (ctxt->nameNr >= ctxt->nameMax) { |
1824 | 23.5k | const xmlChar * *tmp; |
1825 | 23.5k | xmlStartTag *tmp2; |
1826 | 23.5k | ctxt->nameMax *= 2; |
1827 | 23.5k | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1828 | 23.5k | ctxt->nameMax * |
1829 | 23.5k | sizeof(ctxt->nameTab[0])); |
1830 | 23.5k | if (tmp == NULL) { |
1831 | 0 | ctxt->nameMax /= 2; |
1832 | 0 | goto mem_error; |
1833 | 0 | } |
1834 | 23.5k | ctxt->nameTab = tmp; |
1835 | 23.5k | tmp2 = (xmlStartTag *) xmlRealloc((void * *)ctxt->pushTab, |
1836 | 23.5k | ctxt->nameMax * |
1837 | 23.5k | sizeof(ctxt->pushTab[0])); |
1838 | 23.5k | if (tmp2 == NULL) { |
1839 | 0 | ctxt->nameMax /= 2; |
1840 | 0 | goto mem_error; |
1841 | 0 | } |
1842 | 23.5k | ctxt->pushTab = tmp2; |
1843 | 7.80M | } else if (ctxt->pushTab == NULL) { |
1844 | 344k | ctxt->pushTab = (xmlStartTag *) xmlMalloc(ctxt->nameMax * |
1845 | 344k | sizeof(ctxt->pushTab[0])); |
1846 | 344k | if (ctxt->pushTab == NULL) |
1847 | 0 | goto mem_error; |
1848 | 344k | } |
1849 | 7.82M | ctxt->nameTab[ctxt->nameNr] = value; |
1850 | 7.82M | ctxt->name = value; |
1851 | 7.82M | tag = &ctxt->pushTab[ctxt->nameNr]; |
1852 | 7.82M | tag->prefix = prefix; |
1853 | 7.82M | tag->URI = URI; |
1854 | 7.82M | tag->line = line; |
1855 | 7.82M | tag->nsNr = nsNr; |
1856 | 7.82M | return (ctxt->nameNr++); |
1857 | 0 | mem_error: |
1858 | 0 | xmlErrMemory(ctxt, NULL); |
1859 | 0 | return (-1); |
1860 | 7.82M | } |
1861 | | #ifdef LIBXML_PUSH_ENABLED |
1862 | | /** |
1863 | | * nameNsPop: |
1864 | | * @ctxt: an XML parser context |
1865 | | * |
1866 | | * Pops the top element/prefix/URI name from the name stack |
1867 | | * |
1868 | | * Returns the name just removed |
1869 | | */ |
1870 | | static const xmlChar * |
1871 | | nameNsPop(xmlParserCtxtPtr ctxt) |
1872 | 1.64M | { |
1873 | 1.64M | const xmlChar *ret; |
1874 | | |
1875 | 1.64M | if (ctxt->nameNr <= 0) |
1876 | 0 | return (NULL); |
1877 | 1.64M | ctxt->nameNr--; |
1878 | 1.64M | if (ctxt->nameNr > 0) |
1879 | 1.60M | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1880 | 32.2k | else |
1881 | 32.2k | ctxt->name = NULL; |
1882 | 1.64M | ret = ctxt->nameTab[ctxt->nameNr]; |
1883 | 1.64M | ctxt->nameTab[ctxt->nameNr] = NULL; |
1884 | 1.64M | return (ret); |
1885 | 1.64M | } |
1886 | | #endif /* LIBXML_PUSH_ENABLED */ |
1887 | | |
1888 | | /** |
1889 | | * namePush: |
1890 | | * @ctxt: an XML parser context |
1891 | | * @value: the element name |
1892 | | * |
1893 | | * Pushes a new element name on top of the name stack |
1894 | | * |
1895 | | * Returns -1 in case of error, the index in the stack otherwise |
1896 | | */ |
1897 | | int |
1898 | | namePush(xmlParserCtxtPtr ctxt, const xmlChar * value) |
1899 | 0 | { |
1900 | 0 | if (ctxt == NULL) return (-1); |
1901 | | |
1902 | 0 | if (ctxt->nameNr >= ctxt->nameMax) { |
1903 | 0 | const xmlChar * *tmp; |
1904 | 0 | tmp = (const xmlChar * *) xmlRealloc((xmlChar * *)ctxt->nameTab, |
1905 | 0 | ctxt->nameMax * 2 * |
1906 | 0 | sizeof(ctxt->nameTab[0])); |
1907 | 0 | if (tmp == NULL) { |
1908 | 0 | goto mem_error; |
1909 | 0 | } |
1910 | 0 | ctxt->nameTab = tmp; |
1911 | 0 | ctxt->nameMax *= 2; |
1912 | 0 | } |
1913 | 0 | ctxt->nameTab[ctxt->nameNr] = value; |
1914 | 0 | ctxt->name = value; |
1915 | 0 | return (ctxt->nameNr++); |
1916 | 0 | mem_error: |
1917 | 0 | xmlErrMemory(ctxt, NULL); |
1918 | 0 | return (-1); |
1919 | 0 | } |
1920 | | /** |
1921 | | * namePop: |
1922 | | * @ctxt: an XML parser context |
1923 | | * |
1924 | | * Pops the top element name from the name stack |
1925 | | * |
1926 | | * Returns the name just removed |
1927 | | */ |
1928 | | const xmlChar * |
1929 | | namePop(xmlParserCtxtPtr ctxt) |
1930 | 5.33M | { |
1931 | 5.33M | const xmlChar *ret; |
1932 | | |
1933 | 5.33M | if ((ctxt == NULL) || (ctxt->nameNr <= 0)) |
1934 | 0 | return (NULL); |
1935 | 5.33M | ctxt->nameNr--; |
1936 | 5.33M | if (ctxt->nameNr > 0) |
1937 | 5.05M | ctxt->name = ctxt->nameTab[ctxt->nameNr - 1]; |
1938 | 276k | else |
1939 | 276k | ctxt->name = NULL; |
1940 | 5.33M | ret = ctxt->nameTab[ctxt->nameNr]; |
1941 | 5.33M | ctxt->nameTab[ctxt->nameNr] = NULL; |
1942 | 5.33M | return (ret); |
1943 | 5.33M | } |
1944 | | |
1945 | 10.6M | static int spacePush(xmlParserCtxtPtr ctxt, int val) { |
1946 | 10.6M | if (ctxt->spaceNr >= ctxt->spaceMax) { |
1947 | 15.8k | int *tmp; |
1948 | | |
1949 | 15.8k | ctxt->spaceMax *= 2; |
1950 | 15.8k | tmp = (int *) xmlRealloc(ctxt->spaceTab, |
1951 | 15.8k | ctxt->spaceMax * sizeof(ctxt->spaceTab[0])); |
1952 | 15.8k | if (tmp == NULL) { |
1953 | 0 | xmlErrMemory(ctxt, NULL); |
1954 | 0 | ctxt->spaceMax /=2; |
1955 | 0 | return(-1); |
1956 | 0 | } |
1957 | 15.8k | ctxt->spaceTab = tmp; |
1958 | 15.8k | } |
1959 | 10.6M | ctxt->spaceTab[ctxt->spaceNr] = val; |
1960 | 10.6M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr]; |
1961 | 10.6M | return(ctxt->spaceNr++); |
1962 | 10.6M | } |
1963 | | |
1964 | 10.1M | static int spacePop(xmlParserCtxtPtr ctxt) { |
1965 | 10.1M | int ret; |
1966 | 10.1M | if (ctxt->spaceNr <= 0) return(0); |
1967 | 10.1M | ctxt->spaceNr--; |
1968 | 10.1M | if (ctxt->spaceNr > 0) |
1969 | 10.1M | ctxt->space = &ctxt->spaceTab[ctxt->spaceNr - 1]; |
1970 | 58.9k | else |
1971 | 58.9k | ctxt->space = &ctxt->spaceTab[0]; |
1972 | 10.1M | ret = ctxt->spaceTab[ctxt->spaceNr]; |
1973 | 10.1M | ctxt->spaceTab[ctxt->spaceNr] = -1; |
1974 | 10.1M | return(ret); |
1975 | 10.1M | } |
1976 | | |
1977 | | /* |
1978 | | * Macros for accessing the content. Those should be used only by the parser, |
1979 | | * and not exported. |
1980 | | * |
1981 | | * Dirty macros, i.e. one often need to make assumption on the context to |
1982 | | * use them |
1983 | | * |
1984 | | * CUR_PTR return the current pointer to the xmlChar to be parsed. |
1985 | | * To be used with extreme caution since operations consuming |
1986 | | * characters may move the input buffer to a different location ! |
1987 | | * CUR returns the current xmlChar value, i.e. a 8 bit value if compiled |
1988 | | * This should be used internally by the parser |
1989 | | * only to compare to ASCII values otherwise it would break when |
1990 | | * running with UTF-8 encoding. |
1991 | | * RAW same as CUR but in the input buffer, bypass any token |
1992 | | * extraction that may have been done |
1993 | | * NXT(n) returns the n'th next xmlChar. Same as CUR is should be used only |
1994 | | * to compare on ASCII based substring. |
1995 | | * SKIP(n) Skip n xmlChar, and must also be used only to skip ASCII defined |
1996 | | * strings without newlines within the parser. |
1997 | | * NEXT1(l) Skip 1 xmlChar, and must also be used only to skip 1 non-newline ASCII |
1998 | | * defined char within the parser. |
1999 | | * Clean macros, not dependent of an ASCII context, expect UTF-8 encoding |
2000 | | * |
2001 | | * NEXT Skip to the next character, this does the proper decoding |
2002 | | * in UTF-8 mode. It also pop-up unfinished entities on the fly. |
2003 | | * NEXTL(l) Skip the current unicode character of l xmlChars long. |
2004 | | * CUR_CHAR(l) returns the current unicode character (int), set l |
2005 | | * to the number of xmlChars used for the encoding [0-5]. |
2006 | | * CUR_SCHAR same but operate on a string instead of the context |
2007 | | * COPY_BUF copy the current unicode char to the target buffer, increment |
2008 | | * the index |
2009 | | * GROW, SHRINK handling of input buffers |
2010 | | */ |
2011 | | |
2012 | 405M | #define RAW (*ctxt->input->cur) |
2013 | 333M | #define CUR (*ctxt->input->cur) |
2014 | 266M | #define NXT(val) ctxt->input->cur[(val)] |
2015 | 21.8M | #define CUR_PTR ctxt->input->cur |
2016 | 552k | #define BASE_PTR ctxt->input->base |
2017 | | |
2018 | | #define CMP4( s, c1, c2, c3, c4 ) \ |
2019 | 99.1M | ( ((unsigned char *) s)[ 0 ] == c1 && ((unsigned char *) s)[ 1 ] == c2 && \ |
2020 | 49.7M | ((unsigned char *) s)[ 2 ] == c3 && ((unsigned char *) s)[ 3 ] == c4 ) |
2021 | | #define CMP5( s, c1, c2, c3, c4, c5 ) \ |
2022 | 90.1M | ( CMP4( s, c1, c2, c3, c4 ) && ((unsigned char *) s)[ 4 ] == c5 ) |
2023 | | #define CMP6( s, c1, c2, c3, c4, c5, c6 ) \ |
2024 | 75.5M | ( CMP5( s, c1, c2, c3, c4, c5 ) && ((unsigned char *) s)[ 5 ] == c6 ) |
2025 | | #define CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) \ |
2026 | 63.8M | ( CMP6( s, c1, c2, c3, c4, c5, c6 ) && ((unsigned char *) s)[ 6 ] == c7 ) |
2027 | | #define CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) \ |
2028 | 50.8M | ( CMP7( s, c1, c2, c3, c4, c5, c6, c7 ) && ((unsigned char *) s)[ 7 ] == c8 ) |
2029 | | #define CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) \ |
2030 | 21.9M | ( CMP8( s, c1, c2, c3, c4, c5, c6, c7, c8 ) && \ |
2031 | 21.9M | ((unsigned char *) s)[ 8 ] == c9 ) |
2032 | | #define CMP10( s, c1, c2, c3, c4, c5, c6, c7, c8, c9, c10 ) \ |
2033 | 128k | ( CMP9( s, c1, c2, c3, c4, c5, c6, c7, c8, c9 ) && \ |
2034 | 128k | ((unsigned char *) s)[ 9 ] == c10 ) |
2035 | | |
2036 | 113M | #define SKIP(val) do { \ |
2037 | 113M | ctxt->input->cur += (val),ctxt->input->col+=(val); \ |
2038 | 113M | if (*ctxt->input->cur == 0) \ |
2039 | 113M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2040 | 113M | } while (0) |
2041 | | |
2042 | 51.1k | #define SKIPL(val) do { \ |
2043 | 51.1k | int skipl; \ |
2044 | 12.0M | for(skipl=0; skipl<val; skipl++) { \ |
2045 | 11.9M | if (*(ctxt->input->cur) == '\n') { \ |
2046 | 239k | ctxt->input->line++; ctxt->input->col = 1; \ |
2047 | 11.7M | } else ctxt->input->col++; \ |
2048 | 11.9M | ctxt->input->cur++; \ |
2049 | 11.9M | } \ |
2050 | 51.1k | if (*ctxt->input->cur == 0) \ |
2051 | 51.1k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2052 | 51.1k | } while (0) |
2053 | | |
2054 | 114M | #define SHRINK if ((ctxt->progressive == 0) && \ |
2055 | 114M | (ctxt->input->cur - ctxt->input->base > 2 * INPUT_CHUNK) && \ |
2056 | 114M | (ctxt->input->end - ctxt->input->cur < 2 * INPUT_CHUNK)) \ |
2057 | 114M | xmlSHRINK (ctxt); |
2058 | | |
2059 | 1.66M | static void xmlSHRINK (xmlParserCtxtPtr ctxt) { |
2060 | | /* Don't shrink memory buffers. */ |
2061 | 1.66M | if ((ctxt->input->buf) && |
2062 | 1.66M | ((ctxt->input->buf->encoder) || (ctxt->input->buf->readcallback))) |
2063 | 8.66k | xmlParserInputShrink(ctxt->input); |
2064 | 1.66M | if (*ctxt->input->cur == 0) |
2065 | 70.2k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2066 | 1.66M | } |
2067 | | |
2068 | 419M | #define GROW if ((ctxt->progressive == 0) && \ |
2069 | 419M | (ctxt->input->end - ctxt->input->cur < INPUT_CHUNK)) \ |
2070 | 419M | xmlGROW (ctxt); |
2071 | | |
2072 | 72.1M | static void xmlGROW (xmlParserCtxtPtr ctxt) { |
2073 | 72.1M | ptrdiff_t curEnd = ctxt->input->end - ctxt->input->cur; |
2074 | 72.1M | ptrdiff_t curBase = ctxt->input->cur - ctxt->input->base; |
2075 | | |
2076 | 72.1M | if (((curEnd > XML_MAX_LOOKUP_LIMIT) || |
2077 | 72.1M | (curBase > XML_MAX_LOOKUP_LIMIT)) && |
2078 | 72.1M | ((ctxt->input->buf) && |
2079 | 0 | (ctxt->input->buf->readcallback != NULL)) && |
2080 | 72.1M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
2081 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
2082 | 0 | xmlHaltParser(ctxt); |
2083 | 0 | return; |
2084 | 0 | } |
2085 | 72.1M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2086 | 72.1M | if ((ctxt->input->cur > ctxt->input->end) || |
2087 | 72.1M | (ctxt->input->cur < ctxt->input->base)) { |
2088 | 0 | xmlHaltParser(ctxt); |
2089 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "cur index out of bound"); |
2090 | 0 | return; |
2091 | 0 | } |
2092 | 72.1M | if ((ctxt->input->cur != NULL) && (*ctxt->input->cur == 0)) |
2093 | 943k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2094 | 72.1M | } |
2095 | | |
2096 | 116M | #define SKIP_BLANKS xmlSkipBlankChars(ctxt) |
2097 | | |
2098 | 247M | #define NEXT xmlNextChar(ctxt) |
2099 | | |
2100 | 18.1M | #define NEXT1 { \ |
2101 | 18.1M | ctxt->input->col++; \ |
2102 | 18.1M | ctxt->input->cur++; \ |
2103 | 18.1M | if (*ctxt->input->cur == 0) \ |
2104 | 18.1M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); \ |
2105 | 18.1M | } |
2106 | | |
2107 | 241M | #define NEXTL(l) do { \ |
2108 | 241M | if (*(ctxt->input->cur) == '\n') { \ |
2109 | 4.94M | ctxt->input->line++; ctxt->input->col = 1; \ |
2110 | 236M | } else ctxt->input->col++; \ |
2111 | 241M | ctxt->input->cur += l; \ |
2112 | 241M | } while (0) |
2113 | | |
2114 | 250M | #define CUR_CHAR(l) xmlCurrentChar(ctxt, &l) |
2115 | 2.36G | #define CUR_SCHAR(s, l) xmlStringCurrentChar(ctxt, s, &l) |
2116 | | |
2117 | | #define COPY_BUF(l,b,i,v) \ |
2118 | 2.51G | if (l == 1) b[i++] = v; \ |
2119 | 2.51G | else i += xmlCopyCharMultiByte(&b[i],v) |
2120 | | |
2121 | | /** |
2122 | | * xmlSkipBlankChars: |
2123 | | * @ctxt: the XML parser context |
2124 | | * |
2125 | | * skip all blanks character found at that point in the input streams. |
2126 | | * It pops up finished entities in the process if allowable at that point. |
2127 | | * |
2128 | | * Returns the number of space chars skipped |
2129 | | */ |
2130 | | |
2131 | | int |
2132 | 116M | xmlSkipBlankChars(xmlParserCtxtPtr ctxt) { |
2133 | 116M | int res = 0; |
2134 | | |
2135 | | /* |
2136 | | * It's Okay to use CUR/NEXT here since all the blanks are on |
2137 | | * the ASCII range. |
2138 | | */ |
2139 | 116M | if (((ctxt->inputNr == 1) && (ctxt->instate != XML_PARSER_DTD)) || |
2140 | 116M | (ctxt->instate == XML_PARSER_START)) { |
2141 | 46.4M | const xmlChar *cur; |
2142 | | /* |
2143 | | * if we are in the document content, go really fast |
2144 | | */ |
2145 | 46.4M | cur = ctxt->input->cur; |
2146 | 46.4M | while (IS_BLANK_CH(*cur)) { |
2147 | 19.5M | if (*cur == '\n') { |
2148 | 952k | ctxt->input->line++; ctxt->input->col = 1; |
2149 | 18.5M | } else { |
2150 | 18.5M | ctxt->input->col++; |
2151 | 18.5M | } |
2152 | 19.5M | cur++; |
2153 | 19.5M | if (res < INT_MAX) |
2154 | 19.5M | res++; |
2155 | 19.5M | if (*cur == 0) { |
2156 | 79.0k | ctxt->input->cur = cur; |
2157 | 79.0k | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2158 | 79.0k | cur = ctxt->input->cur; |
2159 | 79.0k | } |
2160 | 19.5M | } |
2161 | 46.4M | ctxt->input->cur = cur; |
2162 | 70.2M | } else { |
2163 | 70.2M | int expandPE = ((ctxt->external != 0) || (ctxt->inputNr != 1)); |
2164 | | |
2165 | 240M | while (ctxt->instate != XML_PARSER_EOF) { |
2166 | 240M | if (IS_BLANK_CH(CUR)) { /* CHECKED tstblanks.xml */ |
2167 | 104M | NEXT; |
2168 | 136M | } else if (CUR == '%') { |
2169 | | /* |
2170 | | * Need to handle support of entities branching here |
2171 | | */ |
2172 | 37.4M | if ((expandPE == 0) || (IS_BLANK_CH(NXT(1))) || (NXT(1) == 0)) |
2173 | 574k | break; |
2174 | 36.8M | xmlParsePEReference(ctxt); |
2175 | 98.8M | } else if (CUR == 0) { |
2176 | 29.2M | unsigned long consumed; |
2177 | 29.2M | xmlEntityPtr ent; |
2178 | | |
2179 | 29.2M | if (ctxt->inputNr <= 1) |
2180 | 34.2k | break; |
2181 | | |
2182 | 29.2M | consumed = ctxt->input->consumed; |
2183 | 29.2M | xmlSaturatedAddSizeT(&consumed, |
2184 | 29.2M | ctxt->input->cur - ctxt->input->base); |
2185 | | |
2186 | | /* |
2187 | | * Add to sizeentities when parsing an external entity |
2188 | | * for the first time. |
2189 | | */ |
2190 | 29.2M | ent = ctxt->input->entity; |
2191 | 29.2M | if ((ent->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
2192 | 29.2M | ((ent->flags & XML_ENT_PARSED) == 0)) { |
2193 | 18.0k | ent->flags |= XML_ENT_PARSED; |
2194 | | |
2195 | 18.0k | xmlSaturatedAdd(&ctxt->sizeentities, consumed); |
2196 | 18.0k | } |
2197 | | |
2198 | 29.2M | xmlParserEntityCheck(ctxt, consumed); |
2199 | | |
2200 | 29.2M | xmlPopInput(ctxt); |
2201 | 69.5M | } else { |
2202 | 69.5M | break; |
2203 | 69.5M | } |
2204 | | |
2205 | | /* |
2206 | | * Also increase the counter when entering or exiting a PERef. |
2207 | | * The spec says: "When a parameter-entity reference is recognized |
2208 | | * in the DTD and included, its replacement text MUST be enlarged |
2209 | | * by the attachment of one leading and one following space (#x20) |
2210 | | * character." |
2211 | | */ |
2212 | 170M | if (res < INT_MAX) |
2213 | 170M | res++; |
2214 | 170M | } |
2215 | 70.2M | } |
2216 | 116M | return(res); |
2217 | 116M | } |
2218 | | |
2219 | | /************************************************************************ |
2220 | | * * |
2221 | | * Commodity functions to handle entities * |
2222 | | * * |
2223 | | ************************************************************************/ |
2224 | | |
2225 | | /** |
2226 | | * xmlPopInput: |
2227 | | * @ctxt: an XML parser context |
2228 | | * |
2229 | | * xmlPopInput: the current input pointed by ctxt->input came to an end |
2230 | | * pop it and return the next char. |
2231 | | * |
2232 | | * Returns the current xmlChar in the parser context |
2233 | | */ |
2234 | | xmlChar |
2235 | 29.2M | xmlPopInput(xmlParserCtxtPtr ctxt) { |
2236 | 29.2M | xmlParserInputPtr input; |
2237 | | |
2238 | 29.2M | if ((ctxt == NULL) || (ctxt->inputNr <= 1)) return(0); |
2239 | 29.2M | if (xmlParserDebugEntities) |
2240 | 0 | xmlGenericError(xmlGenericErrorContext, |
2241 | 0 | "Popping input %d\n", ctxt->inputNr); |
2242 | 29.2M | if ((ctxt->inputNr > 1) && (ctxt->inSubset == 0) && |
2243 | 29.2M | (ctxt->instate != XML_PARSER_EOF)) |
2244 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
2245 | 0 | "Unfinished entity outside the DTD"); |
2246 | 29.2M | input = inputPop(ctxt); |
2247 | 29.2M | if (input->entity != NULL) |
2248 | 29.2M | input->entity->flags &= ~XML_ENT_EXPANDING; |
2249 | 29.2M | xmlFreeInputStream(input); |
2250 | 29.2M | if (*ctxt->input->cur == 0) |
2251 | 13.8M | xmlParserInputGrow(ctxt->input, INPUT_CHUNK); |
2252 | 29.2M | return(CUR); |
2253 | 29.2M | } |
2254 | | |
2255 | | /** |
2256 | | * xmlPushInput: |
2257 | | * @ctxt: an XML parser context |
2258 | | * @input: an XML parser input fragment (entity, XML fragment ...). |
2259 | | * |
2260 | | * xmlPushInput: switch to a new input stream which is stacked on top |
2261 | | * of the previous one(s). |
2262 | | * Returns -1 in case of error or the index in the input stack |
2263 | | */ |
2264 | | int |
2265 | 29.2M | xmlPushInput(xmlParserCtxtPtr ctxt, xmlParserInputPtr input) { |
2266 | 29.2M | int ret; |
2267 | 29.2M | if (input == NULL) return(-1); |
2268 | | |
2269 | 29.2M | if (xmlParserDebugEntities) { |
2270 | 0 | if ((ctxt->input != NULL) && (ctxt->input->filename)) |
2271 | 0 | xmlGenericError(xmlGenericErrorContext, |
2272 | 0 | "%s(%d): ", ctxt->input->filename, |
2273 | 0 | ctxt->input->line); |
2274 | 0 | xmlGenericError(xmlGenericErrorContext, |
2275 | 0 | "Pushing input %d : %.30s\n", ctxt->inputNr+1, input->cur); |
2276 | 0 | } |
2277 | 29.2M | if (((ctxt->inputNr > 40) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2278 | 29.2M | (ctxt->inputNr > 100)) { |
2279 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2280 | 0 | while (ctxt->inputNr > 1) |
2281 | 0 | xmlFreeInputStream(inputPop(ctxt)); |
2282 | 0 | return(-1); |
2283 | 0 | } |
2284 | 29.2M | ret = inputPush(ctxt, input); |
2285 | 29.2M | if (ctxt->instate == XML_PARSER_EOF) |
2286 | 0 | return(-1); |
2287 | 29.2M | GROW; |
2288 | 29.2M | return(ret); |
2289 | 29.2M | } |
2290 | | |
2291 | | /** |
2292 | | * xmlParseCharRef: |
2293 | | * @ctxt: an XML parser context |
2294 | | * |
2295 | | * DEPRECATED: Internal function, don't use. |
2296 | | * |
2297 | | * Parse a numeric character reference. Always consumes '&'. |
2298 | | * |
2299 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2300 | | * '&#x' [0-9a-fA-F]+ ';' |
2301 | | * |
2302 | | * [ WFC: Legal Character ] |
2303 | | * Characters referred to using character references must match the |
2304 | | * production for Char. |
2305 | | * |
2306 | | * Returns the value parsed (as an int), 0 in case of error |
2307 | | */ |
2308 | | int |
2309 | 487k | xmlParseCharRef(xmlParserCtxtPtr ctxt) { |
2310 | 487k | int val = 0; |
2311 | 487k | int count = 0; |
2312 | | |
2313 | | /* |
2314 | | * Using RAW/CUR/NEXT is okay since we are working on ASCII range here |
2315 | | */ |
2316 | 487k | if ((RAW == '&') && (NXT(1) == '#') && |
2317 | 487k | (NXT(2) == 'x')) { |
2318 | 192k | SKIP(3); |
2319 | 192k | GROW; |
2320 | 643k | while (RAW != ';') { /* loop blocked by count */ |
2321 | 470k | if (count++ > 20) { |
2322 | 20.4k | count = 0; |
2323 | 20.4k | GROW; |
2324 | 20.4k | if (ctxt->instate == XML_PARSER_EOF) |
2325 | 0 | return(0); |
2326 | 20.4k | } |
2327 | 470k | if ((RAW >= '0') && (RAW <= '9')) |
2328 | 291k | val = val * 16 + (CUR - '0'); |
2329 | 178k | else if ((RAW >= 'a') && (RAW <= 'f') && (count < 20)) |
2330 | 139k | val = val * 16 + (CUR - 'a') + 10; |
2331 | 39.1k | else if ((RAW >= 'A') && (RAW <= 'F') && (count < 20)) |
2332 | 19.6k | val = val * 16 + (CUR - 'A') + 10; |
2333 | 19.5k | else { |
2334 | 19.5k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2335 | 19.5k | val = 0; |
2336 | 19.5k | break; |
2337 | 19.5k | } |
2338 | 450k | if (val > 0x110000) |
2339 | 225k | val = 0x110000; |
2340 | | |
2341 | 450k | NEXT; |
2342 | 450k | count++; |
2343 | 450k | } |
2344 | 192k | if (RAW == ';') { |
2345 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2346 | 172k | ctxt->input->col++; |
2347 | 172k | ctxt->input->cur++; |
2348 | 172k | } |
2349 | 295k | } else if ((RAW == '&') && (NXT(1) == '#')) { |
2350 | 295k | SKIP(2); |
2351 | 295k | GROW; |
2352 | 1.30M | while (RAW != ';') { /* loop blocked by count */ |
2353 | 1.04M | if (count++ > 20) { |
2354 | 22.4k | count = 0; |
2355 | 22.4k | GROW; |
2356 | 22.4k | if (ctxt->instate == XML_PARSER_EOF) |
2357 | 0 | return(0); |
2358 | 22.4k | } |
2359 | 1.04M | if ((RAW >= '0') && (RAW <= '9')) |
2360 | 1.01M | val = val * 10 + (CUR - '0'); |
2361 | 33.5k | else { |
2362 | 33.5k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2363 | 33.5k | val = 0; |
2364 | 33.5k | break; |
2365 | 33.5k | } |
2366 | 1.01M | if (val > 0x110000) |
2367 | 244k | val = 0x110000; |
2368 | | |
2369 | 1.01M | NEXT; |
2370 | 1.01M | count++; |
2371 | 1.01M | } |
2372 | 295k | if (RAW == ';') { |
2373 | | /* on purpose to avoid reentrancy problems with NEXT and SKIP */ |
2374 | 261k | ctxt->input->col++; |
2375 | 261k | ctxt->input->cur++; |
2376 | 261k | } |
2377 | 295k | } else { |
2378 | 0 | if (RAW == '&') |
2379 | 0 | SKIP(1); |
2380 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2381 | 0 | } |
2382 | | |
2383 | | /* |
2384 | | * [ WFC: Legal Character ] |
2385 | | * Characters referred to using character references must match the |
2386 | | * production for Char. |
2387 | | */ |
2388 | 487k | if (val >= 0x110000) { |
2389 | 1.58k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2390 | 1.58k | "xmlParseCharRef: character reference out of bounds\n", |
2391 | 1.58k | val); |
2392 | 485k | } else if (IS_CHAR(val)) { |
2393 | 427k | return(val); |
2394 | 427k | } else { |
2395 | 58.3k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2396 | 58.3k | "xmlParseCharRef: invalid xmlChar value %d\n", |
2397 | 58.3k | val); |
2398 | 58.3k | } |
2399 | 59.9k | return(0); |
2400 | 487k | } |
2401 | | |
2402 | | /** |
2403 | | * xmlParseStringCharRef: |
2404 | | * @ctxt: an XML parser context |
2405 | | * @str: a pointer to an index in the string |
2406 | | * |
2407 | | * parse Reference declarations, variant parsing from a string rather |
2408 | | * than an an input flow. |
2409 | | * |
2410 | | * [66] CharRef ::= '&#' [0-9]+ ';' | |
2411 | | * '&#x' [0-9a-fA-F]+ ';' |
2412 | | * |
2413 | | * [ WFC: Legal Character ] |
2414 | | * Characters referred to using character references must match the |
2415 | | * production for Char. |
2416 | | * |
2417 | | * Returns the value parsed (as an int), 0 in case of error, str will be |
2418 | | * updated to the current value of the index |
2419 | | */ |
2420 | | static int |
2421 | 252k | xmlParseStringCharRef(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
2422 | 252k | const xmlChar *ptr; |
2423 | 252k | xmlChar cur; |
2424 | 252k | int val = 0; |
2425 | | |
2426 | 252k | if ((str == NULL) || (*str == NULL)) return(0); |
2427 | 252k | ptr = *str; |
2428 | 252k | cur = *ptr; |
2429 | 252k | if ((cur == '&') && (ptr[1] == '#') && (ptr[2] == 'x')) { |
2430 | 43.3k | ptr += 3; |
2431 | 43.3k | cur = *ptr; |
2432 | 111k | while (cur != ';') { /* Non input consuming loop */ |
2433 | 69.7k | if ((cur >= '0') && (cur <= '9')) |
2434 | 27.0k | val = val * 16 + (cur - '0'); |
2435 | 42.6k | else if ((cur >= 'a') && (cur <= 'f')) |
2436 | 9.98k | val = val * 16 + (cur - 'a') + 10; |
2437 | 32.7k | else if ((cur >= 'A') && (cur <= 'F')) |
2438 | 30.9k | val = val * 16 + (cur - 'A') + 10; |
2439 | 1.71k | else { |
2440 | 1.71k | xmlFatalErr(ctxt, XML_ERR_INVALID_HEX_CHARREF, NULL); |
2441 | 1.71k | val = 0; |
2442 | 1.71k | break; |
2443 | 1.71k | } |
2444 | 67.9k | if (val > 0x110000) |
2445 | 16.4k | val = 0x110000; |
2446 | | |
2447 | 67.9k | ptr++; |
2448 | 67.9k | cur = *ptr; |
2449 | 67.9k | } |
2450 | 43.3k | if (cur == ';') |
2451 | 41.5k | ptr++; |
2452 | 209k | } else if ((cur == '&') && (ptr[1] == '#')){ |
2453 | 209k | ptr += 2; |
2454 | 209k | cur = *ptr; |
2455 | 711k | while (cur != ';') { /* Non input consuming loops */ |
2456 | 505k | if ((cur >= '0') && (cur <= '9')) |
2457 | 501k | val = val * 10 + (cur - '0'); |
2458 | 3.59k | else { |
2459 | 3.59k | xmlFatalErr(ctxt, XML_ERR_INVALID_DEC_CHARREF, NULL); |
2460 | 3.59k | val = 0; |
2461 | 3.59k | break; |
2462 | 3.59k | } |
2463 | 501k | if (val > 0x110000) |
2464 | 2.46k | val = 0x110000; |
2465 | | |
2466 | 501k | ptr++; |
2467 | 501k | cur = *ptr; |
2468 | 501k | } |
2469 | 209k | if (cur == ';') |
2470 | 206k | ptr++; |
2471 | 209k | } else { |
2472 | 0 | xmlFatalErr(ctxt, XML_ERR_INVALID_CHARREF, NULL); |
2473 | 0 | return(0); |
2474 | 0 | } |
2475 | 252k | *str = ptr; |
2476 | | |
2477 | | /* |
2478 | | * [ WFC: Legal Character ] |
2479 | | * Characters referred to using character references must match the |
2480 | | * production for Char. |
2481 | | */ |
2482 | 252k | if (val >= 0x110000) { |
2483 | 386 | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2484 | 386 | "xmlParseStringCharRef: character reference out of bounds\n", |
2485 | 386 | val); |
2486 | 252k | } else if (IS_CHAR(val)) { |
2487 | 246k | return(val); |
2488 | 246k | } else { |
2489 | 6.56k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
2490 | 6.56k | "xmlParseStringCharRef: invalid xmlChar value %d\n", |
2491 | 6.56k | val); |
2492 | 6.56k | } |
2493 | 6.95k | return(0); |
2494 | 252k | } |
2495 | | |
2496 | | /** |
2497 | | * xmlParserHandlePEReference: |
2498 | | * @ctxt: the parser context |
2499 | | * |
2500 | | * [69] PEReference ::= '%' Name ';' |
2501 | | * |
2502 | | * [ WFC: No Recursion ] |
2503 | | * A parsed entity must not contain a recursive |
2504 | | * reference to itself, either directly or indirectly. |
2505 | | * |
2506 | | * [ WFC: Entity Declared ] |
2507 | | * In a document without any DTD, a document with only an internal DTD |
2508 | | * subset which contains no parameter entity references, or a document |
2509 | | * with "standalone='yes'", ... ... The declaration of a parameter |
2510 | | * entity must precede any reference to it... |
2511 | | * |
2512 | | * [ VC: Entity Declared ] |
2513 | | * In a document with an external subset or external parameter entities |
2514 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
2515 | | * must precede any reference to it... |
2516 | | * |
2517 | | * [ WFC: In DTD ] |
2518 | | * Parameter-entity references may only appear in the DTD. |
2519 | | * NOTE: misleading but this is handled. |
2520 | | * |
2521 | | * A PEReference may have been detected in the current input stream |
2522 | | * the handling is done accordingly to |
2523 | | * http://www.w3.org/TR/REC-xml#entproc |
2524 | | * i.e. |
2525 | | * - Included in literal in entity values |
2526 | | * - Included as Parameter Entity reference within DTDs |
2527 | | */ |
2528 | | void |
2529 | 0 | xmlParserHandlePEReference(xmlParserCtxtPtr ctxt) { |
2530 | 0 | switch(ctxt->instate) { |
2531 | 0 | case XML_PARSER_CDATA_SECTION: |
2532 | 0 | return; |
2533 | 0 | case XML_PARSER_COMMENT: |
2534 | 0 | return; |
2535 | 0 | case XML_PARSER_START_TAG: |
2536 | 0 | return; |
2537 | 0 | case XML_PARSER_END_TAG: |
2538 | 0 | return; |
2539 | 0 | case XML_PARSER_EOF: |
2540 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_AT_EOF, NULL); |
2541 | 0 | return; |
2542 | 0 | case XML_PARSER_PROLOG: |
2543 | 0 | case XML_PARSER_START: |
2544 | 0 | case XML_PARSER_MISC: |
2545 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_PROLOG, NULL); |
2546 | 0 | return; |
2547 | 0 | case XML_PARSER_ENTITY_DECL: |
2548 | 0 | case XML_PARSER_CONTENT: |
2549 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
2550 | 0 | case XML_PARSER_PI: |
2551 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
2552 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
2553 | | /* we just ignore it there */ |
2554 | 0 | return; |
2555 | 0 | case XML_PARSER_EPILOG: |
2556 | 0 | xmlFatalErr(ctxt, XML_ERR_PEREF_IN_EPILOG, NULL); |
2557 | 0 | return; |
2558 | 0 | case XML_PARSER_ENTITY_VALUE: |
2559 | | /* |
2560 | | * NOTE: in the case of entity values, we don't do the |
2561 | | * substitution here since we need the literal |
2562 | | * entity value to be able to save the internal |
2563 | | * subset of the document. |
2564 | | * This will be handled by xmlStringDecodeEntities |
2565 | | */ |
2566 | 0 | return; |
2567 | 0 | case XML_PARSER_DTD: |
2568 | | /* |
2569 | | * [WFC: Well-Formedness Constraint: PEs in Internal Subset] |
2570 | | * In the internal DTD subset, parameter-entity references |
2571 | | * can occur only where markup declarations can occur, not |
2572 | | * within markup declarations. |
2573 | | * In that case this is handled in xmlParseMarkupDecl |
2574 | | */ |
2575 | 0 | if ((ctxt->external == 0) && (ctxt->inputNr == 1)) |
2576 | 0 | return; |
2577 | 0 | if (IS_BLANK_CH(NXT(1)) || NXT(1) == 0) |
2578 | 0 | return; |
2579 | 0 | break; |
2580 | 0 | case XML_PARSER_IGNORE: |
2581 | 0 | return; |
2582 | 0 | } |
2583 | | |
2584 | 0 | xmlParsePEReference(ctxt); |
2585 | 0 | } |
2586 | | |
2587 | | /* |
2588 | | * Macro used to grow the current buffer. |
2589 | | * buffer##_size is expected to be a size_t |
2590 | | * mem_error: is expected to handle memory allocation failures |
2591 | | */ |
2592 | 1.86M | #define growBuffer(buffer, n) { \ |
2593 | 1.86M | xmlChar *tmp; \ |
2594 | 1.86M | size_t new_size = buffer##_size * 2 + n; \ |
2595 | 1.86M | if (new_size < buffer##_size) goto mem_error; \ |
2596 | 1.86M | tmp = (xmlChar *) xmlRealloc(buffer, new_size); \ |
2597 | 1.86M | if (tmp == NULL) goto mem_error; \ |
2598 | 1.86M | buffer = tmp; \ |
2599 | 1.86M | buffer##_size = new_size; \ |
2600 | 1.86M | } |
2601 | | |
2602 | | /** |
2603 | | * xmlStringDecodeEntitiesInt: |
2604 | | * @ctxt: the parser context |
2605 | | * @str: the input string |
2606 | | * @len: the string length |
2607 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2608 | | * @end: an end marker xmlChar, 0 if none |
2609 | | * @end2: an end marker xmlChar, 0 if none |
2610 | | * @end3: an end marker xmlChar, 0 if none |
2611 | | * @check: whether to perform entity checks |
2612 | | */ |
2613 | | static xmlChar * |
2614 | | xmlStringDecodeEntitiesInt(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2615 | | int what, xmlChar end, xmlChar end2, xmlChar end3, |
2616 | 30.4M | int check) { |
2617 | 30.4M | xmlChar *buffer = NULL; |
2618 | 30.4M | size_t buffer_size = 0; |
2619 | 30.4M | size_t nbchars = 0; |
2620 | | |
2621 | 30.4M | xmlChar *current = NULL; |
2622 | 30.4M | xmlChar *rep = NULL; |
2623 | 30.4M | const xmlChar *last; |
2624 | 30.4M | xmlEntityPtr ent; |
2625 | 30.4M | int c,l; |
2626 | | |
2627 | 30.4M | if (str == NULL) |
2628 | 15.2k | return(NULL); |
2629 | 30.4M | last = str + len; |
2630 | | |
2631 | 30.4M | if (((ctxt->depth > 40) && |
2632 | 30.4M | ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
2633 | 30.4M | (ctxt->depth > 100)) { |
2634 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_LOOP, |
2635 | 0 | "Maximum entity nesting depth exceeded"); |
2636 | 0 | return(NULL); |
2637 | 0 | } |
2638 | | |
2639 | | /* |
2640 | | * allocate a translation buffer. |
2641 | | */ |
2642 | 30.4M | buffer_size = XML_PARSER_BIG_BUFFER_SIZE; |
2643 | 30.4M | buffer = (xmlChar *) xmlMallocAtomic(buffer_size); |
2644 | 30.4M | if (buffer == NULL) goto mem_error; |
2645 | | |
2646 | | /* |
2647 | | * OK loop until we reach one of the ending char or a size limit. |
2648 | | * we are operating on already parsed values. |
2649 | | */ |
2650 | 30.4M | if (str < last) |
2651 | 30.3M | c = CUR_SCHAR(str, l); |
2652 | 87.0k | else |
2653 | 87.0k | c = 0; |
2654 | 1.78G | while ((c != 0) && (c != end) && /* non input consuming loop */ |
2655 | 1.78G | (c != end2) && (c != end3) && |
2656 | 1.78G | (ctxt->instate != XML_PARSER_EOF)) { |
2657 | | |
2658 | 1.75G | if (c == 0) break; |
2659 | 1.75G | if ((c == '&') && (str[1] == '#')) { |
2660 | 252k | int val = xmlParseStringCharRef(ctxt, &str); |
2661 | 252k | if (val == 0) |
2662 | 6.95k | goto int_error; |
2663 | 246k | COPY_BUF(0,buffer,nbchars,val); |
2664 | 246k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2665 | 288 | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2666 | 288 | } |
2667 | 1.75G | } else if ((c == '&') && (what & XML_SUBSTITUTE_REF)) { |
2668 | 29.4M | if (xmlParserDebugEntities) |
2669 | 0 | xmlGenericError(xmlGenericErrorContext, |
2670 | 0 | "String decoding Entity Reference: %.30s\n", |
2671 | 0 | str); |
2672 | 29.4M | ent = xmlParseStringEntityRef(ctxt, &str); |
2673 | 29.4M | if ((ent != NULL) && |
2674 | 29.4M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
2675 | 160k | if (ent->content != NULL) { |
2676 | 160k | COPY_BUF(0,buffer,nbchars,ent->content[0]); |
2677 | 160k | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2678 | 8.03k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2679 | 8.03k | } |
2680 | 160k | } else { |
2681 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
2682 | 0 | "predefined entity has no content\n"); |
2683 | 0 | goto int_error; |
2684 | 0 | } |
2685 | 29.2M | } else if ((ent != NULL) && (ent->content != NULL)) { |
2686 | 27.5M | if ((check) && (xmlParserEntityCheck(ctxt, ent->length))) |
2687 | 209 | goto int_error; |
2688 | | |
2689 | 27.5M | if (ent->flags & XML_ENT_EXPANDING) { |
2690 | 376 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2691 | 376 | xmlHaltParser(ctxt); |
2692 | 376 | ent->content[0] = 0; |
2693 | 376 | goto int_error; |
2694 | 376 | } |
2695 | | |
2696 | 27.5M | ent->flags |= XML_ENT_EXPANDING; |
2697 | 27.5M | ctxt->depth++; |
2698 | 27.5M | rep = xmlStringDecodeEntitiesInt(ctxt, ent->content, |
2699 | 27.5M | ent->length, what, 0, 0, 0, check); |
2700 | 27.5M | ctxt->depth--; |
2701 | 27.5M | ent->flags &= ~XML_ENT_EXPANDING; |
2702 | | |
2703 | 27.5M | if (rep == NULL) { |
2704 | 3.73k | ent->content[0] = 0; |
2705 | 3.73k | goto int_error; |
2706 | 3.73k | } |
2707 | | |
2708 | 27.5M | current = rep; |
2709 | 4.25G | while (*current != 0) { /* non input consuming loop */ |
2710 | 4.22G | buffer[nbchars++] = *current++; |
2711 | 4.22G | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2712 | 3.10M | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2713 | 3.10M | } |
2714 | 4.22G | } |
2715 | 27.5M | xmlFree(rep); |
2716 | 27.5M | rep = NULL; |
2717 | 27.5M | } else if (ent != NULL) { |
2718 | 99.8k | int i = xmlStrlen(ent->name); |
2719 | 99.8k | const xmlChar *cur = ent->name; |
2720 | | |
2721 | 99.8k | buffer[nbchars++] = '&'; |
2722 | 99.8k | if (nbchars + i + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2723 | 1.68k | growBuffer(buffer, i + XML_PARSER_BUFFER_SIZE); |
2724 | 1.68k | } |
2725 | 399k | for (;i > 0;i--) |
2726 | 299k | buffer[nbchars++] = *cur++; |
2727 | 99.8k | buffer[nbchars++] = ';'; |
2728 | 99.8k | } |
2729 | 1.72G | } else if (c == '%' && (what & XML_SUBSTITUTE_PEREF)) { |
2730 | 706k | if (xmlParserDebugEntities) |
2731 | 0 | xmlGenericError(xmlGenericErrorContext, |
2732 | 0 | "String decoding PE Reference: %.30s\n", str); |
2733 | 706k | ent = xmlParseStringPEReference(ctxt, &str); |
2734 | 706k | if (ent != NULL) { |
2735 | 445k | if (ent->content == NULL) { |
2736 | | /* |
2737 | | * Note: external parsed entities will not be loaded, |
2738 | | * it is not required for a non-validating parser to |
2739 | | * complete external PEReferences coming from the |
2740 | | * internal subset |
2741 | | */ |
2742 | 4.99k | if (((ctxt->options & XML_PARSE_NOENT) != 0) || |
2743 | 4.99k | ((ctxt->options & XML_PARSE_DTDVALID) != 0) || |
2744 | 4.99k | (ctxt->validate != 0)) { |
2745 | 4.38k | xmlLoadEntityContent(ctxt, ent); |
2746 | 4.38k | } else { |
2747 | 608 | xmlWarningMsg(ctxt, XML_ERR_ENTITY_PROCESSING, |
2748 | 608 | "not validating will not read content for PE entity %s\n", |
2749 | 608 | ent->name, NULL); |
2750 | 608 | } |
2751 | 4.99k | } |
2752 | | |
2753 | 445k | if ((check) && (xmlParserEntityCheck(ctxt, ent->length))) |
2754 | 99 | goto int_error; |
2755 | | |
2756 | 444k | if (ent->flags & XML_ENT_EXPANDING) { |
2757 | 243 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
2758 | 243 | xmlHaltParser(ctxt); |
2759 | 243 | if (ent->content != NULL) |
2760 | 114 | ent->content[0] = 0; |
2761 | 243 | goto int_error; |
2762 | 243 | } |
2763 | | |
2764 | 444k | ent->flags |= XML_ENT_EXPANDING; |
2765 | 444k | ctxt->depth++; |
2766 | 444k | rep = xmlStringDecodeEntitiesInt(ctxt, ent->content, |
2767 | 444k | ent->length, what, 0, 0, 0, check); |
2768 | 444k | ctxt->depth--; |
2769 | 444k | ent->flags &= ~XML_ENT_EXPANDING; |
2770 | | |
2771 | 444k | if (rep == NULL) { |
2772 | 3.33k | if (ent->content != NULL) |
2773 | 324 | ent->content[0] = 0; |
2774 | 3.33k | goto int_error; |
2775 | 3.33k | } |
2776 | 441k | current = rep; |
2777 | 907M | while (*current != 0) { /* non input consuming loop */ |
2778 | 907M | buffer[nbchars++] = *current++; |
2779 | 907M | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2780 | 93.7k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2781 | 93.7k | } |
2782 | 907M | } |
2783 | 441k | xmlFree(rep); |
2784 | 441k | rep = NULL; |
2785 | 441k | } |
2786 | 1.72G | } else { |
2787 | 1.72G | COPY_BUF(l,buffer,nbchars,c); |
2788 | 1.72G | str += l; |
2789 | 1.72G | if (nbchars + XML_PARSER_BUFFER_SIZE > buffer_size) { |
2790 | 388k | growBuffer(buffer, XML_PARSER_BUFFER_SIZE); |
2791 | 388k | } |
2792 | 1.72G | } |
2793 | 1.75G | if (str < last) |
2794 | 1.72G | c = CUR_SCHAR(str, l); |
2795 | 30.3M | else |
2796 | 30.3M | c = 0; |
2797 | 1.75G | } |
2798 | 30.3M | buffer[nbchars] = 0; |
2799 | 30.3M | return(buffer); |
2800 | | |
2801 | 0 | mem_error: |
2802 | 0 | xmlErrMemory(ctxt, NULL); |
2803 | 14.9k | int_error: |
2804 | 14.9k | if (rep != NULL) |
2805 | 0 | xmlFree(rep); |
2806 | 14.9k | if (buffer != NULL) |
2807 | 14.9k | xmlFree(buffer); |
2808 | 14.9k | return(NULL); |
2809 | 0 | } |
2810 | | |
2811 | | /** |
2812 | | * xmlStringLenDecodeEntities: |
2813 | | * @ctxt: the parser context |
2814 | | * @str: the input string |
2815 | | * @len: the string length |
2816 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2817 | | * @end: an end marker xmlChar, 0 if none |
2818 | | * @end2: an end marker xmlChar, 0 if none |
2819 | | * @end3: an end marker xmlChar, 0 if none |
2820 | | * |
2821 | | * DEPRECATED: Internal function, don't use. |
2822 | | * |
2823 | | * Takes a entity string content and process to do the adequate substitutions. |
2824 | | * |
2825 | | * [67] Reference ::= EntityRef | CharRef |
2826 | | * |
2827 | | * [69] PEReference ::= '%' Name ';' |
2828 | | * |
2829 | | * Returns A newly allocated string with the substitution done. The caller |
2830 | | * must deallocate it ! |
2831 | | */ |
2832 | | xmlChar * |
2833 | | xmlStringLenDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2834 | | int what, xmlChar end, xmlChar end2, |
2835 | 11.8k | xmlChar end3) { |
2836 | 11.8k | if ((ctxt == NULL) || (str == NULL) || (len < 0)) |
2837 | 0 | return(NULL); |
2838 | 11.8k | return(xmlStringDecodeEntitiesInt(ctxt, str, len, what, |
2839 | 11.8k | end, end2, end3, 0)); |
2840 | 11.8k | } |
2841 | | |
2842 | | /** |
2843 | | * xmlStringDecodeEntities: |
2844 | | * @ctxt: the parser context |
2845 | | * @str: the input string |
2846 | | * @what: combination of XML_SUBSTITUTE_REF and XML_SUBSTITUTE_PEREF |
2847 | | * @end: an end marker xmlChar, 0 if none |
2848 | | * @end2: an end marker xmlChar, 0 if none |
2849 | | * @end3: an end marker xmlChar, 0 if none |
2850 | | * |
2851 | | * DEPRECATED: Internal function, don't use. |
2852 | | * |
2853 | | * Takes a entity string content and process to do the adequate substitutions. |
2854 | | * |
2855 | | * [67] Reference ::= EntityRef | CharRef |
2856 | | * |
2857 | | * [69] PEReference ::= '%' Name ';' |
2858 | | * |
2859 | | * Returns A newly allocated string with the substitution done. The caller |
2860 | | * must deallocate it ! |
2861 | | */ |
2862 | | xmlChar * |
2863 | | xmlStringDecodeEntities(xmlParserCtxtPtr ctxt, const xmlChar *str, int what, |
2864 | 210k | xmlChar end, xmlChar end2, xmlChar end3) { |
2865 | 210k | if ((ctxt == NULL) || (str == NULL)) return(NULL); |
2866 | 210k | return(xmlStringDecodeEntitiesInt(ctxt, str, xmlStrlen(str), what, |
2867 | 210k | end, end2, end3, 0)); |
2868 | 210k | } |
2869 | | |
2870 | | /************************************************************************ |
2871 | | * * |
2872 | | * Commodity functions, cleanup needed ? * |
2873 | | * * |
2874 | | ************************************************************************/ |
2875 | | |
2876 | | /** |
2877 | | * areBlanks: |
2878 | | * @ctxt: an XML parser context |
2879 | | * @str: a xmlChar * |
2880 | | * @len: the size of @str |
2881 | | * @blank_chars: we know the chars are blanks |
2882 | | * |
2883 | | * Is this a sequence of blank chars that one can ignore ? |
2884 | | * |
2885 | | * Returns 1 if ignorable 0 otherwise. |
2886 | | */ |
2887 | | |
2888 | | static int areBlanks(xmlParserCtxtPtr ctxt, const xmlChar *str, int len, |
2889 | 5.03M | int blank_chars) { |
2890 | 5.03M | int i, ret; |
2891 | 5.03M | xmlNodePtr lastChild; |
2892 | | |
2893 | | /* |
2894 | | * Don't spend time trying to differentiate them, the same callback is |
2895 | | * used ! |
2896 | | */ |
2897 | 5.03M | if (ctxt->sax->ignorableWhitespace == ctxt->sax->characters) |
2898 | 374k | return(0); |
2899 | | |
2900 | | /* |
2901 | | * Check for xml:space value. |
2902 | | */ |
2903 | 4.66M | if ((ctxt->space == NULL) || (*(ctxt->space) == 1) || |
2904 | 4.66M | (*(ctxt->space) == -2)) |
2905 | 1.93M | return(0); |
2906 | | |
2907 | | /* |
2908 | | * Check that the string is made of blanks |
2909 | | */ |
2910 | 2.72M | if (blank_chars == 0) { |
2911 | 5.84M | for (i = 0;i < len;i++) |
2912 | 5.05M | if (!(IS_BLANK_CH(str[i]))) return(0); |
2913 | 1.14M | } |
2914 | | |
2915 | | /* |
2916 | | * Look if the element is mixed content in the DTD if available |
2917 | | */ |
2918 | 2.37M | if (ctxt->node == NULL) return(0); |
2919 | 2.30M | if (ctxt->myDoc != NULL) { |
2920 | 2.30M | ret = xmlIsMixedElement(ctxt->myDoc, ctxt->node->name); |
2921 | 2.30M | if (ret == 0) return(1); |
2922 | 2.13M | if (ret == 1) return(0); |
2923 | 2.13M | } |
2924 | | |
2925 | | /* |
2926 | | * Otherwise, heuristic :-\ |
2927 | | */ |
2928 | 2.11M | if ((RAW != '<') && (RAW != 0xD)) return(0); |
2929 | 2.08M | if ((ctxt->node->children == NULL) && |
2930 | 2.08M | (RAW == '<') && (NXT(1) == '/')) return(0); |
2931 | | |
2932 | 2.08M | lastChild = xmlGetLastChild(ctxt->node); |
2933 | 2.08M | if (lastChild == NULL) { |
2934 | 416k | if ((ctxt->node->type != XML_ELEMENT_NODE) && |
2935 | 416k | (ctxt->node->content != NULL)) return(0); |
2936 | 1.66M | } else if (xmlNodeIsText(lastChild)) |
2937 | 29.2k | return(0); |
2938 | 1.63M | else if ((ctxt->node->children != NULL) && |
2939 | 1.63M | (xmlNodeIsText(ctxt->node->children))) |
2940 | 19.5k | return(0); |
2941 | 2.03M | return(1); |
2942 | 2.08M | } |
2943 | | |
2944 | | /************************************************************************ |
2945 | | * * |
2946 | | * Extra stuff for namespace support * |
2947 | | * Relates to http://www.w3.org/TR/WD-xml-names * |
2948 | | * * |
2949 | | ************************************************************************/ |
2950 | | |
2951 | | /** |
2952 | | * xmlSplitQName: |
2953 | | * @ctxt: an XML parser context |
2954 | | * @name: an XML parser context |
2955 | | * @prefix: a xmlChar ** |
2956 | | * |
2957 | | * parse an UTF8 encoded XML qualified name string |
2958 | | * |
2959 | | * [NS 5] QName ::= (Prefix ':')? LocalPart |
2960 | | * |
2961 | | * [NS 6] Prefix ::= NCName |
2962 | | * |
2963 | | * [NS 7] LocalPart ::= NCName |
2964 | | * |
2965 | | * Returns the local part, and prefix is updated |
2966 | | * to get the Prefix if any. |
2967 | | */ |
2968 | | |
2969 | | xmlChar * |
2970 | 9.96M | xmlSplitQName(xmlParserCtxtPtr ctxt, const xmlChar *name, xmlChar **prefix) { |
2971 | 9.96M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
2972 | 9.96M | xmlChar *buffer = NULL; |
2973 | 9.96M | int len = 0; |
2974 | 9.96M | int max = XML_MAX_NAMELEN; |
2975 | 9.96M | xmlChar *ret = NULL; |
2976 | 9.96M | const xmlChar *cur = name; |
2977 | 9.96M | int c; |
2978 | | |
2979 | 9.96M | if (prefix == NULL) return(NULL); |
2980 | 9.96M | *prefix = NULL; |
2981 | | |
2982 | 9.96M | if (cur == NULL) return(NULL); |
2983 | | |
2984 | | #ifndef XML_XML_NAMESPACE |
2985 | | /* xml: prefix is not really a namespace */ |
2986 | | if ((cur[0] == 'x') && (cur[1] == 'm') && |
2987 | | (cur[2] == 'l') && (cur[3] == ':')) |
2988 | | return(xmlStrdup(name)); |
2989 | | #endif |
2990 | | |
2991 | | /* nasty but well=formed */ |
2992 | 9.96M | if (cur[0] == ':') |
2993 | 4.74k | return(xmlStrdup(name)); |
2994 | | |
2995 | 9.95M | c = *cur++; |
2996 | 48.3M | while ((c != 0) && (c != ':') && (len < max)) { /* tested bigname.xml */ |
2997 | 38.3M | buf[len++] = c; |
2998 | 38.3M | c = *cur++; |
2999 | 38.3M | } |
3000 | 9.95M | if (len >= max) { |
3001 | | /* |
3002 | | * Okay someone managed to make a huge name, so he's ready to pay |
3003 | | * for the processing speed. |
3004 | | */ |
3005 | 4.69k | max = len * 2; |
3006 | | |
3007 | 4.69k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3008 | 4.69k | if (buffer == NULL) { |
3009 | 0 | xmlErrMemory(ctxt, NULL); |
3010 | 0 | return(NULL); |
3011 | 0 | } |
3012 | 4.69k | memcpy(buffer, buf, len); |
3013 | 4.93M | while ((c != 0) && (c != ':')) { /* tested bigname.xml */ |
3014 | 4.92M | if (len + 10 > max) { |
3015 | 5.66k | xmlChar *tmp; |
3016 | | |
3017 | 5.66k | max *= 2; |
3018 | 5.66k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3019 | 5.66k | if (tmp == NULL) { |
3020 | 0 | xmlFree(buffer); |
3021 | 0 | xmlErrMemory(ctxt, NULL); |
3022 | 0 | return(NULL); |
3023 | 0 | } |
3024 | 5.66k | buffer = tmp; |
3025 | 5.66k | } |
3026 | 4.92M | buffer[len++] = c; |
3027 | 4.92M | c = *cur++; |
3028 | 4.92M | } |
3029 | 4.69k | buffer[len] = 0; |
3030 | 4.69k | } |
3031 | | |
3032 | 9.95M | if ((c == ':') && (*cur == 0)) { |
3033 | 9.51k | if (buffer != NULL) |
3034 | 234 | xmlFree(buffer); |
3035 | 9.51k | *prefix = NULL; |
3036 | 9.51k | return(xmlStrdup(name)); |
3037 | 9.51k | } |
3038 | | |
3039 | 9.94M | if (buffer == NULL) |
3040 | 9.94M | ret = xmlStrndup(buf, len); |
3041 | 4.45k | else { |
3042 | 4.45k | ret = buffer; |
3043 | 4.45k | buffer = NULL; |
3044 | 4.45k | max = XML_MAX_NAMELEN; |
3045 | 4.45k | } |
3046 | | |
3047 | | |
3048 | 9.94M | if (c == ':') { |
3049 | 658k | c = *cur; |
3050 | 658k | *prefix = ret; |
3051 | 658k | if (c == 0) { |
3052 | 0 | return(xmlStrndup(BAD_CAST "", 0)); |
3053 | 0 | } |
3054 | 658k | len = 0; |
3055 | | |
3056 | | /* |
3057 | | * Check that the first character is proper to start |
3058 | | * a new name |
3059 | | */ |
3060 | 658k | if (!(((c >= 0x61) && (c <= 0x7A)) || |
3061 | 658k | ((c >= 0x41) && (c <= 0x5A)) || |
3062 | 658k | (c == '_') || (c == ':'))) { |
3063 | 9.94k | int l; |
3064 | 9.94k | int first = CUR_SCHAR(cur, l); |
3065 | | |
3066 | 9.94k | if (!IS_LETTER(first) && (first != '_')) { |
3067 | 5.42k | xmlFatalErrMsgStr(ctxt, XML_NS_ERR_QNAME, |
3068 | 5.42k | "Name %s is not XML Namespace compliant\n", |
3069 | 5.42k | name); |
3070 | 5.42k | } |
3071 | 9.94k | } |
3072 | 658k | cur++; |
3073 | | |
3074 | 4.25M | while ((c != 0) && (len < max)) { /* tested bigname2.xml */ |
3075 | 3.59M | buf[len++] = c; |
3076 | 3.59M | c = *cur++; |
3077 | 3.59M | } |
3078 | 658k | if (len >= max) { |
3079 | | /* |
3080 | | * Okay someone managed to make a huge name, so he's ready to pay |
3081 | | * for the processing speed. |
3082 | | */ |
3083 | 3.69k | max = len * 2; |
3084 | | |
3085 | 3.69k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3086 | 3.69k | if (buffer == NULL) { |
3087 | 0 | xmlErrMemory(ctxt, NULL); |
3088 | 0 | return(NULL); |
3089 | 0 | } |
3090 | 3.69k | memcpy(buffer, buf, len); |
3091 | 2.62M | while (c != 0) { /* tested bigname2.xml */ |
3092 | 2.62M | if (len + 10 > max) { |
3093 | 3.17k | xmlChar *tmp; |
3094 | | |
3095 | 3.17k | max *= 2; |
3096 | 3.17k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3097 | 3.17k | if (tmp == NULL) { |
3098 | 0 | xmlErrMemory(ctxt, NULL); |
3099 | 0 | xmlFree(buffer); |
3100 | 0 | return(NULL); |
3101 | 0 | } |
3102 | 3.17k | buffer = tmp; |
3103 | 3.17k | } |
3104 | 2.62M | buffer[len++] = c; |
3105 | 2.62M | c = *cur++; |
3106 | 2.62M | } |
3107 | 3.69k | buffer[len] = 0; |
3108 | 3.69k | } |
3109 | | |
3110 | 658k | if (buffer == NULL) |
3111 | 655k | ret = xmlStrndup(buf, len); |
3112 | 3.69k | else { |
3113 | 3.69k | ret = buffer; |
3114 | 3.69k | } |
3115 | 658k | } |
3116 | | |
3117 | 9.94M | return(ret); |
3118 | 9.94M | } |
3119 | | |
3120 | | /************************************************************************ |
3121 | | * * |
3122 | | * The parser itself * |
3123 | | * Relates to http://www.w3.org/TR/REC-xml * |
3124 | | * * |
3125 | | ************************************************************************/ |
3126 | | |
3127 | | /************************************************************************ |
3128 | | * * |
3129 | | * Routines to parse Name, NCName and NmToken * |
3130 | | * * |
3131 | | ************************************************************************/ |
3132 | | #ifdef DEBUG |
3133 | | static unsigned long nbParseName = 0; |
3134 | | static unsigned long nbParseNmToken = 0; |
3135 | | static unsigned long nbParseNCName = 0; |
3136 | | static unsigned long nbParseNCNameComplex = 0; |
3137 | | static unsigned long nbParseNameComplex = 0; |
3138 | | static unsigned long nbParseStringName = 0; |
3139 | | #endif |
3140 | | |
3141 | | /* |
3142 | | * The two following functions are related to the change of accepted |
3143 | | * characters for Name and NmToken in the Revision 5 of XML-1.0 |
3144 | | * They correspond to the modified production [4] and the new production [4a] |
3145 | | * changes in that revision. Also note that the macros used for the |
3146 | | * productions Letter, Digit, CombiningChar and Extender are not needed |
3147 | | * anymore. |
3148 | | * We still keep compatibility to pre-revision5 parsing semantic if the |
3149 | | * new XML_PARSE_OLD10 option is given to the parser. |
3150 | | */ |
3151 | | static int |
3152 | 31.0M | xmlIsNameStartChar(xmlParserCtxtPtr ctxt, int c) { |
3153 | 31.0M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3154 | | /* |
3155 | | * Use the new checks of production [4] [4a] amd [5] of the |
3156 | | * Update 5 of XML-1.0 |
3157 | | */ |
3158 | 29.4M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3159 | 29.4M | (((c >= 'a') && (c <= 'z')) || |
3160 | 29.4M | ((c >= 'A') && (c <= 'Z')) || |
3161 | 29.4M | (c == '_') || (c == ':') || |
3162 | 29.4M | ((c >= 0xC0) && (c <= 0xD6)) || |
3163 | 29.4M | ((c >= 0xD8) && (c <= 0xF6)) || |
3164 | 29.4M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3165 | 29.4M | ((c >= 0x370) && (c <= 0x37D)) || |
3166 | 29.4M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3167 | 29.4M | ((c >= 0x200C) && (c <= 0x200D)) || |
3168 | 29.4M | ((c >= 0x2070) && (c <= 0x218F)) || |
3169 | 29.4M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3170 | 29.4M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3171 | 29.4M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3172 | 29.4M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3173 | 29.4M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3174 | 29.2M | return(1); |
3175 | 29.4M | } else { |
3176 | 1.62M | if (IS_LETTER(c) || (c == '_') || (c == ':')) |
3177 | 1.51M | return(1); |
3178 | 1.62M | } |
3179 | 342k | return(0); |
3180 | 31.0M | } |
3181 | | |
3182 | | static int |
3183 | 595M | xmlIsNameChar(xmlParserCtxtPtr ctxt, int c) { |
3184 | 595M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3185 | | /* |
3186 | | * Use the new checks of production [4] [4a] amd [5] of the |
3187 | | * Update 5 of XML-1.0 |
3188 | | */ |
3189 | 584M | if ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3190 | 584M | (((c >= 'a') && (c <= 'z')) || |
3191 | 584M | ((c >= 'A') && (c <= 'Z')) || |
3192 | 584M | ((c >= '0') && (c <= '9')) || /* !start */ |
3193 | 584M | (c == '_') || (c == ':') || |
3194 | 584M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3195 | 584M | ((c >= 0xC0) && (c <= 0xD6)) || |
3196 | 584M | ((c >= 0xD8) && (c <= 0xF6)) || |
3197 | 584M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3198 | 584M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3199 | 584M | ((c >= 0x370) && (c <= 0x37D)) || |
3200 | 584M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3201 | 584M | ((c >= 0x200C) && (c <= 0x200D)) || |
3202 | 584M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3203 | 584M | ((c >= 0x2070) && (c <= 0x218F)) || |
3204 | 584M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3205 | 584M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3206 | 584M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3207 | 584M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3208 | 584M | ((c >= 0x10000) && (c <= 0xEFFFF)))) |
3209 | 554M | return(1); |
3210 | 584M | } else { |
3211 | 10.6M | if ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3212 | 10.6M | (c == '.') || (c == '-') || |
3213 | 10.6M | (c == '_') || (c == ':') || |
3214 | 10.6M | (IS_COMBINING(c)) || |
3215 | 10.6M | (IS_EXTENDER(c))) |
3216 | 8.59M | return(1); |
3217 | 10.6M | } |
3218 | 32.1M | return(0); |
3219 | 595M | } |
3220 | | |
3221 | | static xmlChar * xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, |
3222 | | int *len, int *alloc, int normalize); |
3223 | | |
3224 | | static const xmlChar * |
3225 | 1.23M | xmlParseNameComplex(xmlParserCtxtPtr ctxt) { |
3226 | 1.23M | int len = 0, l; |
3227 | 1.23M | int c; |
3228 | 1.23M | int count = 0; |
3229 | 1.23M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3230 | 546k | XML_MAX_TEXT_LENGTH : |
3231 | 1.23M | XML_MAX_NAME_LENGTH; |
3232 | | |
3233 | | #ifdef DEBUG |
3234 | | nbParseNameComplex++; |
3235 | | #endif |
3236 | | |
3237 | | /* |
3238 | | * Handler for more complex cases |
3239 | | */ |
3240 | 1.23M | GROW; |
3241 | 1.23M | if (ctxt->instate == XML_PARSER_EOF) |
3242 | 92 | return(NULL); |
3243 | 1.23M | c = CUR_CHAR(l); |
3244 | 1.23M | if ((ctxt->options & XML_PARSE_OLD10) == 0) { |
3245 | | /* |
3246 | | * Use the new checks of production [4] [4a] amd [5] of the |
3247 | | * Update 5 of XML-1.0 |
3248 | | */ |
3249 | 757k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3250 | 757k | (!(((c >= 'a') && (c <= 'z')) || |
3251 | 733k | ((c >= 'A') && (c <= 'Z')) || |
3252 | 733k | (c == '_') || (c == ':') || |
3253 | 733k | ((c >= 0xC0) && (c <= 0xD6)) || |
3254 | 733k | ((c >= 0xD8) && (c <= 0xF6)) || |
3255 | 733k | ((c >= 0xF8) && (c <= 0x2FF)) || |
3256 | 733k | ((c >= 0x370) && (c <= 0x37D)) || |
3257 | 733k | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3258 | 733k | ((c >= 0x200C) && (c <= 0x200D)) || |
3259 | 733k | ((c >= 0x2070) && (c <= 0x218F)) || |
3260 | 733k | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3261 | 733k | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3262 | 733k | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3263 | 733k | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3264 | 733k | ((c >= 0x10000) && (c <= 0xEFFFF))))) { |
3265 | 256k | return(NULL); |
3266 | 256k | } |
3267 | 501k | len += l; |
3268 | 501k | NEXTL(l); |
3269 | 501k | c = CUR_CHAR(l); |
3270 | 5.30M | while ((c != ' ') && (c != '>') && (c != '/') && /* accelerators */ |
3271 | 5.30M | (((c >= 'a') && (c <= 'z')) || |
3272 | 5.28M | ((c >= 'A') && (c <= 'Z')) || |
3273 | 5.28M | ((c >= '0') && (c <= '9')) || /* !start */ |
3274 | 5.28M | (c == '_') || (c == ':') || |
3275 | 5.28M | (c == '-') || (c == '.') || (c == 0xB7) || /* !start */ |
3276 | 5.28M | ((c >= 0xC0) && (c <= 0xD6)) || |
3277 | 5.28M | ((c >= 0xD8) && (c <= 0xF6)) || |
3278 | 5.28M | ((c >= 0xF8) && (c <= 0x2FF)) || |
3279 | 5.28M | ((c >= 0x300) && (c <= 0x36F)) || /* !start */ |
3280 | 5.28M | ((c >= 0x370) && (c <= 0x37D)) || |
3281 | 5.28M | ((c >= 0x37F) && (c <= 0x1FFF)) || |
3282 | 5.28M | ((c >= 0x200C) && (c <= 0x200D)) || |
3283 | 5.28M | ((c >= 0x203F) && (c <= 0x2040)) || /* !start */ |
3284 | 5.28M | ((c >= 0x2070) && (c <= 0x218F)) || |
3285 | 5.28M | ((c >= 0x2C00) && (c <= 0x2FEF)) || |
3286 | 5.28M | ((c >= 0x3001) && (c <= 0xD7FF)) || |
3287 | 5.28M | ((c >= 0xF900) && (c <= 0xFDCF)) || |
3288 | 5.28M | ((c >= 0xFDF0) && (c <= 0xFFFD)) || |
3289 | 5.28M | ((c >= 0x10000) && (c <= 0xEFFFF)) |
3290 | 5.28M | )) { |
3291 | 4.80M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3292 | 33.3k | count = 0; |
3293 | 33.3k | GROW; |
3294 | 33.3k | if (ctxt->instate == XML_PARSER_EOF) |
3295 | 0 | return(NULL); |
3296 | 33.3k | } |
3297 | 4.80M | if (len <= INT_MAX - l) |
3298 | 4.80M | len += l; |
3299 | 4.80M | NEXTL(l); |
3300 | 4.80M | c = CUR_CHAR(l); |
3301 | 4.80M | } |
3302 | 501k | } else { |
3303 | 474k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3304 | 474k | (!IS_LETTER(c) && (c != '_') && |
3305 | 451k | (c != ':'))) { |
3306 | 250k | return(NULL); |
3307 | 250k | } |
3308 | 224k | len += l; |
3309 | 224k | NEXTL(l); |
3310 | 224k | c = CUR_CHAR(l); |
3311 | | |
3312 | 4.27M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3313 | 4.27M | ((IS_LETTER(c)) || (IS_DIGIT(c)) || |
3314 | 4.24M | (c == '.') || (c == '-') || |
3315 | 4.24M | (c == '_') || (c == ':') || |
3316 | 4.24M | (IS_COMBINING(c)) || |
3317 | 4.24M | (IS_EXTENDER(c)))) { |
3318 | 4.05M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3319 | 29.4k | count = 0; |
3320 | 29.4k | GROW; |
3321 | 29.4k | if (ctxt->instate == XML_PARSER_EOF) |
3322 | 0 | return(NULL); |
3323 | 29.4k | } |
3324 | 4.05M | if (len <= INT_MAX - l) |
3325 | 4.05M | len += l; |
3326 | 4.05M | NEXTL(l); |
3327 | 4.05M | c = CUR_CHAR(l); |
3328 | 4.05M | } |
3329 | 224k | } |
3330 | 725k | if (len > maxLength) { |
3331 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3332 | 0 | return(NULL); |
3333 | 0 | } |
3334 | 725k | if (ctxt->input->cur - ctxt->input->base < len) { |
3335 | | /* |
3336 | | * There were a couple of bugs where PERefs lead to to a change |
3337 | | * of the buffer. Check the buffer size to avoid passing an invalid |
3338 | | * pointer to xmlDictLookup. |
3339 | | */ |
3340 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
3341 | 0 | "unexpected change of input buffer"); |
3342 | 0 | return (NULL); |
3343 | 0 | } |
3344 | 725k | if ((*ctxt->input->cur == '\n') && (ctxt->input->cur[-1] == '\r')) |
3345 | 3.37k | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - (len + 1), len)); |
3346 | 722k | return(xmlDictLookup(ctxt->dict, ctxt->input->cur - len, len)); |
3347 | 725k | } |
3348 | | |
3349 | | /** |
3350 | | * xmlParseName: |
3351 | | * @ctxt: an XML parser context |
3352 | | * |
3353 | | * DEPRECATED: Internal function, don't use. |
3354 | | * |
3355 | | * parse an XML name. |
3356 | | * |
3357 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3358 | | * CombiningChar | Extender |
3359 | | * |
3360 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3361 | | * |
3362 | | * [6] Names ::= Name (#x20 Name)* |
3363 | | * |
3364 | | * Returns the Name parsed or NULL |
3365 | | */ |
3366 | | |
3367 | | const xmlChar * |
3368 | 64.5M | xmlParseName(xmlParserCtxtPtr ctxt) { |
3369 | 64.5M | const xmlChar *in; |
3370 | 64.5M | const xmlChar *ret; |
3371 | 64.5M | size_t count = 0; |
3372 | 64.5M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3373 | 20.6M | XML_MAX_TEXT_LENGTH : |
3374 | 64.5M | XML_MAX_NAME_LENGTH; |
3375 | | |
3376 | 64.5M | GROW; |
3377 | | |
3378 | | #ifdef DEBUG |
3379 | | nbParseName++; |
3380 | | #endif |
3381 | | |
3382 | | /* |
3383 | | * Accelerator for simple ASCII names |
3384 | | */ |
3385 | 64.5M | in = ctxt->input->cur; |
3386 | 64.5M | if (((*in >= 0x61) && (*in <= 0x7A)) || |
3387 | 64.5M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3388 | 64.5M | (*in == '_') || (*in == ':')) { |
3389 | 63.9M | in++; |
3390 | 287M | while (((*in >= 0x61) && (*in <= 0x7A)) || |
3391 | 287M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3392 | 287M | ((*in >= 0x30) && (*in <= 0x39)) || |
3393 | 287M | (*in == '_') || (*in == '-') || |
3394 | 287M | (*in == ':') || (*in == '.')) |
3395 | 223M | in++; |
3396 | 63.9M | if ((*in > 0) && (*in < 0x80)) { |
3397 | 63.2M | count = in - ctxt->input->cur; |
3398 | 63.2M | if (count > maxLength) { |
3399 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Name"); |
3400 | 0 | return(NULL); |
3401 | 0 | } |
3402 | 63.2M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3403 | 63.2M | ctxt->input->cur = in; |
3404 | 63.2M | ctxt->input->col += count; |
3405 | 63.2M | if (ret == NULL) |
3406 | 0 | xmlErrMemory(ctxt, NULL); |
3407 | 63.2M | return(ret); |
3408 | 63.2M | } |
3409 | 63.9M | } |
3410 | | /* accelerator for special cases */ |
3411 | 1.23M | return(xmlParseNameComplex(ctxt)); |
3412 | 64.5M | } |
3413 | | |
3414 | | static const xmlChar * |
3415 | 455k | xmlParseNCNameComplex(xmlParserCtxtPtr ctxt) { |
3416 | 455k | int len = 0, l; |
3417 | 455k | int c; |
3418 | 455k | int count = 0; |
3419 | 455k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3420 | 185k | XML_MAX_TEXT_LENGTH : |
3421 | 455k | XML_MAX_NAME_LENGTH; |
3422 | 455k | size_t startPosition = 0; |
3423 | | |
3424 | | #ifdef DEBUG |
3425 | | nbParseNCNameComplex++; |
3426 | | #endif |
3427 | | |
3428 | | /* |
3429 | | * Handler for more complex cases |
3430 | | */ |
3431 | 455k | GROW; |
3432 | 455k | startPosition = CUR_PTR - BASE_PTR; |
3433 | 455k | c = CUR_CHAR(l); |
3434 | 455k | if ((c == ' ') || (c == '>') || (c == '/') || /* accelerators */ |
3435 | 455k | (!xmlIsNameStartChar(ctxt, c) || (c == ':'))) { |
3436 | 358k | return(NULL); |
3437 | 358k | } |
3438 | | |
3439 | 3.50M | while ((c != ' ') && (c != '>') && (c != '/') && /* test bigname.xml */ |
3440 | 3.50M | (xmlIsNameChar(ctxt, c) && (c != ':'))) { |
3441 | 3.40M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3442 | 27.6k | count = 0; |
3443 | 27.6k | GROW; |
3444 | 27.6k | if (ctxt->instate == XML_PARSER_EOF) |
3445 | 0 | return(NULL); |
3446 | 27.6k | } |
3447 | 3.40M | if (len <= INT_MAX - l) |
3448 | 3.40M | len += l; |
3449 | 3.40M | NEXTL(l); |
3450 | 3.40M | c = CUR_CHAR(l); |
3451 | 3.40M | if (c == 0) { |
3452 | 11.5k | count = 0; |
3453 | | /* |
3454 | | * when shrinking to extend the buffer we really need to preserve |
3455 | | * the part of the name we already parsed. Hence rolling back |
3456 | | * by current length. |
3457 | | */ |
3458 | 11.5k | ctxt->input->cur -= l; |
3459 | 11.5k | GROW; |
3460 | 11.5k | if (ctxt->instate == XML_PARSER_EOF) |
3461 | 0 | return(NULL); |
3462 | 11.5k | ctxt->input->cur += l; |
3463 | 11.5k | c = CUR_CHAR(l); |
3464 | 11.5k | } |
3465 | 3.40M | } |
3466 | 97.0k | if (len > maxLength) { |
3467 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3468 | 0 | return(NULL); |
3469 | 0 | } |
3470 | 97.0k | return(xmlDictLookup(ctxt->dict, (BASE_PTR + startPosition), len)); |
3471 | 97.0k | } |
3472 | | |
3473 | | /** |
3474 | | * xmlParseNCName: |
3475 | | * @ctxt: an XML parser context |
3476 | | * @len: length of the string parsed |
3477 | | * |
3478 | | * parse an XML name. |
3479 | | * |
3480 | | * [4NS] NCNameChar ::= Letter | Digit | '.' | '-' | '_' | |
3481 | | * CombiningChar | Extender |
3482 | | * |
3483 | | * [5NS] NCName ::= (Letter | '_') (NCNameChar)* |
3484 | | * |
3485 | | * Returns the Name parsed or NULL |
3486 | | */ |
3487 | | |
3488 | | static const xmlChar * |
3489 | 15.1M | xmlParseNCName(xmlParserCtxtPtr ctxt) { |
3490 | 15.1M | const xmlChar *in, *e; |
3491 | 15.1M | const xmlChar *ret; |
3492 | 15.1M | size_t count = 0; |
3493 | 15.1M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3494 | 4.27M | XML_MAX_TEXT_LENGTH : |
3495 | 15.1M | XML_MAX_NAME_LENGTH; |
3496 | | |
3497 | | #ifdef DEBUG |
3498 | | nbParseNCName++; |
3499 | | #endif |
3500 | | |
3501 | | /* |
3502 | | * Accelerator for simple ASCII names |
3503 | | */ |
3504 | 15.1M | in = ctxt->input->cur; |
3505 | 15.1M | e = ctxt->input->end; |
3506 | 15.1M | if ((((*in >= 0x61) && (*in <= 0x7A)) || |
3507 | 15.1M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3508 | 15.1M | (*in == '_')) && (in < e)) { |
3509 | 14.7M | in++; |
3510 | 54.7M | while ((((*in >= 0x61) && (*in <= 0x7A)) || |
3511 | 54.7M | ((*in >= 0x41) && (*in <= 0x5A)) || |
3512 | 54.7M | ((*in >= 0x30) && (*in <= 0x39)) || |
3513 | 54.7M | (*in == '_') || (*in == '-') || |
3514 | 54.7M | (*in == '.')) && (in < e)) |
3515 | 40.0M | in++; |
3516 | 14.7M | if (in >= e) |
3517 | 2.68k | goto complex; |
3518 | 14.7M | if ((*in > 0) && (*in < 0x80)) { |
3519 | 14.6M | count = in - ctxt->input->cur; |
3520 | 14.6M | if (count > maxLength) { |
3521 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3522 | 0 | return(NULL); |
3523 | 0 | } |
3524 | 14.6M | ret = xmlDictLookup(ctxt->dict, ctxt->input->cur, count); |
3525 | 14.6M | ctxt->input->cur = in; |
3526 | 14.6M | ctxt->input->col += count; |
3527 | 14.6M | if (ret == NULL) { |
3528 | 0 | xmlErrMemory(ctxt, NULL); |
3529 | 0 | } |
3530 | 14.6M | return(ret); |
3531 | 14.6M | } |
3532 | 14.7M | } |
3533 | 455k | complex: |
3534 | 455k | return(xmlParseNCNameComplex(ctxt)); |
3535 | 15.1M | } |
3536 | | |
3537 | | /** |
3538 | | * xmlParseNameAndCompare: |
3539 | | * @ctxt: an XML parser context |
3540 | | * |
3541 | | * parse an XML name and compares for match |
3542 | | * (specialized for endtag parsing) |
3543 | | * |
3544 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
3545 | | * and the name for mismatch |
3546 | | */ |
3547 | | |
3548 | | static const xmlChar * |
3549 | 4.79M | xmlParseNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *other) { |
3550 | 4.79M | register const xmlChar *cmp = other; |
3551 | 4.79M | register const xmlChar *in; |
3552 | 4.79M | const xmlChar *ret; |
3553 | | |
3554 | 4.79M | GROW; |
3555 | 4.79M | if (ctxt->instate == XML_PARSER_EOF) |
3556 | 0 | return(NULL); |
3557 | | |
3558 | 4.79M | in = ctxt->input->cur; |
3559 | 24.9M | while (*in != 0 && *in == *cmp) { |
3560 | 20.1M | ++in; |
3561 | 20.1M | ++cmp; |
3562 | 20.1M | } |
3563 | 4.79M | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
3564 | | /* success */ |
3565 | 4.60M | ctxt->input->col += in - ctxt->input->cur; |
3566 | 4.60M | ctxt->input->cur = in; |
3567 | 4.60M | return (const xmlChar*) 1; |
3568 | 4.60M | } |
3569 | | /* failure (or end of input buffer), check with full function */ |
3570 | 198k | ret = xmlParseName (ctxt); |
3571 | | /* strings coming from the dictionary direct compare possible */ |
3572 | 198k | if (ret == other) { |
3573 | 10.4k | return (const xmlChar*) 1; |
3574 | 10.4k | } |
3575 | 187k | return ret; |
3576 | 198k | } |
3577 | | |
3578 | | /** |
3579 | | * xmlParseStringName: |
3580 | | * @ctxt: an XML parser context |
3581 | | * @str: a pointer to the string pointer (IN/OUT) |
3582 | | * |
3583 | | * parse an XML name. |
3584 | | * |
3585 | | * [4] NameChar ::= Letter | Digit | '.' | '-' | '_' | ':' | |
3586 | | * CombiningChar | Extender |
3587 | | * |
3588 | | * [5] Name ::= (Letter | '_' | ':') (NameChar)* |
3589 | | * |
3590 | | * [6] Names ::= Name (#x20 Name)* |
3591 | | * |
3592 | | * Returns the Name parsed or NULL. The @str pointer |
3593 | | * is updated to the current location in the string. |
3594 | | */ |
3595 | | |
3596 | | static xmlChar * |
3597 | 30.6M | xmlParseStringName(xmlParserCtxtPtr ctxt, const xmlChar** str) { |
3598 | 30.6M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3599 | 30.6M | const xmlChar *cur = *str; |
3600 | 30.6M | int len = 0, l; |
3601 | 30.6M | int c; |
3602 | 30.6M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3603 | 4.13M | XML_MAX_TEXT_LENGTH : |
3604 | 30.6M | XML_MAX_NAME_LENGTH; |
3605 | | |
3606 | | #ifdef DEBUG |
3607 | | nbParseStringName++; |
3608 | | #endif |
3609 | | |
3610 | 30.6M | c = CUR_SCHAR(cur, l); |
3611 | 30.6M | if (!xmlIsNameStartChar(ctxt, c)) { |
3612 | 20.8k | return(NULL); |
3613 | 20.8k | } |
3614 | | |
3615 | 30.6M | COPY_BUF(l,buf,len,c); |
3616 | 30.6M | cur += l; |
3617 | 30.6M | c = CUR_SCHAR(cur, l); |
3618 | 272M | while (xmlIsNameChar(ctxt, c)) { |
3619 | 243M | COPY_BUF(l,buf,len,c); |
3620 | 243M | cur += l; |
3621 | 243M | c = CUR_SCHAR(cur, l); |
3622 | 243M | if (len >= XML_MAX_NAMELEN) { /* test bigentname.xml */ |
3623 | | /* |
3624 | | * Okay someone managed to make a huge name, so he's ready to pay |
3625 | | * for the processing speed. |
3626 | | */ |
3627 | 1.35M | xmlChar *buffer; |
3628 | 1.35M | int max = len * 2; |
3629 | | |
3630 | 1.35M | buffer = (xmlChar *) xmlMallocAtomic(max); |
3631 | 1.35M | if (buffer == NULL) { |
3632 | 0 | xmlErrMemory(ctxt, NULL); |
3633 | 0 | return(NULL); |
3634 | 0 | } |
3635 | 1.35M | memcpy(buffer, buf, len); |
3636 | 308M | while (xmlIsNameChar(ctxt, c)) { |
3637 | 306M | if (len + 10 > max) { |
3638 | 1.35M | xmlChar *tmp; |
3639 | | |
3640 | 1.35M | max *= 2; |
3641 | 1.35M | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3642 | 1.35M | if (tmp == NULL) { |
3643 | 0 | xmlErrMemory(ctxt, NULL); |
3644 | 0 | xmlFree(buffer); |
3645 | 0 | return(NULL); |
3646 | 0 | } |
3647 | 1.35M | buffer = tmp; |
3648 | 1.35M | } |
3649 | 306M | COPY_BUF(l,buffer,len,c); |
3650 | 306M | cur += l; |
3651 | 306M | c = CUR_SCHAR(cur, l); |
3652 | 306M | if (len > maxLength) { |
3653 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3654 | 0 | xmlFree(buffer); |
3655 | 0 | return(NULL); |
3656 | 0 | } |
3657 | 306M | } |
3658 | 1.35M | buffer[len] = 0; |
3659 | 1.35M | *str = cur; |
3660 | 1.35M | return(buffer); |
3661 | 1.35M | } |
3662 | 243M | } |
3663 | 29.2M | if (len > maxLength) { |
3664 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NCName"); |
3665 | 0 | return(NULL); |
3666 | 0 | } |
3667 | 29.2M | *str = cur; |
3668 | 29.2M | return(xmlStrndup(buf, len)); |
3669 | 29.2M | } |
3670 | | |
3671 | | /** |
3672 | | * xmlParseNmtoken: |
3673 | | * @ctxt: an XML parser context |
3674 | | * |
3675 | | * DEPRECATED: Internal function, don't use. |
3676 | | * |
3677 | | * parse an XML Nmtoken. |
3678 | | * |
3679 | | * [7] Nmtoken ::= (NameChar)+ |
3680 | | * |
3681 | | * [8] Nmtokens ::= Nmtoken (#x20 Nmtoken)* |
3682 | | * |
3683 | | * Returns the Nmtoken parsed or NULL |
3684 | | */ |
3685 | | |
3686 | | xmlChar * |
3687 | 1.48M | xmlParseNmtoken(xmlParserCtxtPtr ctxt) { |
3688 | 1.48M | xmlChar buf[XML_MAX_NAMELEN + 5]; |
3689 | 1.48M | int len = 0, l; |
3690 | 1.48M | int c; |
3691 | 1.48M | int count = 0; |
3692 | 1.48M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3693 | 591k | XML_MAX_TEXT_LENGTH : |
3694 | 1.48M | XML_MAX_NAME_LENGTH; |
3695 | | |
3696 | | #ifdef DEBUG |
3697 | | nbParseNmToken++; |
3698 | | #endif |
3699 | | |
3700 | 1.48M | GROW; |
3701 | 1.48M | if (ctxt->instate == XML_PARSER_EOF) |
3702 | 5 | return(NULL); |
3703 | 1.48M | c = CUR_CHAR(l); |
3704 | | |
3705 | 8.74M | while (xmlIsNameChar(ctxt, c)) { |
3706 | 7.25M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3707 | 0 | count = 0; |
3708 | 0 | GROW; |
3709 | 0 | } |
3710 | 7.25M | COPY_BUF(l,buf,len,c); |
3711 | 7.25M | NEXTL(l); |
3712 | 7.25M | c = CUR_CHAR(l); |
3713 | 7.25M | if (c == 0) { |
3714 | 1.83k | count = 0; |
3715 | 1.83k | GROW; |
3716 | 1.83k | if (ctxt->instate == XML_PARSER_EOF) |
3717 | 0 | return(NULL); |
3718 | 1.83k | c = CUR_CHAR(l); |
3719 | 1.83k | } |
3720 | 7.25M | if (len >= XML_MAX_NAMELEN) { |
3721 | | /* |
3722 | | * Okay someone managed to make a huge token, so he's ready to pay |
3723 | | * for the processing speed. |
3724 | | */ |
3725 | 1.59k | xmlChar *buffer; |
3726 | 1.59k | int max = len * 2; |
3727 | | |
3728 | 1.59k | buffer = (xmlChar *) xmlMallocAtomic(max); |
3729 | 1.59k | if (buffer == NULL) { |
3730 | 0 | xmlErrMemory(ctxt, NULL); |
3731 | 0 | return(NULL); |
3732 | 0 | } |
3733 | 1.59k | memcpy(buffer, buf, len); |
3734 | 2.13M | while (xmlIsNameChar(ctxt, c)) { |
3735 | 2.12M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
3736 | 21.5k | count = 0; |
3737 | 21.5k | GROW; |
3738 | 21.5k | if (ctxt->instate == XML_PARSER_EOF) { |
3739 | 0 | xmlFree(buffer); |
3740 | 0 | return(NULL); |
3741 | 0 | } |
3742 | 21.5k | } |
3743 | 2.12M | if (len + 10 > max) { |
3744 | 2.88k | xmlChar *tmp; |
3745 | | |
3746 | 2.88k | max *= 2; |
3747 | 2.88k | tmp = (xmlChar *) xmlRealloc(buffer, max); |
3748 | 2.88k | if (tmp == NULL) { |
3749 | 0 | xmlErrMemory(ctxt, NULL); |
3750 | 0 | xmlFree(buffer); |
3751 | 0 | return(NULL); |
3752 | 0 | } |
3753 | 2.88k | buffer = tmp; |
3754 | 2.88k | } |
3755 | 2.12M | COPY_BUF(l,buffer,len,c); |
3756 | 2.12M | NEXTL(l); |
3757 | 2.12M | c = CUR_CHAR(l); |
3758 | 2.12M | if (len > maxLength) { |
3759 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3760 | 0 | xmlFree(buffer); |
3761 | 0 | return(NULL); |
3762 | 0 | } |
3763 | 2.12M | } |
3764 | 1.59k | buffer[len] = 0; |
3765 | 1.59k | return(buffer); |
3766 | 1.59k | } |
3767 | 7.25M | } |
3768 | 1.48M | if (len == 0) |
3769 | 13.6k | return(NULL); |
3770 | 1.46M | if (len > maxLength) { |
3771 | 0 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "NmToken"); |
3772 | 0 | return(NULL); |
3773 | 0 | } |
3774 | 1.46M | return(xmlStrndup(buf, len)); |
3775 | 1.46M | } |
3776 | | |
3777 | | /** |
3778 | | * xmlParseEntityValue: |
3779 | | * @ctxt: an XML parser context |
3780 | | * @orig: if non-NULL store a copy of the original entity value |
3781 | | * |
3782 | | * DEPRECATED: Internal function, don't use. |
3783 | | * |
3784 | | * parse a value for ENTITY declarations |
3785 | | * |
3786 | | * [9] EntityValue ::= '"' ([^%&"] | PEReference | Reference)* '"' | |
3787 | | * "'" ([^%&'] | PEReference | Reference)* "'" |
3788 | | * |
3789 | | * Returns the EntityValue parsed with reference substituted or NULL |
3790 | | */ |
3791 | | |
3792 | | xmlChar * |
3793 | 850k | xmlParseEntityValue(xmlParserCtxtPtr ctxt, xmlChar **orig) { |
3794 | 850k | xmlChar *buf = NULL; |
3795 | 850k | int len = 0; |
3796 | 850k | int size = XML_PARSER_BUFFER_SIZE; |
3797 | 850k | int c, l; |
3798 | 850k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3799 | 306k | XML_MAX_HUGE_LENGTH : |
3800 | 850k | XML_MAX_TEXT_LENGTH; |
3801 | 850k | xmlChar stop; |
3802 | 850k | xmlChar *ret = NULL; |
3803 | 850k | const xmlChar *cur = NULL; |
3804 | 850k | xmlParserInputPtr input; |
3805 | | |
3806 | 850k | if (RAW == '"') stop = '"'; |
3807 | 224k | else if (RAW == '\'') stop = '\''; |
3808 | 0 | else { |
3809 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_STARTED, NULL); |
3810 | 0 | return(NULL); |
3811 | 0 | } |
3812 | 850k | buf = (xmlChar *) xmlMallocAtomic(size); |
3813 | 850k | if (buf == NULL) { |
3814 | 0 | xmlErrMemory(ctxt, NULL); |
3815 | 0 | return(NULL); |
3816 | 0 | } |
3817 | | |
3818 | | /* |
3819 | | * The content of the entity definition is copied in a buffer. |
3820 | | */ |
3821 | | |
3822 | 850k | ctxt->instate = XML_PARSER_ENTITY_VALUE; |
3823 | 850k | input = ctxt->input; |
3824 | 850k | GROW; |
3825 | 850k | if (ctxt->instate == XML_PARSER_EOF) |
3826 | 0 | goto error; |
3827 | 850k | NEXT; |
3828 | 850k | c = CUR_CHAR(l); |
3829 | | /* |
3830 | | * NOTE: 4.4.5 Included in Literal |
3831 | | * When a parameter entity reference appears in a literal entity |
3832 | | * value, ... a single or double quote character in the replacement |
3833 | | * text is always treated as a normal data character and will not |
3834 | | * terminate the literal. |
3835 | | * In practice it means we stop the loop only when back at parsing |
3836 | | * the initial entity and the quote is found |
3837 | | */ |
3838 | 48.3M | while (((IS_CHAR(c)) && ((c != stop) || /* checked */ |
3839 | 48.3M | (ctxt->input != input))) && (ctxt->instate != XML_PARSER_EOF)) { |
3840 | 47.5M | if (len + 5 >= size) { |
3841 | 135k | xmlChar *tmp; |
3842 | | |
3843 | 135k | size *= 2; |
3844 | 135k | tmp = (xmlChar *) xmlRealloc(buf, size); |
3845 | 135k | if (tmp == NULL) { |
3846 | 0 | xmlErrMemory(ctxt, NULL); |
3847 | 0 | goto error; |
3848 | 0 | } |
3849 | 135k | buf = tmp; |
3850 | 135k | } |
3851 | 47.5M | COPY_BUF(l,buf,len,c); |
3852 | 47.5M | NEXTL(l); |
3853 | | |
3854 | 47.5M | GROW; |
3855 | 47.5M | c = CUR_CHAR(l); |
3856 | 47.5M | if (c == 0) { |
3857 | 1.25k | GROW; |
3858 | 1.25k | c = CUR_CHAR(l); |
3859 | 1.25k | } |
3860 | | |
3861 | 47.5M | if (len > maxLength) { |
3862 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
3863 | 0 | "entity value too long\n"); |
3864 | 0 | goto error; |
3865 | 0 | } |
3866 | 47.5M | } |
3867 | 850k | buf[len] = 0; |
3868 | 850k | if (ctxt->instate == XML_PARSER_EOF) |
3869 | 0 | goto error; |
3870 | 850k | if (c != stop) { |
3871 | 1.85k | xmlFatalErr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, NULL); |
3872 | 1.85k | goto error; |
3873 | 1.85k | } |
3874 | 848k | NEXT; |
3875 | | |
3876 | | /* |
3877 | | * Raise problem w.r.t. '&' and '%' being used in non-entities |
3878 | | * reference constructs. Note Charref will be handled in |
3879 | | * xmlStringDecodeEntities() |
3880 | | */ |
3881 | 848k | cur = buf; |
3882 | 38.3M | while (*cur != 0) { /* non input consuming */ |
3883 | 37.5M | if ((*cur == '%') || ((*cur == '&') && (cur[1] != '#'))) { |
3884 | 512k | xmlChar *name; |
3885 | 512k | xmlChar tmp = *cur; |
3886 | 512k | int nameOk = 0; |
3887 | | |
3888 | 512k | cur++; |
3889 | 512k | name = xmlParseStringName(ctxt, &cur); |
3890 | 512k | if (name != NULL) { |
3891 | 508k | nameOk = 1; |
3892 | 508k | xmlFree(name); |
3893 | 508k | } |
3894 | 512k | if ((nameOk == 0) || (*cur != ';')) { |
3895 | 8.10k | xmlFatalErrMsgInt(ctxt, XML_ERR_ENTITY_CHAR_ERROR, |
3896 | 8.10k | "EntityValue: '%c' forbidden except for entities references\n", |
3897 | 8.10k | tmp); |
3898 | 8.10k | goto error; |
3899 | 8.10k | } |
3900 | 504k | if ((tmp == '%') && (ctxt->inSubset == 1) && |
3901 | 504k | (ctxt->inputNr == 1)) { |
3902 | 6.03k | xmlFatalErr(ctxt, XML_ERR_ENTITY_PE_INTERNAL, NULL); |
3903 | 6.03k | goto error; |
3904 | 6.03k | } |
3905 | 498k | if (*cur == 0) |
3906 | 0 | break; |
3907 | 498k | } |
3908 | 37.5M | cur++; |
3909 | 37.5M | } |
3910 | | |
3911 | | /* |
3912 | | * Then PEReference entities are substituted. |
3913 | | * |
3914 | | * NOTE: 4.4.7 Bypassed |
3915 | | * When a general entity reference appears in the EntityValue in |
3916 | | * an entity declaration, it is bypassed and left as is. |
3917 | | * so XML_SUBSTITUTE_REF is not set here. |
3918 | | */ |
3919 | 834k | ++ctxt->depth; |
3920 | 834k | ret = xmlStringDecodeEntitiesInt(ctxt, buf, len, XML_SUBSTITUTE_PEREF, |
3921 | 834k | 0, 0, 0, /* check */ 1); |
3922 | 834k | --ctxt->depth; |
3923 | | |
3924 | 834k | if (orig != NULL) { |
3925 | 834k | *orig = buf; |
3926 | 834k | buf = NULL; |
3927 | 834k | } |
3928 | | |
3929 | 850k | error: |
3930 | 850k | if (buf != NULL) |
3931 | 15.9k | xmlFree(buf); |
3932 | 850k | return(ret); |
3933 | 834k | } |
3934 | | |
3935 | | /** |
3936 | | * xmlParseAttValueComplex: |
3937 | | * @ctxt: an XML parser context |
3938 | | * @len: the resulting attribute len |
3939 | | * @normalize: whether to apply the inner normalization |
3940 | | * |
3941 | | * parse a value for an attribute, this is the fallback function |
3942 | | * of xmlParseAttValue() when the attribute parsing requires handling |
3943 | | * of non-ASCII characters, or normalization compaction. |
3944 | | * |
3945 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
3946 | | */ |
3947 | | static xmlChar * |
3948 | 466k | xmlParseAttValueComplex(xmlParserCtxtPtr ctxt, int *attlen, int normalize) { |
3949 | 466k | xmlChar limit = 0; |
3950 | 466k | xmlChar *buf = NULL; |
3951 | 466k | xmlChar *rep = NULL; |
3952 | 466k | size_t len = 0; |
3953 | 466k | size_t buf_size = 0; |
3954 | 466k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
3955 | 189k | XML_MAX_HUGE_LENGTH : |
3956 | 466k | XML_MAX_TEXT_LENGTH; |
3957 | 466k | int c, l, in_space = 0; |
3958 | 466k | xmlChar *current = NULL; |
3959 | 466k | xmlEntityPtr ent; |
3960 | | |
3961 | 466k | if (NXT(0) == '"') { |
3962 | 312k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3963 | 312k | limit = '"'; |
3964 | 312k | NEXT; |
3965 | 312k | } else if (NXT(0) == '\'') { |
3966 | 154k | limit = '\''; |
3967 | 154k | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
3968 | 154k | NEXT; |
3969 | 154k | } else { |
3970 | 0 | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
3971 | 0 | return(NULL); |
3972 | 0 | } |
3973 | | |
3974 | | /* |
3975 | | * allocate a translation buffer. |
3976 | | */ |
3977 | 466k | buf_size = XML_PARSER_BUFFER_SIZE; |
3978 | 466k | buf = (xmlChar *) xmlMallocAtomic(buf_size); |
3979 | 466k | if (buf == NULL) goto mem_error; |
3980 | | |
3981 | | /* |
3982 | | * OK loop until we reach one of the ending char or a size limit. |
3983 | | */ |
3984 | 466k | c = CUR_CHAR(l); |
3985 | 24.1M | while (((NXT(0) != limit) && /* checked */ |
3986 | 24.1M | (IS_CHAR(c)) && (c != '<')) && |
3987 | 24.1M | (ctxt->instate != XML_PARSER_EOF)) { |
3988 | 23.6M | if (c == '&') { |
3989 | 2.61M | in_space = 0; |
3990 | 2.61M | if (NXT(1) == '#') { |
3991 | 239k | int val = xmlParseCharRef(ctxt); |
3992 | | |
3993 | 239k | if (val == '&') { |
3994 | 3.05k | if (ctxt->replaceEntities) { |
3995 | 1.09k | if (len + 10 > buf_size) { |
3996 | 178 | growBuffer(buf, 10); |
3997 | 178 | } |
3998 | 1.09k | buf[len++] = '&'; |
3999 | 1.96k | } else { |
4000 | | /* |
4001 | | * The reparsing will be done in xmlStringGetNodeList() |
4002 | | * called by the attribute() function in SAX.c |
4003 | | */ |
4004 | 1.96k | if (len + 10 > buf_size) { |
4005 | 200 | growBuffer(buf, 10); |
4006 | 200 | } |
4007 | 1.96k | buf[len++] = '&'; |
4008 | 1.96k | buf[len++] = '#'; |
4009 | 1.96k | buf[len++] = '3'; |
4010 | 1.96k | buf[len++] = '8'; |
4011 | 1.96k | buf[len++] = ';'; |
4012 | 1.96k | } |
4013 | 235k | } else if (val != 0) { |
4014 | 209k | if (len + 10 > buf_size) { |
4015 | 2.00k | growBuffer(buf, 10); |
4016 | 2.00k | } |
4017 | 209k | len += xmlCopyChar(0, &buf[len], val); |
4018 | 209k | } |
4019 | 2.37M | } else { |
4020 | 2.37M | ent = xmlParseEntityRef(ctxt); |
4021 | 2.37M | if ((ent != NULL) && |
4022 | 2.37M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
4023 | 54.5k | if (len + 10 > buf_size) { |
4024 | 208 | growBuffer(buf, 10); |
4025 | 208 | } |
4026 | 54.5k | if ((ctxt->replaceEntities == 0) && |
4027 | 54.5k | (ent->content[0] == '&')) { |
4028 | 17.2k | buf[len++] = '&'; |
4029 | 17.2k | buf[len++] = '#'; |
4030 | 17.2k | buf[len++] = '3'; |
4031 | 17.2k | buf[len++] = '8'; |
4032 | 17.2k | buf[len++] = ';'; |
4033 | 37.3k | } else { |
4034 | 37.3k | buf[len++] = ent->content[0]; |
4035 | 37.3k | } |
4036 | 2.32M | } else if ((ent != NULL) && |
4037 | 2.32M | (ctxt->replaceEntities != 0)) { |
4038 | 1.39M | if (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) { |
4039 | 1.39M | if (xmlParserEntityCheck(ctxt, ent->length)) |
4040 | 0 | goto error; |
4041 | | |
4042 | 1.39M | ++ctxt->depth; |
4043 | 1.39M | rep = xmlStringDecodeEntitiesInt(ctxt, ent->content, |
4044 | 1.39M | ent->length, XML_SUBSTITUTE_REF, 0, 0, 0, |
4045 | 1.39M | /* check */ 1); |
4046 | 1.39M | --ctxt->depth; |
4047 | 1.39M | if (rep != NULL) { |
4048 | 1.38M | current = rep; |
4049 | 288M | while (*current != 0) { /* non input consuming */ |
4050 | 286M | if ((*current == 0xD) || (*current == 0xA) || |
4051 | 286M | (*current == 0x9)) { |
4052 | 637k | buf[len++] = 0x20; |
4053 | 637k | current++; |
4054 | 637k | } else |
4055 | 286M | buf[len++] = *current++; |
4056 | 286M | if (len + 10 > buf_size) { |
4057 | 33.7k | growBuffer(buf, 10); |
4058 | 33.7k | } |
4059 | 286M | } |
4060 | 1.38M | xmlFree(rep); |
4061 | 1.38M | rep = NULL; |
4062 | 1.38M | } |
4063 | 1.39M | } else { |
4064 | 0 | if (len + 10 > buf_size) { |
4065 | 0 | growBuffer(buf, 10); |
4066 | 0 | } |
4067 | 0 | if (ent->content != NULL) |
4068 | 0 | buf[len++] = ent->content[0]; |
4069 | 0 | } |
4070 | 1.39M | } else if (ent != NULL) { |
4071 | 676k | int i = xmlStrlen(ent->name); |
4072 | 676k | const xmlChar *cur = ent->name; |
4073 | | |
4074 | | /* |
4075 | | * We also check for recursion and amplification |
4076 | | * when entities are not substituted. They're |
4077 | | * often expanded later. |
4078 | | */ |
4079 | 676k | if ((ent->etype != XML_INTERNAL_PREDEFINED_ENTITY) && |
4080 | 676k | (ent->content != NULL)) { |
4081 | 652k | if ((ent->flags & XML_ENT_CHECKED) == 0) { |
4082 | 12.3k | unsigned long oldCopy = ctxt->sizeentcopy; |
4083 | | |
4084 | 12.3k | ctxt->sizeentcopy = ent->length; |
4085 | | |
4086 | 12.3k | ++ctxt->depth; |
4087 | 12.3k | rep = xmlStringDecodeEntitiesInt(ctxt, |
4088 | 12.3k | ent->content, ent->length, |
4089 | 12.3k | XML_SUBSTITUTE_REF, 0, 0, 0, |
4090 | 12.3k | /* check */ 1); |
4091 | 12.3k | --ctxt->depth; |
4092 | | |
4093 | | /* |
4094 | | * If we're parsing DTD content, the entity |
4095 | | * might reference other entities which |
4096 | | * weren't defined yet, so the check isn't |
4097 | | * reliable. |
4098 | | */ |
4099 | 12.3k | if (ctxt->inSubset == 0) { |
4100 | 10.8k | ent->flags |= XML_ENT_CHECKED; |
4101 | 10.8k | ent->expandedSize = ctxt->sizeentcopy; |
4102 | 10.8k | } |
4103 | | |
4104 | 12.3k | if (rep != NULL) { |
4105 | 12.2k | xmlFree(rep); |
4106 | 12.2k | rep = NULL; |
4107 | 12.2k | } else { |
4108 | 158 | ent->content[0] = 0; |
4109 | 158 | } |
4110 | | |
4111 | 12.3k | if (xmlParserEntityCheck(ctxt, oldCopy)) |
4112 | 14 | goto error; |
4113 | 640k | } else { |
4114 | 640k | if (xmlParserEntityCheck(ctxt, ent->expandedSize)) |
4115 | 0 | goto error; |
4116 | 640k | } |
4117 | 652k | } |
4118 | | |
4119 | | /* |
4120 | | * Just output the reference |
4121 | | */ |
4122 | 676k | buf[len++] = '&'; |
4123 | 679k | while (len + i + 10 > buf_size) { |
4124 | 6.77k | growBuffer(buf, i + 10); |
4125 | 6.77k | } |
4126 | 1.63M | for (;i > 0;i--) |
4127 | 962k | buf[len++] = *cur++; |
4128 | 676k | buf[len++] = ';'; |
4129 | 676k | } |
4130 | 2.37M | } |
4131 | 21.0M | } else { |
4132 | 21.0M | if ((c == 0x20) || (c == 0xD) || (c == 0xA) || (c == 0x9)) { |
4133 | 2.79M | if ((len != 0) || (!normalize)) { |
4134 | 2.68M | if ((!normalize) || (!in_space)) { |
4135 | 2.37M | COPY_BUF(l,buf,len,0x20); |
4136 | 2.37M | while (len + 10 > buf_size) { |
4137 | 10.0k | growBuffer(buf, 10); |
4138 | 10.0k | } |
4139 | 2.37M | } |
4140 | 2.68M | in_space = 1; |
4141 | 2.68M | } |
4142 | 18.2M | } else { |
4143 | 18.2M | in_space = 0; |
4144 | 18.2M | COPY_BUF(l,buf,len,c); |
4145 | 18.2M | if (len + 10 > buf_size) { |
4146 | 81.0k | growBuffer(buf, 10); |
4147 | 81.0k | } |
4148 | 18.2M | } |
4149 | 21.0M | NEXTL(l); |
4150 | 21.0M | } |
4151 | 23.6M | GROW; |
4152 | 23.6M | c = CUR_CHAR(l); |
4153 | 23.6M | if (len > maxLength) { |
4154 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4155 | 0 | "AttValue length too long\n"); |
4156 | 0 | goto mem_error; |
4157 | 0 | } |
4158 | 23.6M | } |
4159 | 466k | if (ctxt->instate == XML_PARSER_EOF) |
4160 | 571 | goto error; |
4161 | | |
4162 | 465k | if ((in_space) && (normalize)) { |
4163 | 25.2k | while ((len > 0) && (buf[len - 1] == 0x20)) len--; |
4164 | 9.14k | } |
4165 | 465k | buf[len] = 0; |
4166 | 465k | if (RAW == '<') { |
4167 | 94.5k | xmlFatalErr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, NULL); |
4168 | 371k | } else if (RAW != limit) { |
4169 | 71.2k | if ((c != 0) && (!IS_CHAR(c))) { |
4170 | 30.6k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_CHAR, |
4171 | 30.6k | "invalid character in attribute value\n"); |
4172 | 40.5k | } else { |
4173 | 40.5k | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
4174 | 40.5k | "AttValue: ' expected\n"); |
4175 | 40.5k | } |
4176 | 71.2k | } else |
4177 | 299k | NEXT; |
4178 | | |
4179 | 465k | if (attlen != NULL) *attlen = len; |
4180 | 465k | return(buf); |
4181 | | |
4182 | 0 | mem_error: |
4183 | 0 | xmlErrMemory(ctxt, NULL); |
4184 | 585 | error: |
4185 | 585 | if (buf != NULL) |
4186 | 585 | xmlFree(buf); |
4187 | 585 | if (rep != NULL) |
4188 | 0 | xmlFree(rep); |
4189 | 585 | return(NULL); |
4190 | 0 | } |
4191 | | |
4192 | | /** |
4193 | | * xmlParseAttValue: |
4194 | | * @ctxt: an XML parser context |
4195 | | * |
4196 | | * DEPRECATED: Internal function, don't use. |
4197 | | * |
4198 | | * parse a value for an attribute |
4199 | | * Note: the parser won't do substitution of entities here, this |
4200 | | * will be handled later in xmlStringGetNodeList |
4201 | | * |
4202 | | * [10] AttValue ::= '"' ([^<&"] | Reference)* '"' | |
4203 | | * "'" ([^<&'] | Reference)* "'" |
4204 | | * |
4205 | | * 3.3.3 Attribute-Value Normalization: |
4206 | | * Before the value of an attribute is passed to the application or |
4207 | | * checked for validity, the XML processor must normalize it as follows: |
4208 | | * - a character reference is processed by appending the referenced |
4209 | | * character to the attribute value |
4210 | | * - an entity reference is processed by recursively processing the |
4211 | | * replacement text of the entity |
4212 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
4213 | | * appending #x20 to the normalized value, except that only a single |
4214 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
4215 | | * parsed entity or the literal entity value of an internal parsed entity |
4216 | | * - other characters are processed by appending them to the normalized value |
4217 | | * If the declared value is not CDATA, then the XML processor must further |
4218 | | * process the normalized attribute value by discarding any leading and |
4219 | | * trailing space (#x20) characters, and by replacing sequences of space |
4220 | | * (#x20) characters by a single space (#x20) character. |
4221 | | * All attributes for which no declaration has been read should be treated |
4222 | | * by a non-validating parser as if declared CDATA. |
4223 | | * |
4224 | | * Returns the AttValue parsed or NULL. The value has to be freed by the caller. |
4225 | | */ |
4226 | | |
4227 | | |
4228 | | xmlChar * |
4229 | 3.63M | xmlParseAttValue(xmlParserCtxtPtr ctxt) { |
4230 | 3.63M | if ((ctxt == NULL) || (ctxt->input == NULL)) return(NULL); |
4231 | 3.63M | return(xmlParseAttValueInternal(ctxt, NULL, NULL, 0)); |
4232 | 3.63M | } |
4233 | | |
4234 | | /** |
4235 | | * xmlParseSystemLiteral: |
4236 | | * @ctxt: an XML parser context |
4237 | | * |
4238 | | * DEPRECATED: Internal function, don't use. |
4239 | | * |
4240 | | * parse an XML Literal |
4241 | | * |
4242 | | * [11] SystemLiteral ::= ('"' [^"]* '"') | ("'" [^']* "'") |
4243 | | * |
4244 | | * Returns the SystemLiteral parsed or NULL |
4245 | | */ |
4246 | | |
4247 | | xmlChar * |
4248 | 351k | xmlParseSystemLiteral(xmlParserCtxtPtr ctxt) { |
4249 | 351k | xmlChar *buf = NULL; |
4250 | 351k | int len = 0; |
4251 | 351k | int size = XML_PARSER_BUFFER_SIZE; |
4252 | 351k | int cur, l; |
4253 | 351k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4254 | 143k | XML_MAX_TEXT_LENGTH : |
4255 | 351k | XML_MAX_NAME_LENGTH; |
4256 | 351k | xmlChar stop; |
4257 | 351k | int state = ctxt->instate; |
4258 | 351k | int count = 0; |
4259 | | |
4260 | 351k | SHRINK; |
4261 | 351k | if (RAW == '"') { |
4262 | 319k | NEXT; |
4263 | 319k | stop = '"'; |
4264 | 319k | } else if (RAW == '\'') { |
4265 | 22.0k | NEXT; |
4266 | 22.0k | stop = '\''; |
4267 | 22.0k | } else { |
4268 | 10.3k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4269 | 10.3k | return(NULL); |
4270 | 10.3k | } |
4271 | | |
4272 | 341k | buf = (xmlChar *) xmlMallocAtomic(size); |
4273 | 341k | if (buf == NULL) { |
4274 | 0 | xmlErrMemory(ctxt, NULL); |
4275 | 0 | return(NULL); |
4276 | 0 | } |
4277 | 341k | ctxt->instate = XML_PARSER_SYSTEM_LITERAL; |
4278 | 341k | cur = CUR_CHAR(l); |
4279 | 19.2M | while ((IS_CHAR(cur)) && (cur != stop)) { /* checked */ |
4280 | 18.9M | if (len + 5 >= size) { |
4281 | 12.5k | xmlChar *tmp; |
4282 | | |
4283 | 12.5k | size *= 2; |
4284 | 12.5k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4285 | 12.5k | if (tmp == NULL) { |
4286 | 0 | xmlFree(buf); |
4287 | 0 | xmlErrMemory(ctxt, NULL); |
4288 | 0 | ctxt->instate = (xmlParserInputState) state; |
4289 | 0 | return(NULL); |
4290 | 0 | } |
4291 | 12.5k | buf = tmp; |
4292 | 12.5k | } |
4293 | 18.9M | count++; |
4294 | 18.9M | if (count > 50) { |
4295 | 281k | SHRINK; |
4296 | 281k | GROW; |
4297 | 281k | count = 0; |
4298 | 281k | if (ctxt->instate == XML_PARSER_EOF) { |
4299 | 0 | xmlFree(buf); |
4300 | 0 | return(NULL); |
4301 | 0 | } |
4302 | 281k | } |
4303 | 18.9M | COPY_BUF(l,buf,len,cur); |
4304 | 18.9M | NEXTL(l); |
4305 | 18.9M | cur = CUR_CHAR(l); |
4306 | 18.9M | if (cur == 0) { |
4307 | 3.05k | GROW; |
4308 | 3.05k | SHRINK; |
4309 | 3.05k | cur = CUR_CHAR(l); |
4310 | 3.05k | } |
4311 | 18.9M | if (len > maxLength) { |
4312 | 129 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "SystemLiteral"); |
4313 | 129 | xmlFree(buf); |
4314 | 129 | ctxt->instate = (xmlParserInputState) state; |
4315 | 129 | return(NULL); |
4316 | 129 | } |
4317 | 18.9M | } |
4318 | 341k | buf[len] = 0; |
4319 | 341k | ctxt->instate = (xmlParserInputState) state; |
4320 | 341k | if (!IS_CHAR(cur)) { |
4321 | 4.37k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4322 | 337k | } else { |
4323 | 337k | NEXT; |
4324 | 337k | } |
4325 | 341k | return(buf); |
4326 | 341k | } |
4327 | | |
4328 | | /** |
4329 | | * xmlParsePubidLiteral: |
4330 | | * @ctxt: an XML parser context |
4331 | | * |
4332 | | * DEPRECATED: Internal function, don't use. |
4333 | | * |
4334 | | * parse an XML public literal |
4335 | | * |
4336 | | * [12] PubidLiteral ::= '"' PubidChar* '"' | "'" (PubidChar - "'")* "'" |
4337 | | * |
4338 | | * Returns the PubidLiteral parsed or NULL. |
4339 | | */ |
4340 | | |
4341 | | xmlChar * |
4342 | 68.3k | xmlParsePubidLiteral(xmlParserCtxtPtr ctxt) { |
4343 | 68.3k | xmlChar *buf = NULL; |
4344 | 68.3k | int len = 0; |
4345 | 68.3k | int size = XML_PARSER_BUFFER_SIZE; |
4346 | 68.3k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4347 | 26.8k | XML_MAX_TEXT_LENGTH : |
4348 | 68.3k | XML_MAX_NAME_LENGTH; |
4349 | 68.3k | xmlChar cur; |
4350 | 68.3k | xmlChar stop; |
4351 | 68.3k | int count = 0; |
4352 | 68.3k | xmlParserInputState oldstate = ctxt->instate; |
4353 | | |
4354 | 68.3k | SHRINK; |
4355 | 68.3k | if (RAW == '"') { |
4356 | 42.3k | NEXT; |
4357 | 42.3k | stop = '"'; |
4358 | 42.3k | } else if (RAW == '\'') { |
4359 | 24.1k | NEXT; |
4360 | 24.1k | stop = '\''; |
4361 | 24.1k | } else { |
4362 | 1.82k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_STARTED, NULL); |
4363 | 1.82k | return(NULL); |
4364 | 1.82k | } |
4365 | 66.5k | buf = (xmlChar *) xmlMallocAtomic(size); |
4366 | 66.5k | if (buf == NULL) { |
4367 | 0 | xmlErrMemory(ctxt, NULL); |
4368 | 0 | return(NULL); |
4369 | 0 | } |
4370 | 66.5k | ctxt->instate = XML_PARSER_PUBLIC_LITERAL; |
4371 | 66.5k | cur = CUR; |
4372 | 4.23M | while ((IS_PUBIDCHAR_CH(cur)) && (cur != stop)) { /* checked */ |
4373 | 4.17M | if (len + 1 >= size) { |
4374 | 3.82k | xmlChar *tmp; |
4375 | | |
4376 | 3.82k | size *= 2; |
4377 | 3.82k | tmp = (xmlChar *) xmlRealloc(buf, size); |
4378 | 3.82k | if (tmp == NULL) { |
4379 | 0 | xmlErrMemory(ctxt, NULL); |
4380 | 0 | xmlFree(buf); |
4381 | 0 | return(NULL); |
4382 | 0 | } |
4383 | 3.82k | buf = tmp; |
4384 | 3.82k | } |
4385 | 4.17M | buf[len++] = cur; |
4386 | 4.17M | count++; |
4387 | 4.17M | if (count > 50) { |
4388 | 57.3k | SHRINK; |
4389 | 57.3k | GROW; |
4390 | 57.3k | count = 0; |
4391 | 57.3k | if (ctxt->instate == XML_PARSER_EOF) { |
4392 | 0 | xmlFree(buf); |
4393 | 0 | return(NULL); |
4394 | 0 | } |
4395 | 57.3k | } |
4396 | 4.17M | NEXT; |
4397 | 4.17M | cur = CUR; |
4398 | 4.17M | if (cur == 0) { |
4399 | 933 | GROW; |
4400 | 933 | SHRINK; |
4401 | 933 | cur = CUR; |
4402 | 933 | } |
4403 | 4.17M | if (len > maxLength) { |
4404 | 28 | xmlFatalErr(ctxt, XML_ERR_NAME_TOO_LONG, "Public ID"); |
4405 | 28 | xmlFree(buf); |
4406 | 28 | return(NULL); |
4407 | 28 | } |
4408 | 4.17M | } |
4409 | 66.4k | buf[len] = 0; |
4410 | 66.4k | if (cur != stop) { |
4411 | 9.62k | xmlFatalErr(ctxt, XML_ERR_LITERAL_NOT_FINISHED, NULL); |
4412 | 56.8k | } else { |
4413 | 56.8k | NEXT; |
4414 | 56.8k | } |
4415 | 66.4k | ctxt->instate = oldstate; |
4416 | 66.4k | return(buf); |
4417 | 66.5k | } |
4418 | | |
4419 | | static void xmlParseCharDataComplex(xmlParserCtxtPtr ctxt); |
4420 | | |
4421 | | /* |
4422 | | * used for the test in the inner loop of the char data testing |
4423 | | */ |
4424 | | static const unsigned char test_char_data[256] = { |
4425 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4426 | | 0x00, 0x09, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 0x9, CR/LF separated */ |
4427 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4428 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4429 | | 0x20, 0x21, 0x22, 0x23, 0x24, 0x25, 0x00, 0x27, /* & */ |
4430 | | 0x28, 0x29, 0x2A, 0x2B, 0x2C, 0x2D, 0x2E, 0x2F, |
4431 | | 0x30, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36, 0x37, |
4432 | | 0x38, 0x39, 0x3A, 0x3B, 0x00, 0x3D, 0x3E, 0x3F, /* < */ |
4433 | | 0x40, 0x41, 0x42, 0x43, 0x44, 0x45, 0x46, 0x47, |
4434 | | 0x48, 0x49, 0x4A, 0x4B, 0x4C, 0x4D, 0x4E, 0x4F, |
4435 | | 0x50, 0x51, 0x52, 0x53, 0x54, 0x55, 0x56, 0x57, |
4436 | | 0x58, 0x59, 0x5A, 0x5B, 0x5C, 0x00, 0x5E, 0x5F, /* ] */ |
4437 | | 0x60, 0x61, 0x62, 0x63, 0x64, 0x65, 0x66, 0x67, |
4438 | | 0x68, 0x69, 0x6A, 0x6B, 0x6C, 0x6D, 0x6E, 0x6F, |
4439 | | 0x70, 0x71, 0x72, 0x73, 0x74, 0x75, 0x76, 0x77, |
4440 | | 0x78, 0x79, 0x7A, 0x7B, 0x7C, 0x7D, 0x7E, 0x7F, |
4441 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* non-ascii */ |
4442 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4443 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4444 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4445 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4446 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4447 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4448 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4449 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4450 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4451 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4452 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4453 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4454 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4455 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, |
4456 | | 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00 |
4457 | | }; |
4458 | | |
4459 | | /** |
4460 | | * xmlParseCharData: |
4461 | | * @ctxt: an XML parser context |
4462 | | * @cdata: unused |
4463 | | * |
4464 | | * DEPRECATED: Internal function, don't use. |
4465 | | * |
4466 | | * Parse character data. Always makes progress if the first char isn't |
4467 | | * '<' or '&'. |
4468 | | * |
4469 | | * if we are within a CDATA section ']]>' marks an end of section. |
4470 | | * |
4471 | | * The right angle bracket (>) may be represented using the string ">", |
4472 | | * and must, for compatibility, be escaped using ">" or a character |
4473 | | * reference when it appears in the string "]]>" in content, when that |
4474 | | * string is not marking the end of a CDATA section. |
4475 | | * |
4476 | | * [14] CharData ::= [^<&]* - ([^<&]* ']]>' [^<&]*) |
4477 | | */ |
4478 | | |
4479 | | void |
4480 | 16.8M | xmlParseCharData(xmlParserCtxtPtr ctxt, ATTRIBUTE_UNUSED int cdata) { |
4481 | 16.8M | const xmlChar *in; |
4482 | 16.8M | int nbchar = 0; |
4483 | 16.8M | int line = ctxt->input->line; |
4484 | 16.8M | int col = ctxt->input->col; |
4485 | 16.8M | int ccol; |
4486 | | |
4487 | 16.8M | SHRINK; |
4488 | 16.8M | GROW; |
4489 | | /* |
4490 | | * Accelerated common case where input don't need to be |
4491 | | * modified before passing it to the handler. |
4492 | | */ |
4493 | 16.8M | in = ctxt->input->cur; |
4494 | 21.3M | do { |
4495 | 26.8M | get_more_space: |
4496 | 36.8M | while (*in == 0x20) { in++; ctxt->input->col++; } |
4497 | 26.8M | if (*in == 0xA) { |
4498 | 5.62M | do { |
4499 | 5.62M | ctxt->input->line++; ctxt->input->col = 1; |
4500 | 5.62M | in++; |
4501 | 5.62M | } while (*in == 0xA); |
4502 | 5.42M | goto get_more_space; |
4503 | 5.42M | } |
4504 | 21.3M | if (*in == '<') { |
4505 | 4.59M | nbchar = in - ctxt->input->cur; |
4506 | 4.59M | if (nbchar > 0) { |
4507 | 4.59M | const xmlChar *tmp = ctxt->input->cur; |
4508 | 4.59M | ctxt->input->cur = in; |
4509 | | |
4510 | 4.59M | if ((ctxt->sax != NULL) && |
4511 | 4.59M | (ctxt->sax->ignorableWhitespace != |
4512 | 4.59M | ctxt->sax->characters)) { |
4513 | 1.94M | if (areBlanks(ctxt, tmp, nbchar, 1)) { |
4514 | 1.47M | if (ctxt->sax->ignorableWhitespace != NULL) |
4515 | 1.47M | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4516 | 1.47M | tmp, nbchar); |
4517 | 1.47M | } else { |
4518 | 476k | if (ctxt->sax->characters != NULL) |
4519 | 476k | ctxt->sax->characters(ctxt->userData, |
4520 | 476k | tmp, nbchar); |
4521 | 476k | if (*ctxt->space == -1) |
4522 | 114k | *ctxt->space = -2; |
4523 | 476k | } |
4524 | 2.64M | } else if ((ctxt->sax != NULL) && |
4525 | 2.64M | (ctxt->sax->characters != NULL)) { |
4526 | 2.64M | ctxt->sax->characters(ctxt->userData, |
4527 | 2.64M | tmp, nbchar); |
4528 | 2.64M | } |
4529 | 4.59M | } |
4530 | 4.59M | return; |
4531 | 4.59M | } |
4532 | | |
4533 | 21.9M | get_more: |
4534 | 21.9M | ccol = ctxt->input->col; |
4535 | 302M | while (test_char_data[*in]) { |
4536 | 280M | in++; |
4537 | 280M | ccol++; |
4538 | 280M | } |
4539 | 21.9M | ctxt->input->col = ccol; |
4540 | 21.9M | if (*in == 0xA) { |
4541 | 4.90M | do { |
4542 | 4.90M | ctxt->input->line++; ctxt->input->col = 1; |
4543 | 4.90M | in++; |
4544 | 4.90M | } while (*in == 0xA); |
4545 | 4.78M | goto get_more; |
4546 | 4.78M | } |
4547 | 17.1M | if (*in == ']') { |
4548 | 403k | if ((in[1] == ']') && (in[2] == '>')) { |
4549 | 8.61k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4550 | 8.61k | ctxt->input->cur = in + 1; |
4551 | 8.61k | return; |
4552 | 8.61k | } |
4553 | 394k | in++; |
4554 | 394k | ctxt->input->col++; |
4555 | 394k | goto get_more; |
4556 | 403k | } |
4557 | 16.7M | nbchar = in - ctxt->input->cur; |
4558 | 16.7M | if (nbchar > 0) { |
4559 | 11.9M | if ((ctxt->sax != NULL) && |
4560 | 11.9M | (ctxt->sax->ignorableWhitespace != |
4561 | 11.9M | ctxt->sax->characters) && |
4562 | 11.9M | (IS_BLANK_CH(*ctxt->input->cur))) { |
4563 | 2.29M | const xmlChar *tmp = ctxt->input->cur; |
4564 | 2.29M | ctxt->input->cur = in; |
4565 | | |
4566 | 2.29M | if (areBlanks(ctxt, tmp, nbchar, 0)) { |
4567 | 732k | if (ctxt->sax->ignorableWhitespace != NULL) |
4568 | 732k | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4569 | 732k | tmp, nbchar); |
4570 | 1.56M | } else { |
4571 | 1.56M | if (ctxt->sax->characters != NULL) |
4572 | 1.56M | ctxt->sax->characters(ctxt->userData, |
4573 | 1.56M | tmp, nbchar); |
4574 | 1.56M | if (*ctxt->space == -1) |
4575 | 337k | *ctxt->space = -2; |
4576 | 1.56M | } |
4577 | 2.29M | line = ctxt->input->line; |
4578 | 2.29M | col = ctxt->input->col; |
4579 | 9.60M | } else if (ctxt->sax != NULL) { |
4580 | 9.60M | if (ctxt->sax->characters != NULL) |
4581 | 9.60M | ctxt->sax->characters(ctxt->userData, |
4582 | 9.60M | ctxt->input->cur, nbchar); |
4583 | 9.60M | line = ctxt->input->line; |
4584 | 9.60M | col = ctxt->input->col; |
4585 | 9.60M | } |
4586 | 11.9M | } |
4587 | 16.7M | ctxt->input->cur = in; |
4588 | 16.7M | if (*in == 0xD) { |
4589 | 4.57M | in++; |
4590 | 4.57M | if (*in == 0xA) { |
4591 | 4.54M | ctxt->input->cur = in; |
4592 | 4.54M | in++; |
4593 | 4.54M | ctxt->input->line++; ctxt->input->col = 1; |
4594 | 4.54M | continue; /* while */ |
4595 | 4.54M | } |
4596 | 32.3k | in--; |
4597 | 32.3k | } |
4598 | 12.2M | if (*in == '<') { |
4599 | 9.51M | return; |
4600 | 9.51M | } |
4601 | 2.73M | if (*in == '&') { |
4602 | 927k | return; |
4603 | 927k | } |
4604 | 1.80M | SHRINK; |
4605 | 1.80M | GROW; |
4606 | 1.80M | if (ctxt->instate == XML_PARSER_EOF) |
4607 | 0 | return; |
4608 | 1.80M | in = ctxt->input->cur; |
4609 | 6.35M | } while (((*in >= 0x20) && (*in <= 0x7F)) || |
4610 | 6.35M | (*in == 0x09) || (*in == 0x0a)); |
4611 | 1.82M | ctxt->input->line = line; |
4612 | 1.82M | ctxt->input->col = col; |
4613 | 1.82M | xmlParseCharDataComplex(ctxt); |
4614 | 1.82M | } |
4615 | | |
4616 | | /** |
4617 | | * xmlParseCharDataComplex: |
4618 | | * @ctxt: an XML parser context |
4619 | | * @cdata: int indicating whether we are within a CDATA section |
4620 | | * |
4621 | | * Always makes progress if the first char isn't '<' or '&'. |
4622 | | * |
4623 | | * parse a CharData section.this is the fallback function |
4624 | | * of xmlParseCharData() when the parsing requires handling |
4625 | | * of non-ASCII characters. |
4626 | | */ |
4627 | | static void |
4628 | 1.82M | xmlParseCharDataComplex(xmlParserCtxtPtr ctxt) { |
4629 | 1.82M | xmlChar buf[XML_PARSER_BIG_BUFFER_SIZE + 5]; |
4630 | 1.82M | int nbchar = 0; |
4631 | 1.82M | int cur, l; |
4632 | 1.82M | int count = 0; |
4633 | | |
4634 | 1.82M | SHRINK; |
4635 | 1.82M | GROW; |
4636 | 1.82M | cur = CUR_CHAR(l); |
4637 | 33.2M | while ((cur != '<') && /* checked */ |
4638 | 33.2M | (cur != '&') && |
4639 | 33.2M | (IS_CHAR(cur))) /* test also done in xmlCurrentChar() */ { |
4640 | 31.4M | if ((cur == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
4641 | 5.86k | xmlFatalErr(ctxt, XML_ERR_MISPLACED_CDATA_END, NULL); |
4642 | 5.86k | } |
4643 | 31.4M | COPY_BUF(l,buf,nbchar,cur); |
4644 | | /* move current position before possible calling of ctxt->sax->characters */ |
4645 | 31.4M | NEXTL(l); |
4646 | 31.4M | cur = CUR_CHAR(l); |
4647 | 31.4M | if (nbchar >= XML_PARSER_BIG_BUFFER_SIZE) { |
4648 | 75.5k | buf[nbchar] = 0; |
4649 | | |
4650 | | /* |
4651 | | * OK the segment is to be consumed as chars. |
4652 | | */ |
4653 | 75.5k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4654 | 61.3k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4655 | 296 | if (ctxt->sax->ignorableWhitespace != NULL) |
4656 | 296 | ctxt->sax->ignorableWhitespace(ctxt->userData, |
4657 | 296 | buf, nbchar); |
4658 | 61.0k | } else { |
4659 | 61.0k | if (ctxt->sax->characters != NULL) |
4660 | 61.0k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4661 | 61.0k | if ((ctxt->sax->characters != |
4662 | 61.0k | ctxt->sax->ignorableWhitespace) && |
4663 | 61.0k | (*ctxt->space == -1)) |
4664 | 1.94k | *ctxt->space = -2; |
4665 | 61.0k | } |
4666 | 61.3k | } |
4667 | 75.5k | nbchar = 0; |
4668 | | /* something really bad happened in the SAX callback */ |
4669 | 75.5k | if (ctxt->instate != XML_PARSER_CONTENT) |
4670 | 0 | return; |
4671 | 75.5k | } |
4672 | 31.4M | count++; |
4673 | 31.4M | if (count > 50) { |
4674 | 427k | SHRINK; |
4675 | 427k | GROW; |
4676 | 427k | count = 0; |
4677 | 427k | if (ctxt->instate == XML_PARSER_EOF) |
4678 | 0 | return; |
4679 | 427k | } |
4680 | 31.4M | } |
4681 | 1.82M | if (nbchar != 0) { |
4682 | 865k | buf[nbchar] = 0; |
4683 | | /* |
4684 | | * OK the segment is to be consumed as chars. |
4685 | | */ |
4686 | 865k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
4687 | 733k | if (areBlanks(ctxt, buf, nbchar, 0)) { |
4688 | 1.63k | if (ctxt->sax->ignorableWhitespace != NULL) |
4689 | 1.63k | ctxt->sax->ignorableWhitespace(ctxt->userData, buf, nbchar); |
4690 | 732k | } else { |
4691 | 732k | if (ctxt->sax->characters != NULL) |
4692 | 732k | ctxt->sax->characters(ctxt->userData, buf, nbchar); |
4693 | 732k | if ((ctxt->sax->characters != ctxt->sax->ignorableWhitespace) && |
4694 | 732k | (*ctxt->space == -1)) |
4695 | 69.1k | *ctxt->space = -2; |
4696 | 732k | } |
4697 | 733k | } |
4698 | 865k | } |
4699 | 1.82M | if ((ctxt->input->cur < ctxt->input->end) && (!IS_CHAR(cur))) { |
4700 | | /* Generate the error and skip the offending character */ |
4701 | 1.28M | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4702 | 1.28M | "PCDATA invalid Char value %d\n", |
4703 | 1.28M | cur ? cur : CUR); |
4704 | 1.28M | NEXT; |
4705 | 1.28M | } |
4706 | 1.82M | } |
4707 | | |
4708 | | /** |
4709 | | * xmlParseExternalID: |
4710 | | * @ctxt: an XML parser context |
4711 | | * @publicID: a xmlChar** receiving PubidLiteral |
4712 | | * @strict: indicate whether we should restrict parsing to only |
4713 | | * production [75], see NOTE below |
4714 | | * |
4715 | | * DEPRECATED: Internal function, don't use. |
4716 | | * |
4717 | | * Parse an External ID or a Public ID |
4718 | | * |
4719 | | * NOTE: Productions [75] and [83] interact badly since [75] can generate |
4720 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
4721 | | * |
4722 | | * [75] ExternalID ::= 'SYSTEM' S SystemLiteral |
4723 | | * | 'PUBLIC' S PubidLiteral S SystemLiteral |
4724 | | * |
4725 | | * [83] PublicID ::= 'PUBLIC' S PubidLiteral |
4726 | | * |
4727 | | * Returns the function returns SystemLiteral and in the second |
4728 | | * case publicID receives PubidLiteral, is strict is off |
4729 | | * it is possible to return NULL and have publicID set. |
4730 | | */ |
4731 | | |
4732 | | xmlChar * |
4733 | 549k | xmlParseExternalID(xmlParserCtxtPtr ctxt, xmlChar **publicID, int strict) { |
4734 | 549k | xmlChar *URI = NULL; |
4735 | | |
4736 | 549k | SHRINK; |
4737 | | |
4738 | 549k | *publicID = NULL; |
4739 | 549k | if (CMP6(CUR_PTR, 'S', 'Y', 'S', 'T', 'E', 'M')) { |
4740 | 296k | SKIP(6); |
4741 | 296k | if (SKIP_BLANKS == 0) { |
4742 | 999 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4743 | 999 | "Space required after 'SYSTEM'\n"); |
4744 | 999 | } |
4745 | 296k | URI = xmlParseSystemLiteral(ctxt); |
4746 | 296k | if (URI == NULL) { |
4747 | 1.49k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4748 | 1.49k | } |
4749 | 296k | } else if (CMP6(CUR_PTR, 'P', 'U', 'B', 'L', 'I', 'C')) { |
4750 | 68.3k | SKIP(6); |
4751 | 68.3k | if (SKIP_BLANKS == 0) { |
4752 | 1.91k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4753 | 1.91k | "Space required after 'PUBLIC'\n"); |
4754 | 1.91k | } |
4755 | 68.3k | *publicID = xmlParsePubidLiteral(ctxt); |
4756 | 68.3k | if (*publicID == NULL) { |
4757 | 1.84k | xmlFatalErr(ctxt, XML_ERR_PUBID_REQUIRED, NULL); |
4758 | 1.84k | } |
4759 | 68.3k | if (strict) { |
4760 | | /* |
4761 | | * We don't handle [83] so "S SystemLiteral" is required. |
4762 | | */ |
4763 | 53.8k | if (SKIP_BLANKS == 0) { |
4764 | 8.05k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
4765 | 8.05k | "Space required after the Public Identifier\n"); |
4766 | 8.05k | } |
4767 | 53.8k | } else { |
4768 | | /* |
4769 | | * We handle [83] so we return immediately, if |
4770 | | * "S SystemLiteral" is not detected. We skip blanks if no |
4771 | | * system literal was found, but this is harmless since we must |
4772 | | * be at the end of a NotationDecl. |
4773 | | */ |
4774 | 14.4k | if (SKIP_BLANKS == 0) return(NULL); |
4775 | 2.12k | if ((CUR != '\'') && (CUR != '"')) return(NULL); |
4776 | 2.12k | } |
4777 | 55.4k | URI = xmlParseSystemLiteral(ctxt); |
4778 | 55.4k | if (URI == NULL) { |
4779 | 9.02k | xmlFatalErr(ctxt, XML_ERR_URI_REQUIRED, NULL); |
4780 | 9.02k | } |
4781 | 55.4k | } |
4782 | 536k | return(URI); |
4783 | 549k | } |
4784 | | |
4785 | | /** |
4786 | | * xmlParseCommentComplex: |
4787 | | * @ctxt: an XML parser context |
4788 | | * @buf: the already parsed part of the buffer |
4789 | | * @len: number of bytes in the buffer |
4790 | | * @size: allocated size of the buffer |
4791 | | * |
4792 | | * Skip an XML (SGML) comment <!-- .... --> |
4793 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4794 | | * must not occur within comments. " |
4795 | | * This is the slow routine in case the accelerator for ascii didn't work |
4796 | | * |
4797 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4798 | | */ |
4799 | | static void |
4800 | | xmlParseCommentComplex(xmlParserCtxtPtr ctxt, xmlChar *buf, |
4801 | 390k | size_t len, size_t size) { |
4802 | 390k | int q, ql; |
4803 | 390k | int r, rl; |
4804 | 390k | int cur, l; |
4805 | 390k | size_t count = 0; |
4806 | 390k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4807 | 308k | XML_MAX_HUGE_LENGTH : |
4808 | 390k | XML_MAX_TEXT_LENGTH; |
4809 | 390k | int inputid; |
4810 | | |
4811 | 390k | inputid = ctxt->input->id; |
4812 | | |
4813 | 390k | if (buf == NULL) { |
4814 | 63.0k | len = 0; |
4815 | 63.0k | size = XML_PARSER_BUFFER_SIZE; |
4816 | 63.0k | buf = (xmlChar *) xmlMallocAtomic(size); |
4817 | 63.0k | if (buf == NULL) { |
4818 | 0 | xmlErrMemory(ctxt, NULL); |
4819 | 0 | return; |
4820 | 0 | } |
4821 | 63.0k | } |
4822 | 390k | GROW; /* Assure there's enough input data */ |
4823 | 390k | q = CUR_CHAR(ql); |
4824 | 390k | if (q == 0) |
4825 | 248k | goto not_terminated; |
4826 | 142k | if (!IS_CHAR(q)) { |
4827 | 7.86k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4828 | 7.86k | "xmlParseComment: invalid xmlChar value %d\n", |
4829 | 7.86k | q); |
4830 | 7.86k | xmlFree (buf); |
4831 | 7.86k | return; |
4832 | 7.86k | } |
4833 | 134k | NEXTL(ql); |
4834 | 134k | r = CUR_CHAR(rl); |
4835 | 134k | if (r == 0) |
4836 | 2.78k | goto not_terminated; |
4837 | 131k | if (!IS_CHAR(r)) { |
4838 | 1.19k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4839 | 1.19k | "xmlParseComment: invalid xmlChar value %d\n", |
4840 | 1.19k | r); |
4841 | 1.19k | xmlFree (buf); |
4842 | 1.19k | return; |
4843 | 1.19k | } |
4844 | 130k | NEXTL(rl); |
4845 | 130k | cur = CUR_CHAR(l); |
4846 | 130k | if (cur == 0) |
4847 | 2.13k | goto not_terminated; |
4848 | 33.1M | while (IS_CHAR(cur) && /* checked */ |
4849 | 33.1M | ((cur != '>') || |
4850 | 33.1M | (r != '-') || (q != '-'))) { |
4851 | 33.0M | if ((r == '-') && (q == '-')) { |
4852 | 22.4k | xmlFatalErr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, NULL); |
4853 | 22.4k | } |
4854 | 33.0M | if (len + 5 >= size) { |
4855 | 52.5k | xmlChar *new_buf; |
4856 | 52.5k | size_t new_size; |
4857 | | |
4858 | 52.5k | new_size = size * 2; |
4859 | 52.5k | new_buf = (xmlChar *) xmlRealloc(buf, new_size); |
4860 | 52.5k | if (new_buf == NULL) { |
4861 | 0 | xmlFree (buf); |
4862 | 0 | xmlErrMemory(ctxt, NULL); |
4863 | 0 | return; |
4864 | 0 | } |
4865 | 52.5k | buf = new_buf; |
4866 | 52.5k | size = new_size; |
4867 | 52.5k | } |
4868 | 33.0M | COPY_BUF(ql,buf,len,q); |
4869 | 33.0M | q = r; |
4870 | 33.0M | ql = rl; |
4871 | 33.0M | r = cur; |
4872 | 33.0M | rl = l; |
4873 | | |
4874 | 33.0M | count++; |
4875 | 33.0M | if (count > 50) { |
4876 | 617k | SHRINK; |
4877 | 617k | GROW; |
4878 | 617k | count = 0; |
4879 | 617k | if (ctxt->instate == XML_PARSER_EOF) { |
4880 | 0 | xmlFree(buf); |
4881 | 0 | return; |
4882 | 0 | } |
4883 | 617k | } |
4884 | 33.0M | NEXTL(l); |
4885 | 33.0M | cur = CUR_CHAR(l); |
4886 | 33.0M | if (cur == 0) { |
4887 | 22.0k | SHRINK; |
4888 | 22.0k | GROW; |
4889 | 22.0k | cur = CUR_CHAR(l); |
4890 | 22.0k | } |
4891 | | |
4892 | 33.0M | if (len > maxLength) { |
4893 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4894 | 0 | "Comment too big found", NULL); |
4895 | 0 | xmlFree (buf); |
4896 | 0 | return; |
4897 | 0 | } |
4898 | 33.0M | } |
4899 | 128k | buf[len] = 0; |
4900 | 128k | if (cur == 0) { |
4901 | 22.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4902 | 22.0k | "Comment not terminated \n<!--%.50s\n", buf); |
4903 | 105k | } else if (!IS_CHAR(cur)) { |
4904 | 5.21k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
4905 | 5.21k | "xmlParseComment: invalid xmlChar value %d\n", |
4906 | 5.21k | cur); |
4907 | 100k | } else { |
4908 | 100k | if (inputid != ctxt->input->id) { |
4909 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
4910 | 0 | "Comment doesn't start and stop in the same" |
4911 | 0 | " entity\n"); |
4912 | 0 | } |
4913 | 100k | NEXT; |
4914 | 100k | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
4915 | 100k | (!ctxt->disableSAX)) |
4916 | 89.5k | ctxt->sax->comment(ctxt->userData, buf); |
4917 | 100k | } |
4918 | 128k | xmlFree(buf); |
4919 | 128k | return; |
4920 | 253k | not_terminated: |
4921 | 253k | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
4922 | 253k | "Comment not terminated\n", NULL); |
4923 | 253k | xmlFree(buf); |
4924 | 253k | return; |
4925 | 128k | } |
4926 | | |
4927 | | /** |
4928 | | * xmlParseComment: |
4929 | | * @ctxt: an XML parser context |
4930 | | * |
4931 | | * DEPRECATED: Internal function, don't use. |
4932 | | * |
4933 | | * Parse an XML (SGML) comment. Always consumes '<!'. |
4934 | | * |
4935 | | * The spec says that "For compatibility, the string "--" (double-hyphen) |
4936 | | * must not occur within comments. " |
4937 | | * |
4938 | | * [15] Comment ::= '<!--' ((Char - '-') | ('-' (Char - '-')))* '-->' |
4939 | | */ |
4940 | | void |
4941 | 28.3M | xmlParseComment(xmlParserCtxtPtr ctxt) { |
4942 | 28.3M | xmlChar *buf = NULL; |
4943 | 28.3M | size_t size = XML_PARSER_BUFFER_SIZE; |
4944 | 28.3M | size_t len = 0; |
4945 | 28.3M | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
4946 | 5.74M | XML_MAX_HUGE_LENGTH : |
4947 | 28.3M | XML_MAX_TEXT_LENGTH; |
4948 | 28.3M | xmlParserInputState state; |
4949 | 28.3M | const xmlChar *in; |
4950 | 28.3M | size_t nbchar = 0; |
4951 | 28.3M | int ccol; |
4952 | 28.3M | int inputid; |
4953 | | |
4954 | | /* |
4955 | | * Check that there is a comment right here. |
4956 | | */ |
4957 | 28.3M | if ((RAW != '<') || (NXT(1) != '!')) |
4958 | 0 | return; |
4959 | 28.3M | SKIP(2); |
4960 | 28.3M | if ((RAW != '-') || (NXT(1) != '-')) |
4961 | 197 | return; |
4962 | 28.3M | state = ctxt->instate; |
4963 | 28.3M | ctxt->instate = XML_PARSER_COMMENT; |
4964 | 28.3M | inputid = ctxt->input->id; |
4965 | 28.3M | SKIP(2); |
4966 | 28.3M | SHRINK; |
4967 | 28.3M | GROW; |
4968 | | |
4969 | | /* |
4970 | | * Accelerated common case where input don't need to be |
4971 | | * modified before passing it to the handler. |
4972 | | */ |
4973 | 28.3M | in = ctxt->input->cur; |
4974 | 28.3M | do { |
4975 | 28.3M | if (*in == 0xA) { |
4976 | 107k | do { |
4977 | 107k | ctxt->input->line++; ctxt->input->col = 1; |
4978 | 107k | in++; |
4979 | 107k | } while (*in == 0xA); |
4980 | 102k | } |
4981 | 33.0M | get_more: |
4982 | 33.0M | ccol = ctxt->input->col; |
4983 | 216M | while (((*in > '-') && (*in <= 0x7F)) || |
4984 | 216M | ((*in >= 0x20) && (*in < '-')) || |
4985 | 216M | (*in == 0x09)) { |
4986 | 183M | in++; |
4987 | 183M | ccol++; |
4988 | 183M | } |
4989 | 33.0M | ctxt->input->col = ccol; |
4990 | 33.0M | if (*in == 0xA) { |
4991 | 1.15M | do { |
4992 | 1.15M | ctxt->input->line++; ctxt->input->col = 1; |
4993 | 1.15M | in++; |
4994 | 1.15M | } while (*in == 0xA); |
4995 | 1.09M | goto get_more; |
4996 | 1.09M | } |
4997 | 31.9M | nbchar = in - ctxt->input->cur; |
4998 | | /* |
4999 | | * save current set of data |
5000 | | */ |
5001 | 31.9M | if (nbchar > 0) { |
5002 | 4.66M | if ((ctxt->sax != NULL) && |
5003 | 4.66M | (ctxt->sax->comment != NULL)) { |
5004 | 4.66M | if (buf == NULL) { |
5005 | 1.15M | if ((*in == '-') && (in[1] == '-')) |
5006 | 467k | size = nbchar + 1; |
5007 | 689k | else |
5008 | 689k | size = XML_PARSER_BUFFER_SIZE + nbchar; |
5009 | 1.15M | buf = (xmlChar *) xmlMallocAtomic(size); |
5010 | 1.15M | if (buf == NULL) { |
5011 | 0 | xmlErrMemory(ctxt, NULL); |
5012 | 0 | ctxt->instate = state; |
5013 | 0 | return; |
5014 | 0 | } |
5015 | 1.15M | len = 0; |
5016 | 3.50M | } else if (len + nbchar + 1 >= size) { |
5017 | 440k | xmlChar *new_buf; |
5018 | 440k | size += len + nbchar + XML_PARSER_BUFFER_SIZE; |
5019 | 440k | new_buf = (xmlChar *) xmlRealloc(buf, size); |
5020 | 440k | if (new_buf == NULL) { |
5021 | 0 | xmlFree (buf); |
5022 | 0 | xmlErrMemory(ctxt, NULL); |
5023 | 0 | ctxt->instate = state; |
5024 | 0 | return; |
5025 | 0 | } |
5026 | 440k | buf = new_buf; |
5027 | 440k | } |
5028 | 4.66M | memcpy(&buf[len], ctxt->input->cur, nbchar); |
5029 | 4.66M | len += nbchar; |
5030 | 4.66M | buf[len] = 0; |
5031 | 4.66M | } |
5032 | 4.66M | } |
5033 | 31.9M | if (len > maxLength) { |
5034 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_COMMENT_NOT_FINISHED, |
5035 | 0 | "Comment too big found", NULL); |
5036 | 0 | xmlFree (buf); |
5037 | 0 | return; |
5038 | 0 | } |
5039 | 31.9M | ctxt->input->cur = in; |
5040 | 31.9M | if (*in == 0xA) { |
5041 | 0 | in++; |
5042 | 0 | ctxt->input->line++; ctxt->input->col = 1; |
5043 | 0 | } |
5044 | 31.9M | if (*in == 0xD) { |
5045 | 2.32M | in++; |
5046 | 2.32M | if (*in == 0xA) { |
5047 | 2.31M | ctxt->input->cur = in; |
5048 | 2.31M | in++; |
5049 | 2.31M | ctxt->input->line++; ctxt->input->col = 1; |
5050 | 2.31M | goto get_more; |
5051 | 2.31M | } |
5052 | 7.31k | in--; |
5053 | 7.31k | } |
5054 | 29.6M | SHRINK; |
5055 | 29.6M | GROW; |
5056 | 29.6M | if (ctxt->instate == XML_PARSER_EOF) { |
5057 | 0 | xmlFree(buf); |
5058 | 0 | return; |
5059 | 0 | } |
5060 | 29.6M | in = ctxt->input->cur; |
5061 | 29.6M | if (*in == '-') { |
5062 | 29.2M | if (in[1] == '-') { |
5063 | 28.0M | if (in[2] == '>') { |
5064 | 28.0M | if (ctxt->input->id != inputid) { |
5065 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5066 | 0 | "comment doesn't start and stop in the" |
5067 | 0 | " same entity\n"); |
5068 | 0 | } |
5069 | 28.0M | SKIP(3); |
5070 | 28.0M | if ((ctxt->sax != NULL) && (ctxt->sax->comment != NULL) && |
5071 | 28.0M | (!ctxt->disableSAX)) { |
5072 | 25.5M | if (buf != NULL) |
5073 | 730k | ctxt->sax->comment(ctxt->userData, buf); |
5074 | 24.8M | else |
5075 | 24.8M | ctxt->sax->comment(ctxt->userData, BAD_CAST ""); |
5076 | 25.5M | } |
5077 | 28.0M | if (buf != NULL) |
5078 | 829k | xmlFree(buf); |
5079 | 28.0M | if (ctxt->instate != XML_PARSER_EOF) |
5080 | 28.0M | ctxt->instate = state; |
5081 | 28.0M | return; |
5082 | 28.0M | } |
5083 | 58.2k | if (buf != NULL) { |
5084 | 32.9k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5085 | 32.9k | "Double hyphen within comment: " |
5086 | 32.9k | "<!--%.50s\n", |
5087 | 32.9k | buf); |
5088 | 32.9k | } else |
5089 | 25.3k | xmlFatalErrMsgStr(ctxt, XML_ERR_HYPHEN_IN_COMMENT, |
5090 | 25.3k | "Double hyphen within comment\n", NULL); |
5091 | 58.2k | if (ctxt->instate == XML_PARSER_EOF) { |
5092 | 0 | xmlFree(buf); |
5093 | 0 | return; |
5094 | 0 | } |
5095 | 58.2k | in++; |
5096 | 58.2k | ctxt->input->col++; |
5097 | 58.2k | } |
5098 | 1.26M | in++; |
5099 | 1.26M | ctxt->input->col++; |
5100 | 1.26M | goto get_more; |
5101 | 29.2M | } |
5102 | 29.6M | } while (((*in >= 0x20) && (*in <= 0x7F)) || (*in == 0x09) || (*in == 0x0a)); |
5103 | 390k | xmlParseCommentComplex(ctxt, buf, len, size); |
5104 | 390k | ctxt->instate = state; |
5105 | 390k | return; |
5106 | 28.3M | } |
5107 | | |
5108 | | |
5109 | | /** |
5110 | | * xmlParsePITarget: |
5111 | | * @ctxt: an XML parser context |
5112 | | * |
5113 | | * DEPRECATED: Internal function, don't use. |
5114 | | * |
5115 | | * parse the name of a PI |
5116 | | * |
5117 | | * [17] PITarget ::= Name - (('X' | 'x') ('M' | 'm') ('L' | 'l')) |
5118 | | * |
5119 | | * Returns the PITarget name or NULL |
5120 | | */ |
5121 | | |
5122 | | const xmlChar * |
5123 | 188k | xmlParsePITarget(xmlParserCtxtPtr ctxt) { |
5124 | 188k | const xmlChar *name; |
5125 | | |
5126 | 188k | name = xmlParseName(ctxt); |
5127 | 188k | if ((name != NULL) && |
5128 | 188k | ((name[0] == 'x') || (name[0] == 'X')) && |
5129 | 188k | ((name[1] == 'm') || (name[1] == 'M')) && |
5130 | 188k | ((name[2] == 'l') || (name[2] == 'L'))) { |
5131 | 54.1k | int i; |
5132 | 54.1k | if ((name[0] == 'x') && (name[1] == 'm') && |
5133 | 54.1k | (name[2] == 'l') && (name[3] == 0)) { |
5134 | 9.72k | xmlFatalErrMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5135 | 9.72k | "XML declaration allowed only at the start of the document\n"); |
5136 | 9.72k | return(name); |
5137 | 44.4k | } else if (name[3] == 0) { |
5138 | 3.16k | xmlFatalErr(ctxt, XML_ERR_RESERVED_XML_NAME, NULL); |
5139 | 3.16k | return(name); |
5140 | 3.16k | } |
5141 | 88.1k | for (i = 0;;i++) { |
5142 | 88.1k | if (xmlW3CPIs[i] == NULL) break; |
5143 | 64.8k | if (xmlStrEqual(name, (const xmlChar *)xmlW3CPIs[i])) |
5144 | 17.9k | return(name); |
5145 | 64.8k | } |
5146 | 23.3k | xmlWarningMsg(ctxt, XML_ERR_RESERVED_XML_NAME, |
5147 | 23.3k | "xmlParsePITarget: invalid name prefix 'xml'\n", |
5148 | 23.3k | NULL, NULL); |
5149 | 23.3k | } |
5150 | 157k | if ((name != NULL) && (xmlStrchr(name, ':') != NULL)) { |
5151 | 2.92k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5152 | 2.92k | "colons are forbidden from PI names '%s'\n", name, NULL, NULL); |
5153 | 2.92k | } |
5154 | 157k | return(name); |
5155 | 188k | } |
5156 | | |
5157 | | #ifdef LIBXML_CATALOG_ENABLED |
5158 | | /** |
5159 | | * xmlParseCatalogPI: |
5160 | | * @ctxt: an XML parser context |
5161 | | * @catalog: the PI value string |
5162 | | * |
5163 | | * parse an XML Catalog Processing Instruction. |
5164 | | * |
5165 | | * <?oasis-xml-catalog catalog="http://example.com/catalog.xml"?> |
5166 | | * |
5167 | | * Occurs only if allowed by the user and if happening in the Misc |
5168 | | * part of the document before any doctype information |
5169 | | * This will add the given catalog to the parsing context in order |
5170 | | * to be used if there is a resolution need further down in the document |
5171 | | */ |
5172 | | |
5173 | | static void |
5174 | 152 | xmlParseCatalogPI(xmlParserCtxtPtr ctxt, const xmlChar *catalog) { |
5175 | 152 | xmlChar *URL = NULL; |
5176 | 152 | const xmlChar *tmp, *base; |
5177 | 152 | xmlChar marker; |
5178 | | |
5179 | 152 | tmp = catalog; |
5180 | 152 | while (IS_BLANK_CH(*tmp)) tmp++; |
5181 | 152 | if (xmlStrncmp(tmp, BAD_CAST"catalog", 7)) |
5182 | 31 | goto error; |
5183 | 121 | tmp += 7; |
5184 | 954 | while (IS_BLANK_CH(*tmp)) tmp++; |
5185 | 121 | if (*tmp != '=') { |
5186 | 33 | return; |
5187 | 33 | } |
5188 | 88 | tmp++; |
5189 | 435 | while (IS_BLANK_CH(*tmp)) tmp++; |
5190 | 88 | marker = *tmp; |
5191 | 88 | if ((marker != '\'') && (marker != '"')) |
5192 | 33 | goto error; |
5193 | 55 | tmp++; |
5194 | 55 | base = tmp; |
5195 | 460 | while ((*tmp != 0) && (*tmp != marker)) tmp++; |
5196 | 55 | if (*tmp == 0) |
5197 | 3 | goto error; |
5198 | 52 | URL = xmlStrndup(base, tmp - base); |
5199 | 52 | tmp++; |
5200 | 195 | while (IS_BLANK_CH(*tmp)) tmp++; |
5201 | 52 | if (*tmp != 0) |
5202 | 49 | goto error; |
5203 | | |
5204 | 3 | if (URL != NULL) { |
5205 | 3 | ctxt->catalogs = xmlCatalogAddLocal(ctxt->catalogs, URL); |
5206 | 3 | xmlFree(URL); |
5207 | 3 | } |
5208 | 3 | return; |
5209 | | |
5210 | 116 | error: |
5211 | 116 | xmlWarningMsg(ctxt, XML_WAR_CATALOG_PI, |
5212 | 116 | "Catalog PI syntax error: %s\n", |
5213 | 116 | catalog, NULL); |
5214 | 116 | if (URL != NULL) |
5215 | 49 | xmlFree(URL); |
5216 | 116 | } |
5217 | | #endif |
5218 | | |
5219 | | /** |
5220 | | * xmlParsePI: |
5221 | | * @ctxt: an XML parser context |
5222 | | * |
5223 | | * DEPRECATED: Internal function, don't use. |
5224 | | * |
5225 | | * parse an XML Processing Instruction. |
5226 | | * |
5227 | | * [16] PI ::= '<?' PITarget (S (Char* - (Char* '?>' Char*)))? '?>' |
5228 | | * |
5229 | | * The processing is transferred to SAX once parsed. |
5230 | | */ |
5231 | | |
5232 | | void |
5233 | 188k | xmlParsePI(xmlParserCtxtPtr ctxt) { |
5234 | 188k | xmlChar *buf = NULL; |
5235 | 188k | size_t len = 0; |
5236 | 188k | size_t size = XML_PARSER_BUFFER_SIZE; |
5237 | 188k | size_t maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
5238 | 78.5k | XML_MAX_HUGE_LENGTH : |
5239 | 188k | XML_MAX_TEXT_LENGTH; |
5240 | 188k | int cur, l; |
5241 | 188k | const xmlChar *target; |
5242 | 188k | xmlParserInputState state; |
5243 | 188k | int count = 0; |
5244 | | |
5245 | 188k | if ((RAW == '<') && (NXT(1) == '?')) { |
5246 | 188k | int inputid = ctxt->input->id; |
5247 | 188k | state = ctxt->instate; |
5248 | 188k | ctxt->instate = XML_PARSER_PI; |
5249 | | /* |
5250 | | * this is a Processing Instruction. |
5251 | | */ |
5252 | 188k | SKIP(2); |
5253 | 188k | SHRINK; |
5254 | | |
5255 | | /* |
5256 | | * Parse the target name and check for special support like |
5257 | | * namespace. |
5258 | | */ |
5259 | 188k | target = xmlParsePITarget(ctxt); |
5260 | 188k | if (target != NULL) { |
5261 | 175k | if ((RAW == '?') && (NXT(1) == '>')) { |
5262 | 30.6k | if (inputid != ctxt->input->id) { |
5263 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5264 | 0 | "PI declaration doesn't start and stop in" |
5265 | 0 | " the same entity\n"); |
5266 | 0 | } |
5267 | 30.6k | SKIP(2); |
5268 | | |
5269 | | /* |
5270 | | * SAX: PI detected. |
5271 | | */ |
5272 | 30.6k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5273 | 30.6k | (ctxt->sax->processingInstruction != NULL)) |
5274 | 22.5k | ctxt->sax->processingInstruction(ctxt->userData, |
5275 | 22.5k | target, NULL); |
5276 | 30.6k | if (ctxt->instate != XML_PARSER_EOF) |
5277 | 30.6k | ctxt->instate = state; |
5278 | 30.6k | return; |
5279 | 30.6k | } |
5280 | 145k | buf = (xmlChar *) xmlMallocAtomic(size); |
5281 | 145k | if (buf == NULL) { |
5282 | 0 | xmlErrMemory(ctxt, NULL); |
5283 | 0 | ctxt->instate = state; |
5284 | 0 | return; |
5285 | 0 | } |
5286 | 145k | if (SKIP_BLANKS == 0) { |
5287 | 31.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_SPACE_REQUIRED, |
5288 | 31.5k | "ParsePI: PI %s space expected\n", target); |
5289 | 31.5k | } |
5290 | 145k | cur = CUR_CHAR(l); |
5291 | 39.0M | while (IS_CHAR(cur) && /* checked */ |
5292 | 39.0M | ((cur != '?') || (NXT(1) != '>'))) { |
5293 | 38.8M | if (len + 5 >= size) { |
5294 | 50.5k | xmlChar *tmp; |
5295 | 50.5k | size_t new_size = size * 2; |
5296 | 50.5k | tmp = (xmlChar *) xmlRealloc(buf, new_size); |
5297 | 50.5k | if (tmp == NULL) { |
5298 | 0 | xmlErrMemory(ctxt, NULL); |
5299 | 0 | xmlFree(buf); |
5300 | 0 | ctxt->instate = state; |
5301 | 0 | return; |
5302 | 0 | } |
5303 | 50.5k | buf = tmp; |
5304 | 50.5k | size = new_size; |
5305 | 50.5k | } |
5306 | 38.8M | count++; |
5307 | 38.8M | if (count > 50) { |
5308 | 718k | SHRINK; |
5309 | 718k | GROW; |
5310 | 718k | if (ctxt->instate == XML_PARSER_EOF) { |
5311 | 0 | xmlFree(buf); |
5312 | 0 | return; |
5313 | 0 | } |
5314 | 718k | count = 0; |
5315 | 718k | } |
5316 | 38.8M | COPY_BUF(l,buf,len,cur); |
5317 | 38.8M | NEXTL(l); |
5318 | 38.8M | cur = CUR_CHAR(l); |
5319 | 38.8M | if (cur == 0) { |
5320 | 21.1k | SHRINK; |
5321 | 21.1k | GROW; |
5322 | 21.1k | cur = CUR_CHAR(l); |
5323 | 21.1k | } |
5324 | 38.8M | if (len > maxLength) { |
5325 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5326 | 0 | "PI %s too big found", target); |
5327 | 0 | xmlFree(buf); |
5328 | 0 | ctxt->instate = state; |
5329 | 0 | return; |
5330 | 0 | } |
5331 | 38.8M | } |
5332 | 145k | buf[len] = 0; |
5333 | 145k | if (cur != '?') { |
5334 | 27.8k | xmlFatalErrMsgStr(ctxt, XML_ERR_PI_NOT_FINISHED, |
5335 | 27.8k | "ParsePI: PI %s never end ...\n", target); |
5336 | 117k | } else { |
5337 | 117k | if (inputid != ctxt->input->id) { |
5338 | 553 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5339 | 553 | "PI declaration doesn't start and stop in" |
5340 | 553 | " the same entity\n"); |
5341 | 553 | } |
5342 | 117k | SKIP(2); |
5343 | | |
5344 | 117k | #ifdef LIBXML_CATALOG_ENABLED |
5345 | 117k | if (((state == XML_PARSER_MISC) || |
5346 | 117k | (state == XML_PARSER_START)) && |
5347 | 117k | (xmlStrEqual(target, XML_CATALOG_PI))) { |
5348 | 152 | xmlCatalogAllow allow = xmlCatalogGetDefaults(); |
5349 | 152 | if ((allow == XML_CATA_ALLOW_DOCUMENT) || |
5350 | 152 | (allow == XML_CATA_ALLOW_ALL)) |
5351 | 152 | xmlParseCatalogPI(ctxt, buf); |
5352 | 152 | } |
5353 | 117k | #endif |
5354 | | |
5355 | | |
5356 | | /* |
5357 | | * SAX: PI detected. |
5358 | | */ |
5359 | 117k | if ((ctxt->sax) && (!ctxt->disableSAX) && |
5360 | 117k | (ctxt->sax->processingInstruction != NULL)) |
5361 | 93.1k | ctxt->sax->processingInstruction(ctxt->userData, |
5362 | 93.1k | target, buf); |
5363 | 117k | } |
5364 | 145k | xmlFree(buf); |
5365 | 145k | } else { |
5366 | 12.1k | xmlFatalErr(ctxt, XML_ERR_PI_NOT_STARTED, NULL); |
5367 | 12.1k | } |
5368 | 157k | if (ctxt->instate != XML_PARSER_EOF) |
5369 | 157k | ctxt->instate = state; |
5370 | 157k | } |
5371 | 188k | } |
5372 | | |
5373 | | /** |
5374 | | * xmlParseNotationDecl: |
5375 | | * @ctxt: an XML parser context |
5376 | | * |
5377 | | * DEPRECATED: Internal function, don't use. |
5378 | | * |
5379 | | * Parse a notation declaration. Always consumes '<!'. |
5380 | | * |
5381 | | * [82] NotationDecl ::= '<!NOTATION' S Name S (ExternalID | PublicID) S? '>' |
5382 | | * |
5383 | | * Hence there is actually 3 choices: |
5384 | | * 'PUBLIC' S PubidLiteral |
5385 | | * 'PUBLIC' S PubidLiteral S SystemLiteral |
5386 | | * and 'SYSTEM' S SystemLiteral |
5387 | | * |
5388 | | * See the NOTE on xmlParseExternalID(). |
5389 | | */ |
5390 | | |
5391 | | void |
5392 | 28.8k | xmlParseNotationDecl(xmlParserCtxtPtr ctxt) { |
5393 | 28.8k | const xmlChar *name; |
5394 | 28.8k | xmlChar *Pubid; |
5395 | 28.8k | xmlChar *Systemid; |
5396 | | |
5397 | 28.8k | if ((CUR != '<') || (NXT(1) != '!')) |
5398 | 0 | return; |
5399 | 28.8k | SKIP(2); |
5400 | | |
5401 | 28.8k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5402 | 28.2k | int inputid = ctxt->input->id; |
5403 | 28.2k | SHRINK; |
5404 | 28.2k | SKIP(8); |
5405 | 28.2k | if (SKIP_BLANKS == 0) { |
5406 | 740 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5407 | 740 | "Space required after '<!NOTATION'\n"); |
5408 | 740 | return; |
5409 | 740 | } |
5410 | | |
5411 | 27.5k | name = xmlParseName(ctxt); |
5412 | 27.5k | if (name == NULL) { |
5413 | 1.50k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5414 | 1.50k | return; |
5415 | 1.50k | } |
5416 | 26.0k | if (xmlStrchr(name, ':') != NULL) { |
5417 | 1.06k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5418 | 1.06k | "colons are forbidden from notation names '%s'\n", |
5419 | 1.06k | name, NULL, NULL); |
5420 | 1.06k | } |
5421 | 26.0k | if (SKIP_BLANKS == 0) { |
5422 | 2.24k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5423 | 2.24k | "Space required after the NOTATION name'\n"); |
5424 | 2.24k | return; |
5425 | 2.24k | } |
5426 | | |
5427 | | /* |
5428 | | * Parse the IDs. |
5429 | | */ |
5430 | 23.7k | Systemid = xmlParseExternalID(ctxt, &Pubid, 0); |
5431 | 23.7k | SKIP_BLANKS; |
5432 | | |
5433 | 23.7k | if (RAW == '>') { |
5434 | 17.6k | if (inputid != ctxt->input->id) { |
5435 | 11 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5436 | 11 | "Notation declaration doesn't start and stop" |
5437 | 11 | " in the same entity\n"); |
5438 | 11 | } |
5439 | 17.6k | NEXT; |
5440 | 17.6k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5441 | 17.6k | (ctxt->sax->notationDecl != NULL)) |
5442 | 13.8k | ctxt->sax->notationDecl(ctxt->userData, name, Pubid, Systemid); |
5443 | 17.6k | } else { |
5444 | 6.12k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5445 | 6.12k | } |
5446 | 23.7k | if (Systemid != NULL) xmlFree(Systemid); |
5447 | 23.7k | if (Pubid != NULL) xmlFree(Pubid); |
5448 | 23.7k | } |
5449 | 28.8k | } |
5450 | | |
5451 | | /** |
5452 | | * xmlParseEntityDecl: |
5453 | | * @ctxt: an XML parser context |
5454 | | * |
5455 | | * DEPRECATED: Internal function, don't use. |
5456 | | * |
5457 | | * Parse an entity declaration. Always consumes '<!'. |
5458 | | * |
5459 | | * [70] EntityDecl ::= GEDecl | PEDecl |
5460 | | * |
5461 | | * [71] GEDecl ::= '<!ENTITY' S Name S EntityDef S? '>' |
5462 | | * |
5463 | | * [72] PEDecl ::= '<!ENTITY' S '%' S Name S PEDef S? '>' |
5464 | | * |
5465 | | * [73] EntityDef ::= EntityValue | (ExternalID NDataDecl?) |
5466 | | * |
5467 | | * [74] PEDef ::= EntityValue | ExternalID |
5468 | | * |
5469 | | * [76] NDataDecl ::= S 'NDATA' S Name |
5470 | | * |
5471 | | * [ VC: Notation Declared ] |
5472 | | * The Name must match the declared name of a notation. |
5473 | | */ |
5474 | | |
5475 | | void |
5476 | 1.09M | xmlParseEntityDecl(xmlParserCtxtPtr ctxt) { |
5477 | 1.09M | const xmlChar *name = NULL; |
5478 | 1.09M | xmlChar *value = NULL; |
5479 | 1.09M | xmlChar *URI = NULL, *literal = NULL; |
5480 | 1.09M | const xmlChar *ndata = NULL; |
5481 | 1.09M | int isParameter = 0; |
5482 | 1.09M | xmlChar *orig = NULL; |
5483 | | |
5484 | 1.09M | if ((CUR != '<') || (NXT(1) != '!')) |
5485 | 0 | return; |
5486 | 1.09M | SKIP(2); |
5487 | | |
5488 | | /* GROW; done in the caller */ |
5489 | 1.09M | if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
5490 | 1.09M | int inputid = ctxt->input->id; |
5491 | 1.09M | SHRINK; |
5492 | 1.09M | SKIP(6); |
5493 | 1.09M | if (SKIP_BLANKS == 0) { |
5494 | 5.85k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5495 | 5.85k | "Space required after '<!ENTITY'\n"); |
5496 | 5.85k | } |
5497 | | |
5498 | 1.09M | if (RAW == '%') { |
5499 | 503k | NEXT; |
5500 | 503k | if (SKIP_BLANKS == 0) { |
5501 | 802 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5502 | 802 | "Space required after '%%'\n"); |
5503 | 802 | } |
5504 | 503k | isParameter = 1; |
5505 | 503k | } |
5506 | | |
5507 | 1.09M | name = xmlParseName(ctxt); |
5508 | 1.09M | if (name == NULL) { |
5509 | 6.08k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5510 | 6.08k | "xmlParseEntityDecl: no name\n"); |
5511 | 6.08k | return; |
5512 | 6.08k | } |
5513 | 1.08M | if (xmlStrchr(name, ':') != NULL) { |
5514 | 1.82k | xmlNsErr(ctxt, XML_NS_ERR_COLON, |
5515 | 1.82k | "colons are forbidden from entities names '%s'\n", |
5516 | 1.82k | name, NULL, NULL); |
5517 | 1.82k | } |
5518 | 1.08M | if (SKIP_BLANKS == 0) { |
5519 | 5.19k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5520 | 5.19k | "Space required after the entity name\n"); |
5521 | 5.19k | } |
5522 | | |
5523 | 1.08M | ctxt->instate = XML_PARSER_ENTITY_DECL; |
5524 | | /* |
5525 | | * handle the various case of definitions... |
5526 | | */ |
5527 | 1.08M | if (isParameter) { |
5528 | 502k | if ((RAW == '"') || (RAW == '\'')) { |
5529 | 451k | value = xmlParseEntityValue(ctxt, &orig); |
5530 | 451k | if (value) { |
5531 | 441k | if ((ctxt->sax != NULL) && |
5532 | 441k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5533 | 386k | ctxt->sax->entityDecl(ctxt->userData, name, |
5534 | 386k | XML_INTERNAL_PARAMETER_ENTITY, |
5535 | 386k | NULL, NULL, value); |
5536 | 441k | } |
5537 | 451k | } else { |
5538 | 50.3k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5539 | 50.3k | if ((URI == NULL) && (literal == NULL)) { |
5540 | 1.41k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5541 | 1.41k | } |
5542 | 50.3k | if (URI) { |
5543 | 47.8k | xmlURIPtr uri; |
5544 | | |
5545 | 47.8k | uri = xmlParseURI((const char *) URI); |
5546 | 47.8k | if (uri == NULL) { |
5547 | 1.66k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5548 | 1.66k | "Invalid URI: %s\n", URI); |
5549 | | /* |
5550 | | * This really ought to be a well formedness error |
5551 | | * but the XML Core WG decided otherwise c.f. issue |
5552 | | * E26 of the XML erratas. |
5553 | | */ |
5554 | 46.1k | } else { |
5555 | 46.1k | if (uri->fragment != NULL) { |
5556 | | /* |
5557 | | * Okay this is foolish to block those but not |
5558 | | * invalid URIs. |
5559 | | */ |
5560 | 221 | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5561 | 45.9k | } else { |
5562 | 45.9k | if ((ctxt->sax != NULL) && |
5563 | 45.9k | (!ctxt->disableSAX) && |
5564 | 45.9k | (ctxt->sax->entityDecl != NULL)) |
5565 | 43.9k | ctxt->sax->entityDecl(ctxt->userData, name, |
5566 | 43.9k | XML_EXTERNAL_PARAMETER_ENTITY, |
5567 | 43.9k | literal, URI, NULL); |
5568 | 45.9k | } |
5569 | 46.1k | xmlFreeURI(uri); |
5570 | 46.1k | } |
5571 | 47.8k | } |
5572 | 50.3k | } |
5573 | 582k | } else { |
5574 | 582k | if ((RAW == '"') || (RAW == '\'')) { |
5575 | 398k | value = xmlParseEntityValue(ctxt, &orig); |
5576 | 398k | if ((ctxt->sax != NULL) && |
5577 | 398k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5578 | 361k | ctxt->sax->entityDecl(ctxt->userData, name, |
5579 | 361k | XML_INTERNAL_GENERAL_ENTITY, |
5580 | 361k | NULL, NULL, value); |
5581 | | /* |
5582 | | * For expat compatibility in SAX mode. |
5583 | | */ |
5584 | 398k | if ((ctxt->myDoc == NULL) || |
5585 | 398k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
5586 | 4.13k | if (ctxt->myDoc == NULL) { |
5587 | 287 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5588 | 287 | if (ctxt->myDoc == NULL) { |
5589 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5590 | 0 | return; |
5591 | 0 | } |
5592 | 287 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5593 | 287 | } |
5594 | 4.13k | if (ctxt->myDoc->intSubset == NULL) |
5595 | 287 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5596 | 287 | BAD_CAST "fake", NULL, NULL); |
5597 | | |
5598 | 4.13k | xmlSAX2EntityDecl(ctxt, name, XML_INTERNAL_GENERAL_ENTITY, |
5599 | 4.13k | NULL, NULL, value); |
5600 | 4.13k | } |
5601 | 398k | } else { |
5602 | 184k | URI = xmlParseExternalID(ctxt, &literal, 1); |
5603 | 184k | if ((URI == NULL) && (literal == NULL)) { |
5604 | 6.81k | xmlFatalErr(ctxt, XML_ERR_VALUE_REQUIRED, NULL); |
5605 | 6.81k | } |
5606 | 184k | if (URI) { |
5607 | 171k | xmlURIPtr uri; |
5608 | | |
5609 | 171k | uri = xmlParseURI((const char *)URI); |
5610 | 171k | if (uri == NULL) { |
5611 | 6.50k | xmlErrMsgStr(ctxt, XML_ERR_INVALID_URI, |
5612 | 6.50k | "Invalid URI: %s\n", URI); |
5613 | | /* |
5614 | | * This really ought to be a well formedness error |
5615 | | * but the XML Core WG decided otherwise c.f. issue |
5616 | | * E26 of the XML erratas. |
5617 | | */ |
5618 | 165k | } else { |
5619 | 165k | if (uri->fragment != NULL) { |
5620 | | /* |
5621 | | * Okay this is foolish to block those but not |
5622 | | * invalid URIs. |
5623 | | */ |
5624 | 2.14k | xmlFatalErr(ctxt, XML_ERR_URI_FRAGMENT, NULL); |
5625 | 2.14k | } |
5626 | 165k | xmlFreeURI(uri); |
5627 | 165k | } |
5628 | 171k | } |
5629 | 184k | if ((RAW != '>') && (SKIP_BLANKS == 0)) { |
5630 | 8.45k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5631 | 8.45k | "Space required before 'NDATA'\n"); |
5632 | 8.45k | } |
5633 | 184k | if (CMP5(CUR_PTR, 'N', 'D', 'A', 'T', 'A')) { |
5634 | 13.4k | SKIP(5); |
5635 | 13.4k | if (SKIP_BLANKS == 0) { |
5636 | 1.02k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5637 | 1.02k | "Space required after 'NDATA'\n"); |
5638 | 1.02k | } |
5639 | 13.4k | ndata = xmlParseName(ctxt); |
5640 | 13.4k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
5641 | 13.4k | (ctxt->sax->unparsedEntityDecl != NULL)) |
5642 | 11.9k | ctxt->sax->unparsedEntityDecl(ctxt->userData, name, |
5643 | 11.9k | literal, URI, ndata); |
5644 | 170k | } else { |
5645 | 170k | if ((ctxt->sax != NULL) && |
5646 | 170k | (!ctxt->disableSAX) && (ctxt->sax->entityDecl != NULL)) |
5647 | 160k | ctxt->sax->entityDecl(ctxt->userData, name, |
5648 | 160k | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5649 | 160k | literal, URI, NULL); |
5650 | | /* |
5651 | | * For expat compatibility in SAX mode. |
5652 | | * assuming the entity replacement was asked for |
5653 | | */ |
5654 | 170k | if ((ctxt->replaceEntities != 0) && |
5655 | 170k | ((ctxt->myDoc == NULL) || |
5656 | 95.9k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE)))) { |
5657 | 844 | if (ctxt->myDoc == NULL) { |
5658 | 136 | ctxt->myDoc = xmlNewDoc(SAX_COMPAT_MODE); |
5659 | 136 | if (ctxt->myDoc == NULL) { |
5660 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
5661 | 0 | return; |
5662 | 0 | } |
5663 | 136 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
5664 | 136 | } |
5665 | | |
5666 | 844 | if (ctxt->myDoc->intSubset == NULL) |
5667 | 136 | ctxt->myDoc->intSubset = xmlNewDtd(ctxt->myDoc, |
5668 | 136 | BAD_CAST "fake", NULL, NULL); |
5669 | 844 | xmlSAX2EntityDecl(ctxt, name, |
5670 | 844 | XML_EXTERNAL_GENERAL_PARSED_ENTITY, |
5671 | 844 | literal, URI, NULL); |
5672 | 844 | } |
5673 | 170k | } |
5674 | 184k | } |
5675 | 582k | } |
5676 | 1.08M | if (ctxt->instate == XML_PARSER_EOF) |
5677 | 353 | goto done; |
5678 | 1.08M | SKIP_BLANKS; |
5679 | 1.08M | if (RAW != '>') { |
5680 | 14.7k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_NOT_FINISHED, |
5681 | 14.7k | "xmlParseEntityDecl: entity %s not terminated\n", name); |
5682 | 14.7k | xmlHaltParser(ctxt); |
5683 | 1.06M | } else { |
5684 | 1.06M | if (inputid != ctxt->input->id) { |
5685 | 179 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
5686 | 179 | "Entity declaration doesn't start and stop in" |
5687 | 179 | " the same entity\n"); |
5688 | 179 | } |
5689 | 1.06M | NEXT; |
5690 | 1.06M | } |
5691 | 1.08M | if (orig != NULL) { |
5692 | | /* |
5693 | | * Ugly mechanism to save the raw entity value. |
5694 | | */ |
5695 | 834k | xmlEntityPtr cur = NULL; |
5696 | | |
5697 | 834k | if (isParameter) { |
5698 | 442k | if ((ctxt->sax != NULL) && |
5699 | 442k | (ctxt->sax->getParameterEntity != NULL)) |
5700 | 442k | cur = ctxt->sax->getParameterEntity(ctxt->userData, name); |
5701 | 442k | } else { |
5702 | 391k | if ((ctxt->sax != NULL) && |
5703 | 391k | (ctxt->sax->getEntity != NULL)) |
5704 | 391k | cur = ctxt->sax->getEntity(ctxt->userData, name); |
5705 | 391k | if ((cur == NULL) && (ctxt->userData==ctxt)) { |
5706 | 27.9k | cur = xmlSAX2GetEntity(ctxt, name); |
5707 | 27.9k | } |
5708 | 391k | } |
5709 | 834k | if ((cur != NULL) && (cur->orig == NULL)) { |
5710 | 614k | cur->orig = orig; |
5711 | 614k | orig = NULL; |
5712 | 614k | } |
5713 | 834k | } |
5714 | | |
5715 | 1.08M | done: |
5716 | 1.08M | if (value != NULL) xmlFree(value); |
5717 | 1.08M | if (URI != NULL) xmlFree(URI); |
5718 | 1.08M | if (literal != NULL) xmlFree(literal); |
5719 | 1.08M | if (orig != NULL) xmlFree(orig); |
5720 | 1.08M | } |
5721 | 1.09M | } |
5722 | | |
5723 | | /** |
5724 | | * xmlParseDefaultDecl: |
5725 | | * @ctxt: an XML parser context |
5726 | | * @value: Receive a possible fixed default value for the attribute |
5727 | | * |
5728 | | * DEPRECATED: Internal function, don't use. |
5729 | | * |
5730 | | * Parse an attribute default declaration |
5731 | | * |
5732 | | * [60] DefaultDecl ::= '#REQUIRED' | '#IMPLIED' | (('#FIXED' S)? AttValue) |
5733 | | * |
5734 | | * [ VC: Required Attribute ] |
5735 | | * if the default declaration is the keyword #REQUIRED, then the |
5736 | | * attribute must be specified for all elements of the type in the |
5737 | | * attribute-list declaration. |
5738 | | * |
5739 | | * [ VC: Attribute Default Legal ] |
5740 | | * The declared default value must meet the lexical constraints of |
5741 | | * the declared attribute type c.f. xmlValidateAttributeDecl() |
5742 | | * |
5743 | | * [ VC: Fixed Attribute Default ] |
5744 | | * if an attribute has a default value declared with the #FIXED |
5745 | | * keyword, instances of that attribute must match the default value. |
5746 | | * |
5747 | | * [ WFC: No < in Attribute Values ] |
5748 | | * handled in xmlParseAttValue() |
5749 | | * |
5750 | | * returns: XML_ATTRIBUTE_NONE, XML_ATTRIBUTE_REQUIRED, XML_ATTRIBUTE_IMPLIED |
5751 | | * or XML_ATTRIBUTE_FIXED. |
5752 | | */ |
5753 | | |
5754 | | int |
5755 | 4.12M | xmlParseDefaultDecl(xmlParserCtxtPtr ctxt, xmlChar **value) { |
5756 | 4.12M | int val; |
5757 | 4.12M | xmlChar *ret; |
5758 | | |
5759 | 4.12M | *value = NULL; |
5760 | 4.12M | if (CMP9(CUR_PTR, '#', 'R', 'E', 'Q', 'U', 'I', 'R', 'E', 'D')) { |
5761 | 278k | SKIP(9); |
5762 | 278k | return(XML_ATTRIBUTE_REQUIRED); |
5763 | 278k | } |
5764 | 3.84M | if (CMP8(CUR_PTR, '#', 'I', 'M', 'P', 'L', 'I', 'E', 'D')) { |
5765 | 3.52M | SKIP(8); |
5766 | 3.52M | return(XML_ATTRIBUTE_IMPLIED); |
5767 | 3.52M | } |
5768 | 317k | val = XML_ATTRIBUTE_NONE; |
5769 | 317k | if (CMP6(CUR_PTR, '#', 'F', 'I', 'X', 'E', 'D')) { |
5770 | 124k | SKIP(6); |
5771 | 124k | val = XML_ATTRIBUTE_FIXED; |
5772 | 124k | if (SKIP_BLANKS == 0) { |
5773 | 345 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5774 | 345 | "Space required after '#FIXED'\n"); |
5775 | 345 | } |
5776 | 124k | } |
5777 | 317k | ret = xmlParseAttValue(ctxt); |
5778 | 317k | ctxt->instate = XML_PARSER_DTD; |
5779 | 317k | if (ret == NULL) { |
5780 | 6.16k | xmlFatalErrMsg(ctxt, (xmlParserErrors)ctxt->errNo, |
5781 | 6.16k | "Attribute default value declaration error\n"); |
5782 | 6.16k | } else |
5783 | 311k | *value = ret; |
5784 | 317k | return(val); |
5785 | 3.84M | } |
5786 | | |
5787 | | /** |
5788 | | * xmlParseNotationType: |
5789 | | * @ctxt: an XML parser context |
5790 | | * |
5791 | | * DEPRECATED: Internal function, don't use. |
5792 | | * |
5793 | | * parse an Notation attribute type. |
5794 | | * |
5795 | | * Note: the leading 'NOTATION' S part has already being parsed... |
5796 | | * |
5797 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5798 | | * |
5799 | | * [ VC: Notation Attributes ] |
5800 | | * Values of this type must match one of the notation names included |
5801 | | * in the declaration; all notation names in the declaration must be declared. |
5802 | | * |
5803 | | * Returns: the notation attribute tree built while parsing |
5804 | | */ |
5805 | | |
5806 | | xmlEnumerationPtr |
5807 | 11.1k | xmlParseNotationType(xmlParserCtxtPtr ctxt) { |
5808 | 11.1k | const xmlChar *name; |
5809 | 11.1k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5810 | | |
5811 | 11.1k | if (RAW != '(') { |
5812 | 579 | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_STARTED, NULL); |
5813 | 579 | return(NULL); |
5814 | 579 | } |
5815 | 10.5k | SHRINK; |
5816 | 28.0k | do { |
5817 | 28.0k | NEXT; |
5818 | 28.0k | SKIP_BLANKS; |
5819 | 28.0k | name = xmlParseName(ctxt); |
5820 | 28.0k | if (name == NULL) { |
5821 | 466 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
5822 | 466 | "Name expected in NOTATION declaration\n"); |
5823 | 466 | xmlFreeEnumeration(ret); |
5824 | 466 | return(NULL); |
5825 | 466 | } |
5826 | 27.5k | tmp = ret; |
5827 | 84.2k | while (tmp != NULL) { |
5828 | 60.4k | if (xmlStrEqual(name, tmp->name)) { |
5829 | 3.79k | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5830 | 3.79k | "standalone: attribute notation value token %s duplicated\n", |
5831 | 3.79k | name, NULL); |
5832 | 3.79k | if (!xmlDictOwns(ctxt->dict, name)) |
5833 | 0 | xmlFree((xmlChar *) name); |
5834 | 3.79k | break; |
5835 | 3.79k | } |
5836 | 56.7k | tmp = tmp->next; |
5837 | 56.7k | } |
5838 | 27.5k | if (tmp == NULL) { |
5839 | 23.7k | cur = xmlCreateEnumeration(name); |
5840 | 23.7k | if (cur == NULL) { |
5841 | 0 | xmlFreeEnumeration(ret); |
5842 | 0 | return(NULL); |
5843 | 0 | } |
5844 | 23.7k | if (last == NULL) ret = last = cur; |
5845 | 13.4k | else { |
5846 | 13.4k | last->next = cur; |
5847 | 13.4k | last = cur; |
5848 | 13.4k | } |
5849 | 23.7k | } |
5850 | 27.5k | SKIP_BLANKS; |
5851 | 27.5k | } while (RAW == '|'); |
5852 | 10.1k | if (RAW != ')') { |
5853 | 4.29k | xmlFatalErr(ctxt, XML_ERR_NOTATION_NOT_FINISHED, NULL); |
5854 | 4.29k | xmlFreeEnumeration(ret); |
5855 | 4.29k | return(NULL); |
5856 | 4.29k | } |
5857 | 5.82k | NEXT; |
5858 | 5.82k | return(ret); |
5859 | 10.1k | } |
5860 | | |
5861 | | /** |
5862 | | * xmlParseEnumerationType: |
5863 | | * @ctxt: an XML parser context |
5864 | | * |
5865 | | * DEPRECATED: Internal function, don't use. |
5866 | | * |
5867 | | * parse an Enumeration attribute type. |
5868 | | * |
5869 | | * [59] Enumeration ::= '(' S? Nmtoken (S? '|' S? Nmtoken)* S? ')' |
5870 | | * |
5871 | | * [ VC: Enumeration ] |
5872 | | * Values of this type must match one of the Nmtoken tokens in |
5873 | | * the declaration |
5874 | | * |
5875 | | * Returns: the enumeration attribute tree built while parsing |
5876 | | */ |
5877 | | |
5878 | | xmlEnumerationPtr |
5879 | 483k | xmlParseEnumerationType(xmlParserCtxtPtr ctxt) { |
5880 | 483k | xmlChar *name; |
5881 | 483k | xmlEnumerationPtr ret = NULL, last = NULL, cur, tmp; |
5882 | | |
5883 | 483k | if (RAW != '(') { |
5884 | 8.70k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_STARTED, NULL); |
5885 | 8.70k | return(NULL); |
5886 | 8.70k | } |
5887 | 474k | SHRINK; |
5888 | 1.46M | do { |
5889 | 1.46M | NEXT; |
5890 | 1.46M | SKIP_BLANKS; |
5891 | 1.46M | name = xmlParseNmtoken(ctxt); |
5892 | 1.46M | if (name == NULL) { |
5893 | 1.66k | xmlFatalErr(ctxt, XML_ERR_NMTOKEN_REQUIRED, NULL); |
5894 | 1.66k | return(ret); |
5895 | 1.66k | } |
5896 | 1.46M | tmp = ret; |
5897 | 3.96M | while (tmp != NULL) { |
5898 | 2.50M | if (xmlStrEqual(name, tmp->name)) { |
5899 | 1.86k | xmlValidityError(ctxt, XML_DTD_DUP_TOKEN, |
5900 | 1.86k | "standalone: attribute enumeration value token %s duplicated\n", |
5901 | 1.86k | name, NULL); |
5902 | 1.86k | if (!xmlDictOwns(ctxt->dict, name)) |
5903 | 1.86k | xmlFree(name); |
5904 | 1.86k | break; |
5905 | 1.86k | } |
5906 | 2.50M | tmp = tmp->next; |
5907 | 2.50M | } |
5908 | 1.46M | if (tmp == NULL) { |
5909 | 1.46M | cur = xmlCreateEnumeration(name); |
5910 | 1.46M | if (!xmlDictOwns(ctxt->dict, name)) |
5911 | 1.46M | xmlFree(name); |
5912 | 1.46M | if (cur == NULL) { |
5913 | 0 | xmlFreeEnumeration(ret); |
5914 | 0 | return(NULL); |
5915 | 0 | } |
5916 | 1.46M | if (last == NULL) ret = last = cur; |
5917 | 986k | else { |
5918 | 986k | last->next = cur; |
5919 | 986k | last = cur; |
5920 | 986k | } |
5921 | 1.46M | } |
5922 | 1.46M | SKIP_BLANKS; |
5923 | 1.46M | } while (RAW == '|'); |
5924 | 473k | if (RAW != ')') { |
5925 | 3.45k | xmlFatalErr(ctxt, XML_ERR_ATTLIST_NOT_FINISHED, NULL); |
5926 | 3.45k | return(ret); |
5927 | 3.45k | } |
5928 | 469k | NEXT; |
5929 | 469k | return(ret); |
5930 | 473k | } |
5931 | | |
5932 | | /** |
5933 | | * xmlParseEnumeratedType: |
5934 | | * @ctxt: an XML parser context |
5935 | | * @tree: the enumeration tree built while parsing |
5936 | | * |
5937 | | * DEPRECATED: Internal function, don't use. |
5938 | | * |
5939 | | * parse an Enumerated attribute type. |
5940 | | * |
5941 | | * [57] EnumeratedType ::= NotationType | Enumeration |
5942 | | * |
5943 | | * [58] NotationType ::= 'NOTATION' S '(' S? Name (S? '|' S? Name)* S? ')' |
5944 | | * |
5945 | | * |
5946 | | * Returns: XML_ATTRIBUTE_ENUMERATION or XML_ATTRIBUTE_NOTATION |
5947 | | */ |
5948 | | |
5949 | | int |
5950 | 495k | xmlParseEnumeratedType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
5951 | 495k | if (CMP8(CUR_PTR, 'N', 'O', 'T', 'A', 'T', 'I', 'O', 'N')) { |
5952 | 11.4k | SKIP(8); |
5953 | 11.4k | if (SKIP_BLANKS == 0) { |
5954 | 266 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
5955 | 266 | "Space required after 'NOTATION'\n"); |
5956 | 266 | return(0); |
5957 | 266 | } |
5958 | 11.1k | *tree = xmlParseNotationType(ctxt); |
5959 | 11.1k | if (*tree == NULL) return(0); |
5960 | 5.82k | return(XML_ATTRIBUTE_NOTATION); |
5961 | 11.1k | } |
5962 | 483k | *tree = xmlParseEnumerationType(ctxt); |
5963 | 483k | if (*tree == NULL) return(0); |
5964 | 474k | return(XML_ATTRIBUTE_ENUMERATION); |
5965 | 483k | } |
5966 | | |
5967 | | /** |
5968 | | * xmlParseAttributeType: |
5969 | | * @ctxt: an XML parser context |
5970 | | * @tree: the enumeration tree built while parsing |
5971 | | * |
5972 | | * DEPRECATED: Internal function, don't use. |
5973 | | * |
5974 | | * parse the Attribute list def for an element |
5975 | | * |
5976 | | * [54] AttType ::= StringType | TokenizedType | EnumeratedType |
5977 | | * |
5978 | | * [55] StringType ::= 'CDATA' |
5979 | | * |
5980 | | * [56] TokenizedType ::= 'ID' | 'IDREF' | 'IDREFS' | 'ENTITY' | |
5981 | | * 'ENTITIES' | 'NMTOKEN' | 'NMTOKENS' |
5982 | | * |
5983 | | * Validity constraints for attribute values syntax are checked in |
5984 | | * xmlValidateAttributeValue() |
5985 | | * |
5986 | | * [ VC: ID ] |
5987 | | * Values of type ID must match the Name production. A name must not |
5988 | | * appear more than once in an XML document as a value of this type; |
5989 | | * i.e., ID values must uniquely identify the elements which bear them. |
5990 | | * |
5991 | | * [ VC: One ID per Element Type ] |
5992 | | * No element type may have more than one ID attribute specified. |
5993 | | * |
5994 | | * [ VC: ID Attribute Default ] |
5995 | | * An ID attribute must have a declared default of #IMPLIED or #REQUIRED. |
5996 | | * |
5997 | | * [ VC: IDREF ] |
5998 | | * Values of type IDREF must match the Name production, and values |
5999 | | * of type IDREFS must match Names; each IDREF Name must match the value |
6000 | | * of an ID attribute on some element in the XML document; i.e. IDREF |
6001 | | * values must match the value of some ID attribute. |
6002 | | * |
6003 | | * [ VC: Entity Name ] |
6004 | | * Values of type ENTITY must match the Name production, values |
6005 | | * of type ENTITIES must match Names; each Entity Name must match the |
6006 | | * name of an unparsed entity declared in the DTD. |
6007 | | * |
6008 | | * [ VC: Name Token ] |
6009 | | * Values of type NMTOKEN must match the Nmtoken production; values |
6010 | | * of type NMTOKENS must match Nmtokens. |
6011 | | * |
6012 | | * Returns the attribute type |
6013 | | */ |
6014 | | int |
6015 | 4.14M | xmlParseAttributeType(xmlParserCtxtPtr ctxt, xmlEnumerationPtr *tree) { |
6016 | 4.14M | SHRINK; |
6017 | 4.14M | if (CMP5(CUR_PTR, 'C', 'D', 'A', 'T', 'A')) { |
6018 | 1.76M | SKIP(5); |
6019 | 1.76M | return(XML_ATTRIBUTE_CDATA); |
6020 | 2.38M | } else if (CMP6(CUR_PTR, 'I', 'D', 'R', 'E', 'F', 'S')) { |
6021 | 13.2k | SKIP(6); |
6022 | 13.2k | return(XML_ATTRIBUTE_IDREFS); |
6023 | 2.36M | } else if (CMP5(CUR_PTR, 'I', 'D', 'R', 'E', 'F')) { |
6024 | 37.4k | SKIP(5); |
6025 | 37.4k | return(XML_ATTRIBUTE_IDREF); |
6026 | 2.33M | } else if ((RAW == 'I') && (NXT(1) == 'D')) { |
6027 | 1.00M | SKIP(2); |
6028 | 1.00M | return(XML_ATTRIBUTE_ID); |
6029 | 1.32M | } else if (CMP6(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'Y')) { |
6030 | 17.2k | SKIP(6); |
6031 | 17.2k | return(XML_ATTRIBUTE_ENTITY); |
6032 | 1.31M | } else if (CMP8(CUR_PTR, 'E', 'N', 'T', 'I', 'T', 'I', 'E', 'S')) { |
6033 | 1.78k | SKIP(8); |
6034 | 1.78k | return(XML_ATTRIBUTE_ENTITIES); |
6035 | 1.31M | } else if (CMP8(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N', 'S')) { |
6036 | 236k | SKIP(8); |
6037 | 236k | return(XML_ATTRIBUTE_NMTOKENS); |
6038 | 1.07M | } else if (CMP7(CUR_PTR, 'N', 'M', 'T', 'O', 'K', 'E', 'N')) { |
6039 | 579k | SKIP(7); |
6040 | 579k | return(XML_ATTRIBUTE_NMTOKEN); |
6041 | 579k | } |
6042 | 495k | return(xmlParseEnumeratedType(ctxt, tree)); |
6043 | 4.14M | } |
6044 | | |
6045 | | /** |
6046 | | * xmlParseAttributeListDecl: |
6047 | | * @ctxt: an XML parser context |
6048 | | * |
6049 | | * DEPRECATED: Internal function, don't use. |
6050 | | * |
6051 | | * Parse an attribute list declaration for an element. Always consumes '<!'. |
6052 | | * |
6053 | | * [52] AttlistDecl ::= '<!ATTLIST' S Name AttDef* S? '>' |
6054 | | * |
6055 | | * [53] AttDef ::= S Name S AttType S DefaultDecl |
6056 | | * |
6057 | | */ |
6058 | | void |
6059 | 1.29M | xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt) { |
6060 | 1.29M | const xmlChar *elemName; |
6061 | 1.29M | const xmlChar *attrName; |
6062 | 1.29M | xmlEnumerationPtr tree; |
6063 | | |
6064 | 1.29M | if ((CUR != '<') || (NXT(1) != '!')) |
6065 | 0 | return; |
6066 | 1.29M | SKIP(2); |
6067 | | |
6068 | 1.29M | if (CMP7(CUR_PTR, 'A', 'T', 'T', 'L', 'I', 'S', 'T')) { |
6069 | 1.29M | int inputid = ctxt->input->id; |
6070 | | |
6071 | 1.29M | SKIP(7); |
6072 | 1.29M | if (SKIP_BLANKS == 0) { |
6073 | 2.92k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6074 | 2.92k | "Space required after '<!ATTLIST'\n"); |
6075 | 2.92k | } |
6076 | 1.29M | elemName = xmlParseName(ctxt); |
6077 | 1.29M | if (elemName == NULL) { |
6078 | 2.60k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6079 | 2.60k | "ATTLIST: no name for Element\n"); |
6080 | 2.60k | return; |
6081 | 2.60k | } |
6082 | 1.29M | SKIP_BLANKS; |
6083 | 1.29M | GROW; |
6084 | 5.40M | while ((RAW != '>') && (ctxt->instate != XML_PARSER_EOF)) { |
6085 | 4.16M | int type; |
6086 | 4.16M | int def; |
6087 | 4.16M | xmlChar *defaultValue = NULL; |
6088 | | |
6089 | 4.16M | GROW; |
6090 | 4.16M | tree = NULL; |
6091 | 4.16M | attrName = xmlParseName(ctxt); |
6092 | 4.16M | if (attrName == NULL) { |
6093 | 11.3k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6094 | 11.3k | "ATTLIST: no name for Attribute\n"); |
6095 | 11.3k | break; |
6096 | 11.3k | } |
6097 | 4.15M | GROW; |
6098 | 4.15M | if (SKIP_BLANKS == 0) { |
6099 | 5.66k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6100 | 5.66k | "Space required after the attribute name\n"); |
6101 | 5.66k | break; |
6102 | 5.66k | } |
6103 | | |
6104 | 4.14M | type = xmlParseAttributeType(ctxt, &tree); |
6105 | 4.14M | if (type <= 0) { |
6106 | 15.0k | break; |
6107 | 15.0k | } |
6108 | | |
6109 | 4.12M | GROW; |
6110 | 4.12M | if (SKIP_BLANKS == 0) { |
6111 | 6.15k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6112 | 6.15k | "Space required after the attribute type\n"); |
6113 | 6.15k | if (tree != NULL) |
6114 | 4.83k | xmlFreeEnumeration(tree); |
6115 | 6.15k | break; |
6116 | 6.15k | } |
6117 | | |
6118 | 4.12M | def = xmlParseDefaultDecl(ctxt, &defaultValue); |
6119 | 4.12M | if (def <= 0) { |
6120 | 0 | if (defaultValue != NULL) |
6121 | 0 | xmlFree(defaultValue); |
6122 | 0 | if (tree != NULL) |
6123 | 0 | xmlFreeEnumeration(tree); |
6124 | 0 | break; |
6125 | 0 | } |
6126 | 4.12M | if ((type != XML_ATTRIBUTE_CDATA) && (defaultValue != NULL)) |
6127 | 149k | xmlAttrNormalizeSpace(defaultValue, defaultValue); |
6128 | | |
6129 | 4.12M | GROW; |
6130 | 4.12M | if (RAW != '>') { |
6131 | 3.93M | if (SKIP_BLANKS == 0) { |
6132 | 12.9k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6133 | 12.9k | "Space required after the attribute default value\n"); |
6134 | 12.9k | if (defaultValue != NULL) |
6135 | 6.96k | xmlFree(defaultValue); |
6136 | 12.9k | if (tree != NULL) |
6137 | 2.08k | xmlFreeEnumeration(tree); |
6138 | 12.9k | break; |
6139 | 12.9k | } |
6140 | 3.93M | } |
6141 | 4.11M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6142 | 4.11M | (ctxt->sax->attributeDecl != NULL)) |
6143 | 3.63M | ctxt->sax->attributeDecl(ctxt->userData, elemName, attrName, |
6144 | 3.63M | type, def, defaultValue, tree); |
6145 | 474k | else if (tree != NULL) |
6146 | 57.9k | xmlFreeEnumeration(tree); |
6147 | | |
6148 | 4.11M | if ((ctxt->sax2) && (defaultValue != NULL) && |
6149 | 4.11M | (def != XML_ATTRIBUTE_IMPLIED) && |
6150 | 4.11M | (def != XML_ATTRIBUTE_REQUIRED)) { |
6151 | 187k | xmlAddDefAttrs(ctxt, elemName, attrName, defaultValue); |
6152 | 187k | } |
6153 | 4.11M | if (ctxt->sax2) { |
6154 | 2.53M | xmlAddSpecialAttr(ctxt, elemName, attrName, type); |
6155 | 2.53M | } |
6156 | 4.11M | if (defaultValue != NULL) |
6157 | 304k | xmlFree(defaultValue); |
6158 | 4.11M | GROW; |
6159 | 4.11M | } |
6160 | 1.29M | if (RAW == '>') { |
6161 | 1.24M | if (inputid != ctxt->input->id) { |
6162 | 480 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6163 | 480 | "Attribute list declaration doesn't start and" |
6164 | 480 | " stop in the same entity\n"); |
6165 | 480 | } |
6166 | 1.24M | NEXT; |
6167 | 1.24M | } |
6168 | 1.29M | } |
6169 | 1.29M | } |
6170 | | |
6171 | | /** |
6172 | | * xmlParseElementMixedContentDecl: |
6173 | | * @ctxt: an XML parser context |
6174 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6175 | | * |
6176 | | * DEPRECATED: Internal function, don't use. |
6177 | | * |
6178 | | * parse the declaration for a Mixed Element content |
6179 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6180 | | * |
6181 | | * [51] Mixed ::= '(' S? '#PCDATA' (S? '|' S? Name)* S? ')*' | |
6182 | | * '(' S? '#PCDATA' S? ')' |
6183 | | * |
6184 | | * [ VC: Proper Group/PE Nesting ] applies to [51] too (see [49]) |
6185 | | * |
6186 | | * [ VC: No Duplicate Types ] |
6187 | | * The same name must not appear more than once in a single |
6188 | | * mixed-content declaration. |
6189 | | * |
6190 | | * returns: the list of the xmlElementContentPtr describing the element choices |
6191 | | */ |
6192 | | xmlElementContentPtr |
6193 | 580k | xmlParseElementMixedContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6194 | 580k | xmlElementContentPtr ret = NULL, cur = NULL, n; |
6195 | 580k | const xmlChar *elem = NULL; |
6196 | | |
6197 | 580k | GROW; |
6198 | 580k | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6199 | 580k | SKIP(7); |
6200 | 580k | SKIP_BLANKS; |
6201 | 580k | SHRINK; |
6202 | 580k | if (RAW == ')') { |
6203 | 336k | if (ctxt->input->id != inputchk) { |
6204 | 48 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6205 | 48 | "Element content declaration doesn't start and" |
6206 | 48 | " stop in the same entity\n"); |
6207 | 48 | } |
6208 | 336k | NEXT; |
6209 | 336k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6210 | 336k | if (ret == NULL) |
6211 | 0 | return(NULL); |
6212 | 336k | if (RAW == '*') { |
6213 | 323 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6214 | 323 | NEXT; |
6215 | 323 | } |
6216 | 336k | return(ret); |
6217 | 336k | } |
6218 | 243k | if ((RAW == '(') || (RAW == '|')) { |
6219 | 242k | ret = cur = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_PCDATA); |
6220 | 242k | if (ret == NULL) return(NULL); |
6221 | 242k | } |
6222 | 2.59M | while ((RAW == '|') && (ctxt->instate != XML_PARSER_EOF)) { |
6223 | 2.35M | NEXT; |
6224 | 2.35M | if (elem == NULL) { |
6225 | 242k | ret = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6226 | 242k | if (ret == NULL) { |
6227 | 0 | xmlFreeDocElementContent(ctxt->myDoc, cur); |
6228 | 0 | return(NULL); |
6229 | 0 | } |
6230 | 242k | ret->c1 = cur; |
6231 | 242k | if (cur != NULL) |
6232 | 242k | cur->parent = ret; |
6233 | 242k | cur = ret; |
6234 | 2.11M | } else { |
6235 | 2.11M | n = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6236 | 2.11M | if (n == NULL) { |
6237 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6238 | 0 | return(NULL); |
6239 | 0 | } |
6240 | 2.11M | n->c1 = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6241 | 2.11M | if (n->c1 != NULL) |
6242 | 2.11M | n->c1->parent = n; |
6243 | 2.11M | cur->c2 = n; |
6244 | 2.11M | if (n != NULL) |
6245 | 2.11M | n->parent = cur; |
6246 | 2.11M | cur = n; |
6247 | 2.11M | } |
6248 | 2.35M | SKIP_BLANKS; |
6249 | 2.35M | elem = xmlParseName(ctxt); |
6250 | 2.35M | if (elem == NULL) { |
6251 | 1.49k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6252 | 1.49k | "xmlParseElementMixedContentDecl : Name expected\n"); |
6253 | 1.49k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6254 | 1.49k | return(NULL); |
6255 | 1.49k | } |
6256 | 2.35M | SKIP_BLANKS; |
6257 | 2.35M | GROW; |
6258 | 2.35M | } |
6259 | 242k | if ((RAW == ')') && (NXT(1) == '*')) { |
6260 | 240k | if (elem != NULL) { |
6261 | 240k | cur->c2 = xmlNewDocElementContent(ctxt->myDoc, elem, |
6262 | 240k | XML_ELEMENT_CONTENT_ELEMENT); |
6263 | 240k | if (cur->c2 != NULL) |
6264 | 240k | cur->c2->parent = cur; |
6265 | 240k | } |
6266 | 240k | if (ret != NULL) |
6267 | 240k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6268 | 240k | if (ctxt->input->id != inputchk) { |
6269 | 26 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6270 | 26 | "Element content declaration doesn't start and" |
6271 | 26 | " stop in the same entity\n"); |
6272 | 26 | } |
6273 | 240k | SKIP(2); |
6274 | 240k | } else { |
6275 | 2.42k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6276 | 2.42k | xmlFatalErr(ctxt, XML_ERR_MIXED_NOT_STARTED, NULL); |
6277 | 2.42k | return(NULL); |
6278 | 2.42k | } |
6279 | | |
6280 | 242k | } else { |
6281 | 0 | xmlFatalErr(ctxt, XML_ERR_PCDATA_REQUIRED, NULL); |
6282 | 0 | } |
6283 | 240k | return(ret); |
6284 | 580k | } |
6285 | | |
6286 | | /** |
6287 | | * xmlParseElementChildrenContentDeclPriv: |
6288 | | * @ctxt: an XML parser context |
6289 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6290 | | * @depth: the level of recursion |
6291 | | * |
6292 | | * parse the declaration for a Mixed Element content |
6293 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6294 | | * |
6295 | | * |
6296 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6297 | | * |
6298 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6299 | | * |
6300 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6301 | | * |
6302 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6303 | | * |
6304 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6305 | | * TODO Parameter-entity replacement text must be properly nested |
6306 | | * with parenthesized groups. That is to say, if either of the |
6307 | | * opening or closing parentheses in a choice, seq, or Mixed |
6308 | | * construct is contained in the replacement text for a parameter |
6309 | | * entity, both must be contained in the same replacement text. For |
6310 | | * interoperability, if a parameter-entity reference appears in a |
6311 | | * choice, seq, or Mixed construct, its replacement text should not |
6312 | | * be empty, and neither the first nor last non-blank character of |
6313 | | * the replacement text should be a connector (| or ,). |
6314 | | * |
6315 | | * Returns the tree of xmlElementContentPtr describing the element |
6316 | | * hierarchy. |
6317 | | */ |
6318 | | static xmlElementContentPtr |
6319 | | xmlParseElementChildrenContentDeclPriv(xmlParserCtxtPtr ctxt, int inputchk, |
6320 | 600k | int depth) { |
6321 | 600k | xmlElementContentPtr ret = NULL, cur = NULL, last = NULL, op = NULL; |
6322 | 600k | const xmlChar *elem; |
6323 | 600k | xmlChar type = 0; |
6324 | | |
6325 | 600k | if (((depth > 128) && ((ctxt->options & XML_PARSE_HUGE) == 0)) || |
6326 | 600k | (depth > 2048)) { |
6327 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, |
6328 | 0 | "xmlParseElementChildrenContentDecl : depth %d too deep, use XML_PARSE_HUGE\n", |
6329 | 0 | depth); |
6330 | 0 | return(NULL); |
6331 | 0 | } |
6332 | 600k | SKIP_BLANKS; |
6333 | 600k | GROW; |
6334 | 600k | if (RAW == '(') { |
6335 | 35.8k | int inputid = ctxt->input->id; |
6336 | | |
6337 | | /* Recurse on first child */ |
6338 | 35.8k | NEXT; |
6339 | 35.8k | SKIP_BLANKS; |
6340 | 35.8k | cur = ret = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6341 | 35.8k | depth + 1); |
6342 | 35.8k | if (cur == NULL) |
6343 | 4.36k | return(NULL); |
6344 | 31.5k | SKIP_BLANKS; |
6345 | 31.5k | GROW; |
6346 | 564k | } else { |
6347 | 564k | elem = xmlParseName(ctxt); |
6348 | 564k | if (elem == NULL) { |
6349 | 5.22k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6350 | 5.22k | return(NULL); |
6351 | 5.22k | } |
6352 | 559k | cur = ret = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6353 | 559k | if (cur == NULL) { |
6354 | 0 | xmlErrMemory(ctxt, NULL); |
6355 | 0 | return(NULL); |
6356 | 0 | } |
6357 | 559k | GROW; |
6358 | 559k | if (RAW == '?') { |
6359 | 71.0k | cur->ocur = XML_ELEMENT_CONTENT_OPT; |
6360 | 71.0k | NEXT; |
6361 | 488k | } else if (RAW == '*') { |
6362 | 77.0k | cur->ocur = XML_ELEMENT_CONTENT_MULT; |
6363 | 77.0k | NEXT; |
6364 | 411k | } else if (RAW == '+') { |
6365 | 64.3k | cur->ocur = XML_ELEMENT_CONTENT_PLUS; |
6366 | 64.3k | NEXT; |
6367 | 347k | } else { |
6368 | 347k | cur->ocur = XML_ELEMENT_CONTENT_ONCE; |
6369 | 347k | } |
6370 | 559k | GROW; |
6371 | 559k | } |
6372 | 591k | SKIP_BLANKS; |
6373 | 591k | SHRINK; |
6374 | 2.26M | while ((RAW != ')') && (ctxt->instate != XML_PARSER_EOF)) { |
6375 | | /* |
6376 | | * Each loop we parse one separator and one element. |
6377 | | */ |
6378 | 1.68M | if (RAW == ',') { |
6379 | 616k | if (type == 0) type = CUR; |
6380 | | |
6381 | | /* |
6382 | | * Detect "Name | Name , Name" error |
6383 | | */ |
6384 | 389k | else if (type != CUR) { |
6385 | 76 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6386 | 76 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6387 | 76 | type); |
6388 | 76 | if ((last != NULL) && (last != ret)) |
6389 | 76 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6390 | 76 | if (ret != NULL) |
6391 | 76 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6392 | 76 | return(NULL); |
6393 | 76 | } |
6394 | 616k | NEXT; |
6395 | | |
6396 | 616k | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_SEQ); |
6397 | 616k | if (op == NULL) { |
6398 | 0 | if ((last != NULL) && (last != ret)) |
6399 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6400 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6401 | 0 | return(NULL); |
6402 | 0 | } |
6403 | 616k | if (last == NULL) { |
6404 | 227k | op->c1 = ret; |
6405 | 227k | if (ret != NULL) |
6406 | 227k | ret->parent = op; |
6407 | 227k | ret = cur = op; |
6408 | 389k | } else { |
6409 | 389k | cur->c2 = op; |
6410 | 389k | if (op != NULL) |
6411 | 389k | op->parent = cur; |
6412 | 389k | op->c1 = last; |
6413 | 389k | if (last != NULL) |
6414 | 389k | last->parent = op; |
6415 | 389k | cur =op; |
6416 | 389k | last = NULL; |
6417 | 389k | } |
6418 | 1.07M | } else if (RAW == '|') { |
6419 | 1.06M | if (type == 0) type = CUR; |
6420 | | |
6421 | | /* |
6422 | | * Detect "Name , Name | Name" error |
6423 | | */ |
6424 | 902k | else if (type != CUR) { |
6425 | 74 | xmlFatalErrMsgInt(ctxt, XML_ERR_SEPARATOR_REQUIRED, |
6426 | 74 | "xmlParseElementChildrenContentDecl : '%c' expected\n", |
6427 | 74 | type); |
6428 | 74 | if ((last != NULL) && (last != ret)) |
6429 | 74 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6430 | 74 | if (ret != NULL) |
6431 | 74 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6432 | 74 | return(NULL); |
6433 | 74 | } |
6434 | 1.06M | NEXT; |
6435 | | |
6436 | 1.06M | op = xmlNewDocElementContent(ctxt->myDoc, NULL, XML_ELEMENT_CONTENT_OR); |
6437 | 1.06M | if (op == NULL) { |
6438 | 0 | if ((last != NULL) && (last != ret)) |
6439 | 0 | xmlFreeDocElementContent(ctxt->myDoc, last); |
6440 | 0 | if (ret != NULL) |
6441 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6442 | 0 | return(NULL); |
6443 | 0 | } |
6444 | 1.06M | if (last == NULL) { |
6445 | 159k | op->c1 = ret; |
6446 | 159k | if (ret != NULL) |
6447 | 159k | ret->parent = op; |
6448 | 159k | ret = cur = op; |
6449 | 902k | } else { |
6450 | 902k | cur->c2 = op; |
6451 | 902k | if (op != NULL) |
6452 | 902k | op->parent = cur; |
6453 | 902k | op->c1 = last; |
6454 | 902k | if (last != NULL) |
6455 | 902k | last->parent = op; |
6456 | 902k | cur =op; |
6457 | 902k | last = NULL; |
6458 | 902k | } |
6459 | 1.06M | } else { |
6460 | 9.21k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_FINISHED, NULL); |
6461 | 9.21k | if ((last != NULL) && (last != ret)) |
6462 | 4.48k | xmlFreeDocElementContent(ctxt->myDoc, last); |
6463 | 9.21k | if (ret != NULL) |
6464 | 9.21k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6465 | 9.21k | return(NULL); |
6466 | 9.21k | } |
6467 | 1.67M | GROW; |
6468 | 1.67M | SKIP_BLANKS; |
6469 | 1.67M | GROW; |
6470 | 1.67M | if (RAW == '(') { |
6471 | 75.7k | int inputid = ctxt->input->id; |
6472 | | /* Recurse on second child */ |
6473 | 75.7k | NEXT; |
6474 | 75.7k | SKIP_BLANKS; |
6475 | 75.7k | last = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, |
6476 | 75.7k | depth + 1); |
6477 | 75.7k | if (last == NULL) { |
6478 | 1.38k | if (ret != NULL) |
6479 | 1.38k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6480 | 1.38k | return(NULL); |
6481 | 1.38k | } |
6482 | 74.3k | SKIP_BLANKS; |
6483 | 1.60M | } else { |
6484 | 1.60M | elem = xmlParseName(ctxt); |
6485 | 1.60M | if (elem == NULL) { |
6486 | 2.89k | xmlFatalErr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, NULL); |
6487 | 2.89k | if (ret != NULL) |
6488 | 2.89k | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6489 | 2.89k | return(NULL); |
6490 | 2.89k | } |
6491 | 1.59M | last = xmlNewDocElementContent(ctxt->myDoc, elem, XML_ELEMENT_CONTENT_ELEMENT); |
6492 | 1.59M | if (last == NULL) { |
6493 | 0 | if (ret != NULL) |
6494 | 0 | xmlFreeDocElementContent(ctxt->myDoc, ret); |
6495 | 0 | return(NULL); |
6496 | 0 | } |
6497 | 1.59M | if (RAW == '?') { |
6498 | 265k | last->ocur = XML_ELEMENT_CONTENT_OPT; |
6499 | 265k | NEXT; |
6500 | 1.33M | } else if (RAW == '*') { |
6501 | 179k | last->ocur = XML_ELEMENT_CONTENT_MULT; |
6502 | 179k | NEXT; |
6503 | 1.15M | } else if (RAW == '+') { |
6504 | 30.8k | last->ocur = XML_ELEMENT_CONTENT_PLUS; |
6505 | 30.8k | NEXT; |
6506 | 1.12M | } else { |
6507 | 1.12M | last->ocur = XML_ELEMENT_CONTENT_ONCE; |
6508 | 1.12M | } |
6509 | 1.59M | } |
6510 | 1.67M | SKIP_BLANKS; |
6511 | 1.67M | GROW; |
6512 | 1.67M | } |
6513 | 577k | if ((cur != NULL) && (last != NULL)) { |
6514 | 377k | cur->c2 = last; |
6515 | 377k | if (last != NULL) |
6516 | 377k | last->parent = cur; |
6517 | 377k | } |
6518 | 577k | if (ctxt->input->id != inputchk) { |
6519 | 637 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6520 | 637 | "Element content declaration doesn't start and stop in" |
6521 | 637 | " the same entity\n"); |
6522 | 637 | } |
6523 | 577k | NEXT; |
6524 | 577k | if (RAW == '?') { |
6525 | 24.2k | if (ret != NULL) { |
6526 | 24.2k | if ((ret->ocur == XML_ELEMENT_CONTENT_PLUS) || |
6527 | 24.2k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6528 | 21 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6529 | 24.2k | else |
6530 | 24.2k | ret->ocur = XML_ELEMENT_CONTENT_OPT; |
6531 | 24.2k | } |
6532 | 24.2k | NEXT; |
6533 | 553k | } else if (RAW == '*') { |
6534 | 157k | if (ret != NULL) { |
6535 | 157k | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6536 | 157k | cur = ret; |
6537 | | /* |
6538 | | * Some normalization: |
6539 | | * (a | b* | c?)* == (a | b | c)* |
6540 | | */ |
6541 | 868k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6542 | 711k | if ((cur->c1 != NULL) && |
6543 | 711k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6544 | 711k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) |
6545 | 18.8k | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6546 | 711k | if ((cur->c2 != NULL) && |
6547 | 711k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6548 | 711k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) |
6549 | 3.11k | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6550 | 711k | cur = cur->c2; |
6551 | 711k | } |
6552 | 157k | } |
6553 | 157k | NEXT; |
6554 | 395k | } else if (RAW == '+') { |
6555 | 70.8k | if (ret != NULL) { |
6556 | 70.8k | int found = 0; |
6557 | | |
6558 | 70.8k | if ((ret->ocur == XML_ELEMENT_CONTENT_OPT) || |
6559 | 70.8k | (ret->ocur == XML_ELEMENT_CONTENT_MULT)) |
6560 | 51 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6561 | 70.7k | else |
6562 | 70.7k | ret->ocur = XML_ELEMENT_CONTENT_PLUS; |
6563 | | /* |
6564 | | * Some normalization: |
6565 | | * (a | b*)+ == (a | b)* |
6566 | | * (a | b?)+ == (a | b)* |
6567 | | */ |
6568 | 112k | while ((cur != NULL) && (cur->type == XML_ELEMENT_CONTENT_OR)) { |
6569 | 41.9k | if ((cur->c1 != NULL) && |
6570 | 41.9k | ((cur->c1->ocur == XML_ELEMENT_CONTENT_OPT) || |
6571 | 41.9k | (cur->c1->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6572 | 505 | cur->c1->ocur = XML_ELEMENT_CONTENT_ONCE; |
6573 | 505 | found = 1; |
6574 | 505 | } |
6575 | 41.9k | if ((cur->c2 != NULL) && |
6576 | 41.9k | ((cur->c2->ocur == XML_ELEMENT_CONTENT_OPT) || |
6577 | 41.9k | (cur->c2->ocur == XML_ELEMENT_CONTENT_MULT))) { |
6578 | 413 | cur->c2->ocur = XML_ELEMENT_CONTENT_ONCE; |
6579 | 413 | found = 1; |
6580 | 413 | } |
6581 | 41.9k | cur = cur->c2; |
6582 | 41.9k | } |
6583 | 70.8k | if (found) |
6584 | 548 | ret->ocur = XML_ELEMENT_CONTENT_MULT; |
6585 | 70.8k | } |
6586 | 70.8k | NEXT; |
6587 | 70.8k | } |
6588 | 577k | return(ret); |
6589 | 591k | } |
6590 | | |
6591 | | /** |
6592 | | * xmlParseElementChildrenContentDecl: |
6593 | | * @ctxt: an XML parser context |
6594 | | * @inputchk: the input used for the current entity, needed for boundary checks |
6595 | | * |
6596 | | * DEPRECATED: Internal function, don't use. |
6597 | | * |
6598 | | * parse the declaration for a Mixed Element content |
6599 | | * The leading '(' and spaces have been skipped in xmlParseElementContentDecl |
6600 | | * |
6601 | | * [47] children ::= (choice | seq) ('?' | '*' | '+')? |
6602 | | * |
6603 | | * [48] cp ::= (Name | choice | seq) ('?' | '*' | '+')? |
6604 | | * |
6605 | | * [49] choice ::= '(' S? cp ( S? '|' S? cp )* S? ')' |
6606 | | * |
6607 | | * [50] seq ::= '(' S? cp ( S? ',' S? cp )* S? ')' |
6608 | | * |
6609 | | * [ VC: Proper Group/PE Nesting ] applies to [49] and [50] |
6610 | | * TODO Parameter-entity replacement text must be properly nested |
6611 | | * with parenthesized groups. That is to say, if either of the |
6612 | | * opening or closing parentheses in a choice, seq, or Mixed |
6613 | | * construct is contained in the replacement text for a parameter |
6614 | | * entity, both must be contained in the same replacement text. For |
6615 | | * interoperability, if a parameter-entity reference appears in a |
6616 | | * choice, seq, or Mixed construct, its replacement text should not |
6617 | | * be empty, and neither the first nor last non-blank character of |
6618 | | * the replacement text should be a connector (| or ,). |
6619 | | * |
6620 | | * Returns the tree of xmlElementContentPtr describing the element |
6621 | | * hierarchy. |
6622 | | */ |
6623 | | xmlElementContentPtr |
6624 | 0 | xmlParseElementChildrenContentDecl(xmlParserCtxtPtr ctxt, int inputchk) { |
6625 | | /* stub left for API/ABI compat */ |
6626 | 0 | return(xmlParseElementChildrenContentDeclPriv(ctxt, inputchk, 1)); |
6627 | 0 | } |
6628 | | |
6629 | | /** |
6630 | | * xmlParseElementContentDecl: |
6631 | | * @ctxt: an XML parser context |
6632 | | * @name: the name of the element being defined. |
6633 | | * @result: the Element Content pointer will be stored here if any |
6634 | | * |
6635 | | * DEPRECATED: Internal function, don't use. |
6636 | | * |
6637 | | * parse the declaration for an Element content either Mixed or Children, |
6638 | | * the cases EMPTY and ANY are handled directly in xmlParseElementDecl |
6639 | | * |
6640 | | * [46] contentspec ::= 'EMPTY' | 'ANY' | Mixed | children |
6641 | | * |
6642 | | * returns: the type of element content XML_ELEMENT_TYPE_xxx |
6643 | | */ |
6644 | | |
6645 | | int |
6646 | | xmlParseElementContentDecl(xmlParserCtxtPtr ctxt, const xmlChar *name, |
6647 | 1.06M | xmlElementContentPtr *result) { |
6648 | | |
6649 | 1.06M | xmlElementContentPtr tree = NULL; |
6650 | 1.06M | int inputid = ctxt->input->id; |
6651 | 1.06M | int res; |
6652 | | |
6653 | 1.06M | *result = NULL; |
6654 | | |
6655 | 1.06M | if (RAW != '(') { |
6656 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6657 | 0 | "xmlParseElementContentDecl : %s '(' expected\n", name); |
6658 | 0 | return(-1); |
6659 | 0 | } |
6660 | 1.06M | NEXT; |
6661 | 1.06M | GROW; |
6662 | 1.06M | if (ctxt->instate == XML_PARSER_EOF) |
6663 | 0 | return(-1); |
6664 | 1.06M | SKIP_BLANKS; |
6665 | 1.06M | if (CMP7(CUR_PTR, '#', 'P', 'C', 'D', 'A', 'T', 'A')) { |
6666 | 580k | tree = xmlParseElementMixedContentDecl(ctxt, inputid); |
6667 | 580k | res = XML_ELEMENT_TYPE_MIXED; |
6668 | 580k | } else { |
6669 | 489k | tree = xmlParseElementChildrenContentDeclPriv(ctxt, inputid, 1); |
6670 | 489k | res = XML_ELEMENT_TYPE_ELEMENT; |
6671 | 489k | } |
6672 | 1.06M | SKIP_BLANKS; |
6673 | 1.06M | *result = tree; |
6674 | 1.06M | return(res); |
6675 | 1.06M | } |
6676 | | |
6677 | | /** |
6678 | | * xmlParseElementDecl: |
6679 | | * @ctxt: an XML parser context |
6680 | | * |
6681 | | * DEPRECATED: Internal function, don't use. |
6682 | | * |
6683 | | * Parse an element declaration. Always consumes '<!'. |
6684 | | * |
6685 | | * [45] elementdecl ::= '<!ELEMENT' S Name S contentspec S? '>' |
6686 | | * |
6687 | | * [ VC: Unique Element Type Declaration ] |
6688 | | * No element type may be declared more than once |
6689 | | * |
6690 | | * Returns the type of the element, or -1 in case of error |
6691 | | */ |
6692 | | int |
6693 | 1.48M | xmlParseElementDecl(xmlParserCtxtPtr ctxt) { |
6694 | 1.48M | const xmlChar *name; |
6695 | 1.48M | int ret = -1; |
6696 | 1.48M | xmlElementContentPtr content = NULL; |
6697 | | |
6698 | 1.48M | if ((CUR != '<') || (NXT(1) != '!')) |
6699 | 0 | return(ret); |
6700 | 1.48M | SKIP(2); |
6701 | | |
6702 | | /* GROW; done in the caller */ |
6703 | 1.48M | if (CMP7(CUR_PTR, 'E', 'L', 'E', 'M', 'E', 'N', 'T')) { |
6704 | 1.48M | int inputid = ctxt->input->id; |
6705 | | |
6706 | 1.48M | SKIP(7); |
6707 | 1.48M | if (SKIP_BLANKS == 0) { |
6708 | 1.32k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6709 | 1.32k | "Space required after 'ELEMENT'\n"); |
6710 | 1.32k | return(-1); |
6711 | 1.32k | } |
6712 | 1.48M | name = xmlParseName(ctxt); |
6713 | 1.48M | if (name == NULL) { |
6714 | 2.49k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
6715 | 2.49k | "xmlParseElementDecl: no name for Element\n"); |
6716 | 2.49k | return(-1); |
6717 | 2.49k | } |
6718 | 1.47M | if (SKIP_BLANKS == 0) { |
6719 | 4.72k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
6720 | 4.72k | "Space required after the element name\n"); |
6721 | 4.72k | } |
6722 | 1.47M | if (CMP5(CUR_PTR, 'E', 'M', 'P', 'T', 'Y')) { |
6723 | 394k | SKIP(5); |
6724 | | /* |
6725 | | * Element must always be empty. |
6726 | | */ |
6727 | 394k | ret = XML_ELEMENT_TYPE_EMPTY; |
6728 | 1.08M | } else if ((RAW == 'A') && (NXT(1) == 'N') && |
6729 | 1.08M | (NXT(2) == 'Y')) { |
6730 | 6.12k | SKIP(3); |
6731 | | /* |
6732 | | * Element is a generic container. |
6733 | | */ |
6734 | 6.12k | ret = XML_ELEMENT_TYPE_ANY; |
6735 | 1.07M | } else if (RAW == '(') { |
6736 | 1.06M | ret = xmlParseElementContentDecl(ctxt, name, &content); |
6737 | 1.06M | } else { |
6738 | | /* |
6739 | | * [ WFC: PEs in Internal Subset ] error handling. |
6740 | | */ |
6741 | 7.95k | if ((RAW == '%') && (ctxt->external == 0) && |
6742 | 7.95k | (ctxt->inputNr == 1)) { |
6743 | 624 | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_IN_INT_SUBSET, |
6744 | 624 | "PEReference: forbidden within markup decl in internal subset\n"); |
6745 | 7.33k | } else { |
6746 | 7.33k | xmlFatalErrMsg(ctxt, XML_ERR_ELEMCONTENT_NOT_STARTED, |
6747 | 7.33k | "xmlParseElementDecl: 'EMPTY', 'ANY' or '(' expected\n"); |
6748 | 7.33k | } |
6749 | 7.95k | return(-1); |
6750 | 7.95k | } |
6751 | | |
6752 | 1.47M | SKIP_BLANKS; |
6753 | | |
6754 | 1.47M | if (RAW != '>') { |
6755 | 18.5k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
6756 | 18.5k | if (content != NULL) { |
6757 | 2.39k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6758 | 2.39k | } |
6759 | 1.45M | } else { |
6760 | 1.45M | if (inputid != ctxt->input->id) { |
6761 | 152 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6762 | 152 | "Element declaration doesn't start and stop in" |
6763 | 152 | " the same entity\n"); |
6764 | 152 | } |
6765 | | |
6766 | 1.45M | NEXT; |
6767 | 1.45M | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
6768 | 1.45M | (ctxt->sax->elementDecl != NULL)) { |
6769 | 1.29M | if (content != NULL) |
6770 | 937k | content->parent = NULL; |
6771 | 1.29M | ctxt->sax->elementDecl(ctxt->userData, name, ret, |
6772 | 1.29M | content); |
6773 | 1.29M | if ((content != NULL) && (content->parent == NULL)) { |
6774 | | /* |
6775 | | * this is a trick: if xmlAddElementDecl is called, |
6776 | | * instead of copying the full tree it is plugged directly |
6777 | | * if called from the parser. Avoid duplicating the |
6778 | | * interfaces or change the API/ABI |
6779 | | */ |
6780 | 194k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6781 | 194k | } |
6782 | 1.29M | } else if (content != NULL) { |
6783 | 107k | xmlFreeDocElementContent(ctxt->myDoc, content); |
6784 | 107k | } |
6785 | 1.45M | } |
6786 | 1.47M | } |
6787 | 1.47M | return(ret); |
6788 | 1.48M | } |
6789 | | |
6790 | | /** |
6791 | | * xmlParseConditionalSections |
6792 | | * @ctxt: an XML parser context |
6793 | | * |
6794 | | * Parse a conditional section. Always consumes '<!['. |
6795 | | * |
6796 | | * [61] conditionalSect ::= includeSect | ignoreSect |
6797 | | * [62] includeSect ::= '<![' S? 'INCLUDE' S? '[' extSubsetDecl ']]>' |
6798 | | * [63] ignoreSect ::= '<![' S? 'IGNORE' S? '[' ignoreSectContents* ']]>' |
6799 | | * [64] ignoreSectContents ::= Ignore ('<![' ignoreSectContents ']]>' Ignore)* |
6800 | | * [65] Ignore ::= Char* - (Char* ('<![' | ']]>') Char*) |
6801 | | */ |
6802 | | |
6803 | | static void |
6804 | 7.42k | xmlParseConditionalSections(xmlParserCtxtPtr ctxt) { |
6805 | 7.42k | int *inputIds = NULL; |
6806 | 7.42k | size_t inputIdsSize = 0; |
6807 | 7.42k | size_t depth = 0; |
6808 | | |
6809 | 53.9k | while (ctxt->instate != XML_PARSER_EOF) { |
6810 | 53.7k | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6811 | 24.4k | int id = ctxt->input->id; |
6812 | | |
6813 | 24.4k | SKIP(3); |
6814 | 24.4k | SKIP_BLANKS; |
6815 | | |
6816 | 24.4k | if (CMP7(CUR_PTR, 'I', 'N', 'C', 'L', 'U', 'D', 'E')) { |
6817 | 20.5k | SKIP(7); |
6818 | 20.5k | SKIP_BLANKS; |
6819 | 20.5k | if (RAW != '[') { |
6820 | 48 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6821 | 48 | xmlHaltParser(ctxt); |
6822 | 48 | goto error; |
6823 | 48 | } |
6824 | 20.4k | if (ctxt->input->id != id) { |
6825 | 52 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6826 | 52 | "All markup of the conditional section is" |
6827 | 52 | " not in the same entity\n"); |
6828 | 52 | } |
6829 | 20.4k | NEXT; |
6830 | | |
6831 | 20.4k | if (inputIdsSize <= depth) { |
6832 | 5.85k | int *tmp; |
6833 | | |
6834 | 5.85k | inputIdsSize = (inputIdsSize == 0 ? 4 : inputIdsSize * 2); |
6835 | 5.85k | tmp = (int *) xmlRealloc(inputIds, |
6836 | 5.85k | inputIdsSize * sizeof(int)); |
6837 | 5.85k | if (tmp == NULL) { |
6838 | 0 | xmlErrMemory(ctxt, NULL); |
6839 | 0 | goto error; |
6840 | 0 | } |
6841 | 5.85k | inputIds = tmp; |
6842 | 5.85k | } |
6843 | 20.4k | inputIds[depth] = id; |
6844 | 20.4k | depth++; |
6845 | 20.4k | } else if (CMP6(CUR_PTR, 'I', 'G', 'N', 'O', 'R', 'E')) { |
6846 | 3.14k | size_t ignoreDepth = 0; |
6847 | | |
6848 | 3.14k | SKIP(6); |
6849 | 3.14k | SKIP_BLANKS; |
6850 | 3.14k | if (RAW != '[') { |
6851 | 78 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID, NULL); |
6852 | 78 | xmlHaltParser(ctxt); |
6853 | 78 | goto error; |
6854 | 78 | } |
6855 | 3.06k | if (ctxt->input->id != id) { |
6856 | 129 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6857 | 129 | "All markup of the conditional section is" |
6858 | 129 | " not in the same entity\n"); |
6859 | 129 | } |
6860 | 3.06k | NEXT; |
6861 | | |
6862 | 6.91M | while (RAW != 0) { |
6863 | 6.91M | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
6864 | 6.63k | SKIP(3); |
6865 | 6.63k | ignoreDepth++; |
6866 | | /* Check for integer overflow */ |
6867 | 6.63k | if (ignoreDepth == 0) { |
6868 | 0 | xmlErrMemory(ctxt, NULL); |
6869 | 0 | goto error; |
6870 | 0 | } |
6871 | 6.91M | } else if ((RAW == ']') && (NXT(1) == ']') && |
6872 | 6.91M | (NXT(2) == '>')) { |
6873 | 5.95k | if (ignoreDepth == 0) |
6874 | 1.71k | break; |
6875 | 4.24k | SKIP(3); |
6876 | 4.24k | ignoreDepth--; |
6877 | 6.90M | } else { |
6878 | 6.90M | NEXT; |
6879 | 6.90M | } |
6880 | 6.91M | } |
6881 | | |
6882 | 3.06k | if (RAW == 0) { |
6883 | 1.35k | xmlFatalErr(ctxt, XML_ERR_CONDSEC_NOT_FINISHED, NULL); |
6884 | 1.35k | goto error; |
6885 | 1.35k | } |
6886 | 1.71k | if (ctxt->input->id != id) { |
6887 | 36 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6888 | 36 | "All markup of the conditional section is" |
6889 | 36 | " not in the same entity\n"); |
6890 | 36 | } |
6891 | 1.71k | SKIP(3); |
6892 | 1.71k | } else { |
6893 | 828 | xmlFatalErr(ctxt, XML_ERR_CONDSEC_INVALID_KEYWORD, NULL); |
6894 | 828 | xmlHaltParser(ctxt); |
6895 | 828 | goto error; |
6896 | 828 | } |
6897 | 29.2k | } else if ((depth > 0) && |
6898 | 29.2k | (RAW == ']') && (NXT(1) == ']') && (NXT(2) == '>')) { |
6899 | 10.0k | depth--; |
6900 | 10.0k | if (ctxt->input->id != inputIds[depth]) { |
6901 | 381 | xmlFatalErrMsg(ctxt, XML_ERR_ENTITY_BOUNDARY, |
6902 | 381 | "All markup of the conditional section is not" |
6903 | 381 | " in the same entity\n"); |
6904 | 381 | } |
6905 | 10.0k | SKIP(3); |
6906 | 19.1k | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
6907 | 17.2k | xmlParseMarkupDecl(ctxt); |
6908 | 17.2k | } else { |
6909 | 1.94k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
6910 | 1.94k | xmlHaltParser(ctxt); |
6911 | 1.94k | goto error; |
6912 | 1.94k | } |
6913 | | |
6914 | 49.4k | if (depth == 0) |
6915 | 2.92k | break; |
6916 | | |
6917 | 46.5k | SKIP_BLANKS; |
6918 | 46.5k | GROW; |
6919 | 46.5k | } |
6920 | | |
6921 | 7.42k | error: |
6922 | 7.42k | xmlFree(inputIds); |
6923 | 7.42k | } |
6924 | | |
6925 | | /** |
6926 | | * xmlParseMarkupDecl: |
6927 | | * @ctxt: an XML parser context |
6928 | | * |
6929 | | * DEPRECATED: Internal function, don't use. |
6930 | | * |
6931 | | * Parse markup declarations. Always consumes '<!' or '<?'. |
6932 | | * |
6933 | | * [29] markupdecl ::= elementdecl | AttlistDecl | EntityDecl | |
6934 | | * NotationDecl | PI | Comment |
6935 | | * |
6936 | | * [ VC: Proper Declaration/PE Nesting ] |
6937 | | * Parameter-entity replacement text must be properly nested with |
6938 | | * markup declarations. That is to say, if either the first character |
6939 | | * or the last character of a markup declaration (markupdecl above) is |
6940 | | * contained in the replacement text for a parameter-entity reference, |
6941 | | * both must be contained in the same replacement text. |
6942 | | * |
6943 | | * [ WFC: PEs in Internal Subset ] |
6944 | | * In the internal DTD subset, parameter-entity references can occur |
6945 | | * only where markup declarations can occur, not within markup declarations. |
6946 | | * (This does not apply to references that occur in external parameter |
6947 | | * entities or to the external subset.) |
6948 | | */ |
6949 | | void |
6950 | 32.0M | xmlParseMarkupDecl(xmlParserCtxtPtr ctxt) { |
6951 | 32.0M | GROW; |
6952 | 32.0M | if (CUR == '<') { |
6953 | 32.0M | if (NXT(1) == '!') { |
6954 | 31.9M | switch (NXT(2)) { |
6955 | 2.57M | case 'E': |
6956 | 2.57M | if (NXT(3) == 'L') |
6957 | 1.48M | xmlParseElementDecl(ctxt); |
6958 | 1.09M | else if (NXT(3) == 'N') |
6959 | 1.09M | xmlParseEntityDecl(ctxt); |
6960 | 949 | else |
6961 | 949 | SKIP(2); |
6962 | 2.57M | break; |
6963 | 1.29M | case 'A': |
6964 | 1.29M | xmlParseAttributeListDecl(ctxt); |
6965 | 1.29M | break; |
6966 | 28.8k | case 'N': |
6967 | 28.8k | xmlParseNotationDecl(ctxt); |
6968 | 28.8k | break; |
6969 | 28.0M | case '-': |
6970 | 28.0M | xmlParseComment(ctxt); |
6971 | 28.0M | break; |
6972 | 17.1k | default: |
6973 | | /* there is an error but it will be detected later */ |
6974 | 17.1k | SKIP(2); |
6975 | 17.1k | break; |
6976 | 31.9M | } |
6977 | 31.9M | } else if (NXT(1) == '?') { |
6978 | 56.1k | xmlParsePI(ctxt); |
6979 | 56.1k | } |
6980 | 32.0M | } |
6981 | | |
6982 | | /* |
6983 | | * detect requirement to exit there and act accordingly |
6984 | | * and avoid having instate overridden later on |
6985 | | */ |
6986 | 32.0M | if (ctxt->instate == XML_PARSER_EOF) |
6987 | 15.4k | return; |
6988 | | |
6989 | 31.9M | ctxt->instate = XML_PARSER_DTD; |
6990 | 31.9M | } |
6991 | | |
6992 | | /** |
6993 | | * xmlParseTextDecl: |
6994 | | * @ctxt: an XML parser context |
6995 | | * |
6996 | | * DEPRECATED: Internal function, don't use. |
6997 | | * |
6998 | | * parse an XML declaration header for external entities |
6999 | | * |
7000 | | * [77] TextDecl ::= '<?xml' VersionInfo? EncodingDecl S? '?>' |
7001 | | */ |
7002 | | |
7003 | | void |
7004 | 50.1k | xmlParseTextDecl(xmlParserCtxtPtr ctxt) { |
7005 | 50.1k | xmlChar *version; |
7006 | 50.1k | const xmlChar *encoding; |
7007 | 50.1k | int oldstate; |
7008 | | |
7009 | | /* |
7010 | | * We know that '<?xml' is here. |
7011 | | */ |
7012 | 50.1k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
7013 | 49.9k | SKIP(5); |
7014 | 49.9k | } else { |
7015 | 196 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_STARTED, NULL); |
7016 | 196 | return; |
7017 | 196 | } |
7018 | | |
7019 | | /* Avoid expansion of parameter entities when skipping blanks. */ |
7020 | 49.9k | oldstate = ctxt->instate; |
7021 | 49.9k | ctxt->instate = XML_PARSER_START; |
7022 | | |
7023 | 49.9k | if (SKIP_BLANKS == 0) { |
7024 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7025 | 0 | "Space needed after '<?xml'\n"); |
7026 | 0 | } |
7027 | | |
7028 | | /* |
7029 | | * We may have the VersionInfo here. |
7030 | | */ |
7031 | 49.9k | version = xmlParseVersionInfo(ctxt); |
7032 | 49.9k | if (version == NULL) |
7033 | 18.3k | version = xmlCharStrdup(XML_DEFAULT_VERSION); |
7034 | 31.6k | else { |
7035 | 31.6k | if (SKIP_BLANKS == 0) { |
7036 | 3.12k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
7037 | 3.12k | "Space needed here\n"); |
7038 | 3.12k | } |
7039 | 31.6k | } |
7040 | 49.9k | ctxt->input->version = version; |
7041 | | |
7042 | | /* |
7043 | | * We must have the encoding declaration |
7044 | | */ |
7045 | 49.9k | encoding = xmlParseEncodingDecl(ctxt); |
7046 | 49.9k | if (ctxt->instate == XML_PARSER_EOF) |
7047 | 0 | return; |
7048 | 49.9k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7049 | | /* |
7050 | | * The XML REC instructs us to stop parsing right here |
7051 | | */ |
7052 | 1.61k | ctxt->instate = oldstate; |
7053 | 1.61k | return; |
7054 | 1.61k | } |
7055 | 48.3k | if ((encoding == NULL) && (ctxt->errNo == XML_ERR_OK)) { |
7056 | 4.35k | xmlFatalErrMsg(ctxt, XML_ERR_MISSING_ENCODING, |
7057 | 4.35k | "Missing encoding in text declaration\n"); |
7058 | 4.35k | } |
7059 | | |
7060 | 48.3k | SKIP_BLANKS; |
7061 | 48.3k | if ((RAW == '?') && (NXT(1) == '>')) { |
7062 | 10.5k | SKIP(2); |
7063 | 37.7k | } else if (RAW == '>') { |
7064 | | /* Deprecated old WD ... */ |
7065 | 860 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7066 | 860 | NEXT; |
7067 | 36.9k | } else { |
7068 | 36.9k | int c; |
7069 | | |
7070 | 36.9k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
7071 | 3.92M | while ((c = CUR) != 0) { |
7072 | 3.90M | NEXT; |
7073 | 3.90M | if (c == '>') |
7074 | 16.9k | break; |
7075 | 3.90M | } |
7076 | 36.9k | } |
7077 | | |
7078 | 48.3k | ctxt->instate = oldstate; |
7079 | 48.3k | } |
7080 | | |
7081 | | /** |
7082 | | * xmlParseExternalSubset: |
7083 | | * @ctxt: an XML parser context |
7084 | | * @ExternalID: the external identifier |
7085 | | * @SystemID: the system identifier (or URL) |
7086 | | * |
7087 | | * parse Markup declarations from an external subset |
7088 | | * |
7089 | | * [30] extSubset ::= textDecl? extSubsetDecl |
7090 | | * |
7091 | | * [31] extSubsetDecl ::= (markupdecl | conditionalSect | PEReference | S) * |
7092 | | */ |
7093 | | void |
7094 | | xmlParseExternalSubset(xmlParserCtxtPtr ctxt, const xmlChar *ExternalID, |
7095 | 33.4k | const xmlChar *SystemID) { |
7096 | 33.4k | xmlDetectSAX2(ctxt); |
7097 | 33.4k | GROW; |
7098 | | |
7099 | 33.4k | if ((ctxt->encoding == NULL) && |
7100 | 33.4k | (ctxt->input->end - ctxt->input->cur >= 4)) { |
7101 | 33.3k | xmlChar start[4]; |
7102 | 33.3k | xmlCharEncoding enc; |
7103 | | |
7104 | 33.3k | start[0] = RAW; |
7105 | 33.3k | start[1] = NXT(1); |
7106 | 33.3k | start[2] = NXT(2); |
7107 | 33.3k | start[3] = NXT(3); |
7108 | 33.3k | enc = xmlDetectCharEncoding(start, 4); |
7109 | 33.3k | if (enc != XML_CHAR_ENCODING_NONE) |
7110 | 6.00k | xmlSwitchEncoding(ctxt, enc); |
7111 | 33.3k | } |
7112 | | |
7113 | 33.4k | if (CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) { |
7114 | 5.65k | xmlParseTextDecl(ctxt); |
7115 | 5.65k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
7116 | | /* |
7117 | | * The XML REC instructs us to stop parsing right here |
7118 | | */ |
7119 | 72 | xmlHaltParser(ctxt); |
7120 | 72 | return; |
7121 | 72 | } |
7122 | 5.65k | } |
7123 | 33.3k | if (ctxt->myDoc == NULL) { |
7124 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
7125 | 0 | if (ctxt->myDoc == NULL) { |
7126 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
7127 | 0 | return; |
7128 | 0 | } |
7129 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
7130 | 0 | } |
7131 | 33.3k | if ((ctxt->myDoc != NULL) && (ctxt->myDoc->intSubset == NULL)) |
7132 | 0 | xmlCreateIntSubset(ctxt->myDoc, NULL, ExternalID, SystemID); |
7133 | | |
7134 | 33.3k | ctxt->instate = XML_PARSER_DTD; |
7135 | 33.3k | ctxt->external = 1; |
7136 | 33.3k | SKIP_BLANKS; |
7137 | 1.16M | while ((ctxt->instate != XML_PARSER_EOF) && (RAW != 0)) { |
7138 | 1.13M | GROW; |
7139 | 1.13M | if ((RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
7140 | 7.42k | xmlParseConditionalSections(ctxt); |
7141 | 1.12M | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
7142 | 1.12M | xmlParseMarkupDecl(ctxt); |
7143 | 1.12M | } else { |
7144 | 7.04k | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7145 | 7.04k | xmlHaltParser(ctxt); |
7146 | 7.04k | return; |
7147 | 7.04k | } |
7148 | 1.12M | SKIP_BLANKS; |
7149 | 1.12M | } |
7150 | | |
7151 | 26.3k | if (RAW != 0) { |
7152 | 0 | xmlFatalErr(ctxt, XML_ERR_EXT_SUBSET_NOT_FINISHED, NULL); |
7153 | 0 | } |
7154 | | |
7155 | 26.3k | } |
7156 | | |
7157 | | /** |
7158 | | * xmlParseReference: |
7159 | | * @ctxt: an XML parser context |
7160 | | * |
7161 | | * DEPRECATED: Internal function, don't use. |
7162 | | * |
7163 | | * parse and handle entity references in content, depending on the SAX |
7164 | | * interface, this may end-up in a call to character() if this is a |
7165 | | * CharRef, a predefined entity, if there is no reference() callback. |
7166 | | * or if the parser was asked to switch to that mode. |
7167 | | * |
7168 | | * Always consumes '&'. |
7169 | | * |
7170 | | * [67] Reference ::= EntityRef | CharRef |
7171 | | */ |
7172 | | void |
7173 | 4.69M | xmlParseReference(xmlParserCtxtPtr ctxt) { |
7174 | 4.69M | xmlEntityPtr ent; |
7175 | 4.69M | xmlChar *val; |
7176 | 4.69M | int was_checked; |
7177 | 4.69M | xmlNodePtr list = NULL; |
7178 | 4.69M | xmlParserErrors ret = XML_ERR_OK; |
7179 | | |
7180 | | |
7181 | 4.69M | if (RAW != '&') |
7182 | 0 | return; |
7183 | | |
7184 | | /* |
7185 | | * Simple case of a CharRef |
7186 | | */ |
7187 | 4.69M | if (NXT(1) == '#') { |
7188 | 248k | int i = 0; |
7189 | 248k | xmlChar out[16]; |
7190 | 248k | int hex = NXT(2); |
7191 | 248k | int value = xmlParseCharRef(ctxt); |
7192 | | |
7193 | 248k | if (value == 0) |
7194 | 33.5k | return; |
7195 | 214k | if (ctxt->charset != XML_CHAR_ENCODING_UTF8) { |
7196 | | /* |
7197 | | * So we are using non-UTF-8 buffers |
7198 | | * Check that the char fit on 8bits, if not |
7199 | | * generate a CharRef. |
7200 | | */ |
7201 | 134k | if (value <= 0xFF) { |
7202 | 129k | out[0] = value; |
7203 | 129k | out[1] = 0; |
7204 | 129k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7205 | 129k | (!ctxt->disableSAX)) |
7206 | 90.8k | ctxt->sax->characters(ctxt->userData, out, 1); |
7207 | 129k | } else { |
7208 | 5.11k | if ((hex == 'x') || (hex == 'X')) |
7209 | 665 | snprintf((char *)out, sizeof(out), "#x%X", value); |
7210 | 4.45k | else |
7211 | 4.45k | snprintf((char *)out, sizeof(out), "#%d", value); |
7212 | 5.11k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7213 | 5.11k | (!ctxt->disableSAX)) |
7214 | 2.64k | ctxt->sax->reference(ctxt->userData, out); |
7215 | 5.11k | } |
7216 | 134k | } else { |
7217 | | /* |
7218 | | * Just encode the value in UTF-8 |
7219 | | */ |
7220 | 80.0k | COPY_BUF(0 ,out, i, value); |
7221 | 80.0k | out[i] = 0; |
7222 | 80.0k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7223 | 80.0k | (!ctxt->disableSAX)) |
7224 | 65.0k | ctxt->sax->characters(ctxt->userData, out, i); |
7225 | 80.0k | } |
7226 | 214k | return; |
7227 | 248k | } |
7228 | | |
7229 | | /* |
7230 | | * We are seeing an entity reference |
7231 | | */ |
7232 | 4.45M | ent = xmlParseEntityRef(ctxt); |
7233 | 4.45M | if (ent == NULL) return; |
7234 | 3.42M | if (!ctxt->wellFormed) |
7235 | 1.09M | return; |
7236 | 2.32M | was_checked = ent->flags & XML_ENT_PARSED; |
7237 | | |
7238 | | /* special case of predefined entities */ |
7239 | 2.32M | if ((ent->name == NULL) || |
7240 | 2.32M | (ent->etype == XML_INTERNAL_PREDEFINED_ENTITY)) { |
7241 | 140k | val = ent->content; |
7242 | 140k | if (val == NULL) return; |
7243 | | /* |
7244 | | * inline the entity. |
7245 | | */ |
7246 | 140k | if ((ctxt->sax != NULL) && (ctxt->sax->characters != NULL) && |
7247 | 140k | (!ctxt->disableSAX)) |
7248 | 140k | ctxt->sax->characters(ctxt->userData, val, xmlStrlen(val)); |
7249 | 140k | return; |
7250 | 140k | } |
7251 | | |
7252 | | /* |
7253 | | * The first reference to the entity trigger a parsing phase |
7254 | | * where the ent->children is filled with the result from |
7255 | | * the parsing. |
7256 | | * Note: external parsed entities will not be loaded, it is not |
7257 | | * required for a non-validating parser, unless the parsing option |
7258 | | * of validating, or substituting entities were given. Doing so is |
7259 | | * far more secure as the parser will only process data coming from |
7260 | | * the document entity by default. |
7261 | | */ |
7262 | 2.18M | if (((ent->flags & XML_ENT_PARSED) == 0) && |
7263 | 2.18M | ((ent->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY) || |
7264 | 133k | (ctxt->options & (XML_PARSE_NOENT | XML_PARSE_DTDVALID)))) { |
7265 | 125k | unsigned long oldsizeentcopy = ctxt->sizeentcopy; |
7266 | | |
7267 | | /* |
7268 | | * This is a bit hackish but this seems the best |
7269 | | * way to make sure both SAX and DOM entity support |
7270 | | * behaves okay. |
7271 | | */ |
7272 | 125k | void *user_data; |
7273 | 125k | if (ctxt->userData == ctxt) |
7274 | 125k | user_data = NULL; |
7275 | 0 | else |
7276 | 0 | user_data = ctxt->userData; |
7277 | | |
7278 | | /* Avoid overflow as much as possible */ |
7279 | 125k | ctxt->sizeentcopy = 0; |
7280 | | |
7281 | 125k | if (ent->flags & XML_ENT_EXPANDING) { |
7282 | 299 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7283 | 299 | xmlHaltParser(ctxt); |
7284 | 299 | return; |
7285 | 299 | } |
7286 | | |
7287 | 125k | ent->flags |= XML_ENT_EXPANDING; |
7288 | | |
7289 | | /* |
7290 | | * Check that this entity is well formed |
7291 | | * 4.3.2: An internal general parsed entity is well-formed |
7292 | | * if its replacement text matches the production labeled |
7293 | | * content. |
7294 | | */ |
7295 | 125k | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7296 | 60.9k | ctxt->depth++; |
7297 | 60.9k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, ent->content, |
7298 | 60.9k | user_data, &list); |
7299 | 60.9k | ctxt->depth--; |
7300 | | |
7301 | 64.5k | } else if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7302 | 64.5k | ctxt->depth++; |
7303 | 64.5k | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, ctxt->sax, |
7304 | 64.5k | user_data, ctxt->depth, ent->URI, |
7305 | 64.5k | ent->ExternalID, &list); |
7306 | 64.5k | ctxt->depth--; |
7307 | 64.5k | } else { |
7308 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7309 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7310 | 0 | "invalid entity type found\n", NULL); |
7311 | 0 | } |
7312 | | |
7313 | 125k | ent->flags &= ~XML_ENT_EXPANDING; |
7314 | 125k | ent->flags |= XML_ENT_PARSED | XML_ENT_CHECKED; |
7315 | 125k | ent->expandedSize = ctxt->sizeentcopy; |
7316 | 125k | if (ret == XML_ERR_ENTITY_LOOP) { |
7317 | 5.99k | xmlHaltParser(ctxt); |
7318 | 5.99k | xmlFreeNodeList(list); |
7319 | 5.99k | return; |
7320 | 5.99k | } |
7321 | 119k | if (xmlParserEntityCheck(ctxt, oldsizeentcopy)) { |
7322 | 0 | xmlFreeNodeList(list); |
7323 | 0 | return; |
7324 | 0 | } |
7325 | | |
7326 | 119k | if ((ret == XML_ERR_OK) && (list != NULL)) { |
7327 | 53.3k | ent->children = list; |
7328 | | /* |
7329 | | * Prune it directly in the generated document |
7330 | | * except for single text nodes. |
7331 | | */ |
7332 | 53.3k | if ((ctxt->replaceEntities == 0) || |
7333 | 53.3k | (ctxt->parseMode == XML_PARSE_READER) || |
7334 | 53.3k | ((list->type == XML_TEXT_NODE) && |
7335 | 44.3k | (list->next == NULL))) { |
7336 | 44.3k | ent->owner = 1; |
7337 | 443k | while (list != NULL) { |
7338 | 398k | list->parent = (xmlNodePtr) ent; |
7339 | 398k | if (list->doc != ent->doc) |
7340 | 0 | xmlSetTreeDoc(list, ent->doc); |
7341 | 398k | if (list->next == NULL) |
7342 | 44.3k | ent->last = list; |
7343 | 398k | list = list->next; |
7344 | 398k | } |
7345 | 44.3k | list = NULL; |
7346 | 44.3k | } else { |
7347 | 9.07k | ent->owner = 0; |
7348 | 596k | while (list != NULL) { |
7349 | 587k | list->parent = (xmlNodePtr) ctxt->node; |
7350 | 587k | list->doc = ctxt->myDoc; |
7351 | 587k | if (list->next == NULL) |
7352 | 9.07k | ent->last = list; |
7353 | 587k | list = list->next; |
7354 | 587k | } |
7355 | 9.07k | list = ent->children; |
7356 | | #ifdef LIBXML_LEGACY_ENABLED |
7357 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7358 | | xmlAddEntityReference(ent, list, NULL); |
7359 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7360 | 9.07k | } |
7361 | 66.0k | } else if ((ret != XML_ERR_OK) && |
7362 | 66.0k | (ret != XML_WAR_UNDECLARED_ENTITY)) { |
7363 | 33.0k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7364 | 33.0k | "Entity '%s' failed to parse\n", ent->name); |
7365 | 33.0k | if (ent->content != NULL) |
7366 | 5.95k | ent->content[0] = 0; |
7367 | 33.0k | } else if (list != NULL) { |
7368 | 0 | xmlFreeNodeList(list); |
7369 | 0 | list = NULL; |
7370 | 0 | } |
7371 | | |
7372 | | /* Prevent entity from being parsed and expanded twice (Bug 760367). */ |
7373 | 119k | was_checked = 0; |
7374 | 119k | } |
7375 | | |
7376 | | /* |
7377 | | * Now that the entity content has been gathered |
7378 | | * provide it to the application, this can take different forms based |
7379 | | * on the parsing modes. |
7380 | | */ |
7381 | 2.18M | if (ent->children == NULL) { |
7382 | | /* |
7383 | | * Probably running in SAX mode and the callbacks don't |
7384 | | * build the entity content. So unless we already went |
7385 | | * though parsing for first checking go though the entity |
7386 | | * content to generate callbacks associated to the entity |
7387 | | */ |
7388 | 727k | if (was_checked != 0) { |
7389 | 653k | void *user_data; |
7390 | | /* |
7391 | | * This is a bit hackish but this seems the best |
7392 | | * way to make sure both SAX and DOM entity support |
7393 | | * behaves okay. |
7394 | | */ |
7395 | 653k | if (ctxt->userData == ctxt) |
7396 | 653k | user_data = NULL; |
7397 | 0 | else |
7398 | 0 | user_data = ctxt->userData; |
7399 | | |
7400 | 653k | if (ent->etype == XML_INTERNAL_GENERAL_ENTITY) { |
7401 | 32.1k | ctxt->depth++; |
7402 | 32.1k | ret = xmlParseBalancedChunkMemoryInternal(ctxt, |
7403 | 32.1k | ent->content, user_data, NULL); |
7404 | 32.1k | ctxt->depth--; |
7405 | 621k | } else if (ent->etype == |
7406 | 621k | XML_EXTERNAL_GENERAL_PARSED_ENTITY) { |
7407 | 621k | unsigned long oldsizeentities = ctxt->sizeentities; |
7408 | | |
7409 | 621k | ctxt->depth++; |
7410 | 621k | ret = xmlParseExternalEntityPrivate(ctxt->myDoc, ctxt, |
7411 | 621k | ctxt->sax, user_data, ctxt->depth, |
7412 | 621k | ent->URI, ent->ExternalID, NULL); |
7413 | 621k | ctxt->depth--; |
7414 | | |
7415 | | /* Undo the change to sizeentities */ |
7416 | 621k | ctxt->sizeentities = oldsizeentities; |
7417 | 621k | } else { |
7418 | 0 | ret = XML_ERR_ENTITY_PE_INTERNAL; |
7419 | 0 | xmlErrMsgStr(ctxt, XML_ERR_INTERNAL_ERROR, |
7420 | 0 | "invalid entity type found\n", NULL); |
7421 | 0 | } |
7422 | 653k | if (ret == XML_ERR_ENTITY_LOOP) { |
7423 | 0 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
7424 | 0 | return; |
7425 | 0 | } |
7426 | 653k | if (xmlParserEntityCheck(ctxt, 0)) |
7427 | 0 | return; |
7428 | 653k | } |
7429 | 727k | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7430 | 727k | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7431 | | /* |
7432 | | * Entity reference callback comes second, it's somewhat |
7433 | | * superfluous but a compatibility to historical behaviour |
7434 | | */ |
7435 | 170k | ctxt->sax->reference(ctxt->userData, ent->name); |
7436 | 170k | } |
7437 | 727k | return; |
7438 | 727k | } |
7439 | | |
7440 | | /* |
7441 | | * We also check for amplification if entities aren't substituted. |
7442 | | * They might be expanded later. |
7443 | | */ |
7444 | 1.45M | if ((was_checked != 0) && |
7445 | 1.45M | (xmlParserEntityCheck(ctxt, ent->expandedSize))) |
7446 | 282 | return; |
7447 | | |
7448 | | /* |
7449 | | * If we didn't get any children for the entity being built |
7450 | | */ |
7451 | 1.45M | if ((ctxt->sax != NULL) && (ctxt->sax->reference != NULL) && |
7452 | 1.45M | (ctxt->replaceEntities == 0) && (!ctxt->disableSAX)) { |
7453 | | /* |
7454 | | * Create a node. |
7455 | | */ |
7456 | 299k | ctxt->sax->reference(ctxt->userData, ent->name); |
7457 | 299k | return; |
7458 | 299k | } |
7459 | | |
7460 | 1.15M | if (ctxt->replaceEntities) { |
7461 | | /* |
7462 | | * There is a problem on the handling of _private for entities |
7463 | | * (bug 155816): Should we copy the content of the field from |
7464 | | * the entity (possibly overwriting some value set by the user |
7465 | | * when a copy is created), should we leave it alone, or should |
7466 | | * we try to take care of different situations? The problem |
7467 | | * is exacerbated by the usage of this field by the xmlReader. |
7468 | | * To fix this bug, we look at _private on the created node |
7469 | | * and, if it's NULL, we copy in whatever was in the entity. |
7470 | | * If it's not NULL we leave it alone. This is somewhat of a |
7471 | | * hack - maybe we should have further tests to determine |
7472 | | * what to do. |
7473 | | */ |
7474 | 1.15M | if (ctxt->node != NULL) { |
7475 | | /* |
7476 | | * Seems we are generating the DOM content, do |
7477 | | * a simple tree copy for all references except the first |
7478 | | * In the first occurrence list contains the replacement. |
7479 | | */ |
7480 | 1.15M | if (((list == NULL) && (ent->owner == 0)) || |
7481 | 1.15M | (ctxt->parseMode == XML_PARSE_READER)) { |
7482 | 363k | xmlNodePtr nw = NULL, cur, firstChild = NULL; |
7483 | | |
7484 | | /* |
7485 | | * when operating on a reader, the entities definitions |
7486 | | * are always owning the entities subtree. |
7487 | | if (ctxt->parseMode == XML_PARSE_READER) |
7488 | | ent->owner = 1; |
7489 | | */ |
7490 | | |
7491 | 363k | cur = ent->children; |
7492 | 784k | while (cur != NULL) { |
7493 | 784k | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7494 | 784k | if (nw != NULL) { |
7495 | 784k | if (nw->_private == NULL) |
7496 | 784k | nw->_private = cur->_private; |
7497 | 784k | if (firstChild == NULL){ |
7498 | 363k | firstChild = nw; |
7499 | 363k | } |
7500 | 784k | nw = xmlAddChild(ctxt->node, nw); |
7501 | 784k | } |
7502 | 784k | if (cur == ent->last) { |
7503 | | /* |
7504 | | * needed to detect some strange empty |
7505 | | * node cases in the reader tests |
7506 | | */ |
7507 | 363k | if ((ctxt->parseMode == XML_PARSE_READER) && |
7508 | 363k | (nw != NULL) && |
7509 | 363k | (nw->type == XML_ELEMENT_NODE) && |
7510 | 363k | (nw->children == NULL)) |
7511 | 7.58k | nw->extra = 1; |
7512 | | |
7513 | 363k | break; |
7514 | 363k | } |
7515 | 421k | cur = cur->next; |
7516 | 421k | } |
7517 | | #ifdef LIBXML_LEGACY_ENABLED |
7518 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7519 | | xmlAddEntityReference(ent, firstChild, nw); |
7520 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7521 | 792k | } else if ((list == NULL) || (ctxt->inputNr > 0)) { |
7522 | 792k | xmlNodePtr nw = NULL, cur, next, last, |
7523 | 792k | firstChild = NULL; |
7524 | | |
7525 | | /* |
7526 | | * Copy the entity child list and make it the new |
7527 | | * entity child list. The goal is to make sure any |
7528 | | * ID or REF referenced will be the one from the |
7529 | | * document content and not the entity copy. |
7530 | | */ |
7531 | 792k | cur = ent->children; |
7532 | 792k | ent->children = NULL; |
7533 | 792k | last = ent->last; |
7534 | 792k | ent->last = NULL; |
7535 | 2.95M | while (cur != NULL) { |
7536 | 2.95M | next = cur->next; |
7537 | 2.95M | cur->next = NULL; |
7538 | 2.95M | cur->parent = NULL; |
7539 | 2.95M | nw = xmlDocCopyNode(cur, ctxt->myDoc, 1); |
7540 | 2.95M | if (nw != NULL) { |
7541 | 2.95M | if (nw->_private == NULL) |
7542 | 2.95M | nw->_private = cur->_private; |
7543 | 2.95M | if (firstChild == NULL){ |
7544 | 792k | firstChild = cur; |
7545 | 792k | } |
7546 | 2.95M | xmlAddChild((xmlNodePtr) ent, nw); |
7547 | 2.95M | } |
7548 | 2.95M | xmlAddChild(ctxt->node, cur); |
7549 | 2.95M | if (cur == last) |
7550 | 792k | break; |
7551 | 2.16M | cur = next; |
7552 | 2.16M | } |
7553 | 792k | if (ent->owner == 0) |
7554 | 9.07k | ent->owner = 1; |
7555 | | #ifdef LIBXML_LEGACY_ENABLED |
7556 | | if (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY) |
7557 | | xmlAddEntityReference(ent, firstChild, nw); |
7558 | | #endif /* LIBXML_LEGACY_ENABLED */ |
7559 | 792k | } else { |
7560 | 0 | const xmlChar *nbktext; |
7561 | | |
7562 | | /* |
7563 | | * the name change is to avoid coalescing of the |
7564 | | * node with a possible previous text one which |
7565 | | * would make ent->children a dangling pointer |
7566 | | */ |
7567 | 0 | nbktext = xmlDictLookup(ctxt->dict, BAD_CAST "nbktext", |
7568 | 0 | -1); |
7569 | 0 | if (ent->children->type == XML_TEXT_NODE) |
7570 | 0 | ent->children->name = nbktext; |
7571 | 0 | if ((ent->last != ent->children) && |
7572 | 0 | (ent->last->type == XML_TEXT_NODE)) |
7573 | 0 | ent->last->name = nbktext; |
7574 | 0 | xmlAddChildList(ctxt->node, ent->children); |
7575 | 0 | } |
7576 | | |
7577 | | /* |
7578 | | * This is to avoid a nasty side effect, see |
7579 | | * characters() in SAX.c |
7580 | | */ |
7581 | 1.15M | ctxt->nodemem = 0; |
7582 | 1.15M | ctxt->nodelen = 0; |
7583 | 1.15M | return; |
7584 | 1.15M | } |
7585 | 1.15M | } |
7586 | 1.15M | } |
7587 | | |
7588 | | /** |
7589 | | * xmlParseEntityRef: |
7590 | | * @ctxt: an XML parser context |
7591 | | * |
7592 | | * DEPRECATED: Internal function, don't use. |
7593 | | * |
7594 | | * Parse an entitiy reference. Always consumes '&'. |
7595 | | * |
7596 | | * [68] EntityRef ::= '&' Name ';' |
7597 | | * |
7598 | | * [ WFC: Entity Declared ] |
7599 | | * In a document without any DTD, a document with only an internal DTD |
7600 | | * subset which contains no parameter entity references, or a document |
7601 | | * with "standalone='yes'", the Name given in the entity reference |
7602 | | * must match that in an entity declaration, except that well-formed |
7603 | | * documents need not declare any of the following entities: amp, lt, |
7604 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7605 | | * any reference to it. Similarly, the declaration of a general entity |
7606 | | * must precede any reference to it which appears in a default value in an |
7607 | | * attribute-list declaration. Note that if entities are declared in the |
7608 | | * external subset or in external parameter entities, a non-validating |
7609 | | * processor is not obligated to read and process their declarations; |
7610 | | * for such documents, the rule that an entity must be declared is a |
7611 | | * well-formedness constraint only if standalone='yes'. |
7612 | | * |
7613 | | * [ WFC: Parsed Entity ] |
7614 | | * An entity reference must not contain the name of an unparsed entity |
7615 | | * |
7616 | | * Returns the xmlEntityPtr if found, or NULL otherwise. |
7617 | | */ |
7618 | | xmlEntityPtr |
7619 | 6.82M | xmlParseEntityRef(xmlParserCtxtPtr ctxt) { |
7620 | 6.82M | const xmlChar *name; |
7621 | 6.82M | xmlEntityPtr ent = NULL; |
7622 | | |
7623 | 6.82M | GROW; |
7624 | 6.82M | if (ctxt->instate == XML_PARSER_EOF) |
7625 | 0 | return(NULL); |
7626 | | |
7627 | 6.82M | if (RAW != '&') |
7628 | 0 | return(NULL); |
7629 | 6.82M | NEXT; |
7630 | 6.82M | name = xmlParseName(ctxt); |
7631 | 6.82M | if (name == NULL) { |
7632 | 98.4k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7633 | 98.4k | "xmlParseEntityRef: no name\n"); |
7634 | 98.4k | return(NULL); |
7635 | 98.4k | } |
7636 | 6.72M | if (RAW != ';') { |
7637 | 76.7k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7638 | 76.7k | return(NULL); |
7639 | 76.7k | } |
7640 | 6.65M | NEXT; |
7641 | | |
7642 | | /* |
7643 | | * Predefined entities override any extra definition |
7644 | | */ |
7645 | 6.65M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7646 | 4.73M | ent = xmlGetPredefinedEntity(name); |
7647 | 4.73M | if (ent != NULL) |
7648 | 244k | return(ent); |
7649 | 4.73M | } |
7650 | | |
7651 | | /* |
7652 | | * Ask first SAX for entity resolution, otherwise try the |
7653 | | * entities which may have stored in the parser context. |
7654 | | */ |
7655 | 6.40M | if (ctxt->sax != NULL) { |
7656 | 6.40M | if (ctxt->sax->getEntity != NULL) |
7657 | 6.40M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7658 | 6.40M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7659 | 6.40M | (ctxt->options & XML_PARSE_OLDSAX)) |
7660 | 39.3k | ent = xmlGetPredefinedEntity(name); |
7661 | 6.40M | if ((ctxt->wellFormed == 1 ) && (ent == NULL) && |
7662 | 6.40M | (ctxt->userData==ctxt)) { |
7663 | 107k | ent = xmlSAX2GetEntity(ctxt, name); |
7664 | 107k | } |
7665 | 6.40M | } |
7666 | 6.40M | if (ctxt->instate == XML_PARSER_EOF) |
7667 | 0 | return(NULL); |
7668 | | /* |
7669 | | * [ WFC: Entity Declared ] |
7670 | | * In a document without any DTD, a document with only an |
7671 | | * internal DTD subset which contains no parameter entity |
7672 | | * references, or a document with "standalone='yes'", the |
7673 | | * Name given in the entity reference must match that in an |
7674 | | * entity declaration, except that well-formed documents |
7675 | | * need not declare any of the following entities: amp, lt, |
7676 | | * gt, apos, quot. |
7677 | | * The declaration of a parameter entity must precede any |
7678 | | * reference to it. |
7679 | | * Similarly, the declaration of a general entity must |
7680 | | * precede any reference to it which appears in a default |
7681 | | * value in an attribute-list declaration. Note that if |
7682 | | * entities are declared in the external subset or in |
7683 | | * external parameter entities, a non-validating processor |
7684 | | * is not obligated to read and process their declarations; |
7685 | | * for such documents, the rule that an entity must be |
7686 | | * declared is a well-formedness constraint only if |
7687 | | * standalone='yes'. |
7688 | | */ |
7689 | 6.40M | if (ent == NULL) { |
7690 | 1.10M | if ((ctxt->standalone == 1) || |
7691 | 1.10M | ((ctxt->hasExternalSubset == 0) && |
7692 | 1.04M | (ctxt->hasPErefs == 0))) { |
7693 | 719k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7694 | 719k | "Entity '%s' not defined\n", name); |
7695 | 719k | } else { |
7696 | 380k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7697 | 380k | "Entity '%s' not defined\n", name); |
7698 | 380k | if ((ctxt->inSubset == 0) && |
7699 | 380k | (ctxt->sax != NULL) && |
7700 | 380k | (ctxt->sax->reference != NULL)) { |
7701 | 367k | ctxt->sax->reference(ctxt->userData, name); |
7702 | 367k | } |
7703 | 380k | } |
7704 | 1.10M | ctxt->valid = 0; |
7705 | 1.10M | } |
7706 | | |
7707 | | /* |
7708 | | * [ WFC: Parsed Entity ] |
7709 | | * An entity reference must not contain the name of an |
7710 | | * unparsed entity |
7711 | | */ |
7712 | 5.30M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7713 | 1.97k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7714 | 1.97k | "Entity reference to unparsed entity %s\n", name); |
7715 | 1.97k | } |
7716 | | |
7717 | | /* |
7718 | | * [ WFC: No External Entity References ] |
7719 | | * Attribute values cannot contain direct or indirect |
7720 | | * entity references to external entities. |
7721 | | */ |
7722 | 5.30M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7723 | 5.30M | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7724 | 16.3k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7725 | 16.3k | "Attribute references external entity '%s'\n", name); |
7726 | 16.3k | } |
7727 | | /* |
7728 | | * [ WFC: No < in Attribute Values ] |
7729 | | * The replacement text of any entity referred to directly or |
7730 | | * indirectly in an attribute value (other than "<") must |
7731 | | * not contain a <. |
7732 | | */ |
7733 | 5.28M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7734 | 5.28M | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
7735 | 2.05M | if ((ent->flags & XML_ENT_CHECKED_LT) == 0) { |
7736 | 28.3k | if ((ent->content != NULL) && (xmlStrchr(ent->content, '<'))) |
7737 | 618 | ent->flags |= XML_ENT_CONTAINS_LT; |
7738 | 28.3k | ent->flags |= XML_ENT_CHECKED_LT; |
7739 | 28.3k | } |
7740 | 2.05M | if (ent->flags & XML_ENT_CONTAINS_LT) |
7741 | 22.4k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7742 | 22.4k | "'<' in entity '%s' is not allowed in attributes " |
7743 | 22.4k | "values\n", name); |
7744 | 2.05M | } |
7745 | | |
7746 | | /* |
7747 | | * Internal check, no parameter entities here ... |
7748 | | */ |
7749 | 3.23M | else { |
7750 | 3.23M | switch (ent->etype) { |
7751 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7752 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7753 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7754 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7755 | 0 | name); |
7756 | 0 | break; |
7757 | 3.23M | default: |
7758 | 3.23M | break; |
7759 | 3.23M | } |
7760 | 3.23M | } |
7761 | | |
7762 | | /* |
7763 | | * [ WFC: No Recursion ] |
7764 | | * A parsed entity must not contain a recursive reference |
7765 | | * to itself, either directly or indirectly. |
7766 | | * Done somewhere else |
7767 | | */ |
7768 | 6.40M | return(ent); |
7769 | 6.40M | } |
7770 | | |
7771 | | /** |
7772 | | * xmlParseStringEntityRef: |
7773 | | * @ctxt: an XML parser context |
7774 | | * @str: a pointer to an index in the string |
7775 | | * |
7776 | | * parse ENTITY references declarations, but this version parses it from |
7777 | | * a string value. |
7778 | | * |
7779 | | * [68] EntityRef ::= '&' Name ';' |
7780 | | * |
7781 | | * [ WFC: Entity Declared ] |
7782 | | * In a document without any DTD, a document with only an internal DTD |
7783 | | * subset which contains no parameter entity references, or a document |
7784 | | * with "standalone='yes'", the Name given in the entity reference |
7785 | | * must match that in an entity declaration, except that well-formed |
7786 | | * documents need not declare any of the following entities: amp, lt, |
7787 | | * gt, apos, quot. The declaration of a parameter entity must precede |
7788 | | * any reference to it. Similarly, the declaration of a general entity |
7789 | | * must precede any reference to it which appears in a default value in an |
7790 | | * attribute-list declaration. Note that if entities are declared in the |
7791 | | * external subset or in external parameter entities, a non-validating |
7792 | | * processor is not obligated to read and process their declarations; |
7793 | | * for such documents, the rule that an entity must be declared is a |
7794 | | * well-formedness constraint only if standalone='yes'. |
7795 | | * |
7796 | | * [ WFC: Parsed Entity ] |
7797 | | * An entity reference must not contain the name of an unparsed entity |
7798 | | * |
7799 | | * Returns the xmlEntityPtr if found, or NULL otherwise. The str pointer |
7800 | | * is updated to the current location in the string. |
7801 | | */ |
7802 | | static xmlEntityPtr |
7803 | 29.4M | xmlParseStringEntityRef(xmlParserCtxtPtr ctxt, const xmlChar ** str) { |
7804 | 29.4M | xmlChar *name; |
7805 | 29.4M | const xmlChar *ptr; |
7806 | 29.4M | xmlChar cur; |
7807 | 29.4M | xmlEntityPtr ent = NULL; |
7808 | | |
7809 | 29.4M | if ((str == NULL) || (*str == NULL)) |
7810 | 0 | return(NULL); |
7811 | 29.4M | ptr = *str; |
7812 | 29.4M | cur = *ptr; |
7813 | 29.4M | if (cur != '&') |
7814 | 0 | return(NULL); |
7815 | | |
7816 | 29.4M | ptr++; |
7817 | 29.4M | name = xmlParseStringName(ctxt, &ptr); |
7818 | 29.4M | if (name == NULL) { |
7819 | 10.2k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
7820 | 10.2k | "xmlParseStringEntityRef: no name\n"); |
7821 | 10.2k | *str = ptr; |
7822 | 10.2k | return(NULL); |
7823 | 10.2k | } |
7824 | 29.4M | if (*ptr != ';') { |
7825 | 10.0k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
7826 | 10.0k | xmlFree(name); |
7827 | 10.0k | *str = ptr; |
7828 | 10.0k | return(NULL); |
7829 | 10.0k | } |
7830 | 29.4M | ptr++; |
7831 | | |
7832 | | |
7833 | | /* |
7834 | | * Predefined entities override any extra definition |
7835 | | */ |
7836 | 29.4M | if ((ctxt->options & XML_PARSE_OLDSAX) == 0) { |
7837 | 26.5M | ent = xmlGetPredefinedEntity(name); |
7838 | 26.5M | if (ent != NULL) { |
7839 | 104k | xmlFree(name); |
7840 | 104k | *str = ptr; |
7841 | 104k | return(ent); |
7842 | 104k | } |
7843 | 26.5M | } |
7844 | | |
7845 | | /* |
7846 | | * Ask first SAX for entity resolution, otherwise try the |
7847 | | * entities which may have stored in the parser context. |
7848 | | */ |
7849 | 29.3M | if (ctxt->sax != NULL) { |
7850 | 29.3M | if (ctxt->sax->getEntity != NULL) |
7851 | 29.3M | ent = ctxt->sax->getEntity(ctxt->userData, name); |
7852 | 29.3M | if ((ent == NULL) && (ctxt->options & XML_PARSE_OLDSAX)) |
7853 | 1.06M | ent = xmlGetPredefinedEntity(name); |
7854 | 29.3M | if ((ent == NULL) && (ctxt->userData==ctxt)) { |
7855 | 1.64M | ent = xmlSAX2GetEntity(ctxt, name); |
7856 | 1.64M | } |
7857 | 29.3M | } |
7858 | 29.3M | if (ctxt->instate == XML_PARSER_EOF) { |
7859 | 0 | xmlFree(name); |
7860 | 0 | return(NULL); |
7861 | 0 | } |
7862 | | |
7863 | | /* |
7864 | | * [ WFC: Entity Declared ] |
7865 | | * In a document without any DTD, a document with only an |
7866 | | * internal DTD subset which contains no parameter entity |
7867 | | * references, or a document with "standalone='yes'", the |
7868 | | * Name given in the entity reference must match that in an |
7869 | | * entity declaration, except that well-formed documents |
7870 | | * need not declare any of the following entities: amp, lt, |
7871 | | * gt, apos, quot. |
7872 | | * The declaration of a parameter entity must precede any |
7873 | | * reference to it. |
7874 | | * Similarly, the declaration of a general entity must |
7875 | | * precede any reference to it which appears in a default |
7876 | | * value in an attribute-list declaration. Note that if |
7877 | | * entities are declared in the external subset or in |
7878 | | * external parameter entities, a non-validating processor |
7879 | | * is not obligated to read and process their declarations; |
7880 | | * for such documents, the rule that an entity must be |
7881 | | * declared is a well-formedness constraint only if |
7882 | | * standalone='yes'. |
7883 | | */ |
7884 | 29.3M | if (ent == NULL) { |
7885 | 1.64M | if ((ctxt->standalone == 1) || |
7886 | 1.64M | ((ctxt->hasExternalSubset == 0) && |
7887 | 1.64M | (ctxt->hasPErefs == 0))) { |
7888 | 1.58M | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
7889 | 1.58M | "Entity '%s' not defined\n", name); |
7890 | 1.58M | } else { |
7891 | 54.4k | xmlErrMsgStr(ctxt, XML_WAR_UNDECLARED_ENTITY, |
7892 | 54.4k | "Entity '%s' not defined\n", |
7893 | 54.4k | name); |
7894 | 54.4k | } |
7895 | | /* TODO ? check regressions ctxt->valid = 0; */ |
7896 | 1.64M | } |
7897 | | |
7898 | | /* |
7899 | | * [ WFC: Parsed Entity ] |
7900 | | * An entity reference must not contain the name of an |
7901 | | * unparsed entity |
7902 | | */ |
7903 | 27.6M | else if (ent->etype == XML_EXTERNAL_GENERAL_UNPARSED_ENTITY) { |
7904 | 3.62k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNPARSED_ENTITY, |
7905 | 3.62k | "Entity reference to unparsed entity %s\n", name); |
7906 | 3.62k | } |
7907 | | |
7908 | | /* |
7909 | | * [ WFC: No External Entity References ] |
7910 | | * Attribute values cannot contain direct or indirect |
7911 | | * entity references to external entities. |
7912 | | */ |
7913 | 27.6M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7914 | 27.6M | (ent->etype == XML_EXTERNAL_GENERAL_PARSED_ENTITY)) { |
7915 | 8.12k | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_EXTERNAL, |
7916 | 8.12k | "Attribute references external entity '%s'\n", name); |
7917 | 8.12k | } |
7918 | | /* |
7919 | | * [ WFC: No < in Attribute Values ] |
7920 | | * The replacement text of any entity referred to directly or |
7921 | | * indirectly in an attribute value (other than "<") must |
7922 | | * not contain a <. |
7923 | | */ |
7924 | 27.6M | else if ((ctxt->instate == XML_PARSER_ATTRIBUTE_VALUE) && |
7925 | 27.6M | (ent->etype != XML_INTERNAL_PREDEFINED_ENTITY)) { |
7926 | 27.3M | if ((ent->flags & XML_ENT_CHECKED_LT) == 0) { |
7927 | 11.6k | if ((ent->content != NULL) && (xmlStrchr(ent->content, '<'))) |
7928 | 1.03k | ent->flags |= XML_ENT_CONTAINS_LT; |
7929 | 11.6k | ent->flags |= XML_ENT_CHECKED_LT; |
7930 | 11.6k | } |
7931 | 27.3M | if (ent->flags & XML_ENT_CONTAINS_LT) |
7932 | 94.2k | xmlFatalErrMsgStr(ctxt, XML_ERR_LT_IN_ATTRIBUTE, |
7933 | 94.2k | "'<' in entity '%s' is not allowed in attributes " |
7934 | 94.2k | "values\n", name); |
7935 | 27.3M | } |
7936 | | |
7937 | | /* |
7938 | | * Internal check, no parameter entities here ... |
7939 | | */ |
7940 | 289k | else { |
7941 | 289k | switch (ent->etype) { |
7942 | 0 | case XML_INTERNAL_PARAMETER_ENTITY: |
7943 | 0 | case XML_EXTERNAL_PARAMETER_ENTITY: |
7944 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_ENTITY_IS_PARAMETER, |
7945 | 0 | "Attempt to reference the parameter entity '%s'\n", |
7946 | 0 | name); |
7947 | 0 | break; |
7948 | 289k | default: |
7949 | 289k | break; |
7950 | 289k | } |
7951 | 289k | } |
7952 | | |
7953 | | /* |
7954 | | * [ WFC: No Recursion ] |
7955 | | * A parsed entity must not contain a recursive reference |
7956 | | * to itself, either directly or indirectly. |
7957 | | * Done somewhere else |
7958 | | */ |
7959 | | |
7960 | 29.3M | xmlFree(name); |
7961 | 29.3M | *str = ptr; |
7962 | 29.3M | return(ent); |
7963 | 29.3M | } |
7964 | | |
7965 | | /** |
7966 | | * xmlParsePEReference: |
7967 | | * @ctxt: an XML parser context |
7968 | | * |
7969 | | * DEPRECATED: Internal function, don't use. |
7970 | | * |
7971 | | * Parse a parameter entity reference. Always consumes '%'. |
7972 | | * |
7973 | | * The entity content is handled directly by pushing it's content as |
7974 | | * a new input stream. |
7975 | | * |
7976 | | * [69] PEReference ::= '%' Name ';' |
7977 | | * |
7978 | | * [ WFC: No Recursion ] |
7979 | | * A parsed entity must not contain a recursive |
7980 | | * reference to itself, either directly or indirectly. |
7981 | | * |
7982 | | * [ WFC: Entity Declared ] |
7983 | | * In a document without any DTD, a document with only an internal DTD |
7984 | | * subset which contains no parameter entity references, or a document |
7985 | | * with "standalone='yes'", ... ... The declaration of a parameter |
7986 | | * entity must precede any reference to it... |
7987 | | * |
7988 | | * [ VC: Entity Declared ] |
7989 | | * In a document with an external subset or external parameter entities |
7990 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
7991 | | * must precede any reference to it... |
7992 | | * |
7993 | | * [ WFC: In DTD ] |
7994 | | * Parameter-entity references may only appear in the DTD. |
7995 | | * NOTE: misleading but this is handled. |
7996 | | */ |
7997 | | void |
7998 | | xmlParsePEReference(xmlParserCtxtPtr ctxt) |
7999 | 36.9M | { |
8000 | 36.9M | const xmlChar *name; |
8001 | 36.9M | xmlEntityPtr entity = NULL; |
8002 | 36.9M | xmlParserInputPtr input; |
8003 | | |
8004 | 36.9M | if (RAW != '%') |
8005 | 0 | return; |
8006 | 36.9M | NEXT; |
8007 | 36.9M | name = xmlParseName(ctxt); |
8008 | 36.9M | if (name == NULL) { |
8009 | 14.5k | xmlFatalErrMsg(ctxt, XML_ERR_PEREF_NO_NAME, "PEReference: no name\n"); |
8010 | 14.5k | return; |
8011 | 14.5k | } |
8012 | 36.9M | if (xmlParserDebugEntities) |
8013 | 0 | xmlGenericError(xmlGenericErrorContext, |
8014 | 0 | "PEReference: %s\n", name); |
8015 | 36.9M | if (RAW != ';') { |
8016 | 765k | xmlFatalErr(ctxt, XML_ERR_PEREF_SEMICOL_MISSING, NULL); |
8017 | 765k | return; |
8018 | 765k | } |
8019 | | |
8020 | 36.1M | NEXT; |
8021 | | |
8022 | | /* |
8023 | | * Request the entity from SAX |
8024 | | */ |
8025 | 36.1M | if ((ctxt->sax != NULL) && |
8026 | 36.1M | (ctxt->sax->getParameterEntity != NULL)) |
8027 | 36.1M | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8028 | 36.1M | if (ctxt->instate == XML_PARSER_EOF) |
8029 | 0 | return; |
8030 | 36.1M | if (entity == NULL) { |
8031 | | /* |
8032 | | * [ WFC: Entity Declared ] |
8033 | | * In a document without any DTD, a document with only an |
8034 | | * internal DTD subset which contains no parameter entity |
8035 | | * references, or a document with "standalone='yes'", ... |
8036 | | * ... The declaration of a parameter entity must precede |
8037 | | * any reference to it... |
8038 | | */ |
8039 | 6.94M | if ((ctxt->standalone == 1) || |
8040 | 6.94M | ((ctxt->hasExternalSubset == 0) && |
8041 | 6.94M | (ctxt->hasPErefs == 0))) { |
8042 | 3.12k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8043 | 3.12k | "PEReference: %%%s; not found\n", |
8044 | 3.12k | name); |
8045 | 6.94M | } else { |
8046 | | /* |
8047 | | * [ VC: Entity Declared ] |
8048 | | * In a document with an external subset or external |
8049 | | * parameter entities with "standalone='no'", ... |
8050 | | * ... The declaration of a parameter entity must |
8051 | | * precede any reference to it... |
8052 | | */ |
8053 | 6.94M | if ((ctxt->validate) && (ctxt->vctxt.error != NULL)) { |
8054 | 1.47M | xmlValidityError(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8055 | 1.47M | "PEReference: %%%s; not found\n", |
8056 | 1.47M | name, NULL); |
8057 | 1.47M | } else |
8058 | 5.46M | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8059 | 5.46M | "PEReference: %%%s; not found\n", |
8060 | 5.46M | name, NULL); |
8061 | 6.94M | ctxt->valid = 0; |
8062 | 6.94M | } |
8063 | 29.2M | } else { |
8064 | | /* |
8065 | | * Internal checking in case the entity quest barfed |
8066 | | */ |
8067 | 29.2M | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8068 | 29.2M | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8069 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8070 | 0 | "Internal: %%%s; is not a parameter entity\n", |
8071 | 0 | name, NULL); |
8072 | 29.2M | } else { |
8073 | 29.2M | xmlChar start[4]; |
8074 | 29.2M | xmlCharEncoding enc; |
8075 | 29.2M | unsigned long parentConsumed; |
8076 | 29.2M | xmlEntityPtr oldEnt; |
8077 | | |
8078 | 29.2M | if ((entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
8079 | 29.2M | ((ctxt->options & XML_PARSE_NOENT) == 0) && |
8080 | 29.2M | ((ctxt->options & XML_PARSE_DTDVALID) == 0) && |
8081 | 29.2M | ((ctxt->options & XML_PARSE_DTDLOAD) == 0) && |
8082 | 29.2M | ((ctxt->options & XML_PARSE_DTDATTR) == 0) && |
8083 | 29.2M | (ctxt->replaceEntities == 0) && |
8084 | 29.2M | (ctxt->validate == 0)) |
8085 | 1.85k | return; |
8086 | | |
8087 | 29.2M | if (entity->flags & XML_ENT_EXPANDING) { |
8088 | 614 | xmlFatalErr(ctxt, XML_ERR_ENTITY_LOOP, NULL); |
8089 | 614 | xmlHaltParser(ctxt); |
8090 | 614 | return; |
8091 | 614 | } |
8092 | | |
8093 | | /* Must be computed from old input before pushing new input. */ |
8094 | 29.2M | parentConsumed = ctxt->input->parentConsumed; |
8095 | 29.2M | oldEnt = ctxt->input->entity; |
8096 | 29.2M | if ((oldEnt == NULL) || |
8097 | 29.2M | ((oldEnt->etype == XML_EXTERNAL_PARAMETER_ENTITY) && |
8098 | 28.7M | ((oldEnt->flags & XML_ENT_PARSED) == 0))) { |
8099 | 1.13M | xmlSaturatedAdd(&parentConsumed, ctxt->input->consumed); |
8100 | 1.13M | xmlSaturatedAddSizeT(&parentConsumed, |
8101 | 1.13M | ctxt->input->cur - ctxt->input->base); |
8102 | 1.13M | } |
8103 | | |
8104 | 29.2M | input = xmlNewEntityInputStream(ctxt, entity); |
8105 | 29.2M | if (xmlPushInput(ctxt, input) < 0) { |
8106 | 11.4k | xmlFreeInputStream(input); |
8107 | 11.4k | return; |
8108 | 11.4k | } |
8109 | | |
8110 | 29.2M | entity->flags |= XML_ENT_EXPANDING; |
8111 | | |
8112 | 29.2M | input->parentConsumed = parentConsumed; |
8113 | | |
8114 | 29.2M | if (entity->etype == XML_EXTERNAL_PARAMETER_ENTITY) { |
8115 | | /* |
8116 | | * Get the 4 first bytes and decode the charset |
8117 | | * if enc != XML_CHAR_ENCODING_NONE |
8118 | | * plug some encoding conversion routines. |
8119 | | * Note that, since we may have some non-UTF8 |
8120 | | * encoding (like UTF16, bug 135229), the 'length' |
8121 | | * is not known, but we can calculate based upon |
8122 | | * the amount of data in the buffer. |
8123 | | */ |
8124 | 104k | GROW |
8125 | 104k | if (ctxt->instate == XML_PARSER_EOF) |
8126 | 0 | return; |
8127 | 104k | if ((ctxt->input->end - ctxt->input->cur)>=4) { |
8128 | 103k | start[0] = RAW; |
8129 | 103k | start[1] = NXT(1); |
8130 | 103k | start[2] = NXT(2); |
8131 | 103k | start[3] = NXT(3); |
8132 | 103k | enc = xmlDetectCharEncoding(start, 4); |
8133 | 103k | if (enc != XML_CHAR_ENCODING_NONE) { |
8134 | 50.2k | xmlSwitchEncoding(ctxt, enc); |
8135 | 50.2k | } |
8136 | 103k | } |
8137 | | |
8138 | 104k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && |
8139 | 104k | (IS_BLANK_CH(NXT(5)))) { |
8140 | 42.6k | xmlParseTextDecl(ctxt); |
8141 | 42.6k | } |
8142 | 104k | } |
8143 | 29.2M | } |
8144 | 29.2M | } |
8145 | 36.1M | ctxt->hasPErefs = 1; |
8146 | 36.1M | } |
8147 | | |
8148 | | /** |
8149 | | * xmlLoadEntityContent: |
8150 | | * @ctxt: an XML parser context |
8151 | | * @entity: an unloaded system entity |
8152 | | * |
8153 | | * Load the original content of the given system entity from the |
8154 | | * ExternalID/SystemID given. This is to be used for Included in Literal |
8155 | | * http://www.w3.org/TR/REC-xml/#inliteral processing of entities references |
8156 | | * |
8157 | | * Returns 0 in case of success and -1 in case of failure |
8158 | | */ |
8159 | | static int |
8160 | 4.38k | xmlLoadEntityContent(xmlParserCtxtPtr ctxt, xmlEntityPtr entity) { |
8161 | 4.38k | xmlParserInputPtr input; |
8162 | 4.38k | xmlBufferPtr buf; |
8163 | 4.38k | int l, c; |
8164 | 4.38k | int count = 0; |
8165 | | |
8166 | 4.38k | if ((ctxt == NULL) || (entity == NULL) || |
8167 | 4.38k | ((entity->etype != XML_EXTERNAL_PARAMETER_ENTITY) && |
8168 | 4.38k | (entity->etype != XML_EXTERNAL_GENERAL_PARSED_ENTITY)) || |
8169 | 4.38k | (entity->content != NULL)) { |
8170 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8171 | 0 | "xmlLoadEntityContent parameter error"); |
8172 | 0 | return(-1); |
8173 | 0 | } |
8174 | | |
8175 | 4.38k | if (xmlParserDebugEntities) |
8176 | 0 | xmlGenericError(xmlGenericErrorContext, |
8177 | 0 | "Reading %s entity content input\n", entity->name); |
8178 | | |
8179 | 4.38k | buf = xmlBufferCreate(); |
8180 | 4.38k | if (buf == NULL) { |
8181 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8182 | 0 | "xmlLoadEntityContent parameter error"); |
8183 | 0 | return(-1); |
8184 | 0 | } |
8185 | 4.38k | xmlBufferSetAllocationScheme(buf, XML_BUFFER_ALLOC_DOUBLEIT); |
8186 | | |
8187 | 4.38k | input = xmlNewEntityInputStream(ctxt, entity); |
8188 | 4.38k | if (input == NULL) { |
8189 | 1.17k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8190 | 1.17k | "xmlLoadEntityContent input error"); |
8191 | 1.17k | xmlBufferFree(buf); |
8192 | 1.17k | return(-1); |
8193 | 1.17k | } |
8194 | | |
8195 | | /* |
8196 | | * Push the entity as the current input, read char by char |
8197 | | * saving to the buffer until the end of the entity or an error |
8198 | | */ |
8199 | 3.21k | if (xmlPushInput(ctxt, input) < 0) { |
8200 | 0 | xmlBufferFree(buf); |
8201 | 0 | xmlFreeInputStream(input); |
8202 | 0 | return(-1); |
8203 | 0 | } |
8204 | | |
8205 | 3.21k | GROW; |
8206 | 3.21k | c = CUR_CHAR(l); |
8207 | 14.0M | while ((ctxt->input == input) && (ctxt->input->cur < ctxt->input->end) && |
8208 | 14.0M | (IS_CHAR(c))) { |
8209 | 14.0M | xmlBufferAdd(buf, ctxt->input->cur, l); |
8210 | 14.0M | if (count++ > XML_PARSER_CHUNK_SIZE) { |
8211 | 135k | count = 0; |
8212 | 135k | GROW; |
8213 | 135k | if (ctxt->instate == XML_PARSER_EOF) { |
8214 | 0 | xmlBufferFree(buf); |
8215 | 0 | return(-1); |
8216 | 0 | } |
8217 | 135k | } |
8218 | 14.0M | NEXTL(l); |
8219 | 14.0M | c = CUR_CHAR(l); |
8220 | 14.0M | if (c == 0) { |
8221 | 2.82k | count = 0; |
8222 | 2.82k | GROW; |
8223 | 2.82k | if (ctxt->instate == XML_PARSER_EOF) { |
8224 | 0 | xmlBufferFree(buf); |
8225 | 0 | return(-1); |
8226 | 0 | } |
8227 | 2.82k | c = CUR_CHAR(l); |
8228 | 2.82k | } |
8229 | 14.0M | } |
8230 | | |
8231 | 3.21k | if ((ctxt->input == input) && (ctxt->input->cur >= ctxt->input->end)) { |
8232 | 1.85k | xmlSaturatedAdd(&ctxt->sizeentities, ctxt->input->consumed); |
8233 | 1.85k | xmlPopInput(ctxt); |
8234 | 1.85k | } else if (!IS_CHAR(c)) { |
8235 | 1.35k | xmlFatalErrMsgInt(ctxt, XML_ERR_INVALID_CHAR, |
8236 | 1.35k | "xmlLoadEntityContent: invalid char value %d\n", |
8237 | 1.35k | c); |
8238 | 1.35k | xmlBufferFree(buf); |
8239 | 1.35k | return(-1); |
8240 | 1.35k | } |
8241 | 1.85k | entity->content = buf->content; |
8242 | 1.85k | entity->length = buf->use; |
8243 | 1.85k | buf->content = NULL; |
8244 | 1.85k | xmlBufferFree(buf); |
8245 | | |
8246 | 1.85k | return(0); |
8247 | 3.21k | } |
8248 | | |
8249 | | /** |
8250 | | * xmlParseStringPEReference: |
8251 | | * @ctxt: an XML parser context |
8252 | | * @str: a pointer to an index in the string |
8253 | | * |
8254 | | * parse PEReference declarations |
8255 | | * |
8256 | | * [69] PEReference ::= '%' Name ';' |
8257 | | * |
8258 | | * [ WFC: No Recursion ] |
8259 | | * A parsed entity must not contain a recursive |
8260 | | * reference to itself, either directly or indirectly. |
8261 | | * |
8262 | | * [ WFC: Entity Declared ] |
8263 | | * In a document without any DTD, a document with only an internal DTD |
8264 | | * subset which contains no parameter entity references, or a document |
8265 | | * with "standalone='yes'", ... ... The declaration of a parameter |
8266 | | * entity must precede any reference to it... |
8267 | | * |
8268 | | * [ VC: Entity Declared ] |
8269 | | * In a document with an external subset or external parameter entities |
8270 | | * with "standalone='no'", ... ... The declaration of a parameter entity |
8271 | | * must precede any reference to it... |
8272 | | * |
8273 | | * [ WFC: In DTD ] |
8274 | | * Parameter-entity references may only appear in the DTD. |
8275 | | * NOTE: misleading but this is handled. |
8276 | | * |
8277 | | * Returns the string of the entity content. |
8278 | | * str is updated to the current value of the index |
8279 | | */ |
8280 | | static xmlEntityPtr |
8281 | 706k | xmlParseStringPEReference(xmlParserCtxtPtr ctxt, const xmlChar **str) { |
8282 | 706k | const xmlChar *ptr; |
8283 | 706k | xmlChar cur; |
8284 | 706k | xmlChar *name; |
8285 | 706k | xmlEntityPtr entity = NULL; |
8286 | | |
8287 | 706k | if ((str == NULL) || (*str == NULL)) return(NULL); |
8288 | 706k | ptr = *str; |
8289 | 706k | cur = *ptr; |
8290 | 706k | if (cur != '%') |
8291 | 0 | return(NULL); |
8292 | 706k | ptr++; |
8293 | 706k | name = xmlParseStringName(ctxt, &ptr); |
8294 | 706k | if (name == NULL) { |
8295 | 6.96k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8296 | 6.96k | "xmlParseStringPEReference: no name\n"); |
8297 | 6.96k | *str = ptr; |
8298 | 6.96k | return(NULL); |
8299 | 6.96k | } |
8300 | 699k | cur = *ptr; |
8301 | 699k | if (cur != ';') { |
8302 | 3.10k | xmlFatalErr(ctxt, XML_ERR_ENTITYREF_SEMICOL_MISSING, NULL); |
8303 | 3.10k | xmlFree(name); |
8304 | 3.10k | *str = ptr; |
8305 | 3.10k | return(NULL); |
8306 | 3.10k | } |
8307 | 696k | ptr++; |
8308 | | |
8309 | | /* |
8310 | | * Request the entity from SAX |
8311 | | */ |
8312 | 696k | if ((ctxt->sax != NULL) && |
8313 | 696k | (ctxt->sax->getParameterEntity != NULL)) |
8314 | 696k | entity = ctxt->sax->getParameterEntity(ctxt->userData, name); |
8315 | 696k | if (ctxt->instate == XML_PARSER_EOF) { |
8316 | 0 | xmlFree(name); |
8317 | 0 | *str = ptr; |
8318 | 0 | return(NULL); |
8319 | 0 | } |
8320 | 696k | if (entity == NULL) { |
8321 | | /* |
8322 | | * [ WFC: Entity Declared ] |
8323 | | * In a document without any DTD, a document with only an |
8324 | | * internal DTD subset which contains no parameter entity |
8325 | | * references, or a document with "standalone='yes'", ... |
8326 | | * ... The declaration of a parameter entity must precede |
8327 | | * any reference to it... |
8328 | | */ |
8329 | 251k | if ((ctxt->standalone == 1) || |
8330 | 251k | ((ctxt->hasExternalSubset == 0) && (ctxt->hasPErefs == 0))) { |
8331 | 471 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNDECLARED_ENTITY, |
8332 | 471 | "PEReference: %%%s; not found\n", name); |
8333 | 250k | } else { |
8334 | | /* |
8335 | | * [ VC: Entity Declared ] |
8336 | | * In a document with an external subset or external |
8337 | | * parameter entities with "standalone='no'", ... |
8338 | | * ... The declaration of a parameter entity must |
8339 | | * precede any reference to it... |
8340 | | */ |
8341 | 250k | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8342 | 250k | "PEReference: %%%s; not found\n", |
8343 | 250k | name, NULL); |
8344 | 250k | ctxt->valid = 0; |
8345 | 250k | } |
8346 | 445k | } else { |
8347 | | /* |
8348 | | * Internal checking in case the entity quest barfed |
8349 | | */ |
8350 | 445k | if ((entity->etype != XML_INTERNAL_PARAMETER_ENTITY) && |
8351 | 445k | (entity->etype != XML_EXTERNAL_PARAMETER_ENTITY)) { |
8352 | 0 | xmlWarningMsg(ctxt, XML_WAR_UNDECLARED_ENTITY, |
8353 | 0 | "%%%s; is not a parameter entity\n", |
8354 | 0 | name, NULL); |
8355 | 0 | } |
8356 | 445k | } |
8357 | 696k | ctxt->hasPErefs = 1; |
8358 | 696k | xmlFree(name); |
8359 | 696k | *str = ptr; |
8360 | 696k | return(entity); |
8361 | 696k | } |
8362 | | |
8363 | | /** |
8364 | | * xmlParseDocTypeDecl: |
8365 | | * @ctxt: an XML parser context |
8366 | | * |
8367 | | * DEPRECATED: Internal function, don't use. |
8368 | | * |
8369 | | * parse a DOCTYPE declaration |
8370 | | * |
8371 | | * [28] doctypedecl ::= '<!DOCTYPE' S Name (S ExternalID)? S? |
8372 | | * ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8373 | | * |
8374 | | * [ VC: Root Element Type ] |
8375 | | * The Name in the document type declaration must match the element |
8376 | | * type of the root element. |
8377 | | */ |
8378 | | |
8379 | | void |
8380 | 291k | xmlParseDocTypeDecl(xmlParserCtxtPtr ctxt) { |
8381 | 291k | const xmlChar *name = NULL; |
8382 | 291k | xmlChar *ExternalID = NULL; |
8383 | 291k | xmlChar *URI = NULL; |
8384 | | |
8385 | | /* |
8386 | | * We know that '<!DOCTYPE' has been detected. |
8387 | | */ |
8388 | 291k | SKIP(9); |
8389 | | |
8390 | 291k | SKIP_BLANKS; |
8391 | | |
8392 | | /* |
8393 | | * Parse the DOCTYPE name. |
8394 | | */ |
8395 | 291k | name = xmlParseName(ctxt); |
8396 | 291k | if (name == NULL) { |
8397 | 485 | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8398 | 485 | "xmlParseDocTypeDecl : no DOCTYPE name !\n"); |
8399 | 485 | } |
8400 | 291k | ctxt->intSubName = name; |
8401 | | |
8402 | 291k | SKIP_BLANKS; |
8403 | | |
8404 | | /* |
8405 | | * Check for SystemID and ExternalID |
8406 | | */ |
8407 | 291k | URI = xmlParseExternalID(ctxt, &ExternalID, 1); |
8408 | | |
8409 | 291k | if ((URI != NULL) || (ExternalID != NULL)) { |
8410 | 116k | ctxt->hasExternalSubset = 1; |
8411 | 116k | } |
8412 | 291k | ctxt->extSubURI = URI; |
8413 | 291k | ctxt->extSubSystem = ExternalID; |
8414 | | |
8415 | 291k | SKIP_BLANKS; |
8416 | | |
8417 | | /* |
8418 | | * Create and update the internal subset. |
8419 | | */ |
8420 | 291k | if ((ctxt->sax != NULL) && (ctxt->sax->internalSubset != NULL) && |
8421 | 291k | (!ctxt->disableSAX)) |
8422 | 287k | ctxt->sax->internalSubset(ctxt->userData, name, ExternalID, URI); |
8423 | 291k | if (ctxt->instate == XML_PARSER_EOF) |
8424 | 0 | return; |
8425 | | |
8426 | | /* |
8427 | | * Is there any internal subset declarations ? |
8428 | | * they are handled separately in xmlParseInternalSubset() |
8429 | | */ |
8430 | 291k | if (RAW == '[') |
8431 | 233k | return; |
8432 | | |
8433 | | /* |
8434 | | * We should be at the end of the DOCTYPE declaration. |
8435 | | */ |
8436 | 58.3k | if (RAW != '>') { |
8437 | 5.99k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8438 | 5.99k | } |
8439 | 58.3k | NEXT; |
8440 | 58.3k | } |
8441 | | |
8442 | | /** |
8443 | | * xmlParseInternalSubset: |
8444 | | * @ctxt: an XML parser context |
8445 | | * |
8446 | | * parse the internal subset declaration |
8447 | | * |
8448 | | * [28 end] ('[' (markupdecl | PEReference | S)* ']' S?)? '>' |
8449 | | */ |
8450 | | |
8451 | | static void |
8452 | 233k | xmlParseInternalSubset(xmlParserCtxtPtr ctxt) { |
8453 | | /* |
8454 | | * Is there any DTD definition ? |
8455 | | */ |
8456 | 233k | if (RAW == '[') { |
8457 | 233k | int baseInputNr = ctxt->inputNr; |
8458 | 233k | ctxt->instate = XML_PARSER_DTD; |
8459 | 233k | NEXT; |
8460 | | /* |
8461 | | * Parse the succession of Markup declarations and |
8462 | | * PEReferences. |
8463 | | * Subsequence (markupdecl | PEReference | S)* |
8464 | | */ |
8465 | 233k | SKIP_BLANKS; |
8466 | 31.1M | while (((RAW != ']') || (ctxt->inputNr > baseInputNr)) && |
8467 | 31.1M | (ctxt->instate != XML_PARSER_EOF)) { |
8468 | | |
8469 | | /* |
8470 | | * Conditional sections are allowed from external entities included |
8471 | | * by PE References in the internal subset. |
8472 | | */ |
8473 | 30.9M | if ((ctxt->inputNr > 1) && (ctxt->input->filename != NULL) && |
8474 | 30.9M | (RAW == '<') && (NXT(1) == '!') && (NXT(2) == '[')) { |
8475 | 0 | xmlParseConditionalSections(ctxt); |
8476 | 30.9M | } else if ((RAW == '<') && ((NXT(1) == '!') || (NXT(1) == '?'))) { |
8477 | 30.8M | xmlParseMarkupDecl(ctxt); |
8478 | 30.8M | } else if (RAW == '%') { |
8479 | 62.6k | xmlParsePEReference(ctxt); |
8480 | 62.6k | } else { |
8481 | 45.4k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
8482 | 45.4k | "xmlParseInternalSubset: error detected in" |
8483 | 45.4k | " Markup declaration\n"); |
8484 | 45.4k | xmlHaltParser(ctxt); |
8485 | 45.4k | return; |
8486 | 45.4k | } |
8487 | 30.9M | SKIP_BLANKS; |
8488 | 30.9M | } |
8489 | 187k | if (RAW == ']') { |
8490 | 175k | NEXT; |
8491 | 175k | SKIP_BLANKS; |
8492 | 175k | } |
8493 | 187k | } |
8494 | | |
8495 | | /* |
8496 | | * We should be at the end of the DOCTYPE declaration. |
8497 | | */ |
8498 | 187k | if (RAW != '>') { |
8499 | 12.8k | xmlFatalErr(ctxt, XML_ERR_DOCTYPE_NOT_FINISHED, NULL); |
8500 | 12.8k | return; |
8501 | 12.8k | } |
8502 | 174k | NEXT; |
8503 | 174k | } |
8504 | | |
8505 | | #ifdef LIBXML_SAX1_ENABLED |
8506 | | /** |
8507 | | * xmlParseAttribute: |
8508 | | * @ctxt: an XML parser context |
8509 | | * @value: a xmlChar ** used to store the value of the attribute |
8510 | | * |
8511 | | * DEPRECATED: Internal function, don't use. |
8512 | | * |
8513 | | * parse an attribute |
8514 | | * |
8515 | | * [41] Attribute ::= Name Eq AttValue |
8516 | | * |
8517 | | * [ WFC: No External Entity References ] |
8518 | | * Attribute values cannot contain direct or indirect entity references |
8519 | | * to external entities. |
8520 | | * |
8521 | | * [ WFC: No < in Attribute Values ] |
8522 | | * The replacement text of any entity referred to directly or indirectly in |
8523 | | * an attribute value (other than "<") must not contain a <. |
8524 | | * |
8525 | | * [ VC: Attribute Value Type ] |
8526 | | * The attribute must have been declared; the value must be of the type |
8527 | | * declared for it. |
8528 | | * |
8529 | | * [25] Eq ::= S? '=' S? |
8530 | | * |
8531 | | * With namespace: |
8532 | | * |
8533 | | * [NS 11] Attribute ::= QName Eq AttValue |
8534 | | * |
8535 | | * Also the case QName == xmlns:??? is handled independently as a namespace |
8536 | | * definition. |
8537 | | * |
8538 | | * Returns the attribute name, and the value in *value. |
8539 | | */ |
8540 | | |
8541 | | const xmlChar * |
8542 | 3.63M | xmlParseAttribute(xmlParserCtxtPtr ctxt, xmlChar **value) { |
8543 | 3.63M | const xmlChar *name; |
8544 | 3.63M | xmlChar *val; |
8545 | | |
8546 | 3.63M | *value = NULL; |
8547 | 3.63M | GROW; |
8548 | 3.63M | name = xmlParseName(ctxt); |
8549 | 3.63M | if (name == NULL) { |
8550 | 195k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8551 | 195k | "error parsing attribute name\n"); |
8552 | 195k | return(NULL); |
8553 | 195k | } |
8554 | | |
8555 | | /* |
8556 | | * read the value |
8557 | | */ |
8558 | 3.43M | SKIP_BLANKS; |
8559 | 3.43M | if (RAW == '=') { |
8560 | 3.31M | NEXT; |
8561 | 3.31M | SKIP_BLANKS; |
8562 | 3.31M | val = xmlParseAttValue(ctxt); |
8563 | 3.31M | ctxt->instate = XML_PARSER_CONTENT; |
8564 | 3.31M | } else { |
8565 | 120k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
8566 | 120k | "Specification mandates value for attribute %s\n", name); |
8567 | 120k | return(name); |
8568 | 120k | } |
8569 | | |
8570 | | /* |
8571 | | * Check that xml:lang conforms to the specification |
8572 | | * No more registered as an error, just generate a warning now |
8573 | | * since this was deprecated in XML second edition |
8574 | | */ |
8575 | 3.31M | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "xml:lang"))) { |
8576 | 12.6k | if (!xmlCheckLanguageID(val)) { |
8577 | 7.11k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
8578 | 7.11k | "Malformed value for xml:lang : %s\n", |
8579 | 7.11k | val, NULL); |
8580 | 7.11k | } |
8581 | 12.6k | } |
8582 | | |
8583 | | /* |
8584 | | * Check that xml:space conforms to the specification |
8585 | | */ |
8586 | 3.31M | if (xmlStrEqual(name, BAD_CAST "xml:space")) { |
8587 | 1.08k | if (xmlStrEqual(val, BAD_CAST "default")) |
8588 | 136 | *(ctxt->space) = 0; |
8589 | 952 | else if (xmlStrEqual(val, BAD_CAST "preserve")) |
8590 | 516 | *(ctxt->space) = 1; |
8591 | 436 | else { |
8592 | 436 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
8593 | 436 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
8594 | 436 | val, NULL); |
8595 | 436 | } |
8596 | 1.08k | } |
8597 | | |
8598 | 3.31M | *value = val; |
8599 | 3.31M | return(name); |
8600 | 3.43M | } |
8601 | | |
8602 | | /** |
8603 | | * xmlParseStartTag: |
8604 | | * @ctxt: an XML parser context |
8605 | | * |
8606 | | * DEPRECATED: Internal function, don't use. |
8607 | | * |
8608 | | * Parse a start tag. Always consumes '<'. |
8609 | | * |
8610 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
8611 | | * |
8612 | | * [ WFC: Unique Att Spec ] |
8613 | | * No attribute name may appear more than once in the same start-tag or |
8614 | | * empty-element tag. |
8615 | | * |
8616 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
8617 | | * |
8618 | | * [ WFC: Unique Att Spec ] |
8619 | | * No attribute name may appear more than once in the same start-tag or |
8620 | | * empty-element tag. |
8621 | | * |
8622 | | * With namespace: |
8623 | | * |
8624 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
8625 | | * |
8626 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
8627 | | * |
8628 | | * Returns the element name parsed |
8629 | | */ |
8630 | | |
8631 | | const xmlChar * |
8632 | 3.79M | xmlParseStartTag(xmlParserCtxtPtr ctxt) { |
8633 | 3.79M | const xmlChar *name; |
8634 | 3.79M | const xmlChar *attname; |
8635 | 3.79M | xmlChar *attvalue; |
8636 | 3.79M | const xmlChar **atts = ctxt->atts; |
8637 | 3.79M | int nbatts = 0; |
8638 | 3.79M | int maxatts = ctxt->maxatts; |
8639 | 3.79M | int i; |
8640 | | |
8641 | 3.79M | if (RAW != '<') return(NULL); |
8642 | 3.79M | NEXT1; |
8643 | | |
8644 | 3.79M | name = xmlParseName(ctxt); |
8645 | 3.79M | if (name == NULL) { |
8646 | 140k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
8647 | 140k | "xmlParseStartTag: invalid element name\n"); |
8648 | 140k | return(NULL); |
8649 | 140k | } |
8650 | | |
8651 | | /* |
8652 | | * Now parse the attributes, it ends up with the ending |
8653 | | * |
8654 | | * (S Attribute)* S? |
8655 | | */ |
8656 | 3.65M | SKIP_BLANKS; |
8657 | 3.65M | GROW; |
8658 | | |
8659 | 5.09M | while (((RAW != '>') && |
8660 | 5.09M | ((RAW != '/') || (NXT(1) != '>')) && |
8661 | 5.09M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
8662 | 3.63M | attname = xmlParseAttribute(ctxt, &attvalue); |
8663 | 3.63M | if (attname == NULL) { |
8664 | 195k | xmlFatalErrMsg(ctxt, XML_ERR_INTERNAL_ERROR, |
8665 | 195k | "xmlParseStartTag: problem parsing attributes\n"); |
8666 | 195k | break; |
8667 | 195k | } |
8668 | 3.43M | if (attvalue != NULL) { |
8669 | | /* |
8670 | | * [ WFC: Unique Att Spec ] |
8671 | | * No attribute name may appear more than once in the same |
8672 | | * start-tag or empty-element tag. |
8673 | | */ |
8674 | 4.57M | for (i = 0; i < nbatts;i += 2) { |
8675 | 1.28M | if (xmlStrEqual(atts[i], attname)) { |
8676 | 7.74k | xmlErrAttributeDup(ctxt, NULL, attname); |
8677 | 7.74k | xmlFree(attvalue); |
8678 | 7.74k | goto failed; |
8679 | 7.74k | } |
8680 | 1.28M | } |
8681 | | /* |
8682 | | * Add the pair to atts |
8683 | | */ |
8684 | 3.29M | if (atts == NULL) { |
8685 | 80.4k | maxatts = 22; /* allow for 10 attrs by default */ |
8686 | 80.4k | atts = (const xmlChar **) |
8687 | 80.4k | xmlMalloc(maxatts * sizeof(xmlChar *)); |
8688 | 80.4k | if (atts == NULL) { |
8689 | 0 | xmlErrMemory(ctxt, NULL); |
8690 | 0 | if (attvalue != NULL) |
8691 | 0 | xmlFree(attvalue); |
8692 | 0 | goto failed; |
8693 | 0 | } |
8694 | 80.4k | ctxt->atts = atts; |
8695 | 80.4k | ctxt->maxatts = maxatts; |
8696 | 3.21M | } else if (nbatts + 4 > maxatts) { |
8697 | 135 | const xmlChar **n; |
8698 | | |
8699 | 135 | maxatts *= 2; |
8700 | 135 | n = (const xmlChar **) xmlRealloc((void *) atts, |
8701 | 135 | maxatts * sizeof(const xmlChar *)); |
8702 | 135 | if (n == NULL) { |
8703 | 0 | xmlErrMemory(ctxt, NULL); |
8704 | 0 | if (attvalue != NULL) |
8705 | 0 | xmlFree(attvalue); |
8706 | 0 | goto failed; |
8707 | 0 | } |
8708 | 135 | atts = n; |
8709 | 135 | ctxt->atts = atts; |
8710 | 135 | ctxt->maxatts = maxatts; |
8711 | 135 | } |
8712 | 3.29M | atts[nbatts++] = attname; |
8713 | 3.29M | atts[nbatts++] = attvalue; |
8714 | 3.29M | atts[nbatts] = NULL; |
8715 | 3.29M | atts[nbatts + 1] = NULL; |
8716 | 3.29M | } else { |
8717 | 132k | if (attvalue != NULL) |
8718 | 0 | xmlFree(attvalue); |
8719 | 132k | } |
8720 | | |
8721 | 3.43M | failed: |
8722 | | |
8723 | 3.43M | GROW |
8724 | 3.43M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
8725 | 2.00M | break; |
8726 | 1.43M | if (SKIP_BLANKS == 0) { |
8727 | 211k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
8728 | 211k | "attributes construct error\n"); |
8729 | 211k | } |
8730 | 1.43M | SHRINK; |
8731 | 1.43M | GROW; |
8732 | 1.43M | } |
8733 | | |
8734 | | /* |
8735 | | * SAX: Start of Element ! |
8736 | | */ |
8737 | 3.65M | if ((ctxt->sax != NULL) && (ctxt->sax->startElement != NULL) && |
8738 | 3.65M | (!ctxt->disableSAX)) { |
8739 | 3.27M | if (nbatts > 0) |
8740 | 1.86M | ctxt->sax->startElement(ctxt->userData, name, atts); |
8741 | 1.41M | else |
8742 | 1.41M | ctxt->sax->startElement(ctxt->userData, name, NULL); |
8743 | 3.27M | } |
8744 | | |
8745 | 3.65M | if (atts != NULL) { |
8746 | | /* Free only the content strings */ |
8747 | 6.63M | for (i = 1;i < nbatts;i+=2) |
8748 | 3.29M | if (atts[i] != NULL) |
8749 | 3.29M | xmlFree((xmlChar *) atts[i]); |
8750 | 3.34M | } |
8751 | 3.65M | return(name); |
8752 | 3.65M | } |
8753 | | |
8754 | | /** |
8755 | | * xmlParseEndTag1: |
8756 | | * @ctxt: an XML parser context |
8757 | | * @line: line of the start tag |
8758 | | * @nsNr: number of namespaces on the start tag |
8759 | | * |
8760 | | * Parse an end tag. Always consumes '</'. |
8761 | | * |
8762 | | * [42] ETag ::= '</' Name S? '>' |
8763 | | * |
8764 | | * With namespace |
8765 | | * |
8766 | | * [NS 9] ETag ::= '</' QName S? '>' |
8767 | | */ |
8768 | | |
8769 | | static void |
8770 | 1.90M | xmlParseEndTag1(xmlParserCtxtPtr ctxt, int line) { |
8771 | 1.90M | const xmlChar *name; |
8772 | | |
8773 | 1.90M | GROW; |
8774 | 1.90M | if ((RAW != '<') || (NXT(1) != '/')) { |
8775 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_LTSLASH_REQUIRED, |
8776 | 0 | "xmlParseEndTag: '</' not found\n"); |
8777 | 0 | return; |
8778 | 0 | } |
8779 | 1.90M | SKIP(2); |
8780 | | |
8781 | 1.90M | name = xmlParseNameAndCompare(ctxt,ctxt->name); |
8782 | | |
8783 | | /* |
8784 | | * We should definitely be at the ending "S? '>'" part |
8785 | | */ |
8786 | 1.90M | GROW; |
8787 | 1.90M | SKIP_BLANKS; |
8788 | 1.90M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
8789 | 32.3k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
8790 | 32.3k | } else |
8791 | 1.87M | NEXT1; |
8792 | | |
8793 | | /* |
8794 | | * [ WFC: Element Type Match ] |
8795 | | * The Name in an element's end-tag must match the element type in the |
8796 | | * start-tag. |
8797 | | * |
8798 | | */ |
8799 | 1.90M | if (name != (xmlChar*)1) { |
8800 | 93.1k | if (name == NULL) name = BAD_CAST "unparsable"; |
8801 | 93.1k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
8802 | 93.1k | "Opening and ending tag mismatch: %s line %d and %s\n", |
8803 | 93.1k | ctxt->name, line, name); |
8804 | 93.1k | } |
8805 | | |
8806 | | /* |
8807 | | * SAX: End of Tag |
8808 | | */ |
8809 | 1.90M | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
8810 | 1.90M | (!ctxt->disableSAX)) |
8811 | 1.71M | ctxt->sax->endElement(ctxt->userData, ctxt->name); |
8812 | | |
8813 | 1.90M | namePop(ctxt); |
8814 | 1.90M | spacePop(ctxt); |
8815 | 1.90M | return; |
8816 | 1.90M | } |
8817 | | |
8818 | | /** |
8819 | | * xmlParseEndTag: |
8820 | | * @ctxt: an XML parser context |
8821 | | * |
8822 | | * DEPRECATED: Internal function, don't use. |
8823 | | * |
8824 | | * parse an end of tag |
8825 | | * |
8826 | | * [42] ETag ::= '</' Name S? '>' |
8827 | | * |
8828 | | * With namespace |
8829 | | * |
8830 | | * [NS 9] ETag ::= '</' QName S? '>' |
8831 | | */ |
8832 | | |
8833 | | void |
8834 | 0 | xmlParseEndTag(xmlParserCtxtPtr ctxt) { |
8835 | 0 | xmlParseEndTag1(ctxt, 0); |
8836 | 0 | } |
8837 | | #endif /* LIBXML_SAX1_ENABLED */ |
8838 | | |
8839 | | /************************************************************************ |
8840 | | * * |
8841 | | * SAX 2 specific operations * |
8842 | | * * |
8843 | | ************************************************************************/ |
8844 | | |
8845 | | /* |
8846 | | * xmlGetNamespace: |
8847 | | * @ctxt: an XML parser context |
8848 | | * @prefix: the prefix to lookup |
8849 | | * |
8850 | | * Lookup the namespace name for the @prefix (which ca be NULL) |
8851 | | * The prefix must come from the @ctxt->dict dictionary |
8852 | | * |
8853 | | * Returns the namespace name or NULL if not bound |
8854 | | */ |
8855 | | static const xmlChar * |
8856 | 7.06M | xmlGetNamespace(xmlParserCtxtPtr ctxt, const xmlChar *prefix) { |
8857 | 7.06M | int i; |
8858 | | |
8859 | 7.06M | if (prefix == ctxt->str_xml) return(ctxt->str_xml_ns); |
8860 | 7.73M | for (i = ctxt->nsNr - 2;i >= 0;i-=2) |
8861 | 1.34M | if (ctxt->nsTab[i] == prefix) { |
8862 | 581k | if ((prefix == NULL) && (*ctxt->nsTab[i + 1] == 0)) |
8863 | 14.7k | return(NULL); |
8864 | 567k | return(ctxt->nsTab[i + 1]); |
8865 | 581k | } |
8866 | 6.38M | return(NULL); |
8867 | 6.97M | } |
8868 | | |
8869 | | /** |
8870 | | * xmlParseQName: |
8871 | | * @ctxt: an XML parser context |
8872 | | * @prefix: pointer to store the prefix part |
8873 | | * |
8874 | | * parse an XML Namespace QName |
8875 | | * |
8876 | | * [6] QName ::= (Prefix ':')? LocalPart |
8877 | | * [7] Prefix ::= NCName |
8878 | | * [8] LocalPart ::= NCName |
8879 | | * |
8880 | | * Returns the Name parsed or NULL |
8881 | | */ |
8882 | | |
8883 | | static const xmlChar * |
8884 | 14.3M | xmlParseQName(xmlParserCtxtPtr ctxt, const xmlChar **prefix) { |
8885 | 14.3M | const xmlChar *l, *p; |
8886 | | |
8887 | 14.3M | GROW; |
8888 | | |
8889 | 14.3M | l = xmlParseNCName(ctxt); |
8890 | 14.3M | if (l == NULL) { |
8891 | 339k | if (CUR == ':') { |
8892 | 6.61k | l = xmlParseName(ctxt); |
8893 | 6.61k | if (l != NULL) { |
8894 | 6.61k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8895 | 6.61k | "Failed to parse QName '%s'\n", l, NULL, NULL); |
8896 | 6.61k | *prefix = NULL; |
8897 | 6.61k | return(l); |
8898 | 6.61k | } |
8899 | 6.61k | } |
8900 | 333k | return(NULL); |
8901 | 339k | } |
8902 | 14.0M | if (CUR == ':') { |
8903 | 758k | NEXT; |
8904 | 758k | p = l; |
8905 | 758k | l = xmlParseNCName(ctxt); |
8906 | 758k | if (l == NULL) { |
8907 | 18.5k | xmlChar *tmp; |
8908 | | |
8909 | 18.5k | if (ctxt->instate == XML_PARSER_EOF) |
8910 | 0 | return(NULL); |
8911 | 18.5k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8912 | 18.5k | "Failed to parse QName '%s:'\n", p, NULL, NULL); |
8913 | 18.5k | l = xmlParseNmtoken(ctxt); |
8914 | 18.5k | if (l == NULL) { |
8915 | 11.9k | if (ctxt->instate == XML_PARSER_EOF) |
8916 | 0 | return(NULL); |
8917 | 11.9k | tmp = xmlBuildQName(BAD_CAST "", p, NULL, 0); |
8918 | 11.9k | } else { |
8919 | 6.55k | tmp = xmlBuildQName(l, p, NULL, 0); |
8920 | 6.55k | xmlFree((char *)l); |
8921 | 6.55k | } |
8922 | 18.5k | p = xmlDictLookup(ctxt->dict, tmp, -1); |
8923 | 18.5k | if (tmp != NULL) xmlFree(tmp); |
8924 | 18.5k | *prefix = NULL; |
8925 | 18.5k | return(p); |
8926 | 18.5k | } |
8927 | 739k | if (CUR == ':') { |
8928 | 10.6k | xmlChar *tmp; |
8929 | | |
8930 | 10.6k | xmlNsErr(ctxt, XML_NS_ERR_QNAME, |
8931 | 10.6k | "Failed to parse QName '%s:%s:'\n", p, l, NULL); |
8932 | 10.6k | NEXT; |
8933 | 10.6k | tmp = (xmlChar *) xmlParseName(ctxt); |
8934 | 10.6k | if (tmp != NULL) { |
8935 | 8.35k | tmp = xmlBuildQName(tmp, l, NULL, 0); |
8936 | 8.35k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8937 | 8.35k | if (tmp != NULL) xmlFree(tmp); |
8938 | 8.35k | *prefix = p; |
8939 | 8.35k | return(l); |
8940 | 8.35k | } |
8941 | 2.27k | if (ctxt->instate == XML_PARSER_EOF) |
8942 | 0 | return(NULL); |
8943 | 2.27k | tmp = xmlBuildQName(BAD_CAST "", l, NULL, 0); |
8944 | 2.27k | l = xmlDictLookup(ctxt->dict, tmp, -1); |
8945 | 2.27k | if (tmp != NULL) xmlFree(tmp); |
8946 | 2.27k | *prefix = p; |
8947 | 2.27k | return(l); |
8948 | 2.27k | } |
8949 | 729k | *prefix = p; |
8950 | 729k | } else |
8951 | 13.2M | *prefix = NULL; |
8952 | 13.9M | return(l); |
8953 | 14.0M | } |
8954 | | |
8955 | | /** |
8956 | | * xmlParseQNameAndCompare: |
8957 | | * @ctxt: an XML parser context |
8958 | | * @name: the localname |
8959 | | * @prefix: the prefix, if any. |
8960 | | * |
8961 | | * parse an XML name and compares for match |
8962 | | * (specialized for endtag parsing) |
8963 | | * |
8964 | | * Returns NULL for an illegal name, (xmlChar*) 1 for success |
8965 | | * and the name for mismatch |
8966 | | */ |
8967 | | |
8968 | | static const xmlChar * |
8969 | | xmlParseQNameAndCompare(xmlParserCtxtPtr ctxt, xmlChar const *name, |
8970 | 184k | xmlChar const *prefix) { |
8971 | 184k | const xmlChar *cmp; |
8972 | 184k | const xmlChar *in; |
8973 | 184k | const xmlChar *ret; |
8974 | 184k | const xmlChar *prefix2; |
8975 | | |
8976 | 184k | if (prefix == NULL) return(xmlParseNameAndCompare(ctxt, name)); |
8977 | | |
8978 | 184k | GROW; |
8979 | 184k | in = ctxt->input->cur; |
8980 | | |
8981 | 184k | cmp = prefix; |
8982 | 623k | while (*in != 0 && *in == *cmp) { |
8983 | 438k | ++in; |
8984 | 438k | ++cmp; |
8985 | 438k | } |
8986 | 184k | if ((*cmp == 0) && (*in == ':')) { |
8987 | 168k | in++; |
8988 | 168k | cmp = name; |
8989 | 1.34M | while (*in != 0 && *in == *cmp) { |
8990 | 1.17M | ++in; |
8991 | 1.17M | ++cmp; |
8992 | 1.17M | } |
8993 | 168k | if (*cmp == 0 && (*in == '>' || IS_BLANK_CH (*in))) { |
8994 | | /* success */ |
8995 | 152k | ctxt->input->col += in - ctxt->input->cur; |
8996 | 152k | ctxt->input->cur = in; |
8997 | 152k | return((const xmlChar*) 1); |
8998 | 152k | } |
8999 | 168k | } |
9000 | | /* |
9001 | | * all strings coms from the dictionary, equality can be done directly |
9002 | | */ |
9003 | 32.6k | ret = xmlParseQName (ctxt, &prefix2); |
9004 | 32.6k | if ((ret == name) && (prefix == prefix2)) |
9005 | 487 | return((const xmlChar*) 1); |
9006 | 32.2k | return ret; |
9007 | 32.6k | } |
9008 | | |
9009 | | /** |
9010 | | * xmlParseAttValueInternal: |
9011 | | * @ctxt: an XML parser context |
9012 | | * @len: attribute len result |
9013 | | * @alloc: whether the attribute was reallocated as a new string |
9014 | | * @normalize: if 1 then further non-CDATA normalization must be done |
9015 | | * |
9016 | | * parse a value for an attribute. |
9017 | | * NOTE: if no normalization is needed, the routine will return pointers |
9018 | | * directly from the data buffer. |
9019 | | * |
9020 | | * 3.3.3 Attribute-Value Normalization: |
9021 | | * Before the value of an attribute is passed to the application or |
9022 | | * checked for validity, the XML processor must normalize it as follows: |
9023 | | * - a character reference is processed by appending the referenced |
9024 | | * character to the attribute value |
9025 | | * - an entity reference is processed by recursively processing the |
9026 | | * replacement text of the entity |
9027 | | * - a whitespace character (#x20, #xD, #xA, #x9) is processed by |
9028 | | * appending #x20 to the normalized value, except that only a single |
9029 | | * #x20 is appended for a "#xD#xA" sequence that is part of an external |
9030 | | * parsed entity or the literal entity value of an internal parsed entity |
9031 | | * - other characters are processed by appending them to the normalized value |
9032 | | * If the declared value is not CDATA, then the XML processor must further |
9033 | | * process the normalized attribute value by discarding any leading and |
9034 | | * trailing space (#x20) characters, and by replacing sequences of space |
9035 | | * (#x20) characters by a single space (#x20) character. |
9036 | | * All attributes for which no declaration has been read should be treated |
9037 | | * by a non-validating parser as if declared CDATA. |
9038 | | * |
9039 | | * Returns the AttValue parsed or NULL. The value has to be freed by the |
9040 | | * caller if it was copied, this can be detected by val[*len] == 0. |
9041 | | */ |
9042 | | |
9043 | | #define GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) \ |
9044 | 5.58k | const xmlChar *oldbase = ctxt->input->base;\ |
9045 | 5.58k | GROW;\ |
9046 | 5.58k | if (ctxt->instate == XML_PARSER_EOF)\ |
9047 | 5.58k | return(NULL);\ |
9048 | 5.58k | if (oldbase != ctxt->input->base) {\ |
9049 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase;\ |
9050 | 0 | start = start + delta;\ |
9051 | 0 | in = in + delta;\ |
9052 | 0 | }\ |
9053 | 5.58k | end = ctxt->input->end; |
9054 | | |
9055 | | static xmlChar * |
9056 | | xmlParseAttValueInternal(xmlParserCtxtPtr ctxt, int *len, int *alloc, |
9057 | | int normalize) |
9058 | 10.8M | { |
9059 | 10.8M | xmlChar limit = 0; |
9060 | 10.8M | const xmlChar *in = NULL, *start, *end, *last; |
9061 | 10.8M | xmlChar *ret = NULL; |
9062 | 10.8M | int line, col; |
9063 | 10.8M | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9064 | 3.50M | XML_MAX_HUGE_LENGTH : |
9065 | 10.8M | XML_MAX_TEXT_LENGTH; |
9066 | | |
9067 | 10.8M | GROW; |
9068 | 10.8M | in = (xmlChar *) CUR_PTR; |
9069 | 10.8M | line = ctxt->input->line; |
9070 | 10.8M | col = ctxt->input->col; |
9071 | 10.8M | if (*in != '"' && *in != '\'') { |
9072 | 32.3k | xmlFatalErr(ctxt, XML_ERR_ATTRIBUTE_NOT_STARTED, NULL); |
9073 | 32.3k | return (NULL); |
9074 | 32.3k | } |
9075 | 10.8M | ctxt->instate = XML_PARSER_ATTRIBUTE_VALUE; |
9076 | | |
9077 | | /* |
9078 | | * try to handle in this routine the most common case where no |
9079 | | * allocation of a new string is required and where content is |
9080 | | * pure ASCII. |
9081 | | */ |
9082 | 10.8M | limit = *in++; |
9083 | 10.8M | col++; |
9084 | 10.8M | end = ctxt->input->end; |
9085 | 10.8M | start = in; |
9086 | 10.8M | if (in >= end) { |
9087 | 919 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9088 | 919 | } |
9089 | 10.8M | if (normalize) { |
9090 | | /* |
9091 | | * Skip any leading spaces |
9092 | | */ |
9093 | 468k | while ((in < end) && (*in != limit) && |
9094 | 468k | ((*in == 0x20) || (*in == 0x9) || |
9095 | 466k | (*in == 0xA) || (*in == 0xD))) { |
9096 | 138k | if (*in == 0xA) { |
9097 | 21.6k | line++; col = 1; |
9098 | 116k | } else { |
9099 | 116k | col++; |
9100 | 116k | } |
9101 | 138k | in++; |
9102 | 138k | start = in; |
9103 | 138k | if (in >= end) { |
9104 | 140 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9105 | 140 | if ((in - start) > maxLength) { |
9106 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9107 | 0 | "AttValue length too long\n"); |
9108 | 0 | return(NULL); |
9109 | 0 | } |
9110 | 140 | } |
9111 | 138k | } |
9112 | 3.38M | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9113 | 3.38M | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9114 | 3.05M | col++; |
9115 | 3.05M | if ((*in++ == 0x20) && (*in == 0x20)) break; |
9116 | 3.05M | if (in >= end) { |
9117 | 249 | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9118 | 249 | if ((in - start) > maxLength) { |
9119 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9120 | 0 | "AttValue length too long\n"); |
9121 | 0 | return(NULL); |
9122 | 0 | } |
9123 | 249 | } |
9124 | 3.05M | } |
9125 | 330k | last = in; |
9126 | | /* |
9127 | | * skip the trailing blanks |
9128 | | */ |
9129 | 334k | while ((last[-1] == 0x20) && (last > start)) last--; |
9130 | 560k | while ((in < end) && (*in != limit) && |
9131 | 560k | ((*in == 0x20) || (*in == 0x9) || |
9132 | 256k | (*in == 0xA) || (*in == 0xD))) { |
9133 | 230k | if (*in == 0xA) { |
9134 | 13.1k | line++, col = 1; |
9135 | 217k | } else { |
9136 | 217k | col++; |
9137 | 217k | } |
9138 | 230k | in++; |
9139 | 230k | if (in >= end) { |
9140 | 222 | const xmlChar *oldbase = ctxt->input->base; |
9141 | 222 | GROW; |
9142 | 222 | if (ctxt->instate == XML_PARSER_EOF) |
9143 | 0 | return(NULL); |
9144 | 222 | if (oldbase != ctxt->input->base) { |
9145 | 0 | ptrdiff_t delta = ctxt->input->base - oldbase; |
9146 | 0 | start = start + delta; |
9147 | 0 | in = in + delta; |
9148 | 0 | last = last + delta; |
9149 | 0 | } |
9150 | 222 | end = ctxt->input->end; |
9151 | 222 | if ((in - start) > maxLength) { |
9152 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9153 | 0 | "AttValue length too long\n"); |
9154 | 0 | return(NULL); |
9155 | 0 | } |
9156 | 222 | } |
9157 | 230k | } |
9158 | 330k | if ((in - start) > maxLength) { |
9159 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9160 | 0 | "AttValue length too long\n"); |
9161 | 0 | return(NULL); |
9162 | 0 | } |
9163 | 330k | if (*in != limit) goto need_complex; |
9164 | 10.5M | } else { |
9165 | 118M | while ((in < end) && (*in != limit) && (*in >= 0x20) && |
9166 | 118M | (*in <= 0x7f) && (*in != '&') && (*in != '<')) { |
9167 | 108M | in++; |
9168 | 108M | col++; |
9169 | 108M | if (in >= end) { |
9170 | 4.28k | GROW_PARSE_ATT_VALUE_INTERNAL(ctxt, in, start, end) |
9171 | 4.28k | if ((in - start) > maxLength) { |
9172 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9173 | 0 | "AttValue length too long\n"); |
9174 | 0 | return(NULL); |
9175 | 0 | } |
9176 | 4.28k | } |
9177 | 108M | } |
9178 | 10.5M | last = in; |
9179 | 10.5M | if ((in - start) > maxLength) { |
9180 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_ATTRIBUTE_NOT_FINISHED, |
9181 | 0 | "AttValue length too long\n"); |
9182 | 0 | return(NULL); |
9183 | 0 | } |
9184 | 10.5M | if (*in != limit) goto need_complex; |
9185 | 10.5M | } |
9186 | 10.3M | in++; |
9187 | 10.3M | col++; |
9188 | 10.3M | if (len != NULL) { |
9189 | 6.97M | if (alloc) *alloc = 0; |
9190 | 6.97M | *len = last - start; |
9191 | 6.97M | ret = (xmlChar *) start; |
9192 | 6.97M | } else { |
9193 | 3.40M | if (alloc) *alloc = 1; |
9194 | 3.40M | ret = xmlStrndup(start, last - start); |
9195 | 3.40M | } |
9196 | 10.3M | CUR_PTR = in; |
9197 | 10.3M | ctxt->input->line = line; |
9198 | 10.3M | ctxt->input->col = col; |
9199 | 10.3M | return ret; |
9200 | 466k | need_complex: |
9201 | 466k | if (alloc) *alloc = 1; |
9202 | 466k | return xmlParseAttValueComplex(ctxt, len, normalize); |
9203 | 10.8M | } |
9204 | | |
9205 | | /** |
9206 | | * xmlParseAttribute2: |
9207 | | * @ctxt: an XML parser context |
9208 | | * @pref: the element prefix |
9209 | | * @elem: the element name |
9210 | | * @prefix: a xmlChar ** used to store the value of the attribute prefix |
9211 | | * @value: a xmlChar ** used to store the value of the attribute |
9212 | | * @len: an int * to save the length of the attribute |
9213 | | * @alloc: an int * to indicate if the attribute was allocated |
9214 | | * |
9215 | | * parse an attribute in the new SAX2 framework. |
9216 | | * |
9217 | | * Returns the attribute name, and the value in *value, . |
9218 | | */ |
9219 | | |
9220 | | static const xmlChar * |
9221 | | xmlParseAttribute2(xmlParserCtxtPtr ctxt, |
9222 | | const xmlChar * pref, const xmlChar * elem, |
9223 | | const xmlChar ** prefix, xmlChar ** value, |
9224 | | int *len, int *alloc) |
9225 | 7.42M | { |
9226 | 7.42M | const xmlChar *name; |
9227 | 7.42M | xmlChar *val, *internal_val = NULL; |
9228 | 7.42M | int normalize = 0; |
9229 | | |
9230 | 7.42M | *value = NULL; |
9231 | 7.42M | GROW; |
9232 | 7.42M | name = xmlParseQName(ctxt, prefix); |
9233 | 7.42M | if (name == NULL) { |
9234 | 119k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9235 | 119k | "error parsing attribute name\n"); |
9236 | 119k | return (NULL); |
9237 | 119k | } |
9238 | | |
9239 | | /* |
9240 | | * get the type if needed |
9241 | | */ |
9242 | 7.30M | if (ctxt->attsSpecial != NULL) { |
9243 | 851k | int type; |
9244 | | |
9245 | 851k | type = (int) (ptrdiff_t) xmlHashQLookup2(ctxt->attsSpecial, |
9246 | 851k | pref, elem, *prefix, name); |
9247 | 851k | if (type != 0) |
9248 | 331k | normalize = 1; |
9249 | 851k | } |
9250 | | |
9251 | | /* |
9252 | | * read the value |
9253 | | */ |
9254 | 7.30M | SKIP_BLANKS; |
9255 | 7.30M | if (RAW == '=') { |
9256 | 7.24M | NEXT; |
9257 | 7.24M | SKIP_BLANKS; |
9258 | 7.24M | val = xmlParseAttValueInternal(ctxt, len, alloc, normalize); |
9259 | 7.24M | if (val == NULL) |
9260 | 14.7k | return (NULL); |
9261 | 7.23M | if (normalize) { |
9262 | | /* |
9263 | | * Sometimes a second normalisation pass for spaces is needed |
9264 | | * but that only happens if charrefs or entities references |
9265 | | * have been used in the attribute value, i.e. the attribute |
9266 | | * value have been extracted in an allocated string already. |
9267 | | */ |
9268 | 330k | if (*alloc) { |
9269 | 26.5k | const xmlChar *val2; |
9270 | | |
9271 | 26.5k | val2 = xmlAttrNormalizeSpace2(ctxt, val, len); |
9272 | 26.5k | if ((val2 != NULL) && (val2 != val)) { |
9273 | 5.97k | xmlFree(val); |
9274 | 5.97k | val = (xmlChar *) val2; |
9275 | 5.97k | } |
9276 | 26.5k | } |
9277 | 330k | } |
9278 | 7.23M | ctxt->instate = XML_PARSER_CONTENT; |
9279 | 7.23M | } else { |
9280 | 61.5k | xmlFatalErrMsgStr(ctxt, XML_ERR_ATTRIBUTE_WITHOUT_VALUE, |
9281 | 61.5k | "Specification mandates value for attribute %s\n", |
9282 | 61.5k | name); |
9283 | 61.5k | return (name); |
9284 | 61.5k | } |
9285 | | |
9286 | 7.23M | if (*prefix == ctxt->str_xml) { |
9287 | | /* |
9288 | | * Check that xml:lang conforms to the specification |
9289 | | * No more registered as an error, just generate a warning now |
9290 | | * since this was deprecated in XML second edition |
9291 | | */ |
9292 | 49.2k | if ((ctxt->pedantic) && (xmlStrEqual(name, BAD_CAST "lang"))) { |
9293 | 11.1k | internal_val = xmlStrndup(val, *len); |
9294 | 11.1k | if (!xmlCheckLanguageID(internal_val)) { |
9295 | 6.77k | xmlWarningMsg(ctxt, XML_WAR_LANG_VALUE, |
9296 | 6.77k | "Malformed value for xml:lang : %s\n", |
9297 | 6.77k | internal_val, NULL); |
9298 | 6.77k | } |
9299 | 11.1k | } |
9300 | | |
9301 | | /* |
9302 | | * Check that xml:space conforms to the specification |
9303 | | */ |
9304 | 49.2k | if (xmlStrEqual(name, BAD_CAST "space")) { |
9305 | 1.02k | internal_val = xmlStrndup(val, *len); |
9306 | 1.02k | if (xmlStrEqual(internal_val, BAD_CAST "default")) |
9307 | 133 | *(ctxt->space) = 0; |
9308 | 896 | else if (xmlStrEqual(internal_val, BAD_CAST "preserve")) |
9309 | 397 | *(ctxt->space) = 1; |
9310 | 499 | else { |
9311 | 499 | xmlWarningMsg(ctxt, XML_WAR_SPACE_VALUE, |
9312 | 499 | "Invalid value \"%s\" for xml:space : \"default\" or \"preserve\" expected\n", |
9313 | 499 | internal_val, NULL); |
9314 | 499 | } |
9315 | 1.02k | } |
9316 | 49.2k | if (internal_val) { |
9317 | 12.2k | xmlFree(internal_val); |
9318 | 12.2k | } |
9319 | 49.2k | } |
9320 | | |
9321 | 7.23M | *value = val; |
9322 | 7.23M | return (name); |
9323 | 7.30M | } |
9324 | | /** |
9325 | | * xmlParseStartTag2: |
9326 | | * @ctxt: an XML parser context |
9327 | | * |
9328 | | * Parse a start tag. Always consumes '<'. |
9329 | | * |
9330 | | * This routine is called when running SAX2 parsing |
9331 | | * |
9332 | | * [40] STag ::= '<' Name (S Attribute)* S? '>' |
9333 | | * |
9334 | | * [ WFC: Unique Att Spec ] |
9335 | | * No attribute name may appear more than once in the same start-tag or |
9336 | | * empty-element tag. |
9337 | | * |
9338 | | * [44] EmptyElemTag ::= '<' Name (S Attribute)* S? '/>' |
9339 | | * |
9340 | | * [ WFC: Unique Att Spec ] |
9341 | | * No attribute name may appear more than once in the same start-tag or |
9342 | | * empty-element tag. |
9343 | | * |
9344 | | * With namespace: |
9345 | | * |
9346 | | * [NS 8] STag ::= '<' QName (S Attribute)* S? '>' |
9347 | | * |
9348 | | * [NS 10] EmptyElement ::= '<' QName (S Attribute)* S? '/>' |
9349 | | * |
9350 | | * Returns the element name parsed |
9351 | | */ |
9352 | | |
9353 | | static const xmlChar * |
9354 | | xmlParseStartTag2(xmlParserCtxtPtr ctxt, const xmlChar **pref, |
9355 | 6.89M | const xmlChar **URI, int *tlen) { |
9356 | 6.89M | const xmlChar *localname; |
9357 | 6.89M | const xmlChar *prefix; |
9358 | 6.89M | const xmlChar *attname; |
9359 | 6.89M | const xmlChar *aprefix; |
9360 | 6.89M | const xmlChar *nsname; |
9361 | 6.89M | xmlChar *attvalue; |
9362 | 6.89M | const xmlChar **atts = ctxt->atts; |
9363 | 6.89M | int maxatts = ctxt->maxatts; |
9364 | 6.89M | int nratts, nbatts, nbdef, inputid; |
9365 | 6.89M | int i, j, nbNs, attval; |
9366 | 6.89M | unsigned long cur; |
9367 | 6.89M | int nsNr = ctxt->nsNr; |
9368 | | |
9369 | 6.89M | if (RAW != '<') return(NULL); |
9370 | 6.89M | NEXT1; |
9371 | | |
9372 | | /* |
9373 | | * NOTE: it is crucial with the SAX2 API to never call SHRINK beyond that |
9374 | | * point since the attribute values may be stored as pointers to |
9375 | | * the buffer and calling SHRINK would destroy them ! |
9376 | | * The Shrinking is only possible once the full set of attribute |
9377 | | * callbacks have been done. |
9378 | | */ |
9379 | 6.89M | SHRINK; |
9380 | 6.89M | cur = ctxt->input->cur - ctxt->input->base; |
9381 | 6.89M | inputid = ctxt->input->id; |
9382 | 6.89M | nbatts = 0; |
9383 | 6.89M | nratts = 0; |
9384 | 6.89M | nbdef = 0; |
9385 | 6.89M | nbNs = 0; |
9386 | 6.89M | attval = 0; |
9387 | | /* Forget any namespaces added during an earlier parse of this element. */ |
9388 | 6.89M | ctxt->nsNr = nsNr; |
9389 | | |
9390 | 6.89M | localname = xmlParseQName(ctxt, &prefix); |
9391 | 6.89M | if (localname == NULL) { |
9392 | 212k | xmlFatalErrMsg(ctxt, XML_ERR_NAME_REQUIRED, |
9393 | 212k | "StartTag: invalid element name\n"); |
9394 | 212k | return(NULL); |
9395 | 212k | } |
9396 | 6.68M | *tlen = ctxt->input->cur - ctxt->input->base - cur; |
9397 | | |
9398 | | /* |
9399 | | * Now parse the attributes, it ends up with the ending |
9400 | | * |
9401 | | * (S Attribute)* S? |
9402 | | */ |
9403 | 6.68M | SKIP_BLANKS; |
9404 | 6.68M | GROW; |
9405 | | |
9406 | 9.62M | while (((RAW != '>') && |
9407 | 9.62M | ((RAW != '/') || (NXT(1) != '>')) && |
9408 | 9.62M | (IS_BYTE_CHAR(RAW))) && (ctxt->instate != XML_PARSER_EOF)) { |
9409 | 7.42M | int len = -1, alloc = 0; |
9410 | | |
9411 | 7.42M | attname = xmlParseAttribute2(ctxt, prefix, localname, |
9412 | 7.42M | &aprefix, &attvalue, &len, &alloc); |
9413 | 7.42M | if (attname == NULL) { |
9414 | 133k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9415 | 133k | "xmlParseStartTag: problem parsing attributes\n"); |
9416 | 133k | break; |
9417 | 133k | } |
9418 | 7.29M | if (attvalue == NULL) |
9419 | 61.5k | goto next_attr; |
9420 | 7.23M | if (len < 0) len = xmlStrlen(attvalue); |
9421 | | |
9422 | 7.23M | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9423 | 58.1k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9424 | 58.1k | xmlURIPtr uri; |
9425 | | |
9426 | 58.1k | if (URL == NULL) { |
9427 | 0 | xmlErrMemory(ctxt, "dictionary allocation failure"); |
9428 | 0 | if ((attvalue != NULL) && (alloc != 0)) |
9429 | 0 | xmlFree(attvalue); |
9430 | 0 | localname = NULL; |
9431 | 0 | goto done; |
9432 | 0 | } |
9433 | 58.1k | if (*URL != 0) { |
9434 | 53.7k | uri = xmlParseURI((const char *) URL); |
9435 | 53.7k | if (uri == NULL) { |
9436 | 21.0k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9437 | 21.0k | "xmlns: '%s' is not a valid URI\n", |
9438 | 21.0k | URL, NULL, NULL); |
9439 | 32.7k | } else { |
9440 | 32.7k | if (uri->scheme == NULL) { |
9441 | 6.74k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9442 | 6.74k | "xmlns: URI %s is not absolute\n", |
9443 | 6.74k | URL, NULL, NULL); |
9444 | 6.74k | } |
9445 | 32.7k | xmlFreeURI(uri); |
9446 | 32.7k | } |
9447 | 53.7k | if (URL == ctxt->str_xml_ns) { |
9448 | 0 | if (attname != ctxt->str_xml) { |
9449 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9450 | 0 | "xml namespace URI cannot be the default namespace\n", |
9451 | 0 | NULL, NULL, NULL); |
9452 | 0 | } |
9453 | 0 | goto next_attr; |
9454 | 0 | } |
9455 | 53.7k | if ((len == 29) && |
9456 | 53.7k | (xmlStrEqual(URL, |
9457 | 1.94k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9458 | 641 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9459 | 641 | "reuse of the xmlns namespace name is forbidden\n", |
9460 | 641 | NULL, NULL, NULL); |
9461 | 641 | goto next_attr; |
9462 | 641 | } |
9463 | 53.7k | } |
9464 | | /* |
9465 | | * check that it's not a defined namespace |
9466 | | */ |
9467 | 75.2k | for (j = 1;j <= nbNs;j++) |
9468 | 26.6k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9469 | 8.93k | break; |
9470 | 57.5k | if (j <= nbNs) |
9471 | 8.93k | xmlErrAttributeDup(ctxt, NULL, attname); |
9472 | 48.6k | else |
9473 | 48.6k | if (nsPush(ctxt, NULL, URL) > 0) nbNs++; |
9474 | | |
9475 | 7.17M | } else if (aprefix == ctxt->str_xmlns) { |
9476 | 94.1k | const xmlChar *URL = xmlDictLookup(ctxt->dict, attvalue, len); |
9477 | 94.1k | xmlURIPtr uri; |
9478 | | |
9479 | 94.1k | if (attname == ctxt->str_xml) { |
9480 | 1.43k | if (URL != ctxt->str_xml_ns) { |
9481 | 1.43k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9482 | 1.43k | "xml namespace prefix mapped to wrong URI\n", |
9483 | 1.43k | NULL, NULL, NULL); |
9484 | 1.43k | } |
9485 | | /* |
9486 | | * Do not keep a namespace definition node |
9487 | | */ |
9488 | 1.43k | goto next_attr; |
9489 | 1.43k | } |
9490 | 92.7k | if (URL == ctxt->str_xml_ns) { |
9491 | 0 | if (attname != ctxt->str_xml) { |
9492 | 0 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9493 | 0 | "xml namespace URI mapped to wrong prefix\n", |
9494 | 0 | NULL, NULL, NULL); |
9495 | 0 | } |
9496 | 0 | goto next_attr; |
9497 | 0 | } |
9498 | 92.7k | if (attname == ctxt->str_xmlns) { |
9499 | 369 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9500 | 369 | "redefinition of the xmlns prefix is forbidden\n", |
9501 | 369 | NULL, NULL, NULL); |
9502 | 369 | goto next_attr; |
9503 | 369 | } |
9504 | 92.3k | if ((len == 29) && |
9505 | 92.3k | (xmlStrEqual(URL, |
9506 | 2.15k | BAD_CAST "http://www.w3.org/2000/xmlns/"))) { |
9507 | 170 | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9508 | 170 | "reuse of the xmlns namespace name is forbidden\n", |
9509 | 170 | NULL, NULL, NULL); |
9510 | 170 | goto next_attr; |
9511 | 170 | } |
9512 | 92.2k | if ((URL == NULL) || (URL[0] == 0)) { |
9513 | 1.35k | xmlNsErr(ctxt, XML_NS_ERR_XML_NAMESPACE, |
9514 | 1.35k | "xmlns:%s: Empty XML namespace is not allowed\n", |
9515 | 1.35k | attname, NULL, NULL); |
9516 | 1.35k | goto next_attr; |
9517 | 90.8k | } else { |
9518 | 90.8k | uri = xmlParseURI((const char *) URL); |
9519 | 90.8k | if (uri == NULL) { |
9520 | 22.5k | xmlNsErr(ctxt, XML_WAR_NS_URI, |
9521 | 22.5k | "xmlns:%s: '%s' is not a valid URI\n", |
9522 | 22.5k | attname, URL, NULL); |
9523 | 68.2k | } else { |
9524 | 68.2k | if ((ctxt->pedantic) && (uri->scheme == NULL)) { |
9525 | 2.24k | xmlNsWarn(ctxt, XML_WAR_NS_URI_RELATIVE, |
9526 | 2.24k | "xmlns:%s: URI %s is not absolute\n", |
9527 | 2.24k | attname, URL, NULL); |
9528 | 2.24k | } |
9529 | 68.2k | xmlFreeURI(uri); |
9530 | 68.2k | } |
9531 | 90.8k | } |
9532 | | |
9533 | | /* |
9534 | | * check that it's not a defined namespace |
9535 | | */ |
9536 | 147k | for (j = 1;j <= nbNs;j++) |
9537 | 61.2k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9538 | 4.63k | break; |
9539 | 90.8k | if (j <= nbNs) |
9540 | 4.63k | xmlErrAttributeDup(ctxt, aprefix, attname); |
9541 | 86.2k | else |
9542 | 86.2k | if (nsPush(ctxt, attname, URL) > 0) nbNs++; |
9543 | | |
9544 | 7.07M | } else { |
9545 | | /* |
9546 | | * Add the pair to atts |
9547 | | */ |
9548 | 7.07M | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9549 | 103k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9550 | 0 | goto next_attr; |
9551 | 0 | } |
9552 | 103k | maxatts = ctxt->maxatts; |
9553 | 103k | atts = ctxt->atts; |
9554 | 103k | } |
9555 | 7.07M | ctxt->attallocs[nratts++] = alloc; |
9556 | 7.07M | atts[nbatts++] = attname; |
9557 | 7.07M | atts[nbatts++] = aprefix; |
9558 | | /* |
9559 | | * The namespace URI field is used temporarily to point at the |
9560 | | * base of the current input buffer for non-alloced attributes. |
9561 | | * When the input buffer is reallocated, all the pointers become |
9562 | | * invalid, but they can be reconstructed later. |
9563 | | */ |
9564 | 7.07M | if (alloc) |
9565 | 213k | atts[nbatts++] = NULL; |
9566 | 6.86M | else |
9567 | 6.86M | atts[nbatts++] = ctxt->input->base; |
9568 | 7.07M | atts[nbatts++] = attvalue; |
9569 | 7.07M | attvalue += len; |
9570 | 7.07M | atts[nbatts++] = attvalue; |
9571 | | /* |
9572 | | * tag if some deallocation is needed |
9573 | | */ |
9574 | 7.07M | if (alloc != 0) attval = 1; |
9575 | 7.07M | attvalue = NULL; /* moved into atts */ |
9576 | 7.07M | } |
9577 | | |
9578 | 7.29M | next_attr: |
9579 | 7.29M | if ((attvalue != NULL) && (alloc != 0)) { |
9580 | 42.9k | xmlFree(attvalue); |
9581 | 42.9k | attvalue = NULL; |
9582 | 42.9k | } |
9583 | | |
9584 | 7.29M | GROW |
9585 | 7.29M | if (ctxt->instate == XML_PARSER_EOF) |
9586 | 0 | break; |
9587 | 7.29M | if ((RAW == '>') || (((RAW == '/') && (NXT(1) == '>')))) |
9588 | 4.17M | break; |
9589 | 3.12M | if (SKIP_BLANKS == 0) { |
9590 | 180k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
9591 | 180k | "attributes construct error\n"); |
9592 | 180k | break; |
9593 | 180k | } |
9594 | 2.93M | GROW; |
9595 | 2.93M | } |
9596 | | |
9597 | 6.68M | if (ctxt->input->id != inputid) { |
9598 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
9599 | 0 | "Unexpected change of input\n"); |
9600 | 0 | localname = NULL; |
9601 | 0 | goto done; |
9602 | 0 | } |
9603 | | |
9604 | | /* Reconstruct attribute value pointers. */ |
9605 | 13.7M | for (i = 0, j = 0; j < nratts; i += 5, j++) { |
9606 | 7.07M | if (atts[i+2] != NULL) { |
9607 | | /* |
9608 | | * Arithmetic on dangling pointers is technically undefined |
9609 | | * behavior, but well... |
9610 | | */ |
9611 | 6.86M | const xmlChar *old = atts[i+2]; |
9612 | 6.86M | atts[i+2] = NULL; /* Reset repurposed namespace URI */ |
9613 | 6.86M | atts[i+3] = ctxt->input->base + (atts[i+3] - old); /* value */ |
9614 | 6.86M | atts[i+4] = ctxt->input->base + (atts[i+4] - old); /* valuend */ |
9615 | 6.86M | } |
9616 | 7.07M | } |
9617 | | |
9618 | | /* |
9619 | | * The attributes defaulting |
9620 | | */ |
9621 | 6.68M | if (ctxt->attsDefault != NULL) { |
9622 | 1.11M | xmlDefAttrsPtr defaults; |
9623 | | |
9624 | 1.11M | defaults = xmlHashLookup2(ctxt->attsDefault, localname, prefix); |
9625 | 1.11M | if (defaults != NULL) { |
9626 | 368k | for (i = 0;i < defaults->nbAttrs;i++) { |
9627 | 257k | attname = defaults->values[5 * i]; |
9628 | 257k | aprefix = defaults->values[5 * i + 1]; |
9629 | | |
9630 | | /* |
9631 | | * special work for namespaces defaulted defs |
9632 | | */ |
9633 | 257k | if ((attname == ctxt->str_xmlns) && (aprefix == NULL)) { |
9634 | | /* |
9635 | | * check that it's not a defined namespace |
9636 | | */ |
9637 | 12.8k | for (j = 1;j <= nbNs;j++) |
9638 | 3.68k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == NULL) |
9639 | 1.62k | break; |
9640 | 10.8k | if (j <= nbNs) continue; |
9641 | | |
9642 | 9.17k | nsname = xmlGetNamespace(ctxt, NULL); |
9643 | 9.17k | if (nsname != defaults->values[5 * i + 2]) { |
9644 | 5.10k | if (nsPush(ctxt, NULL, |
9645 | 5.10k | defaults->values[5 * i + 2]) > 0) |
9646 | 4.65k | nbNs++; |
9647 | 5.10k | } |
9648 | 246k | } else if (aprefix == ctxt->str_xmlns) { |
9649 | | /* |
9650 | | * check that it's not a defined namespace |
9651 | | */ |
9652 | 29.7k | for (j = 1;j <= nbNs;j++) |
9653 | 6.36k | if (ctxt->nsTab[ctxt->nsNr - 2 * j] == attname) |
9654 | 3.14k | break; |
9655 | 26.5k | if (j <= nbNs) continue; |
9656 | | |
9657 | 23.3k | nsname = xmlGetNamespace(ctxt, attname); |
9658 | 23.3k | if (nsname != defaults->values[5 * i + 2]) { |
9659 | 12.7k | if (nsPush(ctxt, attname, |
9660 | 12.7k | defaults->values[5 * i + 2]) > 0) |
9661 | 12.1k | nbNs++; |
9662 | 12.7k | } |
9663 | 220k | } else { |
9664 | | /* |
9665 | | * check that it's not a defined attribute |
9666 | | */ |
9667 | 606k | for (j = 0;j < nbatts;j+=5) { |
9668 | 388k | if ((attname == atts[j]) && (aprefix == atts[j+1])) |
9669 | 2.34k | break; |
9670 | 388k | } |
9671 | 220k | if (j < nbatts) continue; |
9672 | | |
9673 | 217k | if ((atts == NULL) || (nbatts + 5 > maxatts)) { |
9674 | 5.99k | if (xmlCtxtGrowAttrs(ctxt, nbatts + 5) < 0) { |
9675 | 0 | localname = NULL; |
9676 | 0 | goto done; |
9677 | 0 | } |
9678 | 5.99k | maxatts = ctxt->maxatts; |
9679 | 5.99k | atts = ctxt->atts; |
9680 | 5.99k | } |
9681 | 217k | atts[nbatts++] = attname; |
9682 | 217k | atts[nbatts++] = aprefix; |
9683 | 217k | if (aprefix == NULL) |
9684 | 160k | atts[nbatts++] = NULL; |
9685 | 57.8k | else |
9686 | 57.8k | atts[nbatts++] = xmlGetNamespace(ctxt, aprefix); |
9687 | 217k | atts[nbatts++] = defaults->values[5 * i + 2]; |
9688 | 217k | atts[nbatts++] = defaults->values[5 * i + 3]; |
9689 | 217k | if ((ctxt->standalone == 1) && |
9690 | 217k | (defaults->values[5 * i + 4] != NULL)) { |
9691 | 0 | xmlValidityError(ctxt, XML_DTD_STANDALONE_DEFAULTED, |
9692 | 0 | "standalone: attribute %s on %s defaulted from external subset\n", |
9693 | 0 | attname, localname); |
9694 | 0 | } |
9695 | 217k | nbdef++; |
9696 | 217k | } |
9697 | 257k | } |
9698 | 111k | } |
9699 | 1.11M | } |
9700 | | |
9701 | | /* |
9702 | | * The attributes checkings |
9703 | | */ |
9704 | 13.9M | for (i = 0; i < nbatts;i += 5) { |
9705 | | /* |
9706 | | * The default namespace does not apply to attribute names. |
9707 | | */ |
9708 | 7.29M | if (atts[i + 1] != NULL) { |
9709 | 289k | nsname = xmlGetNamespace(ctxt, atts[i + 1]); |
9710 | 289k | if (nsname == NULL) { |
9711 | 125k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9712 | 125k | "Namespace prefix %s for %s on %s is not defined\n", |
9713 | 125k | atts[i + 1], atts[i], localname); |
9714 | 125k | } |
9715 | 289k | atts[i + 2] = nsname; |
9716 | 289k | } else |
9717 | 7.00M | nsname = NULL; |
9718 | | /* |
9719 | | * [ WFC: Unique Att Spec ] |
9720 | | * No attribute name may appear more than once in the same |
9721 | | * start-tag or empty-element tag. |
9722 | | * As extended by the Namespace in XML REC. |
9723 | | */ |
9724 | 10.6M | for (j = 0; j < i;j += 5) { |
9725 | 3.31M | if (atts[i] == atts[j]) { |
9726 | 29.4k | if (atts[i+1] == atts[j+1]) { |
9727 | 10.1k | xmlErrAttributeDup(ctxt, atts[i+1], atts[i]); |
9728 | 10.1k | break; |
9729 | 10.1k | } |
9730 | 19.2k | if ((nsname != NULL) && (atts[j + 2] == nsname)) { |
9731 | 1.23k | xmlNsErr(ctxt, XML_NS_ERR_ATTRIBUTE_REDEFINED, |
9732 | 1.23k | "Namespaced Attribute %s in '%s' redefined\n", |
9733 | 1.23k | atts[i], nsname, NULL); |
9734 | 1.23k | break; |
9735 | 1.23k | } |
9736 | 19.2k | } |
9737 | 3.31M | } |
9738 | 7.29M | } |
9739 | | |
9740 | 6.68M | nsname = xmlGetNamespace(ctxt, prefix); |
9741 | 6.68M | if ((prefix != NULL) && (nsname == NULL)) { |
9742 | 176k | xmlNsErr(ctxt, XML_NS_ERR_UNDEFINED_NAMESPACE, |
9743 | 176k | "Namespace prefix %s on %s is not defined\n", |
9744 | 176k | prefix, localname, NULL); |
9745 | 176k | } |
9746 | 6.68M | *pref = prefix; |
9747 | 6.68M | *URI = nsname; |
9748 | | |
9749 | | /* |
9750 | | * SAX: Start of Element ! |
9751 | | */ |
9752 | 6.68M | if ((ctxt->sax != NULL) && (ctxt->sax->startElementNs != NULL) && |
9753 | 6.68M | (!ctxt->disableSAX)) { |
9754 | 5.82M | if (nbNs > 0) |
9755 | 79.8k | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9756 | 79.8k | nsname, nbNs, &ctxt->nsTab[ctxt->nsNr - 2 * nbNs], |
9757 | 79.8k | nbatts / 5, nbdef, atts); |
9758 | 5.74M | else |
9759 | 5.74M | ctxt->sax->startElementNs(ctxt->userData, localname, prefix, |
9760 | 5.74M | nsname, 0, NULL, nbatts / 5, nbdef, atts); |
9761 | 5.82M | } |
9762 | | |
9763 | 6.68M | done: |
9764 | | /* |
9765 | | * Free up attribute allocated strings if needed |
9766 | | */ |
9767 | 6.68M | if (attval != 0) { |
9768 | 456k | for (i = 3,j = 0; j < nratts;i += 5,j++) |
9769 | 261k | if ((ctxt->attallocs[j] != 0) && (atts[i] != NULL)) |
9770 | 213k | xmlFree((xmlChar *) atts[i]); |
9771 | 195k | } |
9772 | | |
9773 | 6.68M | return(localname); |
9774 | 6.68M | } |
9775 | | |
9776 | | /** |
9777 | | * xmlParseEndTag2: |
9778 | | * @ctxt: an XML parser context |
9779 | | * @line: line of the start tag |
9780 | | * @nsNr: number of namespaces on the start tag |
9781 | | * |
9782 | | * Parse an end tag. Always consumes '</'. |
9783 | | * |
9784 | | * [42] ETag ::= '</' Name S? '>' |
9785 | | * |
9786 | | * With namespace |
9787 | | * |
9788 | | * [NS 9] ETag ::= '</' QName S? '>' |
9789 | | */ |
9790 | | |
9791 | | static void |
9792 | 3.07M | xmlParseEndTag2(xmlParserCtxtPtr ctxt, const xmlStartTag *tag) { |
9793 | 3.07M | const xmlChar *name; |
9794 | | |
9795 | 3.07M | GROW; |
9796 | 3.07M | if ((RAW != '<') || (NXT(1) != '/')) { |
9797 | 0 | xmlFatalErr(ctxt, XML_ERR_LTSLASH_REQUIRED, NULL); |
9798 | 0 | return; |
9799 | 0 | } |
9800 | 3.07M | SKIP(2); |
9801 | | |
9802 | 3.07M | if (tag->prefix == NULL) |
9803 | 2.89M | name = xmlParseNameAndCompare(ctxt, ctxt->name); |
9804 | 184k | else |
9805 | 184k | name = xmlParseQNameAndCompare(ctxt, ctxt->name, tag->prefix); |
9806 | | |
9807 | | /* |
9808 | | * We should definitely be at the ending "S? '>'" part |
9809 | | */ |
9810 | 3.07M | GROW; |
9811 | 3.07M | if (ctxt->instate == XML_PARSER_EOF) |
9812 | 0 | return; |
9813 | 3.07M | SKIP_BLANKS; |
9814 | 3.07M | if ((!IS_BYTE_CHAR(RAW)) || (RAW != '>')) { |
9815 | 40.5k | xmlFatalErr(ctxt, XML_ERR_GT_REQUIRED, NULL); |
9816 | 40.5k | } else |
9817 | 3.03M | NEXT1; |
9818 | | |
9819 | | /* |
9820 | | * [ WFC: Element Type Match ] |
9821 | | * The Name in an element's end-tag must match the element type in the |
9822 | | * start-tag. |
9823 | | * |
9824 | | */ |
9825 | 3.07M | if (name != (xmlChar*)1) { |
9826 | 126k | if (name == NULL) name = BAD_CAST "unparsable"; |
9827 | 126k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NAME_MISMATCH, |
9828 | 126k | "Opening and ending tag mismatch: %s line %d and %s\n", |
9829 | 126k | ctxt->name, tag->line, name); |
9830 | 126k | } |
9831 | | |
9832 | | /* |
9833 | | * SAX: End of Tag |
9834 | | */ |
9835 | 3.07M | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
9836 | 3.07M | (!ctxt->disableSAX)) |
9837 | 2.60M | ctxt->sax->endElementNs(ctxt->userData, ctxt->name, tag->prefix, |
9838 | 2.60M | tag->URI); |
9839 | | |
9840 | 3.07M | spacePop(ctxt); |
9841 | 3.07M | if (tag->nsNr != 0) |
9842 | 16.3k | nsPop(ctxt, tag->nsNr); |
9843 | 3.07M | } |
9844 | | |
9845 | | /** |
9846 | | * xmlParseCDSect: |
9847 | | * @ctxt: an XML parser context |
9848 | | * |
9849 | | * DEPRECATED: Internal function, don't use. |
9850 | | * |
9851 | | * Parse escaped pure raw content. Always consumes '<!['. |
9852 | | * |
9853 | | * [18] CDSect ::= CDStart CData CDEnd |
9854 | | * |
9855 | | * [19] CDStart ::= '<![CDATA[' |
9856 | | * |
9857 | | * [20] Data ::= (Char* - (Char* ']]>' Char*)) |
9858 | | * |
9859 | | * [21] CDEnd ::= ']]>' |
9860 | | */ |
9861 | | void |
9862 | 26.9k | xmlParseCDSect(xmlParserCtxtPtr ctxt) { |
9863 | 26.9k | xmlChar *buf = NULL; |
9864 | 26.9k | int len = 0; |
9865 | 26.9k | int size = XML_PARSER_BUFFER_SIZE; |
9866 | 26.9k | int r, rl; |
9867 | 26.9k | int s, sl; |
9868 | 26.9k | int cur, l; |
9869 | 26.9k | int count = 0; |
9870 | 26.9k | int maxLength = (ctxt->options & XML_PARSE_HUGE) ? |
9871 | 9.99k | XML_MAX_HUGE_LENGTH : |
9872 | 26.9k | XML_MAX_TEXT_LENGTH; |
9873 | | |
9874 | 26.9k | if ((CUR != '<') || (NXT(1) != '!') || (NXT(2) != '[')) |
9875 | 0 | return; |
9876 | 26.9k | SKIP(3); |
9877 | | |
9878 | 26.9k | if (!CMP6(CUR_PTR, 'C', 'D', 'A', 'T', 'A', '[')) |
9879 | 0 | return; |
9880 | 26.9k | SKIP(6); |
9881 | | |
9882 | 26.9k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
9883 | 26.9k | r = CUR_CHAR(rl); |
9884 | 26.9k | if (!IS_CHAR(r)) { |
9885 | 1.05k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9886 | 1.05k | goto out; |
9887 | 1.05k | } |
9888 | 25.8k | NEXTL(rl); |
9889 | 25.8k | s = CUR_CHAR(sl); |
9890 | 25.8k | if (!IS_CHAR(s)) { |
9891 | 1.17k | xmlFatalErr(ctxt, XML_ERR_CDATA_NOT_FINISHED, NULL); |
9892 | 1.17k | goto out; |
9893 | 1.17k | } |
9894 | 24.7k | NEXTL(sl); |
9895 | 24.7k | cur = CUR_CHAR(l); |
9896 | 24.7k | buf = (xmlChar *) xmlMallocAtomic(size); |
9897 | 24.7k | if (buf == NULL) { |
9898 | 0 | xmlErrMemory(ctxt, NULL); |
9899 | 0 | goto out; |
9900 | 0 | } |
9901 | 13.5M | while (IS_CHAR(cur) && |
9902 | 13.5M | ((r != ']') || (s != ']') || (cur != '>'))) { |
9903 | 13.4M | if (len + 5 >= size) { |
9904 | 26.3k | xmlChar *tmp; |
9905 | | |
9906 | 26.3k | tmp = (xmlChar *) xmlRealloc(buf, size * 2); |
9907 | 26.3k | if (tmp == NULL) { |
9908 | 0 | xmlErrMemory(ctxt, NULL); |
9909 | 0 | goto out; |
9910 | 0 | } |
9911 | 26.3k | buf = tmp; |
9912 | 26.3k | size *= 2; |
9913 | 26.3k | } |
9914 | 13.4M | COPY_BUF(rl,buf,len,r); |
9915 | 13.4M | r = s; |
9916 | 13.4M | rl = sl; |
9917 | 13.4M | s = cur; |
9918 | 13.4M | sl = l; |
9919 | 13.4M | count++; |
9920 | 13.4M | if (count > 50) { |
9921 | 254k | SHRINK; |
9922 | 254k | GROW; |
9923 | 254k | if (ctxt->instate == XML_PARSER_EOF) { |
9924 | 0 | goto out; |
9925 | 0 | } |
9926 | 254k | count = 0; |
9927 | 254k | } |
9928 | 13.4M | NEXTL(l); |
9929 | 13.4M | cur = CUR_CHAR(l); |
9930 | 13.4M | if (len > maxLength) { |
9931 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9932 | 0 | "CData section too big found\n"); |
9933 | 0 | goto out; |
9934 | 0 | } |
9935 | 13.4M | } |
9936 | 24.7k | buf[len] = 0; |
9937 | 24.7k | if (cur != '>') { |
9938 | 4.55k | xmlFatalErrMsgStr(ctxt, XML_ERR_CDATA_NOT_FINISHED, |
9939 | 4.55k | "CData section not finished\n%.50s\n", buf); |
9940 | 4.55k | goto out; |
9941 | 4.55k | } |
9942 | 20.1k | NEXTL(l); |
9943 | | |
9944 | | /* |
9945 | | * OK the buffer is to be consumed as cdata. |
9946 | | */ |
9947 | 20.1k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
9948 | 14.5k | if (ctxt->sax->cdataBlock != NULL) |
9949 | 9.63k | ctxt->sax->cdataBlock(ctxt->userData, buf, len); |
9950 | 4.87k | else if (ctxt->sax->characters != NULL) |
9951 | 4.87k | ctxt->sax->characters(ctxt->userData, buf, len); |
9952 | 14.5k | } |
9953 | | |
9954 | 26.9k | out: |
9955 | 26.9k | if (ctxt->instate != XML_PARSER_EOF) |
9956 | 26.9k | ctxt->instate = XML_PARSER_CONTENT; |
9957 | 26.9k | xmlFree(buf); |
9958 | 26.9k | } |
9959 | | |
9960 | | /** |
9961 | | * xmlParseContentInternal: |
9962 | | * @ctxt: an XML parser context |
9963 | | * |
9964 | | * Parse a content sequence. Stops at EOF or '</'. Leaves checking of |
9965 | | * unexpected EOF to the caller. |
9966 | | */ |
9967 | | |
9968 | | static void |
9969 | 249k | xmlParseContentInternal(xmlParserCtxtPtr ctxt) { |
9970 | 249k | int nameNr = ctxt->nameNr; |
9971 | | |
9972 | 249k | GROW; |
9973 | 17.6M | while ((RAW != 0) && |
9974 | 17.6M | (ctxt->instate != XML_PARSER_EOF)) { |
9975 | 17.4M | const xmlChar *cur = ctxt->input->cur; |
9976 | | |
9977 | | /* |
9978 | | * First case : a Processing Instruction. |
9979 | | */ |
9980 | 17.4M | if ((*cur == '<') && (cur[1] == '?')) { |
9981 | 40.1k | xmlParsePI(ctxt); |
9982 | 40.1k | } |
9983 | | |
9984 | | /* |
9985 | | * Second case : a CDSection |
9986 | | */ |
9987 | | /* 2.6.0 test was *cur not RAW */ |
9988 | 17.4M | else if (CMP9(CUR_PTR, '<', '!', '[', 'C', 'D', 'A', 'T', 'A', '[')) { |
9989 | 26.9k | xmlParseCDSect(ctxt); |
9990 | 26.9k | } |
9991 | | |
9992 | | /* |
9993 | | * Third case : a comment |
9994 | | */ |
9995 | 17.3M | else if ((*cur == '<') && (NXT(1) == '!') && |
9996 | 17.3M | (NXT(2) == '-') && (NXT(3) == '-')) { |
9997 | 159k | xmlParseComment(ctxt); |
9998 | 159k | ctxt->instate = XML_PARSER_CONTENT; |
9999 | 159k | } |
10000 | | |
10001 | | /* |
10002 | | * Fourth case : a sub-element. |
10003 | | */ |
10004 | 17.2M | else if (*cur == '<') { |
10005 | 7.05M | if (NXT(1) == '/') { |
10006 | 2.31M | if (ctxt->nameNr <= nameNr) |
10007 | 38.9k | break; |
10008 | 2.28M | xmlParseElementEnd(ctxt); |
10009 | 4.73M | } else { |
10010 | 4.73M | xmlParseElementStart(ctxt); |
10011 | 4.73M | } |
10012 | 7.05M | } |
10013 | | |
10014 | | /* |
10015 | | * Fifth case : a reference. If if has not been resolved, |
10016 | | * parsing returns it's Name, create the node |
10017 | | */ |
10018 | | |
10019 | 10.1M | else if (*cur == '&') { |
10020 | 2.55M | xmlParseReference(ctxt); |
10021 | 2.55M | } |
10022 | | |
10023 | | /* |
10024 | | * Last case, text. Note that References are handled directly. |
10025 | | */ |
10026 | 7.61M | else { |
10027 | 7.61M | xmlParseCharData(ctxt, 0); |
10028 | 7.61M | } |
10029 | | |
10030 | 17.4M | GROW; |
10031 | 17.4M | SHRINK; |
10032 | 17.4M | } |
10033 | 249k | } |
10034 | | |
10035 | | /** |
10036 | | * xmlParseContent: |
10037 | | * @ctxt: an XML parser context |
10038 | | * |
10039 | | * Parse a content sequence. Stops at EOF or '</'. |
10040 | | * |
10041 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
10042 | | */ |
10043 | | |
10044 | | void |
10045 | 164k | xmlParseContent(xmlParserCtxtPtr ctxt) { |
10046 | 164k | int nameNr = ctxt->nameNr; |
10047 | | |
10048 | 164k | xmlParseContentInternal(ctxt); |
10049 | | |
10050 | 164k | if ((ctxt->instate != XML_PARSER_EOF) && (ctxt->nameNr > nameNr)) { |
10051 | 8.90k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
10052 | 8.90k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
10053 | 8.90k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
10054 | 8.90k | "Premature end of data in tag %s line %d\n", |
10055 | 8.90k | name, line, NULL); |
10056 | 8.90k | } |
10057 | 164k | } |
10058 | | |
10059 | | /** |
10060 | | * xmlParseElement: |
10061 | | * @ctxt: an XML parser context |
10062 | | * |
10063 | | * DEPRECATED: Internal function, don't use. |
10064 | | * |
10065 | | * parse an XML element |
10066 | | * |
10067 | | * [39] element ::= EmptyElemTag | STag content ETag |
10068 | | * |
10069 | | * [ WFC: Element Type Match ] |
10070 | | * The Name in an element's end-tag must match the element type in the |
10071 | | * start-tag. |
10072 | | * |
10073 | | */ |
10074 | | |
10075 | | void |
10076 | 115k | xmlParseElement(xmlParserCtxtPtr ctxt) { |
10077 | 115k | if (xmlParseElementStart(ctxt) != 0) |
10078 | 29.8k | return; |
10079 | | |
10080 | 85.3k | xmlParseContentInternal(ctxt); |
10081 | 85.3k | if (ctxt->instate == XML_PARSER_EOF) |
10082 | 262 | return; |
10083 | | |
10084 | 85.0k | if (CUR == 0) { |
10085 | 49.3k | const xmlChar *name = ctxt->nameTab[ctxt->nameNr - 1]; |
10086 | 49.3k | int line = ctxt->pushTab[ctxt->nameNr - 1].line; |
10087 | 49.3k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_TAG_NOT_FINISHED, |
10088 | 49.3k | "Premature end of data in tag %s line %d\n", |
10089 | 49.3k | name, line, NULL); |
10090 | 49.3k | return; |
10091 | 49.3k | } |
10092 | | |
10093 | 35.7k | xmlParseElementEnd(ctxt); |
10094 | 35.7k | } |
10095 | | |
10096 | | /** |
10097 | | * xmlParseElementStart: |
10098 | | * @ctxt: an XML parser context |
10099 | | * |
10100 | | * Parse the start of an XML element. Returns -1 in case of error, 0 if an |
10101 | | * opening tag was parsed, 1 if an empty element was parsed. |
10102 | | * |
10103 | | * Always consumes '<'. |
10104 | | */ |
10105 | | static int |
10106 | 4.84M | xmlParseElementStart(xmlParserCtxtPtr ctxt) { |
10107 | 4.84M | const xmlChar *name; |
10108 | 4.84M | const xmlChar *prefix = NULL; |
10109 | 4.84M | const xmlChar *URI = NULL; |
10110 | 4.84M | xmlParserNodeInfo node_info; |
10111 | 4.84M | int line, tlen = 0; |
10112 | 4.84M | xmlNodePtr ret; |
10113 | 4.84M | int nsNr = ctxt->nsNr; |
10114 | | |
10115 | 4.84M | if (((unsigned int) ctxt->nameNr > xmlParserMaxDepth) && |
10116 | 4.84M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
10117 | 0 | xmlFatalErrMsgInt(ctxt, XML_ERR_INTERNAL_ERROR, |
10118 | 0 | "Excessive depth in document: %d use XML_PARSE_HUGE option\n", |
10119 | 0 | xmlParserMaxDepth); |
10120 | 0 | xmlHaltParser(ctxt); |
10121 | 0 | return(-1); |
10122 | 0 | } |
10123 | | |
10124 | | /* Capture start position */ |
10125 | 4.84M | if (ctxt->record_info) { |
10126 | 0 | node_info.begin_pos = ctxt->input->consumed + |
10127 | 0 | (CUR_PTR - ctxt->input->base); |
10128 | 0 | node_info.begin_line = ctxt->input->line; |
10129 | 0 | } |
10130 | | |
10131 | 4.84M | if (ctxt->spaceNr == 0) |
10132 | 0 | spacePush(ctxt, -1); |
10133 | 4.84M | else if (*ctxt->space == -2) |
10134 | 928k | spacePush(ctxt, -1); |
10135 | 3.92M | else |
10136 | 3.92M | spacePush(ctxt, *ctxt->space); |
10137 | | |
10138 | 4.84M | line = ctxt->input->line; |
10139 | 4.84M | #ifdef LIBXML_SAX1_ENABLED |
10140 | 4.84M | if (ctxt->sax2) |
10141 | 3.10M | #endif /* LIBXML_SAX1_ENABLED */ |
10142 | 3.10M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
10143 | 1.74M | #ifdef LIBXML_SAX1_ENABLED |
10144 | 1.74M | else |
10145 | 1.74M | name = xmlParseStartTag(ctxt); |
10146 | 4.84M | #endif /* LIBXML_SAX1_ENABLED */ |
10147 | 4.84M | if (ctxt->instate == XML_PARSER_EOF) |
10148 | 191 | return(-1); |
10149 | 4.84M | if (name == NULL) { |
10150 | 334k | spacePop(ctxt); |
10151 | 334k | return(-1); |
10152 | 334k | } |
10153 | 4.51M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
10154 | 4.51M | ret = ctxt->node; |
10155 | | |
10156 | 4.51M | #ifdef LIBXML_VALID_ENABLED |
10157 | | /* |
10158 | | * [ VC: Root Element Type ] |
10159 | | * The Name in the document type declaration must match the element |
10160 | | * type of the root element. |
10161 | | */ |
10162 | 4.51M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
10163 | 4.51M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
10164 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
10165 | 4.51M | #endif /* LIBXML_VALID_ENABLED */ |
10166 | | |
10167 | | /* |
10168 | | * Check for an Empty Element. |
10169 | | */ |
10170 | 4.51M | if ((RAW == '/') && (NXT(1) == '>')) { |
10171 | 1.75M | SKIP(2); |
10172 | 1.75M | if (ctxt->sax2) { |
10173 | 1.20M | if ((ctxt->sax != NULL) && (ctxt->sax->endElementNs != NULL) && |
10174 | 1.20M | (!ctxt->disableSAX)) |
10175 | 920k | ctxt->sax->endElementNs(ctxt->userData, name, prefix, URI); |
10176 | 1.20M | #ifdef LIBXML_SAX1_ENABLED |
10177 | 1.20M | } else { |
10178 | 553k | if ((ctxt->sax != NULL) && (ctxt->sax->endElement != NULL) && |
10179 | 553k | (!ctxt->disableSAX)) |
10180 | 406k | ctxt->sax->endElement(ctxt->userData, name); |
10181 | 553k | #endif /* LIBXML_SAX1_ENABLED */ |
10182 | 553k | } |
10183 | 1.75M | namePop(ctxt); |
10184 | 1.75M | spacePop(ctxt); |
10185 | 1.75M | if (nsNr != ctxt->nsNr) |
10186 | 7.40k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10187 | 1.75M | if ( ret != NULL && ctxt->record_info ) { |
10188 | 0 | node_info.end_pos = ctxt->input->consumed + |
10189 | 0 | (CUR_PTR - ctxt->input->base); |
10190 | 0 | node_info.end_line = ctxt->input->line; |
10191 | 0 | node_info.node = ret; |
10192 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10193 | 0 | } |
10194 | 1.75M | return(1); |
10195 | 1.75M | } |
10196 | 2.75M | if (RAW == '>') { |
10197 | 2.52M | NEXT1; |
10198 | 2.52M | } else { |
10199 | 234k | xmlFatalErrMsgStrIntStr(ctxt, XML_ERR_GT_REQUIRED, |
10200 | 234k | "Couldn't find end of Start Tag %s line %d\n", |
10201 | 234k | name, line, NULL); |
10202 | | |
10203 | | /* |
10204 | | * end of parsing of this node. |
10205 | | */ |
10206 | 234k | nodePop(ctxt); |
10207 | 234k | namePop(ctxt); |
10208 | 234k | spacePop(ctxt); |
10209 | 234k | if (nsNr != ctxt->nsNr) |
10210 | 17.8k | nsPop(ctxt, ctxt->nsNr - nsNr); |
10211 | | |
10212 | | /* |
10213 | | * Capture end position and add node |
10214 | | */ |
10215 | 234k | if ( ret != NULL && ctxt->record_info ) { |
10216 | 0 | node_info.end_pos = ctxt->input->consumed + |
10217 | 0 | (CUR_PTR - ctxt->input->base); |
10218 | 0 | node_info.end_line = ctxt->input->line; |
10219 | 0 | node_info.node = ret; |
10220 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10221 | 0 | } |
10222 | 234k | return(-1); |
10223 | 234k | } |
10224 | | |
10225 | 2.52M | return(0); |
10226 | 2.75M | } |
10227 | | |
10228 | | /** |
10229 | | * xmlParseElementEnd: |
10230 | | * @ctxt: an XML parser context |
10231 | | * |
10232 | | * Parse the end of an XML element. Always consumes '</'. |
10233 | | */ |
10234 | | static void |
10235 | 2.31M | xmlParseElementEnd(xmlParserCtxtPtr ctxt) { |
10236 | 2.31M | xmlParserNodeInfo node_info; |
10237 | 2.31M | xmlNodePtr ret = ctxt->node; |
10238 | | |
10239 | 2.31M | if (ctxt->nameNr <= 0) { |
10240 | 0 | if ((RAW == '<') && (NXT(1) == '/')) |
10241 | 0 | SKIP(2); |
10242 | 0 | return; |
10243 | 0 | } |
10244 | | |
10245 | | /* |
10246 | | * parse the end of tag: '</' should be here. |
10247 | | */ |
10248 | 2.31M | if (ctxt->sax2) { |
10249 | 1.43M | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
10250 | 1.43M | namePop(ctxt); |
10251 | 1.43M | } |
10252 | 877k | #ifdef LIBXML_SAX1_ENABLED |
10253 | 877k | else |
10254 | 877k | xmlParseEndTag1(ctxt, 0); |
10255 | 2.31M | #endif /* LIBXML_SAX1_ENABLED */ |
10256 | | |
10257 | | /* |
10258 | | * Capture end position and add node |
10259 | | */ |
10260 | 2.31M | if ( ret != NULL && ctxt->record_info ) { |
10261 | 0 | node_info.end_pos = ctxt->input->consumed + |
10262 | 0 | (CUR_PTR - ctxt->input->base); |
10263 | 0 | node_info.end_line = ctxt->input->line; |
10264 | 0 | node_info.node = ret; |
10265 | 0 | xmlParserAddNodeInfo(ctxt, &node_info); |
10266 | 0 | } |
10267 | 2.31M | } |
10268 | | |
10269 | | /** |
10270 | | * xmlParseVersionNum: |
10271 | | * @ctxt: an XML parser context |
10272 | | * |
10273 | | * DEPRECATED: Internal function, don't use. |
10274 | | * |
10275 | | * parse the XML version value. |
10276 | | * |
10277 | | * [26] VersionNum ::= '1.' [0-9]+ |
10278 | | * |
10279 | | * In practice allow [0-9].[0-9]+ at that level |
10280 | | * |
10281 | | * Returns the string giving the XML version number, or NULL |
10282 | | */ |
10283 | | xmlChar * |
10284 | 252k | xmlParseVersionNum(xmlParserCtxtPtr ctxt) { |
10285 | 252k | xmlChar *buf = NULL; |
10286 | 252k | int len = 0; |
10287 | 252k | int size = 10; |
10288 | 252k | xmlChar cur; |
10289 | | |
10290 | 252k | buf = (xmlChar *) xmlMallocAtomic(size); |
10291 | 252k | if (buf == NULL) { |
10292 | 0 | xmlErrMemory(ctxt, NULL); |
10293 | 0 | return(NULL); |
10294 | 0 | } |
10295 | 252k | cur = CUR; |
10296 | 252k | if (!((cur >= '0') && (cur <= '9'))) { |
10297 | 1.78k | xmlFree(buf); |
10298 | 1.78k | return(NULL); |
10299 | 1.78k | } |
10300 | 250k | buf[len++] = cur; |
10301 | 250k | NEXT; |
10302 | 250k | cur=CUR; |
10303 | 250k | if (cur != '.') { |
10304 | 2.38k | xmlFree(buf); |
10305 | 2.38k | return(NULL); |
10306 | 2.38k | } |
10307 | 248k | buf[len++] = cur; |
10308 | 248k | NEXT; |
10309 | 248k | cur=CUR; |
10310 | 1.08M | while ((cur >= '0') && (cur <= '9')) { |
10311 | 836k | if (len + 1 >= size) { |
10312 | 1.86k | xmlChar *tmp; |
10313 | | |
10314 | 1.86k | size *= 2; |
10315 | 1.86k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10316 | 1.86k | if (tmp == NULL) { |
10317 | 0 | xmlFree(buf); |
10318 | 0 | xmlErrMemory(ctxt, NULL); |
10319 | 0 | return(NULL); |
10320 | 0 | } |
10321 | 1.86k | buf = tmp; |
10322 | 1.86k | } |
10323 | 836k | buf[len++] = cur; |
10324 | 836k | NEXT; |
10325 | 836k | cur=CUR; |
10326 | 836k | } |
10327 | 248k | buf[len] = 0; |
10328 | 248k | return(buf); |
10329 | 248k | } |
10330 | | |
10331 | | /** |
10332 | | * xmlParseVersionInfo: |
10333 | | * @ctxt: an XML parser context |
10334 | | * |
10335 | | * DEPRECATED: Internal function, don't use. |
10336 | | * |
10337 | | * parse the XML version. |
10338 | | * |
10339 | | * [24] VersionInfo ::= S 'version' Eq (' VersionNum ' | " VersionNum ") |
10340 | | * |
10341 | | * [25] Eq ::= S? '=' S? |
10342 | | * |
10343 | | * Returns the version string, e.g. "1.0" |
10344 | | */ |
10345 | | |
10346 | | xmlChar * |
10347 | 285k | xmlParseVersionInfo(xmlParserCtxtPtr ctxt) { |
10348 | 285k | xmlChar *version = NULL; |
10349 | | |
10350 | 285k | if (CMP7(CUR_PTR, 'v', 'e', 'r', 's', 'i', 'o', 'n')) { |
10351 | 259k | SKIP(7); |
10352 | 259k | SKIP_BLANKS; |
10353 | 259k | if (RAW != '=') { |
10354 | 4.03k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10355 | 4.03k | return(NULL); |
10356 | 4.03k | } |
10357 | 255k | NEXT; |
10358 | 255k | SKIP_BLANKS; |
10359 | 255k | if (RAW == '"') { |
10360 | 231k | NEXT; |
10361 | 231k | version = xmlParseVersionNum(ctxt); |
10362 | 231k | if (RAW != '"') { |
10363 | 7.88k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10364 | 7.88k | } else |
10365 | 223k | NEXT; |
10366 | 231k | } else if (RAW == '\''){ |
10367 | 21.3k | NEXT; |
10368 | 21.3k | version = xmlParseVersionNum(ctxt); |
10369 | 21.3k | if (RAW != '\'') { |
10370 | 1.42k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10371 | 1.42k | } else |
10372 | 19.8k | NEXT; |
10373 | 21.3k | } else { |
10374 | 2.77k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10375 | 2.77k | } |
10376 | 255k | } |
10377 | 281k | return(version); |
10378 | 285k | } |
10379 | | |
10380 | | /** |
10381 | | * xmlParseEncName: |
10382 | | * @ctxt: an XML parser context |
10383 | | * |
10384 | | * DEPRECATED: Internal function, don't use. |
10385 | | * |
10386 | | * parse the XML encoding name |
10387 | | * |
10388 | | * [81] EncName ::= [A-Za-z] ([A-Za-z0-9._] | '-')* |
10389 | | * |
10390 | | * Returns the encoding name value or NULL |
10391 | | */ |
10392 | | xmlChar * |
10393 | 117k | xmlParseEncName(xmlParserCtxtPtr ctxt) { |
10394 | 117k | xmlChar *buf = NULL; |
10395 | 117k | int len = 0; |
10396 | 117k | int size = 10; |
10397 | 117k | xmlChar cur; |
10398 | | |
10399 | 117k | cur = CUR; |
10400 | 117k | if (((cur >= 'a') && (cur <= 'z')) || |
10401 | 117k | ((cur >= 'A') && (cur <= 'Z'))) { |
10402 | 115k | buf = (xmlChar *) xmlMallocAtomic(size); |
10403 | 115k | if (buf == NULL) { |
10404 | 0 | xmlErrMemory(ctxt, NULL); |
10405 | 0 | return(NULL); |
10406 | 0 | } |
10407 | | |
10408 | 115k | buf[len++] = cur; |
10409 | 115k | NEXT; |
10410 | 115k | cur = CUR; |
10411 | 1.82M | while (((cur >= 'a') && (cur <= 'z')) || |
10412 | 1.82M | ((cur >= 'A') && (cur <= 'Z')) || |
10413 | 1.82M | ((cur >= '0') && (cur <= '9')) || |
10414 | 1.82M | (cur == '.') || (cur == '_') || |
10415 | 1.82M | (cur == '-')) { |
10416 | 1.71M | if (len + 1 >= size) { |
10417 | 55.8k | xmlChar *tmp; |
10418 | | |
10419 | 55.8k | size *= 2; |
10420 | 55.8k | tmp = (xmlChar *) xmlRealloc(buf, size); |
10421 | 55.8k | if (tmp == NULL) { |
10422 | 0 | xmlErrMemory(ctxt, NULL); |
10423 | 0 | xmlFree(buf); |
10424 | 0 | return(NULL); |
10425 | 0 | } |
10426 | 55.8k | buf = tmp; |
10427 | 55.8k | } |
10428 | 1.71M | buf[len++] = cur; |
10429 | 1.71M | NEXT; |
10430 | 1.71M | cur = CUR; |
10431 | 1.71M | if (cur == 0) { |
10432 | 1.72k | SHRINK; |
10433 | 1.72k | GROW; |
10434 | 1.72k | cur = CUR; |
10435 | 1.72k | } |
10436 | 1.71M | } |
10437 | 115k | buf[len] = 0; |
10438 | 115k | } else { |
10439 | 2.28k | xmlFatalErr(ctxt, XML_ERR_ENCODING_NAME, NULL); |
10440 | 2.28k | } |
10441 | 117k | return(buf); |
10442 | 117k | } |
10443 | | |
10444 | | /** |
10445 | | * xmlParseEncodingDecl: |
10446 | | * @ctxt: an XML parser context |
10447 | | * |
10448 | | * DEPRECATED: Internal function, don't use. |
10449 | | * |
10450 | | * parse the XML encoding declaration |
10451 | | * |
10452 | | * [80] EncodingDecl ::= S 'encoding' Eq ('"' EncName '"' | "'" EncName "'") |
10453 | | * |
10454 | | * this setups the conversion filters. |
10455 | | * |
10456 | | * Returns the encoding value or NULL |
10457 | | */ |
10458 | | |
10459 | | const xmlChar * |
10460 | 198k | xmlParseEncodingDecl(xmlParserCtxtPtr ctxt) { |
10461 | 198k | xmlChar *encoding = NULL; |
10462 | | |
10463 | 198k | SKIP_BLANKS; |
10464 | 198k | if (CMP8(CUR_PTR, 'e', 'n', 'c', 'o', 'd', 'i', 'n', 'g')) { |
10465 | 121k | SKIP(8); |
10466 | 121k | SKIP_BLANKS; |
10467 | 121k | if (RAW != '=') { |
10468 | 1.60k | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10469 | 1.60k | return(NULL); |
10470 | 1.60k | } |
10471 | 120k | NEXT; |
10472 | 120k | SKIP_BLANKS; |
10473 | 120k | if (RAW == '"') { |
10474 | 104k | NEXT; |
10475 | 104k | encoding = xmlParseEncName(ctxt); |
10476 | 104k | if (RAW != '"') { |
10477 | 7.02k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10478 | 7.02k | xmlFree((xmlChar *) encoding); |
10479 | 7.02k | return(NULL); |
10480 | 7.02k | } else |
10481 | 97.3k | NEXT; |
10482 | 104k | } else if (RAW == '\''){ |
10483 | 13.5k | NEXT; |
10484 | 13.5k | encoding = xmlParseEncName(ctxt); |
10485 | 13.5k | if (RAW != '\'') { |
10486 | 869 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10487 | 869 | xmlFree((xmlChar *) encoding); |
10488 | 869 | return(NULL); |
10489 | 869 | } else |
10490 | 12.6k | NEXT; |
10491 | 13.5k | } else { |
10492 | 2.26k | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10493 | 2.26k | } |
10494 | | |
10495 | | /* |
10496 | | * Non standard parsing, allowing the user to ignore encoding |
10497 | | */ |
10498 | 112k | if (ctxt->options & XML_PARSE_IGNORE_ENC) { |
10499 | 52.8k | xmlFree((xmlChar *) encoding); |
10500 | 52.8k | return(NULL); |
10501 | 52.8k | } |
10502 | | |
10503 | | /* |
10504 | | * UTF-16 encoding switch has already taken place at this stage, |
10505 | | * more over the little-endian/big-endian selection is already done |
10506 | | */ |
10507 | 59.4k | if ((encoding != NULL) && |
10508 | 59.4k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-16")) || |
10509 | 58.1k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF16")))) { |
10510 | | /* |
10511 | | * If no encoding was passed to the parser, that we are |
10512 | | * using UTF-16 and no decoder is present i.e. the |
10513 | | * document is apparently UTF-8 compatible, then raise an |
10514 | | * encoding mismatch fatal error |
10515 | | */ |
10516 | 2.55k | if ((ctxt->encoding == NULL) && |
10517 | 2.55k | (ctxt->input->buf != NULL) && |
10518 | 2.55k | (ctxt->input->buf->encoder == NULL)) { |
10519 | 2.11k | xmlFatalErrMsg(ctxt, XML_ERR_INVALID_ENCODING, |
10520 | 2.11k | "Document labelled UTF-16 but has UTF-8 content\n"); |
10521 | 2.11k | } |
10522 | 2.55k | if (ctxt->encoding != NULL) |
10523 | 435 | xmlFree((xmlChar *) ctxt->encoding); |
10524 | 2.55k | ctxt->encoding = encoding; |
10525 | 2.55k | } |
10526 | | /* |
10527 | | * UTF-8 encoding is handled natively |
10528 | | */ |
10529 | 56.9k | else if ((encoding != NULL) && |
10530 | 56.9k | ((!xmlStrcasecmp(encoding, BAD_CAST "UTF-8")) || |
10531 | 55.5k | (!xmlStrcasecmp(encoding, BAD_CAST "UTF8")))) { |
10532 | 20.6k | if (ctxt->encoding != NULL) |
10533 | 963 | xmlFree((xmlChar *) ctxt->encoding); |
10534 | 20.6k | ctxt->encoding = encoding; |
10535 | 20.6k | } |
10536 | 36.3k | else if (encoding != NULL) { |
10537 | 34.9k | xmlCharEncodingHandlerPtr handler; |
10538 | | |
10539 | 34.9k | if (ctxt->input->encoding != NULL) |
10540 | 0 | xmlFree((xmlChar *) ctxt->input->encoding); |
10541 | 34.9k | ctxt->input->encoding = encoding; |
10542 | | |
10543 | 34.9k | handler = xmlFindCharEncodingHandler((const char *) encoding); |
10544 | 34.9k | if (handler != NULL) { |
10545 | 33.9k | if (xmlSwitchToEncoding(ctxt, handler) < 0) { |
10546 | | /* failed to convert */ |
10547 | 1.08k | ctxt->errNo = XML_ERR_UNSUPPORTED_ENCODING; |
10548 | 1.08k | return(NULL); |
10549 | 1.08k | } |
10550 | 33.9k | } else { |
10551 | 1.05k | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
10552 | 1.05k | "Unsupported encoding %s\n", encoding); |
10553 | 1.05k | return(NULL); |
10554 | 1.05k | } |
10555 | 34.9k | } |
10556 | 59.4k | } |
10557 | 133k | return(encoding); |
10558 | 198k | } |
10559 | | |
10560 | | /** |
10561 | | * xmlParseSDDecl: |
10562 | | * @ctxt: an XML parser context |
10563 | | * |
10564 | | * DEPRECATED: Internal function, don't use. |
10565 | | * |
10566 | | * parse the XML standalone declaration |
10567 | | * |
10568 | | * [32] SDDecl ::= S 'standalone' Eq |
10569 | | * (("'" ('yes' | 'no') "'") | ('"' ('yes' | 'no')'"')) |
10570 | | * |
10571 | | * [ VC: Standalone Document Declaration ] |
10572 | | * TODO The standalone document declaration must have the value "no" |
10573 | | * if any external markup declarations contain declarations of: |
10574 | | * - attributes with default values, if elements to which these |
10575 | | * attributes apply appear in the document without specifications |
10576 | | * of values for these attributes, or |
10577 | | * - entities (other than amp, lt, gt, apos, quot), if references |
10578 | | * to those entities appear in the document, or |
10579 | | * - attributes with values subject to normalization, where the |
10580 | | * attribute appears in the document with a value which will change |
10581 | | * as a result of normalization, or |
10582 | | * - element types with element content, if white space occurs directly |
10583 | | * within any instance of those types. |
10584 | | * |
10585 | | * Returns: |
10586 | | * 1 if standalone="yes" |
10587 | | * 0 if standalone="no" |
10588 | | * -2 if standalone attribute is missing or invalid |
10589 | | * (A standalone value of -2 means that the XML declaration was found, |
10590 | | * but no value was specified for the standalone attribute). |
10591 | | */ |
10592 | | |
10593 | | int |
10594 | 128k | xmlParseSDDecl(xmlParserCtxtPtr ctxt) { |
10595 | 128k | int standalone = -2; |
10596 | | |
10597 | 128k | SKIP_BLANKS; |
10598 | 128k | if (CMP10(CUR_PTR, 's', 't', 'a', 'n', 'd', 'a', 'l', 'o', 'n', 'e')) { |
10599 | 23.8k | SKIP(10); |
10600 | 23.8k | SKIP_BLANKS; |
10601 | 23.8k | if (RAW != '=') { |
10602 | 177 | xmlFatalErr(ctxt, XML_ERR_EQUAL_REQUIRED, NULL); |
10603 | 177 | return(standalone); |
10604 | 177 | } |
10605 | 23.6k | NEXT; |
10606 | 23.6k | SKIP_BLANKS; |
10607 | 23.6k | if (RAW == '\''){ |
10608 | 9.32k | NEXT; |
10609 | 9.32k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10610 | 7.55k | standalone = 0; |
10611 | 7.55k | SKIP(2); |
10612 | 7.55k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10613 | 1.76k | (NXT(2) == 's')) { |
10614 | 1.46k | standalone = 1; |
10615 | 1.46k | SKIP(3); |
10616 | 1.46k | } else { |
10617 | 303 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10618 | 303 | } |
10619 | 9.32k | if (RAW != '\'') { |
10620 | 426 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10621 | 426 | } else |
10622 | 8.89k | NEXT; |
10623 | 14.3k | } else if (RAW == '"'){ |
10624 | 14.2k | NEXT; |
10625 | 14.2k | if ((RAW == 'n') && (NXT(1) == 'o')) { |
10626 | 6.17k | standalone = 0; |
10627 | 6.17k | SKIP(2); |
10628 | 8.02k | } else if ((RAW == 'y') && (NXT(1) == 'e') && |
10629 | 8.02k | (NXT(2) == 's')) { |
10630 | 7.61k | standalone = 1; |
10631 | 7.61k | SKIP(3); |
10632 | 7.61k | } else { |
10633 | 417 | xmlFatalErr(ctxt, XML_ERR_STANDALONE_VALUE, NULL); |
10634 | 417 | } |
10635 | 14.2k | if (RAW != '"') { |
10636 | 606 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_CLOSED, NULL); |
10637 | 606 | } else |
10638 | 13.5k | NEXT; |
10639 | 14.2k | } else { |
10640 | 114 | xmlFatalErr(ctxt, XML_ERR_STRING_NOT_STARTED, NULL); |
10641 | 114 | } |
10642 | 23.6k | } |
10643 | 128k | return(standalone); |
10644 | 128k | } |
10645 | | |
10646 | | /** |
10647 | | * xmlParseXMLDecl: |
10648 | | * @ctxt: an XML parser context |
10649 | | * |
10650 | | * DEPRECATED: Internal function, don't use. |
10651 | | * |
10652 | | * parse an XML declaration header |
10653 | | * |
10654 | | * [23] XMLDecl ::= '<?xml' VersionInfo EncodingDecl? SDDecl? S? '?>' |
10655 | | */ |
10656 | | |
10657 | | void |
10658 | 235k | xmlParseXMLDecl(xmlParserCtxtPtr ctxt) { |
10659 | 235k | xmlChar *version; |
10660 | | |
10661 | | /* |
10662 | | * This value for standalone indicates that the document has an |
10663 | | * XML declaration but it does not have a standalone attribute. |
10664 | | * It will be overwritten later if a standalone attribute is found. |
10665 | | */ |
10666 | 235k | ctxt->input->standalone = -2; |
10667 | | |
10668 | | /* |
10669 | | * We know that '<?xml' is here. |
10670 | | */ |
10671 | 235k | SKIP(5); |
10672 | | |
10673 | 235k | if (!IS_BLANK_CH(RAW)) { |
10674 | 0 | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, |
10675 | 0 | "Blank needed after '<?xml'\n"); |
10676 | 0 | } |
10677 | 235k | SKIP_BLANKS; |
10678 | | |
10679 | | /* |
10680 | | * We must have the VersionInfo here. |
10681 | | */ |
10682 | 235k | version = xmlParseVersionInfo(ctxt); |
10683 | 235k | if (version == NULL) { |
10684 | 18.8k | xmlFatalErr(ctxt, XML_ERR_VERSION_MISSING, NULL); |
10685 | 216k | } else { |
10686 | 216k | if (!xmlStrEqual(version, (const xmlChar *) XML_DEFAULT_VERSION)) { |
10687 | | /* |
10688 | | * Changed here for XML-1.0 5th edition |
10689 | | */ |
10690 | 2.14k | if (ctxt->options & XML_PARSE_OLD10) { |
10691 | 733 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10692 | 733 | "Unsupported version '%s'\n", |
10693 | 733 | version); |
10694 | 1.41k | } else { |
10695 | 1.41k | if ((version[0] == '1') && ((version[1] == '.'))) { |
10696 | 1.26k | xmlWarningMsg(ctxt, XML_WAR_UNKNOWN_VERSION, |
10697 | 1.26k | "Unsupported version '%s'\n", |
10698 | 1.26k | version, NULL); |
10699 | 1.26k | } else { |
10700 | 147 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNKNOWN_VERSION, |
10701 | 147 | "Unsupported version '%s'\n", |
10702 | 147 | version); |
10703 | 147 | } |
10704 | 1.41k | } |
10705 | 2.14k | } |
10706 | 216k | if (ctxt->version != NULL) |
10707 | 0 | xmlFree((void *) ctxt->version); |
10708 | 216k | ctxt->version = version; |
10709 | 216k | } |
10710 | | |
10711 | | /* |
10712 | | * We may have the encoding declaration |
10713 | | */ |
10714 | 235k | if (!IS_BLANK_CH(RAW)) { |
10715 | 110k | if ((RAW == '?') && (NXT(1) == '>')) { |
10716 | 87.1k | SKIP(2); |
10717 | 87.1k | return; |
10718 | 87.1k | } |
10719 | 22.9k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10720 | 22.9k | } |
10721 | 148k | xmlParseEncodingDecl(ctxt); |
10722 | 148k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10723 | 148k | (ctxt->instate == XML_PARSER_EOF)) { |
10724 | | /* |
10725 | | * The XML REC instructs us to stop parsing right here |
10726 | | */ |
10727 | 528 | return; |
10728 | 528 | } |
10729 | | |
10730 | | /* |
10731 | | * We may have the standalone status. |
10732 | | */ |
10733 | 147k | if ((ctxt->input->encoding != NULL) && (!IS_BLANK_CH(RAW))) { |
10734 | 20.7k | if ((RAW == '?') && (NXT(1) == '>')) { |
10735 | 19.0k | SKIP(2); |
10736 | 19.0k | return; |
10737 | 19.0k | } |
10738 | 1.75k | xmlFatalErrMsg(ctxt, XML_ERR_SPACE_REQUIRED, "Blank needed here\n"); |
10739 | 1.75k | } |
10740 | | |
10741 | | /* |
10742 | | * We can grow the input buffer freely at that point |
10743 | | */ |
10744 | 128k | GROW; |
10745 | | |
10746 | 128k | SKIP_BLANKS; |
10747 | 128k | ctxt->input->standalone = xmlParseSDDecl(ctxt); |
10748 | | |
10749 | 128k | SKIP_BLANKS; |
10750 | 128k | if ((RAW == '?') && (NXT(1) == '>')) { |
10751 | 89.8k | SKIP(2); |
10752 | 89.8k | } else if (RAW == '>') { |
10753 | | /* Deprecated old WD ... */ |
10754 | 777 | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10755 | 777 | NEXT; |
10756 | 37.9k | } else { |
10757 | 37.9k | int c; |
10758 | | |
10759 | 37.9k | xmlFatalErr(ctxt, XML_ERR_XMLDECL_NOT_FINISHED, NULL); |
10760 | 2.60M | while ((c = CUR) != 0) { |
10761 | 2.59M | NEXT; |
10762 | 2.59M | if (c == '>') |
10763 | 34.6k | break; |
10764 | 2.59M | } |
10765 | 37.9k | } |
10766 | 128k | } |
10767 | | |
10768 | | /** |
10769 | | * xmlParseMisc: |
10770 | | * @ctxt: an XML parser context |
10771 | | * |
10772 | | * DEPRECATED: Internal function, don't use. |
10773 | | * |
10774 | | * parse an XML Misc* optional field. |
10775 | | * |
10776 | | * [27] Misc ::= Comment | PI | S |
10777 | | */ |
10778 | | |
10779 | | void |
10780 | 341k | xmlParseMisc(xmlParserCtxtPtr ctxt) { |
10781 | 380k | while (ctxt->instate != XML_PARSER_EOF) { |
10782 | 380k | SKIP_BLANKS; |
10783 | 380k | GROW; |
10784 | 380k | if ((RAW == '<') && (NXT(1) == '?')) { |
10785 | 23.4k | xmlParsePI(ctxt); |
10786 | 357k | } else if (CMP4(CUR_PTR, '<', '!', '-', '-')) { |
10787 | 15.5k | xmlParseComment(ctxt); |
10788 | 341k | } else { |
10789 | 341k | break; |
10790 | 341k | } |
10791 | 380k | } |
10792 | 341k | } |
10793 | | |
10794 | | /** |
10795 | | * xmlParseDocument: |
10796 | | * @ctxt: an XML parser context |
10797 | | * |
10798 | | * parse an XML document (and build a tree if using the standard SAX |
10799 | | * interface). |
10800 | | * |
10801 | | * [1] document ::= prolog element Misc* |
10802 | | * |
10803 | | * [22] prolog ::= XMLDecl? Misc* (doctypedecl Misc*)? |
10804 | | * |
10805 | | * Returns 0, -1 in case of error. the parser context is augmented |
10806 | | * as a result of the parsing. |
10807 | | */ |
10808 | | |
10809 | | int |
10810 | 153k | xmlParseDocument(xmlParserCtxtPtr ctxt) { |
10811 | 153k | xmlChar start[4]; |
10812 | 153k | xmlCharEncoding enc; |
10813 | | |
10814 | 153k | xmlInitParser(); |
10815 | | |
10816 | 153k | if ((ctxt == NULL) || (ctxt->input == NULL)) |
10817 | 0 | return(-1); |
10818 | | |
10819 | 153k | GROW; |
10820 | | |
10821 | | /* |
10822 | | * SAX: detecting the level. |
10823 | | */ |
10824 | 153k | xmlDetectSAX2(ctxt); |
10825 | | |
10826 | | /* |
10827 | | * SAX: beginning of the document processing. |
10828 | | */ |
10829 | 153k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
10830 | 153k | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
10831 | 153k | if (ctxt->instate == XML_PARSER_EOF) |
10832 | 0 | return(-1); |
10833 | | |
10834 | 153k | if ((ctxt->encoding == NULL) && |
10835 | 153k | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
10836 | | /* |
10837 | | * Get the 4 first bytes and decode the charset |
10838 | | * if enc != XML_CHAR_ENCODING_NONE |
10839 | | * plug some encoding conversion routines. |
10840 | | */ |
10841 | 152k | start[0] = RAW; |
10842 | 152k | start[1] = NXT(1); |
10843 | 152k | start[2] = NXT(2); |
10844 | 152k | start[3] = NXT(3); |
10845 | 152k | enc = xmlDetectCharEncoding(&start[0], 4); |
10846 | 152k | if (enc != XML_CHAR_ENCODING_NONE) { |
10847 | 85.0k | xmlSwitchEncoding(ctxt, enc); |
10848 | 85.0k | } |
10849 | 152k | } |
10850 | | |
10851 | | |
10852 | 153k | if (CUR == 0) { |
10853 | 287 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
10854 | 287 | return(-1); |
10855 | 287 | } |
10856 | | |
10857 | | /* |
10858 | | * Check for the XMLDecl in the Prolog. |
10859 | | * do not GROW here to avoid the detected encoder to decode more |
10860 | | * than just the first line, unless the amount of data is really |
10861 | | * too small to hold "<?xml version="1.0" encoding="foo" |
10862 | | */ |
10863 | 153k | if ((ctxt->input->end - ctxt->input->cur) < 35) { |
10864 | 5.75k | GROW; |
10865 | 5.75k | } |
10866 | 153k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
10867 | | |
10868 | | /* |
10869 | | * Note that we will switch encoding on the fly. |
10870 | | */ |
10871 | 78.5k | xmlParseXMLDecl(ctxt); |
10872 | 78.5k | if ((ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) || |
10873 | 78.5k | (ctxt->instate == XML_PARSER_EOF)) { |
10874 | | /* |
10875 | | * The XML REC instructs us to stop parsing right here |
10876 | | */ |
10877 | 176 | return(-1); |
10878 | 176 | } |
10879 | 78.3k | ctxt->standalone = ctxt->input->standalone; |
10880 | 78.3k | SKIP_BLANKS; |
10881 | 78.3k | } else { |
10882 | 74.8k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
10883 | 74.8k | } |
10884 | 153k | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
10885 | 149k | ctxt->sax->startDocument(ctxt->userData); |
10886 | 153k | if (ctxt->instate == XML_PARSER_EOF) |
10887 | 0 | return(-1); |
10888 | 153k | if ((ctxt->myDoc != NULL) && (ctxt->input != NULL) && |
10889 | 153k | (ctxt->input->buf != NULL) && (ctxt->input->buf->compressed >= 0)) { |
10890 | 0 | ctxt->myDoc->compression = ctxt->input->buf->compressed; |
10891 | 0 | } |
10892 | | |
10893 | | /* |
10894 | | * The Misc part of the Prolog |
10895 | | */ |
10896 | 153k | xmlParseMisc(ctxt); |
10897 | | |
10898 | | /* |
10899 | | * Then possibly doc type declaration(s) and more Misc |
10900 | | * (doctypedecl Misc*)? |
10901 | | */ |
10902 | 153k | GROW; |
10903 | 153k | if (CMP9(CUR_PTR, '<', '!', 'D', 'O', 'C', 'T', 'Y', 'P', 'E')) { |
10904 | | |
10905 | 98.1k | ctxt->inSubset = 1; |
10906 | 98.1k | xmlParseDocTypeDecl(ctxt); |
10907 | 98.1k | if (RAW == '[') { |
10908 | 78.7k | ctxt->instate = XML_PARSER_DTD; |
10909 | 78.7k | xmlParseInternalSubset(ctxt); |
10910 | 78.7k | if (ctxt->instate == XML_PARSER_EOF) |
10911 | 19.9k | return(-1); |
10912 | 78.7k | } |
10913 | | |
10914 | | /* |
10915 | | * Create and update the external subset. |
10916 | | */ |
10917 | 78.1k | ctxt->inSubset = 2; |
10918 | 78.1k | if ((ctxt->sax != NULL) && (ctxt->sax->externalSubset != NULL) && |
10919 | 78.1k | (!ctxt->disableSAX)) |
10920 | 74.2k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
10921 | 74.2k | ctxt->extSubSystem, ctxt->extSubURI); |
10922 | 78.1k | if (ctxt->instate == XML_PARSER_EOF) |
10923 | 4.60k | return(-1); |
10924 | 73.5k | ctxt->inSubset = 0; |
10925 | | |
10926 | 73.5k | xmlCleanSpecialAttr(ctxt); |
10927 | | |
10928 | 73.5k | ctxt->instate = XML_PARSER_PROLOG; |
10929 | 73.5k | xmlParseMisc(ctxt); |
10930 | 73.5k | } |
10931 | | |
10932 | | /* |
10933 | | * Time to start parsing the tree itself |
10934 | | */ |
10935 | 128k | GROW; |
10936 | 128k | if (RAW != '<') { |
10937 | 13.4k | xmlFatalErrMsg(ctxt, XML_ERR_DOCUMENT_EMPTY, |
10938 | 13.4k | "Start tag expected, '<' not found\n"); |
10939 | 115k | } else { |
10940 | 115k | ctxt->instate = XML_PARSER_CONTENT; |
10941 | 115k | xmlParseElement(ctxt); |
10942 | 115k | ctxt->instate = XML_PARSER_EPILOG; |
10943 | | |
10944 | | |
10945 | | /* |
10946 | | * The Misc part at the end |
10947 | | */ |
10948 | 115k | xmlParseMisc(ctxt); |
10949 | | |
10950 | 115k | if (RAW != 0) { |
10951 | 31.3k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
10952 | 31.3k | } |
10953 | 115k | ctxt->instate = XML_PARSER_EOF; |
10954 | 115k | } |
10955 | | |
10956 | | /* |
10957 | | * SAX: end of the document processing. |
10958 | | */ |
10959 | 128k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
10960 | 128k | ctxt->sax->endDocument(ctxt->userData); |
10961 | | |
10962 | | /* |
10963 | | * Remove locally kept entity definitions if the tree was not built |
10964 | | */ |
10965 | 128k | if ((ctxt->myDoc != NULL) && |
10966 | 128k | (xmlStrEqual(ctxt->myDoc->version, SAX_COMPAT_MODE))) { |
10967 | 177 | xmlFreeDoc(ctxt->myDoc); |
10968 | 177 | ctxt->myDoc = NULL; |
10969 | 177 | } |
10970 | | |
10971 | 128k | if ((ctxt->wellFormed) && (ctxt->myDoc != NULL)) { |
10972 | 16.4k | ctxt->myDoc->properties |= XML_DOC_WELLFORMED; |
10973 | 16.4k | if (ctxt->valid) |
10974 | 12.0k | ctxt->myDoc->properties |= XML_DOC_DTDVALID; |
10975 | 16.4k | if (ctxt->nsWellFormed) |
10976 | 15.4k | ctxt->myDoc->properties |= XML_DOC_NSVALID; |
10977 | 16.4k | if (ctxt->options & XML_PARSE_OLD10) |
10978 | 2.46k | ctxt->myDoc->properties |= XML_DOC_OLD10; |
10979 | 16.4k | } |
10980 | 128k | if (! ctxt->wellFormed) { |
10981 | 112k | ctxt->valid = 0; |
10982 | 112k | return(-1); |
10983 | 112k | } |
10984 | 16.4k | return(0); |
10985 | 128k | } |
10986 | | |
10987 | | /** |
10988 | | * xmlParseExtParsedEnt: |
10989 | | * @ctxt: an XML parser context |
10990 | | * |
10991 | | * parse a general parsed entity |
10992 | | * An external general parsed entity is well-formed if it matches the |
10993 | | * production labeled extParsedEnt. |
10994 | | * |
10995 | | * [78] extParsedEnt ::= TextDecl? content |
10996 | | * |
10997 | | * Returns 0, -1 in case of error. the parser context is augmented |
10998 | | * as a result of the parsing. |
10999 | | */ |
11000 | | |
11001 | | int |
11002 | 0 | xmlParseExtParsedEnt(xmlParserCtxtPtr ctxt) { |
11003 | 0 | xmlChar start[4]; |
11004 | 0 | xmlCharEncoding enc; |
11005 | |
|
11006 | 0 | if ((ctxt == NULL) || (ctxt->input == NULL)) |
11007 | 0 | return(-1); |
11008 | | |
11009 | 0 | xmlDetectSAX2(ctxt); |
11010 | |
|
11011 | 0 | GROW; |
11012 | | |
11013 | | /* |
11014 | | * SAX: beginning of the document processing. |
11015 | | */ |
11016 | 0 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11017 | 0 | ctxt->sax->setDocumentLocator(ctxt->userData, &xmlDefaultSAXLocator); |
11018 | | |
11019 | | /* |
11020 | | * Get the 4 first bytes and decode the charset |
11021 | | * if enc != XML_CHAR_ENCODING_NONE |
11022 | | * plug some encoding conversion routines. |
11023 | | */ |
11024 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
11025 | 0 | start[0] = RAW; |
11026 | 0 | start[1] = NXT(1); |
11027 | 0 | start[2] = NXT(2); |
11028 | 0 | start[3] = NXT(3); |
11029 | 0 | enc = xmlDetectCharEncoding(start, 4); |
11030 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
11031 | 0 | xmlSwitchEncoding(ctxt, enc); |
11032 | 0 | } |
11033 | 0 | } |
11034 | | |
11035 | |
|
11036 | 0 | if (CUR == 0) { |
11037 | 0 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11038 | 0 | } |
11039 | | |
11040 | | /* |
11041 | | * Check for the XMLDecl in the Prolog. |
11042 | | */ |
11043 | 0 | GROW; |
11044 | 0 | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
11045 | | |
11046 | | /* |
11047 | | * Note that we will switch encoding on the fly. |
11048 | | */ |
11049 | 0 | xmlParseXMLDecl(ctxt); |
11050 | 0 | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11051 | | /* |
11052 | | * The XML REC instructs us to stop parsing right here |
11053 | | */ |
11054 | 0 | return(-1); |
11055 | 0 | } |
11056 | 0 | SKIP_BLANKS; |
11057 | 0 | } else { |
11058 | 0 | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11059 | 0 | } |
11060 | 0 | if ((ctxt->sax) && (ctxt->sax->startDocument) && (!ctxt->disableSAX)) |
11061 | 0 | ctxt->sax->startDocument(ctxt->userData); |
11062 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11063 | 0 | return(-1); |
11064 | | |
11065 | | /* |
11066 | | * Doing validity checking on chunk doesn't make sense |
11067 | | */ |
11068 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
11069 | 0 | ctxt->validate = 0; |
11070 | 0 | ctxt->loadsubset = 0; |
11071 | 0 | ctxt->depth = 0; |
11072 | |
|
11073 | 0 | xmlParseContent(ctxt); |
11074 | 0 | if (ctxt->instate == XML_PARSER_EOF) |
11075 | 0 | return(-1); |
11076 | | |
11077 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
11078 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
11079 | 0 | } else if (RAW != 0) { |
11080 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
11081 | 0 | } |
11082 | | |
11083 | | /* |
11084 | | * SAX: end of the document processing. |
11085 | | */ |
11086 | 0 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11087 | 0 | ctxt->sax->endDocument(ctxt->userData); |
11088 | |
|
11089 | 0 | if (! ctxt->wellFormed) return(-1); |
11090 | 0 | return(0); |
11091 | 0 | } |
11092 | | |
11093 | | #ifdef LIBXML_PUSH_ENABLED |
11094 | | /************************************************************************ |
11095 | | * * |
11096 | | * Progressive parsing interfaces * |
11097 | | * * |
11098 | | ************************************************************************/ |
11099 | | |
11100 | | /** |
11101 | | * xmlParseLookupChar: |
11102 | | * @ctxt: an XML parser context |
11103 | | * @c: character |
11104 | | * |
11105 | | * Check whether the input buffer contains a character. |
11106 | | */ |
11107 | | static int |
11108 | 4.35M | xmlParseLookupChar(xmlParserCtxtPtr ctxt, int c) { |
11109 | 4.35M | const xmlChar *cur; |
11110 | | |
11111 | 4.35M | if (ctxt->checkIndex == 0) { |
11112 | 4.20M | cur = ctxt->input->cur + 1; |
11113 | 4.20M | } else { |
11114 | 153k | cur = ctxt->input->cur + ctxt->checkIndex; |
11115 | 153k | } |
11116 | | |
11117 | 4.35M | if (memchr(cur, c, ctxt->input->end - cur) == NULL) { |
11118 | 164k | ctxt->checkIndex = ctxt->input->end - ctxt->input->cur; |
11119 | 164k | return(0); |
11120 | 4.19M | } else { |
11121 | 4.19M | ctxt->checkIndex = 0; |
11122 | 4.19M | return(1); |
11123 | 4.19M | } |
11124 | 4.35M | } |
11125 | | |
11126 | | /** |
11127 | | * xmlParseLookupString: |
11128 | | * @ctxt: an XML parser context |
11129 | | * @startDelta: delta to apply at the start |
11130 | | * @str: string |
11131 | | * @strLen: length of string |
11132 | | * |
11133 | | * Check whether the input buffer contains a string. |
11134 | | */ |
11135 | | static const xmlChar * |
11136 | | xmlParseLookupString(xmlParserCtxtPtr ctxt, size_t startDelta, |
11137 | 1.10M | const char *str, size_t strLen) { |
11138 | 1.10M | const xmlChar *cur, *term; |
11139 | | |
11140 | 1.10M | if (ctxt->checkIndex == 0) { |
11141 | 605k | cur = ctxt->input->cur + startDelta; |
11142 | 605k | } else { |
11143 | 502k | cur = ctxt->input->cur + ctxt->checkIndex; |
11144 | 502k | } |
11145 | | |
11146 | 1.10M | term = BAD_CAST strstr((const char *) cur, str); |
11147 | 1.10M | if (term == NULL) { |
11148 | 629k | const xmlChar *end = ctxt->input->end; |
11149 | | |
11150 | | /* Rescan (strLen - 1) characters. */ |
11151 | 629k | if ((size_t) (end - cur) < strLen) |
11152 | 10.4k | end = cur; |
11153 | 619k | else |
11154 | 619k | end -= strLen - 1; |
11155 | 629k | ctxt->checkIndex = end - ctxt->input->cur; |
11156 | 629k | } else { |
11157 | 478k | ctxt->checkIndex = 0; |
11158 | 478k | } |
11159 | | |
11160 | 1.10M | return(term); |
11161 | 1.10M | } |
11162 | | |
11163 | | /** |
11164 | | * xmlParseLookupCharData: |
11165 | | * @ctxt: an XML parser context |
11166 | | * |
11167 | | * Check whether the input buffer contains terminated char data. |
11168 | | */ |
11169 | | static int |
11170 | 6.68M | xmlParseLookupCharData(xmlParserCtxtPtr ctxt) { |
11171 | 6.68M | const xmlChar *cur = ctxt->input->cur + ctxt->checkIndex; |
11172 | 6.68M | const xmlChar *end = ctxt->input->end; |
11173 | | |
11174 | 119M | while (cur < end) { |
11175 | 118M | if ((*cur == '<') || (*cur == '&')) { |
11176 | 5.94M | ctxt->checkIndex = 0; |
11177 | 5.94M | return(1); |
11178 | 5.94M | } |
11179 | 112M | cur++; |
11180 | 112M | } |
11181 | | |
11182 | 731k | ctxt->checkIndex = cur - ctxt->input->cur; |
11183 | 731k | return(0); |
11184 | 6.68M | } |
11185 | | |
11186 | | /** |
11187 | | * xmlParseLookupGt: |
11188 | | * @ctxt: an XML parser context |
11189 | | * |
11190 | | * Check whether there's enough data in the input buffer to finish parsing |
11191 | | * a start tag. This has to take quotes into account. |
11192 | | */ |
11193 | | static int |
11194 | 6.69M | xmlParseLookupGt(xmlParserCtxtPtr ctxt) { |
11195 | 6.69M | const xmlChar *cur; |
11196 | 6.69M | const xmlChar *end = ctxt->input->end; |
11197 | 6.69M | int state = ctxt->endCheckState; |
11198 | | |
11199 | 6.69M | if (ctxt->checkIndex == 0) |
11200 | 5.37M | cur = ctxt->input->cur + 1; |
11201 | 1.32M | else |
11202 | 1.32M | cur = ctxt->input->cur + ctxt->checkIndex; |
11203 | | |
11204 | 285M | while (cur < end) { |
11205 | 284M | if (state) { |
11206 | 160M | if (*cur == state) |
11207 | 7.09M | state = 0; |
11208 | 160M | } else if (*cur == '\'' || *cur == '"') { |
11209 | 7.13M | state = *cur; |
11210 | 116M | } else if (*cur == '>') { |
11211 | 5.30M | ctxt->checkIndex = 0; |
11212 | 5.30M | ctxt->endCheckState = 0; |
11213 | 5.30M | return(1); |
11214 | 5.30M | } |
11215 | 278M | cur++; |
11216 | 278M | } |
11217 | | |
11218 | 1.39M | ctxt->checkIndex = cur - ctxt->input->cur; |
11219 | 1.39M | ctxt->endCheckState = state; |
11220 | 1.39M | return(0); |
11221 | 6.69M | } |
11222 | | |
11223 | | /** |
11224 | | * xmlParseLookupInternalSubset: |
11225 | | * @ctxt: an XML parser context |
11226 | | * |
11227 | | * Check whether there's enough data in the input buffer to finish parsing |
11228 | | * the internal subset. |
11229 | | */ |
11230 | | static int |
11231 | 531k | xmlParseLookupInternalSubset(xmlParserCtxtPtr ctxt) { |
11232 | | /* |
11233 | | * Sorry, but progressive parsing of the internal subset is not |
11234 | | * supported. We first check that the full content of the internal |
11235 | | * subset is available and parsing is launched only at that point. |
11236 | | * Internal subset ends with "']' S? '>'" in an unescaped section and |
11237 | | * not in a ']]>' sequence which are conditional sections. |
11238 | | */ |
11239 | 531k | const xmlChar *cur, *start; |
11240 | 531k | const xmlChar *end = ctxt->input->end; |
11241 | 531k | int state = ctxt->endCheckState; |
11242 | | |
11243 | 531k | if (ctxt->checkIndex == 0) { |
11244 | 151k | cur = ctxt->input->cur + 1; |
11245 | 380k | } else { |
11246 | 380k | cur = ctxt->input->cur + ctxt->checkIndex; |
11247 | 380k | } |
11248 | 531k | start = cur; |
11249 | | |
11250 | 91.0M | while (cur < end) { |
11251 | 90.6M | if (state == '-') { |
11252 | 12.4M | if ((*cur == '-') && |
11253 | 12.4M | (cur[1] == '-') && |
11254 | 12.4M | (cur[2] == '>')) { |
11255 | 103k | state = 0; |
11256 | 103k | cur += 3; |
11257 | 103k | start = cur; |
11258 | 103k | continue; |
11259 | 103k | } |
11260 | 12.4M | } |
11261 | 78.2M | else if (state == ']') { |
11262 | 151k | if (*cur == '>') { |
11263 | 126k | ctxt->checkIndex = 0; |
11264 | 126k | ctxt->endCheckState = 0; |
11265 | 126k | return(1); |
11266 | 126k | } |
11267 | 25.2k | if (IS_BLANK_CH(*cur)) { |
11268 | 16.2k | state = ' '; |
11269 | 16.2k | } else if (*cur != ']') { |
11270 | 3.75k | state = 0; |
11271 | 3.75k | start = cur; |
11272 | 3.75k | continue; |
11273 | 3.75k | } |
11274 | 25.2k | } |
11275 | 78.0M | else if (state == ' ') { |
11276 | 190k | if (*cur == '>') { |
11277 | 538 | ctxt->checkIndex = 0; |
11278 | 538 | ctxt->endCheckState = 0; |
11279 | 538 | return(1); |
11280 | 538 | } |
11281 | 189k | if (!IS_BLANK_CH(*cur)) { |
11282 | 15.7k | state = 0; |
11283 | 15.7k | start = cur; |
11284 | 15.7k | continue; |
11285 | 15.7k | } |
11286 | 189k | } |
11287 | 77.8M | else if (state != 0) { |
11288 | 34.4M | if (*cur == state) { |
11289 | 999k | state = 0; |
11290 | 999k | start = cur + 1; |
11291 | 999k | } |
11292 | 34.4M | } |
11293 | 43.4M | else if (*cur == '<') { |
11294 | 1.22M | if ((cur[1] == '!') && |
11295 | 1.22M | (cur[2] == '-') && |
11296 | 1.22M | (cur[3] == '-')) { |
11297 | 105k | state = '-'; |
11298 | 105k | cur += 4; |
11299 | | /* Don't treat <!--> as comment */ |
11300 | 105k | start = cur; |
11301 | 105k | continue; |
11302 | 105k | } |
11303 | 1.22M | } |
11304 | 42.2M | else if ((*cur == '"') || (*cur == '\'') || (*cur == ']')) { |
11305 | 1.15M | state = *cur; |
11306 | 1.15M | } |
11307 | | |
11308 | 90.2M | cur++; |
11309 | 90.2M | } |
11310 | | |
11311 | | /* |
11312 | | * Rescan the three last characters to detect "<!--" and "-->" |
11313 | | * split across chunks. |
11314 | | */ |
11315 | 404k | if ((state == 0) || (state == '-')) { |
11316 | 233k | if (cur - start < 3) |
11317 | 20.5k | cur = start; |
11318 | 212k | else |
11319 | 212k | cur -= 3; |
11320 | 233k | } |
11321 | 404k | ctxt->checkIndex = cur - ctxt->input->cur; |
11322 | 404k | ctxt->endCheckState = state; |
11323 | 404k | return(0); |
11324 | 531k | } |
11325 | | |
11326 | | /** |
11327 | | * xmlCheckCdataPush: |
11328 | | * @cur: pointer to the block of characters |
11329 | | * @len: length of the block in bytes |
11330 | | * @complete: 1 if complete CDATA block is passed in, 0 if partial block |
11331 | | * |
11332 | | * Check that the block of characters is okay as SCdata content [20] |
11333 | | * |
11334 | | * Returns the number of bytes to pass if okay, a negative index where an |
11335 | | * UTF-8 error occurred otherwise |
11336 | | */ |
11337 | | static int |
11338 | 235k | xmlCheckCdataPush(const xmlChar *utf, int len, int complete) { |
11339 | 235k | int ix; |
11340 | 235k | unsigned char c; |
11341 | 235k | int codepoint; |
11342 | | |
11343 | 235k | if ((utf == NULL) || (len <= 0)) |
11344 | 3.59k | return(0); |
11345 | | |
11346 | 10.8M | for (ix = 0; ix < len;) { /* string is 0-terminated */ |
11347 | 10.8M | c = utf[ix]; |
11348 | 10.8M | if ((c & 0x80) == 0x00) { /* 1-byte code, starts with 10 */ |
11349 | 9.74M | if (c >= 0x20) |
11350 | 8.93M | ix++; |
11351 | 808k | else if ((c == 0xA) || (c == 0xD) || (c == 0x9)) |
11352 | 797k | ix++; |
11353 | 11.5k | else |
11354 | 11.5k | return(-ix); |
11355 | 9.74M | } else if ((c & 0xe0) == 0xc0) {/* 2-byte code, starts with 110 */ |
11356 | 371k | if (ix + 2 > len) return(complete ? -ix : ix); |
11357 | 366k | if ((utf[ix+1] & 0xc0 ) != 0x80) |
11358 | 67.1k | return(-ix); |
11359 | 299k | codepoint = (utf[ix] & 0x1f) << 6; |
11360 | 299k | codepoint |= utf[ix+1] & 0x3f; |
11361 | 299k | if (!xmlIsCharQ(codepoint)) |
11362 | 5.36k | return(-ix); |
11363 | 293k | ix += 2; |
11364 | 702k | } else if ((c & 0xf0) == 0xe0) {/* 3-byte code, starts with 1110 */ |
11365 | 318k | if (ix + 3 > len) return(complete ? -ix : ix); |
11366 | 313k | if (((utf[ix+1] & 0xc0) != 0x80) || |
11367 | 313k | ((utf[ix+2] & 0xc0) != 0x80)) |
11368 | 21.4k | return(-ix); |
11369 | 292k | codepoint = (utf[ix] & 0xf) << 12; |
11370 | 292k | codepoint |= (utf[ix+1] & 0x3f) << 6; |
11371 | 292k | codepoint |= utf[ix+2] & 0x3f; |
11372 | 292k | if (!xmlIsCharQ(codepoint)) |
11373 | 8.83k | return(-ix); |
11374 | 283k | ix += 3; |
11375 | 383k | } else if ((c & 0xf8) == 0xf0) {/* 4-byte code, starts with 11110 */ |
11376 | 362k | if (ix + 4 > len) return(complete ? -ix : ix); |
11377 | 358k | if (((utf[ix+1] & 0xc0) != 0x80) || |
11378 | 358k | ((utf[ix+2] & 0xc0) != 0x80) || |
11379 | 358k | ((utf[ix+3] & 0xc0) != 0x80)) |
11380 | 24.5k | return(-ix); |
11381 | 333k | codepoint = (utf[ix] & 0x7) << 18; |
11382 | 333k | codepoint |= (utf[ix+1] & 0x3f) << 12; |
11383 | 333k | codepoint |= (utf[ix+2] & 0x3f) << 6; |
11384 | 333k | codepoint |= utf[ix+3] & 0x3f; |
11385 | 333k | if (!xmlIsCharQ(codepoint)) |
11386 | 10.8k | return(-ix); |
11387 | 322k | ix += 4; |
11388 | 322k | } else /* unknown encoding */ |
11389 | 21.1k | return(-ix); |
11390 | 10.8M | } |
11391 | 46.2k | return(ix); |
11392 | 231k | } |
11393 | | |
11394 | | /** |
11395 | | * xmlParseTryOrFinish: |
11396 | | * @ctxt: an XML parser context |
11397 | | * @terminate: last chunk indicator |
11398 | | * |
11399 | | * Try to progress on parsing |
11400 | | * |
11401 | | * Returns zero if no parsing was possible |
11402 | | */ |
11403 | | static int |
11404 | 3.99M | xmlParseTryOrFinish(xmlParserCtxtPtr ctxt, int terminate) { |
11405 | 3.99M | int ret = 0; |
11406 | 3.99M | int avail, tlen; |
11407 | 3.99M | xmlChar cur, next; |
11408 | | |
11409 | 3.99M | if (ctxt->input == NULL) |
11410 | 0 | return(0); |
11411 | | |
11412 | | #ifdef DEBUG_PUSH |
11413 | | switch (ctxt->instate) { |
11414 | | case XML_PARSER_EOF: |
11415 | | xmlGenericError(xmlGenericErrorContext, |
11416 | | "PP: try EOF\n"); break; |
11417 | | case XML_PARSER_START: |
11418 | | xmlGenericError(xmlGenericErrorContext, |
11419 | | "PP: try START\n"); break; |
11420 | | case XML_PARSER_MISC: |
11421 | | xmlGenericError(xmlGenericErrorContext, |
11422 | | "PP: try MISC\n");break; |
11423 | | case XML_PARSER_COMMENT: |
11424 | | xmlGenericError(xmlGenericErrorContext, |
11425 | | "PP: try COMMENT\n");break; |
11426 | | case XML_PARSER_PROLOG: |
11427 | | xmlGenericError(xmlGenericErrorContext, |
11428 | | "PP: try PROLOG\n");break; |
11429 | | case XML_PARSER_START_TAG: |
11430 | | xmlGenericError(xmlGenericErrorContext, |
11431 | | "PP: try START_TAG\n");break; |
11432 | | case XML_PARSER_CONTENT: |
11433 | | xmlGenericError(xmlGenericErrorContext, |
11434 | | "PP: try CONTENT\n");break; |
11435 | | case XML_PARSER_CDATA_SECTION: |
11436 | | xmlGenericError(xmlGenericErrorContext, |
11437 | | "PP: try CDATA_SECTION\n");break; |
11438 | | case XML_PARSER_END_TAG: |
11439 | | xmlGenericError(xmlGenericErrorContext, |
11440 | | "PP: try END_TAG\n");break; |
11441 | | case XML_PARSER_ENTITY_DECL: |
11442 | | xmlGenericError(xmlGenericErrorContext, |
11443 | | "PP: try ENTITY_DECL\n");break; |
11444 | | case XML_PARSER_ENTITY_VALUE: |
11445 | | xmlGenericError(xmlGenericErrorContext, |
11446 | | "PP: try ENTITY_VALUE\n");break; |
11447 | | case XML_PARSER_ATTRIBUTE_VALUE: |
11448 | | xmlGenericError(xmlGenericErrorContext, |
11449 | | "PP: try ATTRIBUTE_VALUE\n");break; |
11450 | | case XML_PARSER_DTD: |
11451 | | xmlGenericError(xmlGenericErrorContext, |
11452 | | "PP: try DTD\n");break; |
11453 | | case XML_PARSER_EPILOG: |
11454 | | xmlGenericError(xmlGenericErrorContext, |
11455 | | "PP: try EPILOG\n");break; |
11456 | | case XML_PARSER_PI: |
11457 | | xmlGenericError(xmlGenericErrorContext, |
11458 | | "PP: try PI\n");break; |
11459 | | case XML_PARSER_IGNORE: |
11460 | | xmlGenericError(xmlGenericErrorContext, |
11461 | | "PP: try IGNORE\n");break; |
11462 | | } |
11463 | | #endif |
11464 | | |
11465 | 3.99M | if ((ctxt->input != NULL) && |
11466 | 3.99M | (ctxt->input->cur - ctxt->input->base > 4096)) { |
11467 | 60.5k | xmlParserInputShrink(ctxt->input); |
11468 | 60.5k | } |
11469 | | |
11470 | 33.9M | while (ctxt->instate != XML_PARSER_EOF) { |
11471 | 33.9M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
11472 | 74.7k | return(0); |
11473 | | |
11474 | 33.8M | if (ctxt->input == NULL) break; |
11475 | 33.8M | if (ctxt->input->buf == NULL) |
11476 | 0 | avail = ctxt->input->length - |
11477 | 0 | (ctxt->input->cur - ctxt->input->base); |
11478 | 33.8M | else { |
11479 | | /* |
11480 | | * If we are operating on converted input, try to flush |
11481 | | * remaining chars to avoid them stalling in the non-converted |
11482 | | * buffer. But do not do this in document start where |
11483 | | * encoding="..." may not have been read and we work on a |
11484 | | * guessed encoding. |
11485 | | */ |
11486 | 33.8M | if ((ctxt->instate != XML_PARSER_START) && |
11487 | 33.8M | (ctxt->input->buf->raw != NULL) && |
11488 | 33.8M | (xmlBufIsEmpty(ctxt->input->buf->raw) == 0)) { |
11489 | 133k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
11490 | 133k | ctxt->input); |
11491 | 133k | size_t current = ctxt->input->cur - ctxt->input->base; |
11492 | | |
11493 | 133k | xmlParserInputBufferPush(ctxt->input->buf, 0, ""); |
11494 | 133k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
11495 | 133k | base, current); |
11496 | 133k | } |
11497 | 33.8M | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11498 | 33.8M | (ctxt->input->cur - ctxt->input->base); |
11499 | 33.8M | } |
11500 | 33.8M | if (avail < 1) |
11501 | 226k | goto done; |
11502 | 33.6M | switch (ctxt->instate) { |
11503 | 0 | case XML_PARSER_EOF: |
11504 | | /* |
11505 | | * Document parsing is done ! |
11506 | | */ |
11507 | 0 | goto done; |
11508 | 777k | case XML_PARSER_START: |
11509 | 777k | if (ctxt->charset == XML_CHAR_ENCODING_NONE) { |
11510 | 224k | xmlChar start[4]; |
11511 | 224k | xmlCharEncoding enc; |
11512 | | |
11513 | | /* |
11514 | | * Very first chars read from the document flow. |
11515 | | */ |
11516 | 224k | if (avail < 4) |
11517 | 3.41k | goto done; |
11518 | | |
11519 | | /* |
11520 | | * Get the 4 first bytes and decode the charset |
11521 | | * if enc != XML_CHAR_ENCODING_NONE |
11522 | | * plug some encoding conversion routines, |
11523 | | * else xmlSwitchEncoding will set to (default) |
11524 | | * UTF8. |
11525 | | */ |
11526 | 220k | start[0] = RAW; |
11527 | 220k | start[1] = NXT(1); |
11528 | 220k | start[2] = NXT(2); |
11529 | 220k | start[3] = NXT(3); |
11530 | 220k | enc = xmlDetectCharEncoding(start, 4); |
11531 | 220k | xmlSwitchEncoding(ctxt, enc); |
11532 | 220k | break; |
11533 | 224k | } |
11534 | | |
11535 | 553k | if (avail < 2) |
11536 | 126 | goto done; |
11537 | 552k | cur = ctxt->input->cur[0]; |
11538 | 552k | next = ctxt->input->cur[1]; |
11539 | 552k | if (cur == 0) { |
11540 | 378 | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11541 | 378 | ctxt->sax->setDocumentLocator(ctxt->userData, |
11542 | 378 | &xmlDefaultSAXLocator); |
11543 | 378 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11544 | 378 | xmlHaltParser(ctxt); |
11545 | | #ifdef DEBUG_PUSH |
11546 | | xmlGenericError(xmlGenericErrorContext, |
11547 | | "PP: entering EOF\n"); |
11548 | | #endif |
11549 | 378 | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11550 | 378 | ctxt->sax->endDocument(ctxt->userData); |
11551 | 378 | goto done; |
11552 | 378 | } |
11553 | 552k | if ((cur == '<') && (next == '?')) { |
11554 | | /* PI or XML decl */ |
11555 | 416k | if (avail < 5) goto done; |
11556 | 416k | if ((!terminate) && |
11557 | 416k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11558 | 247k | goto done; |
11559 | 169k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11560 | 169k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11561 | 169k | &xmlDefaultSAXLocator); |
11562 | 169k | if ((ctxt->input->cur[2] == 'x') && |
11563 | 169k | (ctxt->input->cur[3] == 'm') && |
11564 | 169k | (ctxt->input->cur[4] == 'l') && |
11565 | 169k | (IS_BLANK_CH(ctxt->input->cur[5]))) { |
11566 | 156k | ret += 5; |
11567 | | #ifdef DEBUG_PUSH |
11568 | | xmlGenericError(xmlGenericErrorContext, |
11569 | | "PP: Parsing XML Decl\n"); |
11570 | | #endif |
11571 | 156k | xmlParseXMLDecl(ctxt); |
11572 | 156k | if (ctxt->errNo == XML_ERR_UNSUPPORTED_ENCODING) { |
11573 | | /* |
11574 | | * The XML REC instructs us to stop parsing right |
11575 | | * here |
11576 | | */ |
11577 | 352 | xmlHaltParser(ctxt); |
11578 | 352 | return(0); |
11579 | 352 | } |
11580 | 156k | ctxt->standalone = ctxt->input->standalone; |
11581 | 156k | if ((ctxt->encoding == NULL) && |
11582 | 156k | (ctxt->input->encoding != NULL)) |
11583 | 18.5k | ctxt->encoding = xmlStrdup(ctxt->input->encoding); |
11584 | 156k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11585 | 156k | (!ctxt->disableSAX)) |
11586 | 150k | ctxt->sax->startDocument(ctxt->userData); |
11587 | 156k | ctxt->instate = XML_PARSER_MISC; |
11588 | | #ifdef DEBUG_PUSH |
11589 | | xmlGenericError(xmlGenericErrorContext, |
11590 | | "PP: entering MISC\n"); |
11591 | | #endif |
11592 | 156k | } else { |
11593 | 12.5k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11594 | 12.5k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11595 | 12.5k | (!ctxt->disableSAX)) |
11596 | 12.5k | ctxt->sax->startDocument(ctxt->userData); |
11597 | 12.5k | ctxt->instate = XML_PARSER_MISC; |
11598 | | #ifdef DEBUG_PUSH |
11599 | | xmlGenericError(xmlGenericErrorContext, |
11600 | | "PP: entering MISC\n"); |
11601 | | #endif |
11602 | 12.5k | } |
11603 | 169k | } else { |
11604 | 135k | if ((ctxt->sax) && (ctxt->sax->setDocumentLocator)) |
11605 | 135k | ctxt->sax->setDocumentLocator(ctxt->userData, |
11606 | 135k | &xmlDefaultSAXLocator); |
11607 | 135k | ctxt->version = xmlCharStrdup(XML_DEFAULT_VERSION); |
11608 | 135k | if (ctxt->version == NULL) { |
11609 | 0 | xmlErrMemory(ctxt, NULL); |
11610 | 0 | break; |
11611 | 0 | } |
11612 | 135k | if ((ctxt->sax) && (ctxt->sax->startDocument) && |
11613 | 135k | (!ctxt->disableSAX)) |
11614 | 135k | ctxt->sax->startDocument(ctxt->userData); |
11615 | 135k | ctxt->instate = XML_PARSER_MISC; |
11616 | | #ifdef DEBUG_PUSH |
11617 | | xmlGenericError(xmlGenericErrorContext, |
11618 | | "PP: entering MISC\n"); |
11619 | | #endif |
11620 | 135k | } |
11621 | 304k | break; |
11622 | 7.13M | case XML_PARSER_START_TAG: { |
11623 | 7.13M | const xmlChar *name; |
11624 | 7.13M | const xmlChar *prefix = NULL; |
11625 | 7.13M | const xmlChar *URI = NULL; |
11626 | 7.13M | int line = ctxt->input->line; |
11627 | 7.13M | int nsNr = ctxt->nsNr; |
11628 | | |
11629 | 7.13M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11630 | 0 | goto done; |
11631 | 7.13M | cur = ctxt->input->cur[0]; |
11632 | 7.13M | if (cur != '<') { |
11633 | 16.2k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_EMPTY, NULL); |
11634 | 16.2k | xmlHaltParser(ctxt); |
11635 | 16.2k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11636 | 16.2k | ctxt->sax->endDocument(ctxt->userData); |
11637 | 16.2k | goto done; |
11638 | 16.2k | } |
11639 | 7.11M | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11640 | 1.26M | goto done; |
11641 | 5.84M | if (ctxt->spaceNr == 0) |
11642 | 50.1k | spacePush(ctxt, -1); |
11643 | 5.79M | else if (*ctxt->space == -2) |
11644 | 682k | spacePush(ctxt, -1); |
11645 | 5.11M | else |
11646 | 5.11M | spacePush(ctxt, *ctxt->space); |
11647 | 5.84M | #ifdef LIBXML_SAX1_ENABLED |
11648 | 5.84M | if (ctxt->sax2) |
11649 | 3.79M | #endif /* LIBXML_SAX1_ENABLED */ |
11650 | 3.79M | name = xmlParseStartTag2(ctxt, &prefix, &URI, &tlen); |
11651 | 2.05M | #ifdef LIBXML_SAX1_ENABLED |
11652 | 2.05M | else |
11653 | 2.05M | name = xmlParseStartTag(ctxt); |
11654 | 5.84M | #endif /* LIBXML_SAX1_ENABLED */ |
11655 | 5.84M | if (ctxt->instate == XML_PARSER_EOF) |
11656 | 289 | goto done; |
11657 | 5.84M | if (name == NULL) { |
11658 | 19.0k | spacePop(ctxt); |
11659 | 19.0k | xmlHaltParser(ctxt); |
11660 | 19.0k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11661 | 19.0k | ctxt->sax->endDocument(ctxt->userData); |
11662 | 19.0k | goto done; |
11663 | 19.0k | } |
11664 | 5.82M | #ifdef LIBXML_VALID_ENABLED |
11665 | | /* |
11666 | | * [ VC: Root Element Type ] |
11667 | | * The Name in the document type declaration must match |
11668 | | * the element type of the root element. |
11669 | | */ |
11670 | 5.82M | if (ctxt->validate && ctxt->wellFormed && ctxt->myDoc && |
11671 | 5.82M | ctxt->node && (ctxt->node == ctxt->myDoc->children)) |
11672 | 0 | ctxt->valid &= xmlValidateRoot(&ctxt->vctxt, ctxt->myDoc); |
11673 | 5.82M | #endif /* LIBXML_VALID_ENABLED */ |
11674 | | |
11675 | | /* |
11676 | | * Check for an Empty Element. |
11677 | | */ |
11678 | 5.82M | if ((RAW == '/') && (NXT(1) == '>')) { |
11679 | 2.52M | SKIP(2); |
11680 | | |
11681 | 2.52M | if (ctxt->sax2) { |
11682 | 1.78M | if ((ctxt->sax != NULL) && |
11683 | 1.78M | (ctxt->sax->endElementNs != NULL) && |
11684 | 1.78M | (!ctxt->disableSAX)) |
11685 | 1.78M | ctxt->sax->endElementNs(ctxt->userData, name, |
11686 | 1.78M | prefix, URI); |
11687 | 1.78M | if (ctxt->nsNr - nsNr > 0) |
11688 | 8.22k | nsPop(ctxt, ctxt->nsNr - nsNr); |
11689 | 1.78M | #ifdef LIBXML_SAX1_ENABLED |
11690 | 1.78M | } else { |
11691 | 733k | if ((ctxt->sax != NULL) && |
11692 | 733k | (ctxt->sax->endElement != NULL) && |
11693 | 733k | (!ctxt->disableSAX)) |
11694 | 733k | ctxt->sax->endElement(ctxt->userData, name); |
11695 | 733k | #endif /* LIBXML_SAX1_ENABLED */ |
11696 | 733k | } |
11697 | 2.52M | if (ctxt->instate == XML_PARSER_EOF) |
11698 | 0 | goto done; |
11699 | 2.52M | spacePop(ctxt); |
11700 | 2.52M | if (ctxt->nameNr == 0) { |
11701 | 8.87k | ctxt->instate = XML_PARSER_EPILOG; |
11702 | 2.51M | } else { |
11703 | 2.51M | ctxt->instate = XML_PARSER_CONTENT; |
11704 | 2.51M | } |
11705 | 2.52M | break; |
11706 | 2.52M | } |
11707 | 3.30M | if (RAW == '>') { |
11708 | 2.97M | NEXT; |
11709 | 2.97M | } else { |
11710 | 338k | xmlFatalErrMsgStr(ctxt, XML_ERR_GT_REQUIRED, |
11711 | 338k | "Couldn't find end of Start Tag %s\n", |
11712 | 338k | name); |
11713 | 338k | nodePop(ctxt); |
11714 | 338k | spacePop(ctxt); |
11715 | 338k | } |
11716 | 3.30M | nameNsPush(ctxt, name, prefix, URI, line, ctxt->nsNr - nsNr); |
11717 | | |
11718 | 3.30M | ctxt->instate = XML_PARSER_CONTENT; |
11719 | 3.30M | break; |
11720 | 5.82M | } |
11721 | 21.2M | case XML_PARSER_CONTENT: { |
11722 | 21.2M | if ((avail < 2) && (ctxt->inputNr == 1)) |
11723 | 91.1k | goto done; |
11724 | 21.2M | cur = ctxt->input->cur[0]; |
11725 | 21.2M | next = ctxt->input->cur[1]; |
11726 | | |
11727 | 21.2M | if ((cur == '<') && (next == '/')) { |
11728 | 2.66M | ctxt->instate = XML_PARSER_END_TAG; |
11729 | 2.66M | break; |
11730 | 18.5M | } else if ((cur == '<') && (next == '?')) { |
11731 | 51.9k | if ((!terminate) && |
11732 | 51.9k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11733 | 21.6k | goto done; |
11734 | 30.2k | xmlParsePI(ctxt); |
11735 | 30.2k | ctxt->instate = XML_PARSER_CONTENT; |
11736 | 18.4M | } else if ((cur == '<') && (next != '!')) { |
11737 | 5.64M | ctxt->instate = XML_PARSER_START_TAG; |
11738 | 5.64M | break; |
11739 | 12.8M | } else if ((cur == '<') && (next == '!') && |
11740 | 12.8M | (ctxt->input->cur[2] == '-') && |
11741 | 12.8M | (ctxt->input->cur[3] == '-')) { |
11742 | 287k | if ((!terminate) && |
11743 | 287k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11744 | 124k | goto done; |
11745 | 163k | xmlParseComment(ctxt); |
11746 | 163k | ctxt->instate = XML_PARSER_CONTENT; |
11747 | 12.5M | } else if ((cur == '<') && (ctxt->input->cur[1] == '!') && |
11748 | 12.5M | (ctxt->input->cur[2] == '[') && |
11749 | 12.5M | (ctxt->input->cur[3] == 'C') && |
11750 | 12.5M | (ctxt->input->cur[4] == 'D') && |
11751 | 12.5M | (ctxt->input->cur[5] == 'A') && |
11752 | 12.5M | (ctxt->input->cur[6] == 'T') && |
11753 | 12.5M | (ctxt->input->cur[7] == 'A') && |
11754 | 12.5M | (ctxt->input->cur[8] == '[')) { |
11755 | 24.7k | SKIP(9); |
11756 | 24.7k | ctxt->instate = XML_PARSER_CDATA_SECTION; |
11757 | 24.7k | break; |
11758 | 12.5M | } else if ((cur == '<') && (next == '!') && |
11759 | 12.5M | (avail < 9)) { |
11760 | 19.5k | goto done; |
11761 | 12.5M | } else if (cur == '<') { |
11762 | 285k | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, |
11763 | 285k | "detected an error in element content\n"); |
11764 | 285k | SKIP(1); |
11765 | 12.2M | } else if (cur == '&') { |
11766 | 2.22M | if ((!terminate) && (!xmlParseLookupChar(ctxt, ';'))) |
11767 | 80.5k | goto done; |
11768 | 2.14M | xmlParseReference(ctxt); |
11769 | 9.99M | } else { |
11770 | | /* TODO Avoid the extra copy, handle directly !!! */ |
11771 | | /* |
11772 | | * Goal of the following test is: |
11773 | | * - minimize calls to the SAX 'character' callback |
11774 | | * when they are mergeable |
11775 | | * - handle an problem for isBlank when we only parse |
11776 | | * a sequence of blank chars and the next one is |
11777 | | * not available to check against '<' presence. |
11778 | | * - tries to homogenize the differences in SAX |
11779 | | * callbacks between the push and pull versions |
11780 | | * of the parser. |
11781 | | */ |
11782 | 9.99M | if ((ctxt->inputNr == 1) && |
11783 | 9.99M | (avail < XML_PARSER_BIG_BUFFER_SIZE)) { |
11784 | 6.99M | if ((!terminate) && (!xmlParseLookupCharData(ctxt))) |
11785 | 731k | goto done; |
11786 | 6.99M | } |
11787 | 9.26M | ctxt->checkIndex = 0; |
11788 | 9.26M | xmlParseCharData(ctxt, 0); |
11789 | 9.26M | } |
11790 | 11.8M | break; |
11791 | 21.2M | } |
11792 | 11.8M | case XML_PARSER_END_TAG: |
11793 | 2.75M | if (avail < 2) |
11794 | 0 | goto done; |
11795 | 2.75M | if ((!terminate) && (!xmlParseLookupChar(ctxt, '>'))) |
11796 | 83.5k | goto done; |
11797 | 2.66M | if (ctxt->sax2) { |
11798 | 1.64M | xmlParseEndTag2(ctxt, &ctxt->pushTab[ctxt->nameNr - 1]); |
11799 | 1.64M | nameNsPop(ctxt); |
11800 | 1.64M | } |
11801 | 1.02M | #ifdef LIBXML_SAX1_ENABLED |
11802 | 1.02M | else |
11803 | 1.02M | xmlParseEndTag1(ctxt, 0); |
11804 | 2.66M | #endif /* LIBXML_SAX1_ENABLED */ |
11805 | 2.66M | if (ctxt->instate == XML_PARSER_EOF) { |
11806 | | /* Nothing */ |
11807 | 2.66M | } else if (ctxt->nameNr == 0) { |
11808 | 48.2k | ctxt->instate = XML_PARSER_EPILOG; |
11809 | 2.61M | } else { |
11810 | 2.61M | ctxt->instate = XML_PARSER_CONTENT; |
11811 | 2.61M | } |
11812 | 2.66M | break; |
11813 | 302k | case XML_PARSER_CDATA_SECTION: { |
11814 | | /* |
11815 | | * The Push mode need to have the SAX callback for |
11816 | | * cdataBlock merge back contiguous callbacks. |
11817 | | */ |
11818 | 302k | const xmlChar *term; |
11819 | | |
11820 | 302k | if (terminate) { |
11821 | | /* |
11822 | | * Don't call xmlParseLookupString. If 'terminate' |
11823 | | * is set, checkIndex is invalid. |
11824 | | */ |
11825 | 5.92k | term = BAD_CAST strstr((const char *) ctxt->input->cur, |
11826 | 5.92k | "]]>"); |
11827 | 296k | } else { |
11828 | 296k | term = xmlParseLookupString(ctxt, 0, "]]>", 3); |
11829 | 296k | } |
11830 | | |
11831 | 302k | if (term == NULL) { |
11832 | 173k | int tmp, size; |
11833 | | |
11834 | 173k | if (terminate) { |
11835 | | /* Unfinished CDATA section */ |
11836 | 2.64k | size = ctxt->input->end - ctxt->input->cur; |
11837 | 170k | } else { |
11838 | 170k | if (avail < XML_PARSER_BIG_BUFFER_SIZE + 2) |
11839 | 67.2k | goto done; |
11840 | 103k | ctxt->checkIndex = 0; |
11841 | | /* XXX: Why don't we pass the full buffer? */ |
11842 | 103k | size = XML_PARSER_BIG_BUFFER_SIZE; |
11843 | 103k | } |
11844 | 106k | tmp = xmlCheckCdataPush(ctxt->input->cur, size, 0); |
11845 | 106k | if (tmp <= 0) { |
11846 | 73.2k | tmp = -tmp; |
11847 | 73.2k | ctxt->input->cur += tmp; |
11848 | 73.2k | goto encoding_error; |
11849 | 73.2k | } |
11850 | 32.8k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX)) { |
11851 | 32.8k | if (ctxt->sax->cdataBlock != NULL) |
11852 | 20.2k | ctxt->sax->cdataBlock(ctxt->userData, |
11853 | 20.2k | ctxt->input->cur, tmp); |
11854 | 12.5k | else if (ctxt->sax->characters != NULL) |
11855 | 12.5k | ctxt->sax->characters(ctxt->userData, |
11856 | 12.5k | ctxt->input->cur, tmp); |
11857 | 32.8k | } |
11858 | 32.8k | if (ctxt->instate == XML_PARSER_EOF) |
11859 | 0 | goto done; |
11860 | 32.8k | SKIPL(tmp); |
11861 | 129k | } else { |
11862 | 129k | int base = term - CUR_PTR; |
11863 | 129k | int tmp; |
11864 | | |
11865 | 129k | tmp = xmlCheckCdataPush(ctxt->input->cur, base, 1); |
11866 | 129k | if ((tmp < 0) || (tmp != base)) { |
11867 | 110k | tmp = -tmp; |
11868 | 110k | ctxt->input->cur += tmp; |
11869 | 110k | goto encoding_error; |
11870 | 110k | } |
11871 | 18.3k | if ((ctxt->sax != NULL) && (base == 0) && |
11872 | 18.3k | (ctxt->sax->cdataBlock != NULL) && |
11873 | 18.3k | (!ctxt->disableSAX)) { |
11874 | | /* |
11875 | | * Special case to provide identical behaviour |
11876 | | * between pull and push parsers on enpty CDATA |
11877 | | * sections |
11878 | | */ |
11879 | 2.16k | if ((ctxt->input->cur - ctxt->input->base >= 9) && |
11880 | 2.16k | (!strncmp((const char *)&ctxt->input->cur[-9], |
11881 | 2.16k | "<![CDATA[", 9))) |
11882 | 2.15k | ctxt->sax->cdataBlock(ctxt->userData, |
11883 | 2.15k | BAD_CAST "", 0); |
11884 | 16.1k | } else if ((ctxt->sax != NULL) && (base > 0) && |
11885 | 16.1k | (!ctxt->disableSAX)) { |
11886 | 14.7k | if (ctxt->sax->cdataBlock != NULL) |
11887 | 9.48k | ctxt->sax->cdataBlock(ctxt->userData, |
11888 | 9.48k | ctxt->input->cur, base); |
11889 | 5.25k | else if (ctxt->sax->characters != NULL) |
11890 | 5.25k | ctxt->sax->characters(ctxt->userData, |
11891 | 5.25k | ctxt->input->cur, base); |
11892 | 14.7k | } |
11893 | 18.3k | if (ctxt->instate == XML_PARSER_EOF) |
11894 | 0 | goto done; |
11895 | 18.3k | SKIPL(base + 3); |
11896 | 18.3k | ctxt->instate = XML_PARSER_CONTENT; |
11897 | | #ifdef DEBUG_PUSH |
11898 | | xmlGenericError(xmlGenericErrorContext, |
11899 | | "PP: entering CONTENT\n"); |
11900 | | #endif |
11901 | 18.3k | } |
11902 | 51.1k | break; |
11903 | 302k | } |
11904 | 536k | case XML_PARSER_MISC: |
11905 | 722k | case XML_PARSER_PROLOG: |
11906 | 785k | case XML_PARSER_EPILOG: |
11907 | 785k | SKIP_BLANKS; |
11908 | 785k | if (ctxt->input->buf == NULL) |
11909 | 0 | avail = ctxt->input->length - |
11910 | 0 | (ctxt->input->cur - ctxt->input->base); |
11911 | 785k | else |
11912 | 785k | avail = xmlBufUse(ctxt->input->buf->buffer) - |
11913 | 785k | (ctxt->input->cur - ctxt->input->base); |
11914 | 785k | if (avail < 2) |
11915 | 50.7k | goto done; |
11916 | 734k | cur = ctxt->input->cur[0]; |
11917 | 734k | next = ctxt->input->cur[1]; |
11918 | 734k | if ((cur == '<') && (next == '?')) { |
11919 | 57.6k | if ((!terminate) && |
11920 | 57.6k | (!xmlParseLookupString(ctxt, 2, "?>", 2))) |
11921 | 19.4k | goto done; |
11922 | | #ifdef DEBUG_PUSH |
11923 | | xmlGenericError(xmlGenericErrorContext, |
11924 | | "PP: Parsing PI\n"); |
11925 | | #endif |
11926 | 38.1k | xmlParsePI(ctxt); |
11927 | 38.1k | if (ctxt->instate == XML_PARSER_EOF) |
11928 | 0 | goto done; |
11929 | 676k | } else if ((cur == '<') && (next == '!') && |
11930 | 676k | (ctxt->input->cur[2] == '-') && |
11931 | 676k | (ctxt->input->cur[3] == '-')) { |
11932 | 74.0k | if ((!terminate) && |
11933 | 74.0k | (!xmlParseLookupString(ctxt, 4, "-->", 3))) |
11934 | 46.2k | goto done; |
11935 | | #ifdef DEBUG_PUSH |
11936 | | xmlGenericError(xmlGenericErrorContext, |
11937 | | "PP: Parsing Comment\n"); |
11938 | | #endif |
11939 | 27.8k | xmlParseComment(ctxt); |
11940 | 27.8k | if (ctxt->instate == XML_PARSER_EOF) |
11941 | 0 | goto done; |
11942 | 602k | } else if ((ctxt->instate == XML_PARSER_MISC) && |
11943 | 602k | (cur == '<') && (next == '!') && |
11944 | 602k | (ctxt->input->cur[2] == 'D') && |
11945 | 602k | (ctxt->input->cur[3] == 'O') && |
11946 | 602k | (ctxt->input->cur[4] == 'C') && |
11947 | 602k | (ctxt->input->cur[5] == 'T') && |
11948 | 602k | (ctxt->input->cur[6] == 'Y') && |
11949 | 602k | (ctxt->input->cur[7] == 'P') && |
11950 | 602k | (ctxt->input->cur[8] == 'E')) { |
11951 | 318k | if ((!terminate) && (!xmlParseLookupGt(ctxt))) |
11952 | 125k | goto done; |
11953 | | #ifdef DEBUG_PUSH |
11954 | | xmlGenericError(xmlGenericErrorContext, |
11955 | | "PP: Parsing internal subset\n"); |
11956 | | #endif |
11957 | 193k | ctxt->inSubset = 1; |
11958 | 193k | xmlParseDocTypeDecl(ctxt); |
11959 | 193k | if (ctxt->instate == XML_PARSER_EOF) |
11960 | 0 | goto done; |
11961 | 193k | if (RAW == '[') { |
11962 | 155k | ctxt->instate = XML_PARSER_DTD; |
11963 | | #ifdef DEBUG_PUSH |
11964 | | xmlGenericError(xmlGenericErrorContext, |
11965 | | "PP: entering DTD\n"); |
11966 | | #endif |
11967 | 155k | } else { |
11968 | | /* |
11969 | | * Create and update the external subset. |
11970 | | */ |
11971 | 37.8k | ctxt->inSubset = 2; |
11972 | 37.8k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
11973 | 37.8k | (ctxt->sax->externalSubset != NULL)) |
11974 | 35.9k | ctxt->sax->externalSubset(ctxt->userData, |
11975 | 35.9k | ctxt->intSubName, ctxt->extSubSystem, |
11976 | 35.9k | ctxt->extSubURI); |
11977 | 37.8k | ctxt->inSubset = 0; |
11978 | 37.8k | xmlCleanSpecialAttr(ctxt); |
11979 | 37.8k | ctxt->instate = XML_PARSER_PROLOG; |
11980 | | #ifdef DEBUG_PUSH |
11981 | | xmlGenericError(xmlGenericErrorContext, |
11982 | | "PP: entering PROLOG\n"); |
11983 | | #endif |
11984 | 37.8k | } |
11985 | 283k | } else if ((cur == '<') && (next == '!') && |
11986 | 283k | (avail < |
11987 | 41.0k | (ctxt->instate == XML_PARSER_MISC ? 9 : 4))) { |
11988 | 38.9k | goto done; |
11989 | 244k | } else if (ctxt->instate == XML_PARSER_EPILOG) { |
11990 | 10.6k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
11991 | 10.6k | xmlHaltParser(ctxt); |
11992 | | #ifdef DEBUG_PUSH |
11993 | | xmlGenericError(xmlGenericErrorContext, |
11994 | | "PP: entering EOF\n"); |
11995 | | #endif |
11996 | 10.6k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
11997 | 10.6k | ctxt->sax->endDocument(ctxt->userData); |
11998 | 10.6k | goto done; |
11999 | 234k | } else { |
12000 | 234k | ctxt->instate = XML_PARSER_START_TAG; |
12001 | | #ifdef DEBUG_PUSH |
12002 | | xmlGenericError(xmlGenericErrorContext, |
12003 | | "PP: entering START_TAG\n"); |
12004 | | #endif |
12005 | 234k | } |
12006 | 493k | break; |
12007 | 558k | case XML_PARSER_DTD: { |
12008 | 558k | if ((!terminate) && (!xmlParseLookupInternalSubset(ctxt))) |
12009 | 404k | goto done; |
12010 | 154k | xmlParseInternalSubset(ctxt); |
12011 | 154k | if (ctxt->instate == XML_PARSER_EOF) |
12012 | 37.6k | goto done; |
12013 | 116k | ctxt->inSubset = 2; |
12014 | 116k | if ((ctxt->sax != NULL) && (!ctxt->disableSAX) && |
12015 | 116k | (ctxt->sax->externalSubset != NULL)) |
12016 | 112k | ctxt->sax->externalSubset(ctxt->userData, ctxt->intSubName, |
12017 | 112k | ctxt->extSubSystem, ctxt->extSubURI); |
12018 | 116k | ctxt->inSubset = 0; |
12019 | 116k | xmlCleanSpecialAttr(ctxt); |
12020 | 116k | if (ctxt->instate == XML_PARSER_EOF) |
12021 | 3.14k | goto done; |
12022 | 113k | ctxt->instate = XML_PARSER_PROLOG; |
12023 | | #ifdef DEBUG_PUSH |
12024 | | xmlGenericError(xmlGenericErrorContext, |
12025 | | "PP: entering PROLOG\n"); |
12026 | | #endif |
12027 | 113k | break; |
12028 | 116k | } |
12029 | 0 | case XML_PARSER_COMMENT: |
12030 | 0 | xmlGenericError(xmlGenericErrorContext, |
12031 | 0 | "PP: internal error, state == COMMENT\n"); |
12032 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12033 | | #ifdef DEBUG_PUSH |
12034 | | xmlGenericError(xmlGenericErrorContext, |
12035 | | "PP: entering CONTENT\n"); |
12036 | | #endif |
12037 | 0 | break; |
12038 | 0 | case XML_PARSER_IGNORE: |
12039 | 0 | xmlGenericError(xmlGenericErrorContext, |
12040 | 0 | "PP: internal error, state == IGNORE"); |
12041 | 0 | ctxt->instate = XML_PARSER_DTD; |
12042 | | #ifdef DEBUG_PUSH |
12043 | | xmlGenericError(xmlGenericErrorContext, |
12044 | | "PP: entering DTD\n"); |
12045 | | #endif |
12046 | 0 | break; |
12047 | 0 | case XML_PARSER_PI: |
12048 | 0 | xmlGenericError(xmlGenericErrorContext, |
12049 | 0 | "PP: internal error, state == PI\n"); |
12050 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12051 | | #ifdef DEBUG_PUSH |
12052 | | xmlGenericError(xmlGenericErrorContext, |
12053 | | "PP: entering CONTENT\n"); |
12054 | | #endif |
12055 | 0 | break; |
12056 | 0 | case XML_PARSER_ENTITY_DECL: |
12057 | 0 | xmlGenericError(xmlGenericErrorContext, |
12058 | 0 | "PP: internal error, state == ENTITY_DECL\n"); |
12059 | 0 | ctxt->instate = XML_PARSER_DTD; |
12060 | | #ifdef DEBUG_PUSH |
12061 | | xmlGenericError(xmlGenericErrorContext, |
12062 | | "PP: entering DTD\n"); |
12063 | | #endif |
12064 | 0 | break; |
12065 | 0 | case XML_PARSER_ENTITY_VALUE: |
12066 | 0 | xmlGenericError(xmlGenericErrorContext, |
12067 | 0 | "PP: internal error, state == ENTITY_VALUE\n"); |
12068 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
12069 | | #ifdef DEBUG_PUSH |
12070 | | xmlGenericError(xmlGenericErrorContext, |
12071 | | "PP: entering DTD\n"); |
12072 | | #endif |
12073 | 0 | break; |
12074 | 0 | case XML_PARSER_ATTRIBUTE_VALUE: |
12075 | 0 | xmlGenericError(xmlGenericErrorContext, |
12076 | 0 | "PP: internal error, state == ATTRIBUTE_VALUE\n"); |
12077 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12078 | | #ifdef DEBUG_PUSH |
12079 | | xmlGenericError(xmlGenericErrorContext, |
12080 | | "PP: entering START_TAG\n"); |
12081 | | #endif |
12082 | 0 | break; |
12083 | 0 | case XML_PARSER_SYSTEM_LITERAL: |
12084 | 0 | xmlGenericError(xmlGenericErrorContext, |
12085 | 0 | "PP: internal error, state == SYSTEM_LITERAL\n"); |
12086 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12087 | | #ifdef DEBUG_PUSH |
12088 | | xmlGenericError(xmlGenericErrorContext, |
12089 | | "PP: entering START_TAG\n"); |
12090 | | #endif |
12091 | 0 | break; |
12092 | 0 | case XML_PARSER_PUBLIC_LITERAL: |
12093 | 0 | xmlGenericError(xmlGenericErrorContext, |
12094 | 0 | "PP: internal error, state == PUBLIC_LITERAL\n"); |
12095 | 0 | ctxt->instate = XML_PARSER_START_TAG; |
12096 | | #ifdef DEBUG_PUSH |
12097 | | xmlGenericError(xmlGenericErrorContext, |
12098 | | "PP: entering START_TAG\n"); |
12099 | | #endif |
12100 | 0 | break; |
12101 | 33.6M | } |
12102 | 33.6M | } |
12103 | 3.73M | done: |
12104 | | #ifdef DEBUG_PUSH |
12105 | | xmlGenericError(xmlGenericErrorContext, "PP: done %d\n", ret); |
12106 | | #endif |
12107 | 3.73M | return(ret); |
12108 | 184k | encoding_error: |
12109 | 184k | { |
12110 | 184k | char buffer[150]; |
12111 | | |
12112 | 184k | snprintf(buffer, 149, "Bytes: 0x%02X 0x%02X 0x%02X 0x%02X\n", |
12113 | 184k | ctxt->input->cur[0], ctxt->input->cur[1], |
12114 | 184k | ctxt->input->cur[2], ctxt->input->cur[3]); |
12115 | 184k | __xmlErrEncoding(ctxt, XML_ERR_INVALID_CHAR, |
12116 | 184k | "Input is not proper UTF-8, indicate encoding !\n%s", |
12117 | 184k | BAD_CAST buffer, NULL); |
12118 | 184k | } |
12119 | 184k | return(0); |
12120 | 3.99M | } |
12121 | | |
12122 | | /** |
12123 | | * xmlParseChunk: |
12124 | | * @ctxt: an XML parser context |
12125 | | * @chunk: an char array |
12126 | | * @size: the size in byte of the chunk |
12127 | | * @terminate: last chunk indicator |
12128 | | * |
12129 | | * Parse a Chunk of memory |
12130 | | * |
12131 | | * Returns zero if no error, the xmlParserErrors otherwise. |
12132 | | */ |
12133 | | int |
12134 | | xmlParseChunk(xmlParserCtxtPtr ctxt, const char *chunk, int size, |
12135 | 5.09M | int terminate) { |
12136 | 5.09M | int end_in_lf = 0; |
12137 | 5.09M | int remain = 0; |
12138 | | |
12139 | 5.09M | if (ctxt == NULL) |
12140 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12141 | 5.09M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12142 | 1.11M | return(ctxt->errNo); |
12143 | 3.97M | if (ctxt->instate == XML_PARSER_EOF) |
12144 | 149 | return(-1); |
12145 | 3.97M | if (ctxt->input == NULL) |
12146 | 0 | return(-1); |
12147 | | |
12148 | 3.97M | ctxt->progressive = 1; |
12149 | 3.97M | if (ctxt->instate == XML_PARSER_START) |
12150 | 534k | xmlDetectSAX2(ctxt); |
12151 | 3.97M | if ((size > 0) && (chunk != NULL) && (!terminate) && |
12152 | 3.97M | (chunk[size - 1] == '\r')) { |
12153 | 29.4k | end_in_lf = 1; |
12154 | 29.4k | size--; |
12155 | 29.4k | } |
12156 | | |
12157 | 3.99M | xmldecl_done: |
12158 | | |
12159 | 3.99M | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
12160 | 3.99M | (ctxt->input->buf != NULL) && (ctxt->instate != XML_PARSER_EOF)) { |
12161 | 3.77M | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12162 | 3.77M | size_t cur = ctxt->input->cur - ctxt->input->base; |
12163 | 3.77M | int res; |
12164 | | |
12165 | | /* |
12166 | | * Specific handling if we autodetected an encoding, we should not |
12167 | | * push more than the first line ... which depend on the encoding |
12168 | | * And only push the rest once the final encoding was detected |
12169 | | */ |
12170 | 3.77M | if ((ctxt->instate == XML_PARSER_START) && (ctxt->input != NULL) && |
12171 | 3.77M | (ctxt->input->buf != NULL) && (ctxt->input->buf->encoder != NULL)) { |
12172 | 35.2k | unsigned int len = 45; |
12173 | | |
12174 | 35.2k | if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12175 | 35.2k | BAD_CAST "UTF-16")) || |
12176 | 35.2k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12177 | 6.98k | BAD_CAST "UTF16"))) |
12178 | 28.2k | len = 90; |
12179 | 6.98k | else if ((xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12180 | 6.98k | BAD_CAST "UCS-4")) || |
12181 | 6.98k | (xmlStrcasestr(BAD_CAST ctxt->input->buf->encoder->name, |
12182 | 6.56k | BAD_CAST "UCS4"))) |
12183 | 418 | len = 180; |
12184 | | |
12185 | 35.2k | if (ctxt->input->buf->rawconsumed < len) |
12186 | 2.04k | len -= ctxt->input->buf->rawconsumed; |
12187 | | |
12188 | | /* |
12189 | | * Change size for reading the initial declaration only |
12190 | | * if size is greater than len. Otherwise, memmove in xmlBufferAdd |
12191 | | * will blindly copy extra bytes from memory. |
12192 | | */ |
12193 | 35.2k | if ((unsigned int) size > len) { |
12194 | 23.4k | remain = size - len; |
12195 | 23.4k | size = len; |
12196 | 23.4k | } else { |
12197 | 11.7k | remain = 0; |
12198 | 11.7k | } |
12199 | 35.2k | } |
12200 | 3.77M | res = xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12201 | 3.77M | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12202 | 3.77M | if (res < 0) { |
12203 | 445 | ctxt->errNo = XML_PARSER_EOF; |
12204 | 445 | xmlHaltParser(ctxt); |
12205 | 445 | return (XML_PARSER_EOF); |
12206 | 445 | } |
12207 | | #ifdef DEBUG_PUSH |
12208 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12209 | | #endif |
12210 | | |
12211 | 3.77M | } else if (ctxt->instate != XML_PARSER_EOF) { |
12212 | 222k | if ((ctxt->input != NULL) && ctxt->input->buf != NULL) { |
12213 | 222k | xmlParserInputBufferPtr in = ctxt->input->buf; |
12214 | 222k | if ((in->encoder != NULL) && (in->buffer != NULL) && |
12215 | 222k | (in->raw != NULL)) { |
12216 | 13.5k | int nbchars; |
12217 | 13.5k | size_t base = xmlBufGetInputBase(in->buffer, ctxt->input); |
12218 | 13.5k | size_t current = ctxt->input->cur - ctxt->input->base; |
12219 | | |
12220 | 13.5k | nbchars = xmlCharEncInput(in, terminate); |
12221 | 13.5k | xmlBufSetInputBaseCur(in->buffer, ctxt->input, base, current); |
12222 | 13.5k | if (nbchars < 0) { |
12223 | | /* TODO 2.6.0 */ |
12224 | 299 | xmlGenericError(xmlGenericErrorContext, |
12225 | 299 | "xmlParseChunk: encoder error\n"); |
12226 | 299 | xmlHaltParser(ctxt); |
12227 | 299 | return(XML_ERR_INVALID_ENCODING); |
12228 | 299 | } |
12229 | 13.5k | } |
12230 | 222k | } |
12231 | 222k | } |
12232 | | |
12233 | 3.99M | if (remain != 0) { |
12234 | 23.3k | xmlParseTryOrFinish(ctxt, 0); |
12235 | 3.97M | } else { |
12236 | 3.97M | xmlParseTryOrFinish(ctxt, terminate); |
12237 | 3.97M | } |
12238 | 3.99M | if (ctxt->instate == XML_PARSER_EOF) |
12239 | 88.2k | return(ctxt->errNo); |
12240 | | |
12241 | 3.90M | if ((ctxt->input != NULL) && |
12242 | 3.90M | (((ctxt->input->end - ctxt->input->cur) > XML_MAX_LOOKUP_LIMIT) || |
12243 | 3.90M | ((ctxt->input->cur - ctxt->input->base) > XML_MAX_LOOKUP_LIMIT)) && |
12244 | 3.90M | ((ctxt->options & XML_PARSE_HUGE) == 0)) { |
12245 | 0 | xmlFatalErr(ctxt, XML_ERR_INTERNAL_ERROR, "Huge input lookup"); |
12246 | 0 | xmlHaltParser(ctxt); |
12247 | 0 | } |
12248 | 3.90M | if ((ctxt->errNo != XML_ERR_OK) && (ctxt->disableSAX == 1)) |
12249 | 76.1k | return(ctxt->errNo); |
12250 | | |
12251 | 3.83M | if (remain != 0) { |
12252 | 22.9k | chunk += size; |
12253 | 22.9k | size = remain; |
12254 | 22.9k | remain = 0; |
12255 | 22.9k | goto xmldecl_done; |
12256 | 22.9k | } |
12257 | 3.80M | if ((end_in_lf == 1) && (ctxt->input != NULL) && |
12258 | 3.80M | (ctxt->input->buf != NULL)) { |
12259 | 28.9k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, |
12260 | 28.9k | ctxt->input); |
12261 | 28.9k | size_t current = ctxt->input->cur - ctxt->input->base; |
12262 | | |
12263 | 28.9k | xmlParserInputBufferPush(ctxt->input->buf, 1, "\r"); |
12264 | | |
12265 | 28.9k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, |
12266 | 28.9k | base, current); |
12267 | 28.9k | } |
12268 | 3.80M | if (terminate) { |
12269 | | /* |
12270 | | * Check for termination |
12271 | | */ |
12272 | 101k | int cur_avail = 0; |
12273 | | |
12274 | 101k | if (ctxt->input != NULL) { |
12275 | 101k | if (ctxt->input->buf == NULL) |
12276 | 0 | cur_avail = ctxt->input->length - |
12277 | 0 | (ctxt->input->cur - ctxt->input->base); |
12278 | 101k | else |
12279 | 101k | cur_avail = xmlBufUse(ctxt->input->buf->buffer) - |
12280 | 101k | (ctxt->input->cur - ctxt->input->base); |
12281 | 101k | } |
12282 | | |
12283 | 101k | if ((ctxt->instate != XML_PARSER_EOF) && |
12284 | 101k | (ctxt->instate != XML_PARSER_EPILOG)) { |
12285 | 62.1k | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12286 | 62.1k | } |
12287 | 101k | if ((ctxt->instate == XML_PARSER_EPILOG) && (cur_avail > 0)) { |
12288 | 414 | xmlFatalErr(ctxt, XML_ERR_DOCUMENT_END, NULL); |
12289 | 414 | } |
12290 | 101k | if (ctxt->instate != XML_PARSER_EOF) { |
12291 | 101k | if ((ctxt->sax) && (ctxt->sax->endDocument != NULL)) |
12292 | 101k | ctxt->sax->endDocument(ctxt->userData); |
12293 | 101k | } |
12294 | 101k | ctxt->instate = XML_PARSER_EOF; |
12295 | 101k | } |
12296 | 3.80M | if (ctxt->wellFormed == 0) |
12297 | 1.28M | return((xmlParserErrors) ctxt->errNo); |
12298 | 2.52M | else |
12299 | 2.52M | return(0); |
12300 | 3.80M | } |
12301 | | |
12302 | | /************************************************************************ |
12303 | | * * |
12304 | | * I/O front end functions to the parser * |
12305 | | * * |
12306 | | ************************************************************************/ |
12307 | | |
12308 | | /** |
12309 | | * xmlCreatePushParserCtxt: |
12310 | | * @sax: a SAX handler |
12311 | | * @user_data: The user data returned on SAX callbacks |
12312 | | * @chunk: a pointer to an array of chars |
12313 | | * @size: number of chars in the array |
12314 | | * @filename: an optional file name or URI |
12315 | | * |
12316 | | * Create a parser context for using the XML parser in push mode. |
12317 | | * If @buffer and @size are non-NULL, the data is used to detect |
12318 | | * the encoding. The remaining characters will be parsed so they |
12319 | | * don't need to be fed in again through xmlParseChunk. |
12320 | | * To allow content encoding detection, @size should be >= 4 |
12321 | | * The value of @filename is used for fetching external entities |
12322 | | * and error/warning reports. |
12323 | | * |
12324 | | * Returns the new parser context or NULL |
12325 | | */ |
12326 | | |
12327 | | xmlParserCtxtPtr |
12328 | | xmlCreatePushParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12329 | 307k | const char *chunk, int size, const char *filename) { |
12330 | 307k | xmlParserCtxtPtr ctxt; |
12331 | 307k | xmlParserInputPtr inputStream; |
12332 | 307k | xmlParserInputBufferPtr buf; |
12333 | 307k | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
12334 | | |
12335 | | /* |
12336 | | * plug some encoding conversion routines |
12337 | | */ |
12338 | 307k | if ((chunk != NULL) && (size >= 4)) |
12339 | 152k | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
12340 | | |
12341 | 307k | buf = xmlAllocParserInputBuffer(enc); |
12342 | 307k | if (buf == NULL) return(NULL); |
12343 | | |
12344 | 307k | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12345 | 307k | if (ctxt == NULL) { |
12346 | 0 | xmlErrMemory(NULL, "creating parser: out of memory\n"); |
12347 | 0 | xmlFreeParserInputBuffer(buf); |
12348 | 0 | return(NULL); |
12349 | 0 | } |
12350 | 307k | ctxt->dictNames = 1; |
12351 | 307k | if (filename == NULL) { |
12352 | 153k | ctxt->directory = NULL; |
12353 | 153k | } else { |
12354 | 153k | ctxt->directory = xmlParserGetDirectory(filename); |
12355 | 153k | } |
12356 | | |
12357 | 307k | inputStream = xmlNewInputStream(ctxt); |
12358 | 307k | if (inputStream == NULL) { |
12359 | 0 | xmlFreeParserCtxt(ctxt); |
12360 | 0 | xmlFreeParserInputBuffer(buf); |
12361 | 0 | return(NULL); |
12362 | 0 | } |
12363 | | |
12364 | 307k | if (filename == NULL) |
12365 | 153k | inputStream->filename = NULL; |
12366 | 153k | else { |
12367 | 153k | inputStream->filename = (char *) |
12368 | 153k | xmlCanonicPath((const xmlChar *) filename); |
12369 | 153k | if (inputStream->filename == NULL) { |
12370 | 0 | xmlFreeInputStream(inputStream); |
12371 | 0 | xmlFreeParserCtxt(ctxt); |
12372 | 0 | xmlFreeParserInputBuffer(buf); |
12373 | 0 | return(NULL); |
12374 | 0 | } |
12375 | 153k | } |
12376 | 307k | inputStream->buf = buf; |
12377 | 307k | xmlBufResetInput(inputStream->buf->buffer, inputStream); |
12378 | 307k | inputPush(ctxt, inputStream); |
12379 | | |
12380 | | /* |
12381 | | * If the caller didn't provide an initial 'chunk' for determining |
12382 | | * the encoding, we set the context to XML_CHAR_ENCODING_NONE so |
12383 | | * that it can be automatically determined later |
12384 | | */ |
12385 | 307k | ctxt->charset = XML_CHAR_ENCODING_NONE; |
12386 | | |
12387 | 307k | if ((size != 0) && (chunk != NULL) && |
12388 | 307k | (ctxt->input != NULL) && (ctxt->input->buf != NULL)) { |
12389 | 152k | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
12390 | 152k | size_t cur = ctxt->input->cur - ctxt->input->base; |
12391 | | |
12392 | 152k | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
12393 | | |
12394 | 152k | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
12395 | | #ifdef DEBUG_PUSH |
12396 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
12397 | | #endif |
12398 | 152k | } |
12399 | | |
12400 | 307k | if (enc != XML_CHAR_ENCODING_NONE) { |
12401 | 85.0k | xmlSwitchEncoding(ctxt, enc); |
12402 | 85.0k | } |
12403 | | |
12404 | 307k | return(ctxt); |
12405 | 307k | } |
12406 | | #endif /* LIBXML_PUSH_ENABLED */ |
12407 | | |
12408 | | /** |
12409 | | * xmlHaltParser: |
12410 | | * @ctxt: an XML parser context |
12411 | | * |
12412 | | * Blocks further parser processing don't override error |
12413 | | * for internal use |
12414 | | */ |
12415 | | static void |
12416 | 280k | xmlHaltParser(xmlParserCtxtPtr ctxt) { |
12417 | 280k | if (ctxt == NULL) |
12418 | 0 | return; |
12419 | 280k | ctxt->instate = XML_PARSER_EOF; |
12420 | 280k | ctxt->disableSAX = 1; |
12421 | 295k | while (ctxt->inputNr > 1) |
12422 | 15.5k | xmlFreeInputStream(inputPop(ctxt)); |
12423 | 280k | if (ctxt->input != NULL) { |
12424 | | /* |
12425 | | * in case there was a specific allocation deallocate before |
12426 | | * overriding base |
12427 | | */ |
12428 | 280k | if (ctxt->input->free != NULL) { |
12429 | 0 | ctxt->input->free((xmlChar *) ctxt->input->base); |
12430 | 0 | ctxt->input->free = NULL; |
12431 | 0 | } |
12432 | 280k | if (ctxt->input->buf != NULL) { |
12433 | 240k | xmlFreeParserInputBuffer(ctxt->input->buf); |
12434 | 240k | ctxt->input->buf = NULL; |
12435 | 240k | } |
12436 | 280k | ctxt->input->cur = BAD_CAST""; |
12437 | 280k | ctxt->input->length = 0; |
12438 | 280k | ctxt->input->base = ctxt->input->cur; |
12439 | 280k | ctxt->input->end = ctxt->input->cur; |
12440 | 280k | } |
12441 | 280k | } |
12442 | | |
12443 | | /** |
12444 | | * xmlStopParser: |
12445 | | * @ctxt: an XML parser context |
12446 | | * |
12447 | | * Blocks further parser processing |
12448 | | */ |
12449 | | void |
12450 | 154k | xmlStopParser(xmlParserCtxtPtr ctxt) { |
12451 | 154k | if (ctxt == NULL) |
12452 | 0 | return; |
12453 | 154k | xmlHaltParser(ctxt); |
12454 | 154k | ctxt->errNo = XML_ERR_USER_STOP; |
12455 | 154k | } |
12456 | | |
12457 | | /** |
12458 | | * xmlCreateIOParserCtxt: |
12459 | | * @sax: a SAX handler |
12460 | | * @user_data: The user data returned on SAX callbacks |
12461 | | * @ioread: an I/O read function |
12462 | | * @ioclose: an I/O close function |
12463 | | * @ioctx: an I/O handler |
12464 | | * @enc: the charset encoding if known |
12465 | | * |
12466 | | * Create a parser context for using the XML parser with an existing |
12467 | | * I/O stream |
12468 | | * |
12469 | | * Returns the new parser context or NULL |
12470 | | */ |
12471 | | xmlParserCtxtPtr |
12472 | | xmlCreateIOParserCtxt(xmlSAXHandlerPtr sax, void *user_data, |
12473 | | xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
12474 | 0 | void *ioctx, xmlCharEncoding enc) { |
12475 | 0 | xmlParserCtxtPtr ctxt; |
12476 | 0 | xmlParserInputPtr inputStream; |
12477 | 0 | xmlParserInputBufferPtr buf; |
12478 | |
|
12479 | 0 | if (ioread == NULL) return(NULL); |
12480 | | |
12481 | 0 | buf = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, enc); |
12482 | 0 | if (buf == NULL) { |
12483 | 0 | if (ioclose != NULL) |
12484 | 0 | ioclose(ioctx); |
12485 | 0 | return (NULL); |
12486 | 0 | } |
12487 | | |
12488 | 0 | ctxt = xmlNewSAXParserCtxt(sax, user_data); |
12489 | 0 | if (ctxt == NULL) { |
12490 | 0 | xmlFreeParserInputBuffer(buf); |
12491 | 0 | return(NULL); |
12492 | 0 | } |
12493 | | |
12494 | 0 | inputStream = xmlNewIOInputStream(ctxt, buf, enc); |
12495 | 0 | if (inputStream == NULL) { |
12496 | 0 | xmlFreeParserCtxt(ctxt); |
12497 | 0 | return(NULL); |
12498 | 0 | } |
12499 | 0 | inputPush(ctxt, inputStream); |
12500 | |
|
12501 | 0 | return(ctxt); |
12502 | 0 | } |
12503 | | |
12504 | | #ifdef LIBXML_VALID_ENABLED |
12505 | | /************************************************************************ |
12506 | | * * |
12507 | | * Front ends when parsing a DTD * |
12508 | | * * |
12509 | | ************************************************************************/ |
12510 | | |
12511 | | /** |
12512 | | * xmlIOParseDTD: |
12513 | | * @sax: the SAX handler block or NULL |
12514 | | * @input: an Input Buffer |
12515 | | * @enc: the charset encoding if known |
12516 | | * |
12517 | | * Load and parse a DTD |
12518 | | * |
12519 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12520 | | * @input will be freed by the function in any case. |
12521 | | */ |
12522 | | |
12523 | | xmlDtdPtr |
12524 | | xmlIOParseDTD(xmlSAXHandlerPtr sax, xmlParserInputBufferPtr input, |
12525 | 0 | xmlCharEncoding enc) { |
12526 | 0 | xmlDtdPtr ret = NULL; |
12527 | 0 | xmlParserCtxtPtr ctxt; |
12528 | 0 | xmlParserInputPtr pinput = NULL; |
12529 | 0 | xmlChar start[4]; |
12530 | |
|
12531 | 0 | if (input == NULL) |
12532 | 0 | return(NULL); |
12533 | | |
12534 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12535 | 0 | if (ctxt == NULL) { |
12536 | 0 | xmlFreeParserInputBuffer(input); |
12537 | 0 | return(NULL); |
12538 | 0 | } |
12539 | | |
12540 | | /* We are loading a DTD */ |
12541 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12542 | |
|
12543 | 0 | xmlDetectSAX2(ctxt); |
12544 | | |
12545 | | /* |
12546 | | * generate a parser input from the I/O handler |
12547 | | */ |
12548 | |
|
12549 | 0 | pinput = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
12550 | 0 | if (pinput == NULL) { |
12551 | 0 | xmlFreeParserInputBuffer(input); |
12552 | 0 | xmlFreeParserCtxt(ctxt); |
12553 | 0 | return(NULL); |
12554 | 0 | } |
12555 | | |
12556 | | /* |
12557 | | * plug some encoding conversion routines here. |
12558 | | */ |
12559 | 0 | if (xmlPushInput(ctxt, pinput) < 0) { |
12560 | 0 | xmlFreeParserCtxt(ctxt); |
12561 | 0 | return(NULL); |
12562 | 0 | } |
12563 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12564 | 0 | xmlSwitchEncoding(ctxt, enc); |
12565 | 0 | } |
12566 | |
|
12567 | 0 | pinput->filename = NULL; |
12568 | 0 | pinput->line = 1; |
12569 | 0 | pinput->col = 1; |
12570 | 0 | pinput->base = ctxt->input->cur; |
12571 | 0 | pinput->cur = ctxt->input->cur; |
12572 | 0 | pinput->free = NULL; |
12573 | | |
12574 | | /* |
12575 | | * let's parse that entity knowing it's an external subset. |
12576 | | */ |
12577 | 0 | ctxt->inSubset = 2; |
12578 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12579 | 0 | if (ctxt->myDoc == NULL) { |
12580 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12581 | 0 | return(NULL); |
12582 | 0 | } |
12583 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12584 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12585 | 0 | BAD_CAST "none", BAD_CAST "none"); |
12586 | |
|
12587 | 0 | if ((enc == XML_CHAR_ENCODING_NONE) && |
12588 | 0 | ((ctxt->input->end - ctxt->input->cur) >= 4)) { |
12589 | | /* |
12590 | | * Get the 4 first bytes and decode the charset |
12591 | | * if enc != XML_CHAR_ENCODING_NONE |
12592 | | * plug some encoding conversion routines. |
12593 | | */ |
12594 | 0 | start[0] = RAW; |
12595 | 0 | start[1] = NXT(1); |
12596 | 0 | start[2] = NXT(2); |
12597 | 0 | start[3] = NXT(3); |
12598 | 0 | enc = xmlDetectCharEncoding(start, 4); |
12599 | 0 | if (enc != XML_CHAR_ENCODING_NONE) { |
12600 | 0 | xmlSwitchEncoding(ctxt, enc); |
12601 | 0 | } |
12602 | 0 | } |
12603 | |
|
12604 | 0 | xmlParseExternalSubset(ctxt, BAD_CAST "none", BAD_CAST "none"); |
12605 | |
|
12606 | 0 | if (ctxt->myDoc != NULL) { |
12607 | 0 | if (ctxt->wellFormed) { |
12608 | 0 | ret = ctxt->myDoc->extSubset; |
12609 | 0 | ctxt->myDoc->extSubset = NULL; |
12610 | 0 | if (ret != NULL) { |
12611 | 0 | xmlNodePtr tmp; |
12612 | |
|
12613 | 0 | ret->doc = NULL; |
12614 | 0 | tmp = ret->children; |
12615 | 0 | while (tmp != NULL) { |
12616 | 0 | tmp->doc = NULL; |
12617 | 0 | tmp = tmp->next; |
12618 | 0 | } |
12619 | 0 | } |
12620 | 0 | } else { |
12621 | 0 | ret = NULL; |
12622 | 0 | } |
12623 | 0 | xmlFreeDoc(ctxt->myDoc); |
12624 | 0 | ctxt->myDoc = NULL; |
12625 | 0 | } |
12626 | 0 | xmlFreeParserCtxt(ctxt); |
12627 | |
|
12628 | 0 | return(ret); |
12629 | 0 | } |
12630 | | |
12631 | | /** |
12632 | | * xmlSAXParseDTD: |
12633 | | * @sax: the SAX handler block |
12634 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12635 | | * @SystemID: a NAME* containing the URL to the DTD |
12636 | | * |
12637 | | * DEPRECATED: Don't use. |
12638 | | * |
12639 | | * Load and parse an external subset. |
12640 | | * |
12641 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12642 | | */ |
12643 | | |
12644 | | xmlDtdPtr |
12645 | | xmlSAXParseDTD(xmlSAXHandlerPtr sax, const xmlChar *ExternalID, |
12646 | 0 | const xmlChar *SystemID) { |
12647 | 0 | xmlDtdPtr ret = NULL; |
12648 | 0 | xmlParserCtxtPtr ctxt; |
12649 | 0 | xmlParserInputPtr input = NULL; |
12650 | 0 | xmlCharEncoding enc; |
12651 | 0 | xmlChar* systemIdCanonic; |
12652 | |
|
12653 | 0 | if ((ExternalID == NULL) && (SystemID == NULL)) return(NULL); |
12654 | | |
12655 | 0 | ctxt = xmlNewSAXParserCtxt(sax, NULL); |
12656 | 0 | if (ctxt == NULL) { |
12657 | 0 | return(NULL); |
12658 | 0 | } |
12659 | | |
12660 | | /* We are loading a DTD */ |
12661 | 0 | ctxt->options |= XML_PARSE_DTDLOAD; |
12662 | | |
12663 | | /* |
12664 | | * Canonicalise the system ID |
12665 | | */ |
12666 | 0 | systemIdCanonic = xmlCanonicPath(SystemID); |
12667 | 0 | if ((SystemID != NULL) && (systemIdCanonic == NULL)) { |
12668 | 0 | xmlFreeParserCtxt(ctxt); |
12669 | 0 | return(NULL); |
12670 | 0 | } |
12671 | | |
12672 | | /* |
12673 | | * Ask the Entity resolver to load the damn thing |
12674 | | */ |
12675 | | |
12676 | 0 | if ((ctxt->sax != NULL) && (ctxt->sax->resolveEntity != NULL)) |
12677 | 0 | input = ctxt->sax->resolveEntity(ctxt->userData, ExternalID, |
12678 | 0 | systemIdCanonic); |
12679 | 0 | if (input == NULL) { |
12680 | 0 | xmlFreeParserCtxt(ctxt); |
12681 | 0 | if (systemIdCanonic != NULL) |
12682 | 0 | xmlFree(systemIdCanonic); |
12683 | 0 | return(NULL); |
12684 | 0 | } |
12685 | | |
12686 | | /* |
12687 | | * plug some encoding conversion routines here. |
12688 | | */ |
12689 | 0 | if (xmlPushInput(ctxt, input) < 0) { |
12690 | 0 | xmlFreeParserCtxt(ctxt); |
12691 | 0 | if (systemIdCanonic != NULL) |
12692 | 0 | xmlFree(systemIdCanonic); |
12693 | 0 | return(NULL); |
12694 | 0 | } |
12695 | 0 | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
12696 | 0 | enc = xmlDetectCharEncoding(ctxt->input->cur, 4); |
12697 | 0 | xmlSwitchEncoding(ctxt, enc); |
12698 | 0 | } |
12699 | |
|
12700 | 0 | if (input->filename == NULL) |
12701 | 0 | input->filename = (char *) systemIdCanonic; |
12702 | 0 | else |
12703 | 0 | xmlFree(systemIdCanonic); |
12704 | 0 | input->line = 1; |
12705 | 0 | input->col = 1; |
12706 | 0 | input->base = ctxt->input->cur; |
12707 | 0 | input->cur = ctxt->input->cur; |
12708 | 0 | input->free = NULL; |
12709 | | |
12710 | | /* |
12711 | | * let's parse that entity knowing it's an external subset. |
12712 | | */ |
12713 | 0 | ctxt->inSubset = 2; |
12714 | 0 | ctxt->myDoc = xmlNewDoc(BAD_CAST "1.0"); |
12715 | 0 | if (ctxt->myDoc == NULL) { |
12716 | 0 | xmlErrMemory(ctxt, "New Doc failed"); |
12717 | 0 | xmlFreeParserCtxt(ctxt); |
12718 | 0 | return(NULL); |
12719 | 0 | } |
12720 | 0 | ctxt->myDoc->properties = XML_DOC_INTERNAL; |
12721 | 0 | ctxt->myDoc->extSubset = xmlNewDtd(ctxt->myDoc, BAD_CAST "none", |
12722 | 0 | ExternalID, SystemID); |
12723 | 0 | xmlParseExternalSubset(ctxt, ExternalID, SystemID); |
12724 | |
|
12725 | 0 | if (ctxt->myDoc != NULL) { |
12726 | 0 | if (ctxt->wellFormed) { |
12727 | 0 | ret = ctxt->myDoc->extSubset; |
12728 | 0 | ctxt->myDoc->extSubset = NULL; |
12729 | 0 | if (ret != NULL) { |
12730 | 0 | xmlNodePtr tmp; |
12731 | |
|
12732 | 0 | ret->doc = NULL; |
12733 | 0 | tmp = ret->children; |
12734 | 0 | while (tmp != NULL) { |
12735 | 0 | tmp->doc = NULL; |
12736 | 0 | tmp = tmp->next; |
12737 | 0 | } |
12738 | 0 | } |
12739 | 0 | } else { |
12740 | 0 | ret = NULL; |
12741 | 0 | } |
12742 | 0 | xmlFreeDoc(ctxt->myDoc); |
12743 | 0 | ctxt->myDoc = NULL; |
12744 | 0 | } |
12745 | 0 | xmlFreeParserCtxt(ctxt); |
12746 | |
|
12747 | 0 | return(ret); |
12748 | 0 | } |
12749 | | |
12750 | | |
12751 | | /** |
12752 | | * xmlParseDTD: |
12753 | | * @ExternalID: a NAME* containing the External ID of the DTD |
12754 | | * @SystemID: a NAME* containing the URL to the DTD |
12755 | | * |
12756 | | * Load and parse an external subset. |
12757 | | * |
12758 | | * Returns the resulting xmlDtdPtr or NULL in case of error. |
12759 | | */ |
12760 | | |
12761 | | xmlDtdPtr |
12762 | 0 | xmlParseDTD(const xmlChar *ExternalID, const xmlChar *SystemID) { |
12763 | 0 | return(xmlSAXParseDTD(NULL, ExternalID, SystemID)); |
12764 | 0 | } |
12765 | | #endif /* LIBXML_VALID_ENABLED */ |
12766 | | |
12767 | | /************************************************************************ |
12768 | | * * |
12769 | | * Front ends when parsing an Entity * |
12770 | | * * |
12771 | | ************************************************************************/ |
12772 | | |
12773 | | /** |
12774 | | * xmlParseCtxtExternalEntity: |
12775 | | * @ctx: the existing parsing context |
12776 | | * @URL: the URL for the entity to load |
12777 | | * @ID: the System ID for the entity to load |
12778 | | * @lst: the return value for the set of parsed nodes |
12779 | | * |
12780 | | * Parse an external general entity within an existing parsing context |
12781 | | * An external general parsed entity is well-formed if it matches the |
12782 | | * production labeled extParsedEnt. |
12783 | | * |
12784 | | * [78] extParsedEnt ::= TextDecl? content |
12785 | | * |
12786 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12787 | | * the parser error code otherwise |
12788 | | */ |
12789 | | |
12790 | | int |
12791 | | xmlParseCtxtExternalEntity(xmlParserCtxtPtr ctx, const xmlChar *URL, |
12792 | 0 | const xmlChar *ID, xmlNodePtr *lst) { |
12793 | 0 | void *userData; |
12794 | |
|
12795 | 0 | if (ctx == NULL) return(-1); |
12796 | | /* |
12797 | | * If the user provided their own SAX callbacks, then reuse the |
12798 | | * userData callback field, otherwise the expected setup in a |
12799 | | * DOM builder is to have userData == ctxt |
12800 | | */ |
12801 | 0 | if (ctx->userData == ctx) |
12802 | 0 | userData = NULL; |
12803 | 0 | else |
12804 | 0 | userData = ctx->userData; |
12805 | 0 | return xmlParseExternalEntityPrivate(ctx->myDoc, ctx, ctx->sax, |
12806 | 0 | userData, ctx->depth + 1, |
12807 | 0 | URL, ID, lst); |
12808 | 0 | } |
12809 | | |
12810 | | /** |
12811 | | * xmlParseExternalEntityPrivate: |
12812 | | * @doc: the document the chunk pertains to |
12813 | | * @oldctxt: the previous parser context if available |
12814 | | * @sax: the SAX handler block (possibly NULL) |
12815 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
12816 | | * @depth: Used for loop detection, use 0 |
12817 | | * @URL: the URL for the entity to load |
12818 | | * @ID: the System ID for the entity to load |
12819 | | * @list: the return value for the set of parsed nodes |
12820 | | * |
12821 | | * Private version of xmlParseExternalEntity() |
12822 | | * |
12823 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
12824 | | * the parser error code otherwise |
12825 | | */ |
12826 | | |
12827 | | static xmlParserErrors |
12828 | | xmlParseExternalEntityPrivate(xmlDocPtr doc, xmlParserCtxtPtr oldctxt, |
12829 | | xmlSAXHandlerPtr sax, |
12830 | | void *user_data, int depth, const xmlChar *URL, |
12831 | 686k | const xmlChar *ID, xmlNodePtr *list) { |
12832 | 686k | xmlParserCtxtPtr ctxt; |
12833 | 686k | xmlDocPtr newDoc; |
12834 | 686k | xmlNodePtr newRoot; |
12835 | 686k | xmlParserErrors ret = XML_ERR_OK; |
12836 | 686k | xmlChar start[4]; |
12837 | 686k | xmlCharEncoding enc; |
12838 | | |
12839 | 686k | if (((depth > 40) && |
12840 | 686k | ((oldctxt == NULL) || (oldctxt->options & XML_PARSE_HUGE) == 0)) || |
12841 | 686k | (depth > 100)) { |
12842 | 0 | xmlFatalErrMsg(oldctxt, XML_ERR_ENTITY_LOOP, |
12843 | 0 | "Maximum entity nesting depth exceeded"); |
12844 | 0 | return(XML_ERR_ENTITY_LOOP); |
12845 | 0 | } |
12846 | | |
12847 | 686k | if (list != NULL) |
12848 | 64.5k | *list = NULL; |
12849 | 686k | if ((URL == NULL) && (ID == NULL)) |
12850 | 534 | return(XML_ERR_INTERNAL_ERROR); |
12851 | 685k | if (doc == NULL) |
12852 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12853 | | |
12854 | 685k | ctxt = xmlCreateEntityParserCtxtInternal(sax, user_data, URL, ID, NULL, |
12855 | 685k | oldctxt); |
12856 | 685k | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
12857 | 92.0k | if (oldctxt != NULL) { |
12858 | 92.0k | ctxt->nbErrors = oldctxt->nbErrors; |
12859 | 92.0k | ctxt->nbWarnings = oldctxt->nbWarnings; |
12860 | 92.0k | } |
12861 | 92.0k | xmlDetectSAX2(ctxt); |
12862 | | |
12863 | 92.0k | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
12864 | 92.0k | if (newDoc == NULL) { |
12865 | 0 | xmlFreeParserCtxt(ctxt); |
12866 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12867 | 0 | } |
12868 | 92.0k | newDoc->properties = XML_DOC_INTERNAL; |
12869 | 92.0k | if (doc) { |
12870 | 92.0k | newDoc->intSubset = doc->intSubset; |
12871 | 92.0k | newDoc->extSubset = doc->extSubset; |
12872 | 92.0k | if (doc->dict) { |
12873 | 47.5k | newDoc->dict = doc->dict; |
12874 | 47.5k | xmlDictReference(newDoc->dict); |
12875 | 47.5k | } |
12876 | 92.0k | if (doc->URL != NULL) { |
12877 | 60.6k | newDoc->URL = xmlStrdup(doc->URL); |
12878 | 60.6k | } |
12879 | 92.0k | } |
12880 | 92.0k | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
12881 | 92.0k | if (newRoot == NULL) { |
12882 | 0 | if (sax != NULL) |
12883 | 0 | xmlFreeParserCtxt(ctxt); |
12884 | 0 | newDoc->intSubset = NULL; |
12885 | 0 | newDoc->extSubset = NULL; |
12886 | 0 | xmlFreeDoc(newDoc); |
12887 | 0 | return(XML_ERR_INTERNAL_ERROR); |
12888 | 0 | } |
12889 | 92.0k | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
12890 | 92.0k | nodePush(ctxt, newDoc->children); |
12891 | 92.0k | if (doc == NULL) { |
12892 | 0 | ctxt->myDoc = newDoc; |
12893 | 92.0k | } else { |
12894 | 92.0k | ctxt->myDoc = doc; |
12895 | 92.0k | newRoot->doc = doc; |
12896 | 92.0k | } |
12897 | | |
12898 | | /* |
12899 | | * Get the 4 first bytes and decode the charset |
12900 | | * if enc != XML_CHAR_ENCODING_NONE |
12901 | | * plug some encoding conversion routines. |
12902 | | */ |
12903 | 92.0k | GROW; |
12904 | 92.0k | if ((ctxt->input->end - ctxt->input->cur) >= 4) { |
12905 | 87.7k | start[0] = RAW; |
12906 | 87.7k | start[1] = NXT(1); |
12907 | 87.7k | start[2] = NXT(2); |
12908 | 87.7k | start[3] = NXT(3); |
12909 | 87.7k | enc = xmlDetectCharEncoding(start, 4); |
12910 | 87.7k | if (enc != XML_CHAR_ENCODING_NONE) { |
12911 | 3.81k | xmlSwitchEncoding(ctxt, enc); |
12912 | 3.81k | } |
12913 | 87.7k | } |
12914 | | |
12915 | | /* |
12916 | | * Parse a possible text declaration first |
12917 | | */ |
12918 | 92.0k | if ((CMP5(CUR_PTR, '<', '?', 'x', 'm', 'l')) && (IS_BLANK_CH(NXT(5)))) { |
12919 | 1.91k | xmlParseTextDecl(ctxt); |
12920 | | /* |
12921 | | * An XML-1.0 document can't reference an entity not XML-1.0 |
12922 | | */ |
12923 | 1.91k | if ((xmlStrEqual(oldctxt->version, BAD_CAST "1.0")) && |
12924 | 1.91k | (!xmlStrEqual(ctxt->input->version, BAD_CAST "1.0"))) { |
12925 | 102 | xmlFatalErrMsg(ctxt, XML_ERR_VERSION_MISMATCH, |
12926 | 102 | "Version mismatch between document and entity\n"); |
12927 | 102 | } |
12928 | 1.91k | } |
12929 | | |
12930 | 92.0k | ctxt->instate = XML_PARSER_CONTENT; |
12931 | 92.0k | ctxt->depth = depth; |
12932 | 92.0k | if (oldctxt != NULL) { |
12933 | 92.0k | ctxt->_private = oldctxt->_private; |
12934 | 92.0k | ctxt->loadsubset = oldctxt->loadsubset; |
12935 | 92.0k | ctxt->validate = oldctxt->validate; |
12936 | 92.0k | ctxt->valid = oldctxt->valid; |
12937 | 92.0k | ctxt->replaceEntities = oldctxt->replaceEntities; |
12938 | 92.0k | if (oldctxt->validate) { |
12939 | 55.2k | ctxt->vctxt.error = oldctxt->vctxt.error; |
12940 | 55.2k | ctxt->vctxt.warning = oldctxt->vctxt.warning; |
12941 | 55.2k | ctxt->vctxt.userData = oldctxt->vctxt.userData; |
12942 | 55.2k | ctxt->vctxt.flags = oldctxt->vctxt.flags; |
12943 | 55.2k | } |
12944 | 92.0k | ctxt->external = oldctxt->external; |
12945 | 92.0k | if (ctxt->dict) xmlDictFree(ctxt->dict); |
12946 | 92.0k | ctxt->dict = oldctxt->dict; |
12947 | 92.0k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
12948 | 92.0k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
12949 | 92.0k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
12950 | 92.0k | ctxt->dictNames = oldctxt->dictNames; |
12951 | 92.0k | ctxt->attsDefault = oldctxt->attsDefault; |
12952 | 92.0k | ctxt->attsSpecial = oldctxt->attsSpecial; |
12953 | 92.0k | ctxt->linenumbers = oldctxt->linenumbers; |
12954 | 92.0k | ctxt->record_info = oldctxt->record_info; |
12955 | 92.0k | ctxt->node_seq.maximum = oldctxt->node_seq.maximum; |
12956 | 92.0k | ctxt->node_seq.length = oldctxt->node_seq.length; |
12957 | 92.0k | ctxt->node_seq.buffer = oldctxt->node_seq.buffer; |
12958 | 92.0k | } else { |
12959 | | /* |
12960 | | * Doing validity checking on chunk without context |
12961 | | * doesn't make sense |
12962 | | */ |
12963 | 0 | ctxt->_private = NULL; |
12964 | 0 | ctxt->validate = 0; |
12965 | 0 | ctxt->external = 2; |
12966 | 0 | ctxt->loadsubset = 0; |
12967 | 0 | } |
12968 | | |
12969 | 92.0k | xmlParseContent(ctxt); |
12970 | | |
12971 | 92.0k | if ((RAW == '<') && (NXT(1) == '/')) { |
12972 | 3.08k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
12973 | 88.9k | } else if (RAW != 0) { |
12974 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
12975 | 0 | } |
12976 | 92.0k | if (ctxt->node != newDoc->children) { |
12977 | 13.2k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
12978 | 13.2k | } |
12979 | | |
12980 | 92.0k | if (!ctxt->wellFormed) { |
12981 | 26.9k | ret = (xmlParserErrors)ctxt->errNo; |
12982 | 26.9k | if (oldctxt != NULL) { |
12983 | 26.9k | oldctxt->errNo = ctxt->errNo; |
12984 | 26.9k | oldctxt->wellFormed = 0; |
12985 | 26.9k | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
12986 | 26.9k | } |
12987 | 65.0k | } else { |
12988 | 65.0k | if (list != NULL) { |
12989 | 6.42k | xmlNodePtr cur; |
12990 | | |
12991 | | /* |
12992 | | * Return the newly created nodeset after unlinking it from |
12993 | | * they pseudo parent. |
12994 | | */ |
12995 | 6.42k | cur = newDoc->children->children; |
12996 | 6.42k | *list = cur; |
12997 | 877k | while (cur != NULL) { |
12998 | 871k | cur->parent = NULL; |
12999 | 871k | cur = cur->next; |
13000 | 871k | } |
13001 | 6.42k | newDoc->children->children = NULL; |
13002 | 6.42k | } |
13003 | 65.0k | ret = XML_ERR_OK; |
13004 | 65.0k | } |
13005 | | |
13006 | | /* |
13007 | | * Also record the size of the entity parsed |
13008 | | */ |
13009 | 92.0k | if (ctxt->input != NULL && oldctxt != NULL) { |
13010 | 92.0k | unsigned long consumed = ctxt->input->consumed; |
13011 | | |
13012 | 92.0k | xmlSaturatedAddSizeT(&consumed, ctxt->input->cur - ctxt->input->base); |
13013 | | |
13014 | 92.0k | xmlSaturatedAdd(&oldctxt->sizeentities, consumed); |
13015 | 92.0k | xmlSaturatedAdd(&oldctxt->sizeentities, ctxt->sizeentities); |
13016 | | |
13017 | 92.0k | xmlSaturatedAdd(&oldctxt->sizeentcopy, consumed); |
13018 | 92.0k | xmlSaturatedAdd(&oldctxt->sizeentcopy, ctxt->sizeentcopy); |
13019 | 92.0k | } |
13020 | | |
13021 | 92.0k | if (oldctxt != NULL) { |
13022 | 92.0k | ctxt->dict = NULL; |
13023 | 92.0k | ctxt->attsDefault = NULL; |
13024 | 92.0k | ctxt->attsSpecial = NULL; |
13025 | 92.0k | oldctxt->nbErrors = ctxt->nbErrors; |
13026 | 92.0k | oldctxt->nbWarnings = ctxt->nbWarnings; |
13027 | 92.0k | oldctxt->validate = ctxt->validate; |
13028 | 92.0k | oldctxt->valid = ctxt->valid; |
13029 | 92.0k | oldctxt->node_seq.maximum = ctxt->node_seq.maximum; |
13030 | 92.0k | oldctxt->node_seq.length = ctxt->node_seq.length; |
13031 | 92.0k | oldctxt->node_seq.buffer = ctxt->node_seq.buffer; |
13032 | 92.0k | } |
13033 | 92.0k | ctxt->node_seq.maximum = 0; |
13034 | 92.0k | ctxt->node_seq.length = 0; |
13035 | 92.0k | ctxt->node_seq.buffer = NULL; |
13036 | 92.0k | xmlFreeParserCtxt(ctxt); |
13037 | 92.0k | newDoc->intSubset = NULL; |
13038 | 92.0k | newDoc->extSubset = NULL; |
13039 | 92.0k | xmlFreeDoc(newDoc); |
13040 | | |
13041 | 92.0k | return(ret); |
13042 | 92.0k | } |
13043 | | |
13044 | | #ifdef LIBXML_SAX1_ENABLED |
13045 | | /** |
13046 | | * xmlParseExternalEntity: |
13047 | | * @doc: the document the chunk pertains to |
13048 | | * @sax: the SAX handler block (possibly NULL) |
13049 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13050 | | * @depth: Used for loop detection, use 0 |
13051 | | * @URL: the URL for the entity to load |
13052 | | * @ID: the System ID for the entity to load |
13053 | | * @lst: the return value for the set of parsed nodes |
13054 | | * |
13055 | | * Parse an external general entity |
13056 | | * An external general parsed entity is well-formed if it matches the |
13057 | | * production labeled extParsedEnt. |
13058 | | * |
13059 | | * [78] extParsedEnt ::= TextDecl? content |
13060 | | * |
13061 | | * Returns 0 if the entity is well formed, -1 in case of args problem and |
13062 | | * the parser error code otherwise |
13063 | | */ |
13064 | | |
13065 | | int |
13066 | | xmlParseExternalEntity(xmlDocPtr doc, xmlSAXHandlerPtr sax, void *user_data, |
13067 | 0 | int depth, const xmlChar *URL, const xmlChar *ID, xmlNodePtr *lst) { |
13068 | 0 | return(xmlParseExternalEntityPrivate(doc, NULL, sax, user_data, depth, URL, |
13069 | 0 | ID, lst)); |
13070 | 0 | } |
13071 | | |
13072 | | /** |
13073 | | * xmlParseBalancedChunkMemory: |
13074 | | * @doc: the document the chunk pertains to (must not be NULL) |
13075 | | * @sax: the SAX handler block (possibly NULL) |
13076 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13077 | | * @depth: Used for loop detection, use 0 |
13078 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13079 | | * @lst: the return value for the set of parsed nodes |
13080 | | * |
13081 | | * Parse a well-balanced chunk of an XML document |
13082 | | * called by the parser |
13083 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13084 | | * the content production in the XML grammar: |
13085 | | * |
13086 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13087 | | * |
13088 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13089 | | * the parser error code otherwise |
13090 | | */ |
13091 | | |
13092 | | int |
13093 | | xmlParseBalancedChunkMemory(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13094 | 0 | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst) { |
13095 | 0 | return xmlParseBalancedChunkMemoryRecover( doc, sax, user_data, |
13096 | 0 | depth, string, lst, 0 ); |
13097 | 0 | } |
13098 | | #endif /* LIBXML_SAX1_ENABLED */ |
13099 | | |
13100 | | /** |
13101 | | * xmlParseBalancedChunkMemoryInternal: |
13102 | | * @oldctxt: the existing parsing context |
13103 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13104 | | * @user_data: the user data field for the parser context |
13105 | | * @lst: the return value for the set of parsed nodes |
13106 | | * |
13107 | | * |
13108 | | * Parse a well-balanced chunk of an XML document |
13109 | | * called by the parser |
13110 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13111 | | * the content production in the XML grammar: |
13112 | | * |
13113 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13114 | | * |
13115 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13116 | | * error code otherwise |
13117 | | * |
13118 | | * In case recover is set to 1, the nodelist will not be empty even if |
13119 | | * the parsed chunk is not well balanced. |
13120 | | */ |
13121 | | static xmlParserErrors |
13122 | | xmlParseBalancedChunkMemoryInternal(xmlParserCtxtPtr oldctxt, |
13123 | 93.1k | const xmlChar *string, void *user_data, xmlNodePtr *lst) { |
13124 | 93.1k | xmlParserCtxtPtr ctxt; |
13125 | 93.1k | xmlDocPtr newDoc = NULL; |
13126 | 93.1k | xmlNodePtr newRoot; |
13127 | 93.1k | xmlSAXHandlerPtr oldsax = NULL; |
13128 | 93.1k | xmlNodePtr content = NULL; |
13129 | 93.1k | xmlNodePtr last = NULL; |
13130 | 93.1k | int size; |
13131 | 93.1k | xmlParserErrors ret = XML_ERR_OK; |
13132 | 93.1k | #ifdef SAX2 |
13133 | 93.1k | int i; |
13134 | 93.1k | #endif |
13135 | | |
13136 | 93.1k | if (((oldctxt->depth > 40) && ((oldctxt->options & XML_PARSE_HUGE) == 0)) || |
13137 | 93.1k | (oldctxt->depth > 100)) { |
13138 | 99 | xmlFatalErrMsg(oldctxt, XML_ERR_ENTITY_LOOP, |
13139 | 99 | "Maximum entity nesting depth exceeded"); |
13140 | 99 | return(XML_ERR_ENTITY_LOOP); |
13141 | 99 | } |
13142 | | |
13143 | | |
13144 | 93.0k | if (lst != NULL) |
13145 | 60.8k | *lst = NULL; |
13146 | 93.0k | if (string == NULL) |
13147 | 83 | return(XML_ERR_INTERNAL_ERROR); |
13148 | | |
13149 | 92.9k | size = xmlStrlen(string); |
13150 | | |
13151 | 92.9k | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13152 | 92.9k | if (ctxt == NULL) return(XML_WAR_UNDECLARED_ENTITY); |
13153 | 72.2k | ctxt->nbErrors = oldctxt->nbErrors; |
13154 | 72.2k | ctxt->nbWarnings = oldctxt->nbWarnings; |
13155 | 72.2k | if (user_data != NULL) |
13156 | 0 | ctxt->userData = user_data; |
13157 | 72.2k | else |
13158 | 72.2k | ctxt->userData = ctxt; |
13159 | 72.2k | if (ctxt->dict != NULL) xmlDictFree(ctxt->dict); |
13160 | 72.2k | ctxt->dict = oldctxt->dict; |
13161 | 72.2k | ctxt->input_id = oldctxt->input_id; |
13162 | 72.2k | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13163 | 72.2k | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13164 | 72.2k | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13165 | | |
13166 | 72.2k | #ifdef SAX2 |
13167 | | /* propagate namespaces down the entity */ |
13168 | 72.7k | for (i = 0;i < oldctxt->nsNr;i += 2) { |
13169 | 537 | nsPush(ctxt, oldctxt->nsTab[i], oldctxt->nsTab[i+1]); |
13170 | 537 | } |
13171 | 72.2k | #endif |
13172 | | |
13173 | 72.2k | oldsax = ctxt->sax; |
13174 | 72.2k | ctxt->sax = oldctxt->sax; |
13175 | 72.2k | xmlDetectSAX2(ctxt); |
13176 | 72.2k | ctxt->replaceEntities = oldctxt->replaceEntities; |
13177 | 72.2k | ctxt->options = oldctxt->options; |
13178 | | |
13179 | 72.2k | ctxt->_private = oldctxt->_private; |
13180 | 72.2k | if (oldctxt->myDoc == NULL) { |
13181 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13182 | 0 | if (newDoc == NULL) { |
13183 | 0 | ctxt->sax = oldsax; |
13184 | 0 | ctxt->dict = NULL; |
13185 | 0 | xmlFreeParserCtxt(ctxt); |
13186 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13187 | 0 | } |
13188 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13189 | 0 | newDoc->dict = ctxt->dict; |
13190 | 0 | xmlDictReference(newDoc->dict); |
13191 | 0 | ctxt->myDoc = newDoc; |
13192 | 72.2k | } else { |
13193 | 72.2k | ctxt->myDoc = oldctxt->myDoc; |
13194 | 72.2k | content = ctxt->myDoc->children; |
13195 | 72.2k | last = ctxt->myDoc->last; |
13196 | 72.2k | } |
13197 | 72.2k | newRoot = xmlNewDocNode(ctxt->myDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13198 | 72.2k | if (newRoot == NULL) { |
13199 | 0 | ctxt->sax = oldsax; |
13200 | 0 | ctxt->dict = NULL; |
13201 | 0 | xmlFreeParserCtxt(ctxt); |
13202 | 0 | if (newDoc != NULL) { |
13203 | 0 | xmlFreeDoc(newDoc); |
13204 | 0 | } |
13205 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13206 | 0 | } |
13207 | 72.2k | ctxt->myDoc->children = NULL; |
13208 | 72.2k | ctxt->myDoc->last = NULL; |
13209 | 72.2k | xmlAddChild((xmlNodePtr) ctxt->myDoc, newRoot); |
13210 | 72.2k | nodePush(ctxt, ctxt->myDoc->children); |
13211 | 72.2k | ctxt->instate = XML_PARSER_CONTENT; |
13212 | 72.2k | ctxt->depth = oldctxt->depth; |
13213 | | |
13214 | 72.2k | ctxt->validate = 0; |
13215 | 72.2k | ctxt->loadsubset = oldctxt->loadsubset; |
13216 | 72.2k | if ((oldctxt->validate) || (oldctxt->replaceEntities != 0)) { |
13217 | | /* |
13218 | | * ID/IDREF registration will be done in xmlValidateElement below |
13219 | | */ |
13220 | 60.0k | ctxt->loadsubset |= XML_SKIP_IDS; |
13221 | 60.0k | } |
13222 | 72.2k | ctxt->dictNames = oldctxt->dictNames; |
13223 | 72.2k | ctxt->attsDefault = oldctxt->attsDefault; |
13224 | 72.2k | ctxt->attsSpecial = oldctxt->attsSpecial; |
13225 | | |
13226 | 72.2k | xmlParseContent(ctxt); |
13227 | 72.2k | if ((RAW == '<') && (NXT(1) == '/')) { |
13228 | 177 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13229 | 72.0k | } else if (RAW != 0) { |
13230 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13231 | 0 | } |
13232 | 72.2k | if (ctxt->node != ctxt->myDoc->children) { |
13233 | 1.12k | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13234 | 1.12k | } |
13235 | | |
13236 | 72.2k | if (!ctxt->wellFormed) { |
13237 | 11.3k | ret = (xmlParserErrors)ctxt->errNo; |
13238 | 11.3k | oldctxt->errNo = ctxt->errNo; |
13239 | 11.3k | oldctxt->wellFormed = 0; |
13240 | 11.3k | xmlCopyError(&ctxt->lastError, &oldctxt->lastError); |
13241 | 60.9k | } else { |
13242 | 60.9k | ret = XML_ERR_OK; |
13243 | 60.9k | } |
13244 | | |
13245 | 72.2k | if ((lst != NULL) && (ret == XML_ERR_OK)) { |
13246 | 49.1k | xmlNodePtr cur; |
13247 | | |
13248 | | /* |
13249 | | * Return the newly created nodeset after unlinking it from |
13250 | | * they pseudo parent. |
13251 | | */ |
13252 | 49.1k | cur = ctxt->myDoc->children->children; |
13253 | 49.1k | *lst = cur; |
13254 | 164k | while (cur != NULL) { |
13255 | 115k | #ifdef LIBXML_VALID_ENABLED |
13256 | 115k | if ((oldctxt->validate) && (oldctxt->wellFormed) && |
13257 | 115k | (oldctxt->myDoc) && (oldctxt->myDoc->intSubset) && |
13258 | 115k | (cur->type == XML_ELEMENT_NODE)) { |
13259 | 17.7k | oldctxt->valid &= xmlValidateElement(&oldctxt->vctxt, |
13260 | 17.7k | oldctxt->myDoc, cur); |
13261 | 17.7k | } |
13262 | 115k | #endif /* LIBXML_VALID_ENABLED */ |
13263 | 115k | cur->parent = NULL; |
13264 | 115k | cur = cur->next; |
13265 | 115k | } |
13266 | 49.1k | ctxt->myDoc->children->children = NULL; |
13267 | 49.1k | } |
13268 | 72.2k | if (ctxt->myDoc != NULL) { |
13269 | 72.2k | xmlFreeNode(ctxt->myDoc->children); |
13270 | 72.2k | ctxt->myDoc->children = content; |
13271 | 72.2k | ctxt->myDoc->last = last; |
13272 | 72.2k | } |
13273 | | |
13274 | | /* |
13275 | | * Also record the size of the entity parsed |
13276 | | */ |
13277 | 72.2k | if (ctxt->input != NULL && oldctxt != NULL) { |
13278 | 72.2k | unsigned long consumed = ctxt->input->consumed; |
13279 | | |
13280 | 72.2k | xmlSaturatedAddSizeT(&consumed, ctxt->input->cur - ctxt->input->base); |
13281 | | |
13282 | 72.2k | xmlSaturatedAdd(&oldctxt->sizeentcopy, consumed); |
13283 | 72.2k | xmlSaturatedAdd(&oldctxt->sizeentcopy, ctxt->sizeentcopy); |
13284 | 72.2k | } |
13285 | | |
13286 | 72.2k | oldctxt->nbErrors = ctxt->nbErrors; |
13287 | 72.2k | oldctxt->nbWarnings = ctxt->nbWarnings; |
13288 | 72.2k | ctxt->sax = oldsax; |
13289 | 72.2k | ctxt->dict = NULL; |
13290 | 72.2k | ctxt->attsDefault = NULL; |
13291 | 72.2k | ctxt->attsSpecial = NULL; |
13292 | 72.2k | xmlFreeParserCtxt(ctxt); |
13293 | 72.2k | if (newDoc != NULL) { |
13294 | 0 | xmlFreeDoc(newDoc); |
13295 | 0 | } |
13296 | | |
13297 | 72.2k | return(ret); |
13298 | 72.2k | } |
13299 | | |
13300 | | /** |
13301 | | * xmlParseInNodeContext: |
13302 | | * @node: the context node |
13303 | | * @data: the input string |
13304 | | * @datalen: the input string length in bytes |
13305 | | * @options: a combination of xmlParserOption |
13306 | | * @lst: the return value for the set of parsed nodes |
13307 | | * |
13308 | | * Parse a well-balanced chunk of an XML document |
13309 | | * within the context (DTD, namespaces, etc ...) of the given node. |
13310 | | * |
13311 | | * The allowed sequence for the data is a Well Balanced Chunk defined by |
13312 | | * the content production in the XML grammar: |
13313 | | * |
13314 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13315 | | * |
13316 | | * Returns XML_ERR_OK if the chunk is well balanced, and the parser |
13317 | | * error code otherwise |
13318 | | */ |
13319 | | xmlParserErrors |
13320 | | xmlParseInNodeContext(xmlNodePtr node, const char *data, int datalen, |
13321 | 0 | int options, xmlNodePtr *lst) { |
13322 | 0 | #ifdef SAX2 |
13323 | 0 | xmlParserCtxtPtr ctxt; |
13324 | 0 | xmlDocPtr doc = NULL; |
13325 | 0 | xmlNodePtr fake, cur; |
13326 | 0 | int nsnr = 0; |
13327 | |
|
13328 | 0 | xmlParserErrors ret = XML_ERR_OK; |
13329 | | |
13330 | | /* |
13331 | | * check all input parameters, grab the document |
13332 | | */ |
13333 | 0 | if ((lst == NULL) || (node == NULL) || (data == NULL) || (datalen < 0)) |
13334 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13335 | 0 | switch (node->type) { |
13336 | 0 | case XML_ELEMENT_NODE: |
13337 | 0 | case XML_ATTRIBUTE_NODE: |
13338 | 0 | case XML_TEXT_NODE: |
13339 | 0 | case XML_CDATA_SECTION_NODE: |
13340 | 0 | case XML_ENTITY_REF_NODE: |
13341 | 0 | case XML_PI_NODE: |
13342 | 0 | case XML_COMMENT_NODE: |
13343 | 0 | case XML_DOCUMENT_NODE: |
13344 | 0 | case XML_HTML_DOCUMENT_NODE: |
13345 | 0 | break; |
13346 | 0 | default: |
13347 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13348 | |
|
13349 | 0 | } |
13350 | 0 | while ((node != NULL) && (node->type != XML_ELEMENT_NODE) && |
13351 | 0 | (node->type != XML_DOCUMENT_NODE) && |
13352 | 0 | (node->type != XML_HTML_DOCUMENT_NODE)) |
13353 | 0 | node = node->parent; |
13354 | 0 | if (node == NULL) |
13355 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13356 | 0 | if (node->type == XML_ELEMENT_NODE) |
13357 | 0 | doc = node->doc; |
13358 | 0 | else |
13359 | 0 | doc = (xmlDocPtr) node; |
13360 | 0 | if (doc == NULL) |
13361 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13362 | | |
13363 | | /* |
13364 | | * allocate a context and set-up everything not related to the |
13365 | | * node position in the tree |
13366 | | */ |
13367 | 0 | if (doc->type == XML_DOCUMENT_NODE) |
13368 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) data, datalen); |
13369 | 0 | #ifdef LIBXML_HTML_ENABLED |
13370 | 0 | else if (doc->type == XML_HTML_DOCUMENT_NODE) { |
13371 | 0 | ctxt = htmlCreateMemoryParserCtxt((char *) data, datalen); |
13372 | | /* |
13373 | | * When parsing in context, it makes no sense to add implied |
13374 | | * elements like html/body/etc... |
13375 | | */ |
13376 | 0 | options |= HTML_PARSE_NOIMPLIED; |
13377 | 0 | } |
13378 | 0 | #endif |
13379 | 0 | else |
13380 | 0 | return(XML_ERR_INTERNAL_ERROR); |
13381 | | |
13382 | 0 | if (ctxt == NULL) |
13383 | 0 | return(XML_ERR_NO_MEMORY); |
13384 | | |
13385 | | /* |
13386 | | * Use input doc's dict if present, else assure XML_PARSE_NODICT is set. |
13387 | | * We need a dictionary for xmlDetectSAX2, so if there's no doc dict |
13388 | | * we must wait until the last moment to free the original one. |
13389 | | */ |
13390 | 0 | if (doc->dict != NULL) { |
13391 | 0 | if (ctxt->dict != NULL) |
13392 | 0 | xmlDictFree(ctxt->dict); |
13393 | 0 | ctxt->dict = doc->dict; |
13394 | 0 | } else |
13395 | 0 | options |= XML_PARSE_NODICT; |
13396 | |
|
13397 | 0 | if (doc->encoding != NULL) { |
13398 | 0 | xmlCharEncodingHandlerPtr hdlr; |
13399 | |
|
13400 | 0 | if (ctxt->encoding != NULL) |
13401 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
13402 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) doc->encoding); |
13403 | |
|
13404 | 0 | hdlr = xmlFindCharEncodingHandler((const char *) doc->encoding); |
13405 | 0 | if (hdlr != NULL) { |
13406 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
13407 | 0 | } else { |
13408 | 0 | return(XML_ERR_UNSUPPORTED_ENCODING); |
13409 | 0 | } |
13410 | 0 | } |
13411 | | |
13412 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
13413 | 0 | xmlDetectSAX2(ctxt); |
13414 | 0 | ctxt->myDoc = doc; |
13415 | | /* parsing in context, i.e. as within existing content */ |
13416 | 0 | ctxt->input_id = 2; |
13417 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13418 | |
|
13419 | 0 | fake = xmlNewDocComment(node->doc, NULL); |
13420 | 0 | if (fake == NULL) { |
13421 | 0 | xmlFreeParserCtxt(ctxt); |
13422 | 0 | return(XML_ERR_NO_MEMORY); |
13423 | 0 | } |
13424 | 0 | xmlAddChild(node, fake); |
13425 | |
|
13426 | 0 | if (node->type == XML_ELEMENT_NODE) { |
13427 | 0 | nodePush(ctxt, node); |
13428 | | /* |
13429 | | * initialize the SAX2 namespaces stack |
13430 | | */ |
13431 | 0 | cur = node; |
13432 | 0 | while ((cur != NULL) && (cur->type == XML_ELEMENT_NODE)) { |
13433 | 0 | xmlNsPtr ns = cur->nsDef; |
13434 | 0 | const xmlChar *iprefix, *ihref; |
13435 | |
|
13436 | 0 | while (ns != NULL) { |
13437 | 0 | if (ctxt->dict) { |
13438 | 0 | iprefix = xmlDictLookup(ctxt->dict, ns->prefix, -1); |
13439 | 0 | ihref = xmlDictLookup(ctxt->dict, ns->href, -1); |
13440 | 0 | } else { |
13441 | 0 | iprefix = ns->prefix; |
13442 | 0 | ihref = ns->href; |
13443 | 0 | } |
13444 | |
|
13445 | 0 | if (xmlGetNamespace(ctxt, iprefix) == NULL) { |
13446 | 0 | nsPush(ctxt, iprefix, ihref); |
13447 | 0 | nsnr++; |
13448 | 0 | } |
13449 | 0 | ns = ns->next; |
13450 | 0 | } |
13451 | 0 | cur = cur->parent; |
13452 | 0 | } |
13453 | 0 | } |
13454 | |
|
13455 | 0 | if ((ctxt->validate) || (ctxt->replaceEntities != 0)) { |
13456 | | /* |
13457 | | * ID/IDREF registration will be done in xmlValidateElement below |
13458 | | */ |
13459 | 0 | ctxt->loadsubset |= XML_SKIP_IDS; |
13460 | 0 | } |
13461 | |
|
13462 | 0 | #ifdef LIBXML_HTML_ENABLED |
13463 | 0 | if (doc->type == XML_HTML_DOCUMENT_NODE) |
13464 | 0 | __htmlParseContent(ctxt); |
13465 | 0 | else |
13466 | 0 | #endif |
13467 | 0 | xmlParseContent(ctxt); |
13468 | |
|
13469 | 0 | nsPop(ctxt, nsnr); |
13470 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13471 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13472 | 0 | } else if (RAW != 0) { |
13473 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13474 | 0 | } |
13475 | 0 | if ((ctxt->node != NULL) && (ctxt->node != node)) { |
13476 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13477 | 0 | ctxt->wellFormed = 0; |
13478 | 0 | } |
13479 | |
|
13480 | 0 | if (!ctxt->wellFormed) { |
13481 | 0 | if (ctxt->errNo == 0) |
13482 | 0 | ret = XML_ERR_INTERNAL_ERROR; |
13483 | 0 | else |
13484 | 0 | ret = (xmlParserErrors)ctxt->errNo; |
13485 | 0 | } else { |
13486 | 0 | ret = XML_ERR_OK; |
13487 | 0 | } |
13488 | | |
13489 | | /* |
13490 | | * Return the newly created nodeset after unlinking it from |
13491 | | * the pseudo sibling. |
13492 | | */ |
13493 | |
|
13494 | 0 | cur = fake->next; |
13495 | 0 | fake->next = NULL; |
13496 | 0 | node->last = fake; |
13497 | |
|
13498 | 0 | if (cur != NULL) { |
13499 | 0 | cur->prev = NULL; |
13500 | 0 | } |
13501 | |
|
13502 | 0 | *lst = cur; |
13503 | |
|
13504 | 0 | while (cur != NULL) { |
13505 | 0 | cur->parent = NULL; |
13506 | 0 | cur = cur->next; |
13507 | 0 | } |
13508 | |
|
13509 | 0 | xmlUnlinkNode(fake); |
13510 | 0 | xmlFreeNode(fake); |
13511 | | |
13512 | |
|
13513 | 0 | if (ret != XML_ERR_OK) { |
13514 | 0 | xmlFreeNodeList(*lst); |
13515 | 0 | *lst = NULL; |
13516 | 0 | } |
13517 | |
|
13518 | 0 | if (doc->dict != NULL) |
13519 | 0 | ctxt->dict = NULL; |
13520 | 0 | xmlFreeParserCtxt(ctxt); |
13521 | |
|
13522 | 0 | return(ret); |
13523 | | #else /* !SAX2 */ |
13524 | | return(XML_ERR_INTERNAL_ERROR); |
13525 | | #endif |
13526 | 0 | } |
13527 | | |
13528 | | #ifdef LIBXML_SAX1_ENABLED |
13529 | | /** |
13530 | | * xmlParseBalancedChunkMemoryRecover: |
13531 | | * @doc: the document the chunk pertains to (must not be NULL) |
13532 | | * @sax: the SAX handler block (possibly NULL) |
13533 | | * @user_data: The user data returned on SAX callbacks (possibly NULL) |
13534 | | * @depth: Used for loop detection, use 0 |
13535 | | * @string: the input string in UTF8 or ISO-Latin (zero terminated) |
13536 | | * @lst: the return value for the set of parsed nodes |
13537 | | * @recover: return nodes even if the data is broken (use 0) |
13538 | | * |
13539 | | * |
13540 | | * Parse a well-balanced chunk of an XML document |
13541 | | * called by the parser |
13542 | | * The allowed sequence for the Well Balanced Chunk is the one defined by |
13543 | | * the content production in the XML grammar: |
13544 | | * |
13545 | | * [43] content ::= (element | CharData | Reference | CDSect | PI | Comment)* |
13546 | | * |
13547 | | * Returns 0 if the chunk is well balanced, -1 in case of args problem and |
13548 | | * the parser error code otherwise |
13549 | | * |
13550 | | * In case recover is set to 1, the nodelist will not be empty even if |
13551 | | * the parsed chunk is not well balanced, assuming the parsing succeeded to |
13552 | | * some extent. |
13553 | | */ |
13554 | | int |
13555 | | xmlParseBalancedChunkMemoryRecover(xmlDocPtr doc, xmlSAXHandlerPtr sax, |
13556 | | void *user_data, int depth, const xmlChar *string, xmlNodePtr *lst, |
13557 | 0 | int recover) { |
13558 | 0 | xmlParserCtxtPtr ctxt; |
13559 | 0 | xmlDocPtr newDoc; |
13560 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
13561 | 0 | xmlNodePtr content, newRoot; |
13562 | 0 | int size; |
13563 | 0 | int ret = 0; |
13564 | |
|
13565 | 0 | if (depth > 40) { |
13566 | 0 | return(XML_ERR_ENTITY_LOOP); |
13567 | 0 | } |
13568 | | |
13569 | | |
13570 | 0 | if (lst != NULL) |
13571 | 0 | *lst = NULL; |
13572 | 0 | if (string == NULL) |
13573 | 0 | return(-1); |
13574 | | |
13575 | 0 | size = xmlStrlen(string); |
13576 | |
|
13577 | 0 | ctxt = xmlCreateMemoryParserCtxt((char *) string, size); |
13578 | 0 | if (ctxt == NULL) return(-1); |
13579 | 0 | ctxt->userData = ctxt; |
13580 | 0 | if (sax != NULL) { |
13581 | 0 | oldsax = ctxt->sax; |
13582 | 0 | ctxt->sax = sax; |
13583 | 0 | if (user_data != NULL) |
13584 | 0 | ctxt->userData = user_data; |
13585 | 0 | } |
13586 | 0 | newDoc = xmlNewDoc(BAD_CAST "1.0"); |
13587 | 0 | if (newDoc == NULL) { |
13588 | 0 | xmlFreeParserCtxt(ctxt); |
13589 | 0 | return(-1); |
13590 | 0 | } |
13591 | 0 | newDoc->properties = XML_DOC_INTERNAL; |
13592 | 0 | if ((doc != NULL) && (doc->dict != NULL)) { |
13593 | 0 | xmlDictFree(ctxt->dict); |
13594 | 0 | ctxt->dict = doc->dict; |
13595 | 0 | xmlDictReference(ctxt->dict); |
13596 | 0 | ctxt->str_xml = xmlDictLookup(ctxt->dict, BAD_CAST "xml", 3); |
13597 | 0 | ctxt->str_xmlns = xmlDictLookup(ctxt->dict, BAD_CAST "xmlns", 5); |
13598 | 0 | ctxt->str_xml_ns = xmlDictLookup(ctxt->dict, XML_XML_NAMESPACE, 36); |
13599 | 0 | ctxt->dictNames = 1; |
13600 | 0 | } else { |
13601 | 0 | xmlCtxtUseOptionsInternal(ctxt, XML_PARSE_NODICT, NULL); |
13602 | 0 | } |
13603 | | /* doc == NULL is only supported for historic reasons */ |
13604 | 0 | if (doc != NULL) { |
13605 | 0 | newDoc->intSubset = doc->intSubset; |
13606 | 0 | newDoc->extSubset = doc->extSubset; |
13607 | 0 | } |
13608 | 0 | newRoot = xmlNewDocNode(newDoc, NULL, BAD_CAST "pseudoroot", NULL); |
13609 | 0 | if (newRoot == NULL) { |
13610 | 0 | if (sax != NULL) |
13611 | 0 | ctxt->sax = oldsax; |
13612 | 0 | xmlFreeParserCtxt(ctxt); |
13613 | 0 | newDoc->intSubset = NULL; |
13614 | 0 | newDoc->extSubset = NULL; |
13615 | 0 | xmlFreeDoc(newDoc); |
13616 | 0 | return(-1); |
13617 | 0 | } |
13618 | 0 | xmlAddChild((xmlNodePtr) newDoc, newRoot); |
13619 | 0 | nodePush(ctxt, newRoot); |
13620 | | /* doc == NULL is only supported for historic reasons */ |
13621 | 0 | if (doc == NULL) { |
13622 | 0 | ctxt->myDoc = newDoc; |
13623 | 0 | } else { |
13624 | 0 | ctxt->myDoc = newDoc; |
13625 | 0 | newDoc->children->doc = doc; |
13626 | | /* Ensure that doc has XML spec namespace */ |
13627 | 0 | xmlSearchNsByHref(doc, (xmlNodePtr)doc, XML_XML_NAMESPACE); |
13628 | 0 | newDoc->oldNs = doc->oldNs; |
13629 | 0 | } |
13630 | 0 | ctxt->instate = XML_PARSER_CONTENT; |
13631 | 0 | ctxt->input_id = 2; |
13632 | 0 | ctxt->depth = depth; |
13633 | | |
13634 | | /* |
13635 | | * Doing validity checking on chunk doesn't make sense |
13636 | | */ |
13637 | 0 | ctxt->validate = 0; |
13638 | 0 | ctxt->loadsubset = 0; |
13639 | 0 | xmlDetectSAX2(ctxt); |
13640 | |
|
13641 | 0 | if ( doc != NULL ){ |
13642 | 0 | content = doc->children; |
13643 | 0 | doc->children = NULL; |
13644 | 0 | xmlParseContent(ctxt); |
13645 | 0 | doc->children = content; |
13646 | 0 | } |
13647 | 0 | else { |
13648 | 0 | xmlParseContent(ctxt); |
13649 | 0 | } |
13650 | 0 | if ((RAW == '<') && (NXT(1) == '/')) { |
13651 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13652 | 0 | } else if (RAW != 0) { |
13653 | 0 | xmlFatalErr(ctxt, XML_ERR_EXTRA_CONTENT, NULL); |
13654 | 0 | } |
13655 | 0 | if (ctxt->node != newDoc->children) { |
13656 | 0 | xmlFatalErr(ctxt, XML_ERR_NOT_WELL_BALANCED, NULL); |
13657 | 0 | } |
13658 | |
|
13659 | 0 | if (!ctxt->wellFormed) { |
13660 | 0 | if (ctxt->errNo == 0) |
13661 | 0 | ret = 1; |
13662 | 0 | else |
13663 | 0 | ret = ctxt->errNo; |
13664 | 0 | } else { |
13665 | 0 | ret = 0; |
13666 | 0 | } |
13667 | |
|
13668 | 0 | if ((lst != NULL) && ((ret == 0) || (recover == 1))) { |
13669 | 0 | xmlNodePtr cur; |
13670 | | |
13671 | | /* |
13672 | | * Return the newly created nodeset after unlinking it from |
13673 | | * they pseudo parent. |
13674 | | */ |
13675 | 0 | cur = newDoc->children->children; |
13676 | 0 | *lst = cur; |
13677 | 0 | while (cur != NULL) { |
13678 | 0 | xmlSetTreeDoc(cur, doc); |
13679 | 0 | cur->parent = NULL; |
13680 | 0 | cur = cur->next; |
13681 | 0 | } |
13682 | 0 | newDoc->children->children = NULL; |
13683 | 0 | } |
13684 | |
|
13685 | 0 | if (sax != NULL) |
13686 | 0 | ctxt->sax = oldsax; |
13687 | 0 | xmlFreeParserCtxt(ctxt); |
13688 | 0 | newDoc->intSubset = NULL; |
13689 | 0 | newDoc->extSubset = NULL; |
13690 | | /* This leaks the namespace list if doc == NULL */ |
13691 | 0 | newDoc->oldNs = NULL; |
13692 | 0 | xmlFreeDoc(newDoc); |
13693 | |
|
13694 | 0 | return(ret); |
13695 | 0 | } |
13696 | | |
13697 | | /** |
13698 | | * xmlSAXParseEntity: |
13699 | | * @sax: the SAX handler block |
13700 | | * @filename: the filename |
13701 | | * |
13702 | | * DEPRECATED: Don't use. |
13703 | | * |
13704 | | * parse an XML external entity out of context and build a tree. |
13705 | | * It use the given SAX function block to handle the parsing callback. |
13706 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13707 | | * |
13708 | | * [78] extParsedEnt ::= TextDecl? content |
13709 | | * |
13710 | | * This correspond to a "Well Balanced" chunk |
13711 | | * |
13712 | | * Returns the resulting document tree |
13713 | | */ |
13714 | | |
13715 | | xmlDocPtr |
13716 | 0 | xmlSAXParseEntity(xmlSAXHandlerPtr sax, const char *filename) { |
13717 | 0 | xmlDocPtr ret; |
13718 | 0 | xmlParserCtxtPtr ctxt; |
13719 | |
|
13720 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13721 | 0 | if (ctxt == NULL) { |
13722 | 0 | return(NULL); |
13723 | 0 | } |
13724 | 0 | if (sax != NULL) { |
13725 | 0 | if (ctxt->sax != NULL) |
13726 | 0 | xmlFree(ctxt->sax); |
13727 | 0 | ctxt->sax = sax; |
13728 | 0 | ctxt->userData = NULL; |
13729 | 0 | } |
13730 | |
|
13731 | 0 | xmlParseExtParsedEnt(ctxt); |
13732 | |
|
13733 | 0 | if (ctxt->wellFormed) |
13734 | 0 | ret = ctxt->myDoc; |
13735 | 0 | else { |
13736 | 0 | ret = NULL; |
13737 | 0 | xmlFreeDoc(ctxt->myDoc); |
13738 | 0 | ctxt->myDoc = NULL; |
13739 | 0 | } |
13740 | 0 | if (sax != NULL) |
13741 | 0 | ctxt->sax = NULL; |
13742 | 0 | xmlFreeParserCtxt(ctxt); |
13743 | |
|
13744 | 0 | return(ret); |
13745 | 0 | } |
13746 | | |
13747 | | /** |
13748 | | * xmlParseEntity: |
13749 | | * @filename: the filename |
13750 | | * |
13751 | | * parse an XML external entity out of context and build a tree. |
13752 | | * |
13753 | | * [78] extParsedEnt ::= TextDecl? content |
13754 | | * |
13755 | | * This correspond to a "Well Balanced" chunk |
13756 | | * |
13757 | | * Returns the resulting document tree |
13758 | | */ |
13759 | | |
13760 | | xmlDocPtr |
13761 | 0 | xmlParseEntity(const char *filename) { |
13762 | 0 | return(xmlSAXParseEntity(NULL, filename)); |
13763 | 0 | } |
13764 | | #endif /* LIBXML_SAX1_ENABLED */ |
13765 | | |
13766 | | /** |
13767 | | * xmlCreateEntityParserCtxtInternal: |
13768 | | * @URL: the entity URL |
13769 | | * @ID: the entity PUBLIC ID |
13770 | | * @base: a possible base for the target URI |
13771 | | * @pctx: parser context used to set options on new context |
13772 | | * |
13773 | | * Create a parser context for an external entity |
13774 | | * Automatic support for ZLIB/Compress compressed document is provided |
13775 | | * by default if found at compile-time. |
13776 | | * |
13777 | | * Returns the new parser context or NULL |
13778 | | */ |
13779 | | static xmlParserCtxtPtr |
13780 | | xmlCreateEntityParserCtxtInternal(xmlSAXHandlerPtr sax, void *userData, |
13781 | | const xmlChar *URL, const xmlChar *ID, const xmlChar *base, |
13782 | 685k | xmlParserCtxtPtr pctx) { |
13783 | 685k | xmlParserCtxtPtr ctxt; |
13784 | 685k | xmlParserInputPtr inputStream; |
13785 | 685k | char *directory = NULL; |
13786 | 685k | xmlChar *uri; |
13787 | | |
13788 | 685k | ctxt = xmlNewSAXParserCtxt(sax, userData); |
13789 | 685k | if (ctxt == NULL) { |
13790 | 0 | return(NULL); |
13791 | 0 | } |
13792 | | |
13793 | 685k | if (pctx != NULL) { |
13794 | 685k | ctxt->options = pctx->options; |
13795 | 685k | ctxt->_private = pctx->_private; |
13796 | 685k | ctxt->input_id = pctx->input_id; |
13797 | 685k | } |
13798 | | |
13799 | | /* Don't read from stdin. */ |
13800 | 685k | if (xmlStrcmp(URL, BAD_CAST "-") == 0) |
13801 | 0 | URL = BAD_CAST "./-"; |
13802 | | |
13803 | 685k | uri = xmlBuildURI(URL, base); |
13804 | | |
13805 | 685k | if (uri == NULL) { |
13806 | 17.3k | inputStream = xmlLoadExternalEntity((char *)URL, (char *)ID, ctxt); |
13807 | 17.3k | if (inputStream == NULL) { |
13808 | 13.8k | xmlFreeParserCtxt(ctxt); |
13809 | 13.8k | return(NULL); |
13810 | 13.8k | } |
13811 | | |
13812 | 3.51k | inputPush(ctxt, inputStream); |
13813 | | |
13814 | 3.51k | if ((ctxt->directory == NULL) && (directory == NULL)) |
13815 | 3.51k | directory = xmlParserGetDirectory((char *)URL); |
13816 | 3.51k | if ((ctxt->directory == NULL) && (directory != NULL)) |
13817 | 3.51k | ctxt->directory = directory; |
13818 | 668k | } else { |
13819 | 668k | inputStream = xmlLoadExternalEntity((char *)uri, (char *)ID, ctxt); |
13820 | 668k | if (inputStream == NULL) { |
13821 | 579k | xmlFree(uri); |
13822 | 579k | xmlFreeParserCtxt(ctxt); |
13823 | 579k | return(NULL); |
13824 | 579k | } |
13825 | | |
13826 | 88.5k | inputPush(ctxt, inputStream); |
13827 | | |
13828 | 88.5k | if ((ctxt->directory == NULL) && (directory == NULL)) |
13829 | 88.5k | directory = xmlParserGetDirectory((char *)uri); |
13830 | 88.5k | if ((ctxt->directory == NULL) && (directory != NULL)) |
13831 | 88.5k | ctxt->directory = directory; |
13832 | 88.5k | xmlFree(uri); |
13833 | 88.5k | } |
13834 | 92.0k | return(ctxt); |
13835 | 685k | } |
13836 | | |
13837 | | /** |
13838 | | * xmlCreateEntityParserCtxt: |
13839 | | * @URL: the entity URL |
13840 | | * @ID: the entity PUBLIC ID |
13841 | | * @base: a possible base for the target URI |
13842 | | * |
13843 | | * Create a parser context for an external entity |
13844 | | * Automatic support for ZLIB/Compress compressed document is provided |
13845 | | * by default if found at compile-time. |
13846 | | * |
13847 | | * Returns the new parser context or NULL |
13848 | | */ |
13849 | | xmlParserCtxtPtr |
13850 | | xmlCreateEntityParserCtxt(const xmlChar *URL, const xmlChar *ID, |
13851 | 0 | const xmlChar *base) { |
13852 | 0 | return xmlCreateEntityParserCtxtInternal(NULL, NULL, URL, ID, base, NULL); |
13853 | |
|
13854 | 0 | } |
13855 | | |
13856 | | /************************************************************************ |
13857 | | * * |
13858 | | * Front ends when parsing from a file * |
13859 | | * * |
13860 | | ************************************************************************/ |
13861 | | |
13862 | | /** |
13863 | | * xmlCreateURLParserCtxt: |
13864 | | * @filename: the filename or URL |
13865 | | * @options: a combination of xmlParserOption |
13866 | | * |
13867 | | * Create a parser context for a file or URL content. |
13868 | | * Automatic support for ZLIB/Compress compressed document is provided |
13869 | | * by default if found at compile-time and for file accesses |
13870 | | * |
13871 | | * Returns the new parser context or NULL |
13872 | | */ |
13873 | | xmlParserCtxtPtr |
13874 | | xmlCreateURLParserCtxt(const char *filename, int options) |
13875 | 0 | { |
13876 | 0 | xmlParserCtxtPtr ctxt; |
13877 | 0 | xmlParserInputPtr inputStream; |
13878 | 0 | char *directory = NULL; |
13879 | |
|
13880 | 0 | ctxt = xmlNewParserCtxt(); |
13881 | 0 | if (ctxt == NULL) { |
13882 | 0 | xmlErrMemory(NULL, "cannot allocate parser context"); |
13883 | 0 | return(NULL); |
13884 | 0 | } |
13885 | | |
13886 | 0 | if (options) |
13887 | 0 | xmlCtxtUseOptionsInternal(ctxt, options, NULL); |
13888 | 0 | ctxt->linenumbers = 1; |
13889 | |
|
13890 | 0 | inputStream = xmlLoadExternalEntity(filename, NULL, ctxt); |
13891 | 0 | if (inputStream == NULL) { |
13892 | 0 | xmlFreeParserCtxt(ctxt); |
13893 | 0 | return(NULL); |
13894 | 0 | } |
13895 | | |
13896 | 0 | inputPush(ctxt, inputStream); |
13897 | 0 | if ((ctxt->directory == NULL) && (directory == NULL)) |
13898 | 0 | directory = xmlParserGetDirectory(filename); |
13899 | 0 | if ((ctxt->directory == NULL) && (directory != NULL)) |
13900 | 0 | ctxt->directory = directory; |
13901 | |
|
13902 | 0 | return(ctxt); |
13903 | 0 | } |
13904 | | |
13905 | | /** |
13906 | | * xmlCreateFileParserCtxt: |
13907 | | * @filename: the filename |
13908 | | * |
13909 | | * Create a parser context for a file content. |
13910 | | * Automatic support for ZLIB/Compress compressed document is provided |
13911 | | * by default if found at compile-time. |
13912 | | * |
13913 | | * Returns the new parser context or NULL |
13914 | | */ |
13915 | | xmlParserCtxtPtr |
13916 | | xmlCreateFileParserCtxt(const char *filename) |
13917 | 0 | { |
13918 | 0 | return(xmlCreateURLParserCtxt(filename, 0)); |
13919 | 0 | } |
13920 | | |
13921 | | #ifdef LIBXML_SAX1_ENABLED |
13922 | | /** |
13923 | | * xmlSAXParseFileWithData: |
13924 | | * @sax: the SAX handler block |
13925 | | * @filename: the filename |
13926 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
13927 | | * documents |
13928 | | * @data: the userdata |
13929 | | * |
13930 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
13931 | | * |
13932 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
13933 | | * compressed document is provided by default if found at compile-time. |
13934 | | * It use the given SAX function block to handle the parsing callback. |
13935 | | * If sax is NULL, fallback to the default DOM tree building routines. |
13936 | | * |
13937 | | * User data (void *) is stored within the parser context in the |
13938 | | * context's _private member, so it is available nearly everywhere in libxml |
13939 | | * |
13940 | | * Returns the resulting document tree |
13941 | | */ |
13942 | | |
13943 | | xmlDocPtr |
13944 | | xmlSAXParseFileWithData(xmlSAXHandlerPtr sax, const char *filename, |
13945 | 0 | int recovery, void *data) { |
13946 | 0 | xmlDocPtr ret; |
13947 | 0 | xmlParserCtxtPtr ctxt; |
13948 | |
|
13949 | 0 | xmlInitParser(); |
13950 | |
|
13951 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
13952 | 0 | if (ctxt == NULL) { |
13953 | 0 | return(NULL); |
13954 | 0 | } |
13955 | 0 | if (sax != NULL) { |
13956 | 0 | if (ctxt->sax != NULL) |
13957 | 0 | xmlFree(ctxt->sax); |
13958 | 0 | ctxt->sax = sax; |
13959 | 0 | } |
13960 | 0 | xmlDetectSAX2(ctxt); |
13961 | 0 | if (data!=NULL) { |
13962 | 0 | ctxt->_private = data; |
13963 | 0 | } |
13964 | |
|
13965 | 0 | if (ctxt->directory == NULL) |
13966 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
13967 | |
|
13968 | 0 | ctxt->recovery = recovery; |
13969 | |
|
13970 | 0 | xmlParseDocument(ctxt); |
13971 | |
|
13972 | 0 | if ((ctxt->wellFormed) || recovery) { |
13973 | 0 | ret = ctxt->myDoc; |
13974 | 0 | if ((ret != NULL) && (ctxt->input->buf != NULL)) { |
13975 | 0 | if (ctxt->input->buf->compressed > 0) |
13976 | 0 | ret->compression = 9; |
13977 | 0 | else |
13978 | 0 | ret->compression = ctxt->input->buf->compressed; |
13979 | 0 | } |
13980 | 0 | } |
13981 | 0 | else { |
13982 | 0 | ret = NULL; |
13983 | 0 | xmlFreeDoc(ctxt->myDoc); |
13984 | 0 | ctxt->myDoc = NULL; |
13985 | 0 | } |
13986 | 0 | if (sax != NULL) |
13987 | 0 | ctxt->sax = NULL; |
13988 | 0 | xmlFreeParserCtxt(ctxt); |
13989 | |
|
13990 | 0 | return(ret); |
13991 | 0 | } |
13992 | | |
13993 | | /** |
13994 | | * xmlSAXParseFile: |
13995 | | * @sax: the SAX handler block |
13996 | | * @filename: the filename |
13997 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
13998 | | * documents |
13999 | | * |
14000 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
14001 | | * |
14002 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14003 | | * compressed document is provided by default if found at compile-time. |
14004 | | * It use the given SAX function block to handle the parsing callback. |
14005 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14006 | | * |
14007 | | * Returns the resulting document tree |
14008 | | */ |
14009 | | |
14010 | | xmlDocPtr |
14011 | | xmlSAXParseFile(xmlSAXHandlerPtr sax, const char *filename, |
14012 | 0 | int recovery) { |
14013 | 0 | return(xmlSAXParseFileWithData(sax,filename,recovery,NULL)); |
14014 | 0 | } |
14015 | | |
14016 | | /** |
14017 | | * xmlRecoverDoc: |
14018 | | * @cur: a pointer to an array of xmlChar |
14019 | | * |
14020 | | * DEPRECATED: Use xmlReadDoc with XML_PARSE_RECOVER. |
14021 | | * |
14022 | | * parse an XML in-memory document and build a tree. |
14023 | | * In the case the document is not Well Formed, a attempt to build a |
14024 | | * tree is tried anyway |
14025 | | * |
14026 | | * Returns the resulting document tree or NULL in case of failure |
14027 | | */ |
14028 | | |
14029 | | xmlDocPtr |
14030 | 0 | xmlRecoverDoc(const xmlChar *cur) { |
14031 | 0 | return(xmlSAXParseDoc(NULL, cur, 1)); |
14032 | 0 | } |
14033 | | |
14034 | | /** |
14035 | | * xmlParseFile: |
14036 | | * @filename: the filename |
14037 | | * |
14038 | | * DEPRECATED: Use xmlReadFile. |
14039 | | * |
14040 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14041 | | * compressed document is provided by default if found at compile-time. |
14042 | | * |
14043 | | * Returns the resulting document tree if the file was wellformed, |
14044 | | * NULL otherwise. |
14045 | | */ |
14046 | | |
14047 | | xmlDocPtr |
14048 | 0 | xmlParseFile(const char *filename) { |
14049 | 0 | return(xmlSAXParseFile(NULL, filename, 0)); |
14050 | 0 | } |
14051 | | |
14052 | | /** |
14053 | | * xmlRecoverFile: |
14054 | | * @filename: the filename |
14055 | | * |
14056 | | * DEPRECATED: Use xmlReadFile with XML_PARSE_RECOVER. |
14057 | | * |
14058 | | * parse an XML file and build a tree. Automatic support for ZLIB/Compress |
14059 | | * compressed document is provided by default if found at compile-time. |
14060 | | * In the case the document is not Well Formed, it attempts to build |
14061 | | * a tree anyway |
14062 | | * |
14063 | | * Returns the resulting document tree or NULL in case of failure |
14064 | | */ |
14065 | | |
14066 | | xmlDocPtr |
14067 | 0 | xmlRecoverFile(const char *filename) { |
14068 | 0 | return(xmlSAXParseFile(NULL, filename, 1)); |
14069 | 0 | } |
14070 | | |
14071 | | |
14072 | | /** |
14073 | | * xmlSetupParserForBuffer: |
14074 | | * @ctxt: an XML parser context |
14075 | | * @buffer: a xmlChar * buffer |
14076 | | * @filename: a file name |
14077 | | * |
14078 | | * DEPRECATED: Don't use. |
14079 | | * |
14080 | | * Setup the parser context to parse a new buffer; Clears any prior |
14081 | | * contents from the parser context. The buffer parameter must not be |
14082 | | * NULL, but the filename parameter can be |
14083 | | */ |
14084 | | void |
14085 | | xmlSetupParserForBuffer(xmlParserCtxtPtr ctxt, const xmlChar* buffer, |
14086 | | const char* filename) |
14087 | 0 | { |
14088 | 0 | xmlParserInputPtr input; |
14089 | |
|
14090 | 0 | if ((ctxt == NULL) || (buffer == NULL)) |
14091 | 0 | return; |
14092 | | |
14093 | 0 | input = xmlNewInputStream(ctxt); |
14094 | 0 | if (input == NULL) { |
14095 | 0 | xmlErrMemory(NULL, "parsing new buffer: out of memory\n"); |
14096 | 0 | xmlClearParserCtxt(ctxt); |
14097 | 0 | return; |
14098 | 0 | } |
14099 | | |
14100 | 0 | xmlClearParserCtxt(ctxt); |
14101 | 0 | if (filename != NULL) |
14102 | 0 | input->filename = (char *) xmlCanonicPath((const xmlChar *)filename); |
14103 | 0 | input->base = buffer; |
14104 | 0 | input->cur = buffer; |
14105 | 0 | input->end = &buffer[xmlStrlen(buffer)]; |
14106 | 0 | inputPush(ctxt, input); |
14107 | 0 | } |
14108 | | |
14109 | | /** |
14110 | | * xmlSAXUserParseFile: |
14111 | | * @sax: a SAX handler |
14112 | | * @user_data: The user data returned on SAX callbacks |
14113 | | * @filename: a file name |
14114 | | * |
14115 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadFile. |
14116 | | * |
14117 | | * parse an XML file and call the given SAX handler routines. |
14118 | | * Automatic support for ZLIB/Compress compressed document is provided |
14119 | | * |
14120 | | * Returns 0 in case of success or a error number otherwise |
14121 | | */ |
14122 | | int |
14123 | | xmlSAXUserParseFile(xmlSAXHandlerPtr sax, void *user_data, |
14124 | 0 | const char *filename) { |
14125 | 0 | int ret = 0; |
14126 | 0 | xmlParserCtxtPtr ctxt; |
14127 | |
|
14128 | 0 | ctxt = xmlCreateFileParserCtxt(filename); |
14129 | 0 | if (ctxt == NULL) return -1; |
14130 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14131 | 0 | xmlFree(ctxt->sax); |
14132 | 0 | ctxt->sax = sax; |
14133 | 0 | xmlDetectSAX2(ctxt); |
14134 | |
|
14135 | 0 | if (user_data != NULL) |
14136 | 0 | ctxt->userData = user_data; |
14137 | |
|
14138 | 0 | xmlParseDocument(ctxt); |
14139 | |
|
14140 | 0 | if (ctxt->wellFormed) |
14141 | 0 | ret = 0; |
14142 | 0 | else { |
14143 | 0 | if (ctxt->errNo != 0) |
14144 | 0 | ret = ctxt->errNo; |
14145 | 0 | else |
14146 | 0 | ret = -1; |
14147 | 0 | } |
14148 | 0 | if (sax != NULL) |
14149 | 0 | ctxt->sax = NULL; |
14150 | 0 | if (ctxt->myDoc != NULL) { |
14151 | 0 | xmlFreeDoc(ctxt->myDoc); |
14152 | 0 | ctxt->myDoc = NULL; |
14153 | 0 | } |
14154 | 0 | xmlFreeParserCtxt(ctxt); |
14155 | |
|
14156 | 0 | return ret; |
14157 | 0 | } |
14158 | | #endif /* LIBXML_SAX1_ENABLED */ |
14159 | | |
14160 | | /************************************************************************ |
14161 | | * * |
14162 | | * Front ends when parsing from memory * |
14163 | | * * |
14164 | | ************************************************************************/ |
14165 | | |
14166 | | /** |
14167 | | * xmlCreateMemoryParserCtxt: |
14168 | | * @buffer: a pointer to a char array |
14169 | | * @size: the size of the array |
14170 | | * |
14171 | | * Create a parser context for an XML in-memory document. |
14172 | | * |
14173 | | * Returns the new parser context or NULL |
14174 | | */ |
14175 | | xmlParserCtxtPtr |
14176 | 246k | xmlCreateMemoryParserCtxt(const char *buffer, int size) { |
14177 | 246k | xmlParserCtxtPtr ctxt; |
14178 | 246k | xmlParserInputPtr input; |
14179 | 246k | xmlParserInputBufferPtr buf; |
14180 | | |
14181 | 246k | if (buffer == NULL) |
14182 | 0 | return(NULL); |
14183 | 246k | if (size <= 0) |
14184 | 20.7k | return(NULL); |
14185 | | |
14186 | 225k | ctxt = xmlNewParserCtxt(); |
14187 | 225k | if (ctxt == NULL) |
14188 | 0 | return(NULL); |
14189 | | |
14190 | 225k | buf = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
14191 | 225k | if (buf == NULL) { |
14192 | 0 | xmlFreeParserCtxt(ctxt); |
14193 | 0 | return(NULL); |
14194 | 0 | } |
14195 | | |
14196 | 225k | input = xmlNewInputStream(ctxt); |
14197 | 225k | if (input == NULL) { |
14198 | 0 | xmlFreeParserInputBuffer(buf); |
14199 | 0 | xmlFreeParserCtxt(ctxt); |
14200 | 0 | return(NULL); |
14201 | 0 | } |
14202 | | |
14203 | 225k | input->filename = NULL; |
14204 | 225k | input->buf = buf; |
14205 | 225k | xmlBufResetInput(input->buf->buffer, input); |
14206 | | |
14207 | 225k | inputPush(ctxt, input); |
14208 | 225k | return(ctxt); |
14209 | 225k | } |
14210 | | |
14211 | | #ifdef LIBXML_SAX1_ENABLED |
14212 | | /** |
14213 | | * xmlSAXParseMemoryWithData: |
14214 | | * @sax: the SAX handler block |
14215 | | * @buffer: an pointer to a char array |
14216 | | * @size: the size of the array |
14217 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14218 | | * documents |
14219 | | * @data: the userdata |
14220 | | * |
14221 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14222 | | * |
14223 | | * parse an XML in-memory block and use the given SAX function block |
14224 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14225 | | * DOM tree building routines. |
14226 | | * |
14227 | | * User data (void *) is stored within the parser context in the |
14228 | | * context's _private member, so it is available nearly everywhere in libxml |
14229 | | * |
14230 | | * Returns the resulting document tree |
14231 | | */ |
14232 | | |
14233 | | xmlDocPtr |
14234 | | xmlSAXParseMemoryWithData(xmlSAXHandlerPtr sax, const char *buffer, |
14235 | 0 | int size, int recovery, void *data) { |
14236 | 0 | xmlDocPtr ret; |
14237 | 0 | xmlParserCtxtPtr ctxt; |
14238 | |
|
14239 | 0 | xmlInitParser(); |
14240 | |
|
14241 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14242 | 0 | if (ctxt == NULL) return(NULL); |
14243 | 0 | if (sax != NULL) { |
14244 | 0 | if (ctxt->sax != NULL) |
14245 | 0 | xmlFree(ctxt->sax); |
14246 | 0 | ctxt->sax = sax; |
14247 | 0 | } |
14248 | 0 | xmlDetectSAX2(ctxt); |
14249 | 0 | if (data!=NULL) { |
14250 | 0 | ctxt->_private=data; |
14251 | 0 | } |
14252 | |
|
14253 | 0 | ctxt->recovery = recovery; |
14254 | |
|
14255 | 0 | xmlParseDocument(ctxt); |
14256 | |
|
14257 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14258 | 0 | else { |
14259 | 0 | ret = NULL; |
14260 | 0 | xmlFreeDoc(ctxt->myDoc); |
14261 | 0 | ctxt->myDoc = NULL; |
14262 | 0 | } |
14263 | 0 | if (sax != NULL) |
14264 | 0 | ctxt->sax = NULL; |
14265 | 0 | xmlFreeParserCtxt(ctxt); |
14266 | |
|
14267 | 0 | return(ret); |
14268 | 0 | } |
14269 | | |
14270 | | /** |
14271 | | * xmlSAXParseMemory: |
14272 | | * @sax: the SAX handler block |
14273 | | * @buffer: an pointer to a char array |
14274 | | * @size: the size of the array |
14275 | | * @recovery: work in recovery mode, i.e. tries to read not Well Formed |
14276 | | * documents |
14277 | | * |
14278 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14279 | | * |
14280 | | * parse an XML in-memory block and use the given SAX function block |
14281 | | * to handle the parsing callback. If sax is NULL, fallback to the default |
14282 | | * DOM tree building routines. |
14283 | | * |
14284 | | * Returns the resulting document tree |
14285 | | */ |
14286 | | xmlDocPtr |
14287 | | xmlSAXParseMemory(xmlSAXHandlerPtr sax, const char *buffer, |
14288 | 0 | int size, int recovery) { |
14289 | 0 | return xmlSAXParseMemoryWithData(sax, buffer, size, recovery, NULL); |
14290 | 0 | } |
14291 | | |
14292 | | /** |
14293 | | * xmlParseMemory: |
14294 | | * @buffer: an pointer to a char array |
14295 | | * @size: the size of the array |
14296 | | * |
14297 | | * DEPRECATED: Use xmlReadMemory. |
14298 | | * |
14299 | | * parse an XML in-memory block and build a tree. |
14300 | | * |
14301 | | * Returns the resulting document tree |
14302 | | */ |
14303 | | |
14304 | 0 | xmlDocPtr xmlParseMemory(const char *buffer, int size) { |
14305 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 0)); |
14306 | 0 | } |
14307 | | |
14308 | | /** |
14309 | | * xmlRecoverMemory: |
14310 | | * @buffer: an pointer to a char array |
14311 | | * @size: the size of the array |
14312 | | * |
14313 | | * DEPRECATED: Use xmlReadMemory with XML_PARSE_RECOVER. |
14314 | | * |
14315 | | * parse an XML in-memory block and build a tree. |
14316 | | * In the case the document is not Well Formed, an attempt to |
14317 | | * build a tree is tried anyway |
14318 | | * |
14319 | | * Returns the resulting document tree or NULL in case of error |
14320 | | */ |
14321 | | |
14322 | 0 | xmlDocPtr xmlRecoverMemory(const char *buffer, int size) { |
14323 | 0 | return(xmlSAXParseMemory(NULL, buffer, size, 1)); |
14324 | 0 | } |
14325 | | |
14326 | | /** |
14327 | | * xmlSAXUserParseMemory: |
14328 | | * @sax: a SAX handler |
14329 | | * @user_data: The user data returned on SAX callbacks |
14330 | | * @buffer: an in-memory XML document input |
14331 | | * @size: the length of the XML document in bytes |
14332 | | * |
14333 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadMemory. |
14334 | | * |
14335 | | * parse an XML in-memory buffer and call the given SAX handler routines. |
14336 | | * |
14337 | | * Returns 0 in case of success or a error number otherwise |
14338 | | */ |
14339 | | int xmlSAXUserParseMemory(xmlSAXHandlerPtr sax, void *user_data, |
14340 | 0 | const char *buffer, int size) { |
14341 | 0 | int ret = 0; |
14342 | 0 | xmlParserCtxtPtr ctxt; |
14343 | |
|
14344 | 0 | xmlInitParser(); |
14345 | |
|
14346 | 0 | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
14347 | 0 | if (ctxt == NULL) return -1; |
14348 | 0 | if (ctxt->sax != (xmlSAXHandlerPtr) &xmlDefaultSAXHandler) |
14349 | 0 | xmlFree(ctxt->sax); |
14350 | 0 | ctxt->sax = sax; |
14351 | 0 | xmlDetectSAX2(ctxt); |
14352 | |
|
14353 | 0 | if (user_data != NULL) |
14354 | 0 | ctxt->userData = user_data; |
14355 | |
|
14356 | 0 | xmlParseDocument(ctxt); |
14357 | |
|
14358 | 0 | if (ctxt->wellFormed) |
14359 | 0 | ret = 0; |
14360 | 0 | else { |
14361 | 0 | if (ctxt->errNo != 0) |
14362 | 0 | ret = ctxt->errNo; |
14363 | 0 | else |
14364 | 0 | ret = -1; |
14365 | 0 | } |
14366 | 0 | if (sax != NULL) |
14367 | 0 | ctxt->sax = NULL; |
14368 | 0 | if (ctxt->myDoc != NULL) { |
14369 | 0 | xmlFreeDoc(ctxt->myDoc); |
14370 | 0 | ctxt->myDoc = NULL; |
14371 | 0 | } |
14372 | 0 | xmlFreeParserCtxt(ctxt); |
14373 | |
|
14374 | 0 | return ret; |
14375 | 0 | } |
14376 | | #endif /* LIBXML_SAX1_ENABLED */ |
14377 | | |
14378 | | /** |
14379 | | * xmlCreateDocParserCtxt: |
14380 | | * @cur: a pointer to an array of xmlChar |
14381 | | * |
14382 | | * Creates a parser context for an XML in-memory document. |
14383 | | * |
14384 | | * Returns the new parser context or NULL |
14385 | | */ |
14386 | | xmlParserCtxtPtr |
14387 | 0 | xmlCreateDocParserCtxt(const xmlChar *cur) { |
14388 | 0 | int len; |
14389 | |
|
14390 | 0 | if (cur == NULL) |
14391 | 0 | return(NULL); |
14392 | 0 | len = xmlStrlen(cur); |
14393 | 0 | return(xmlCreateMemoryParserCtxt((const char *)cur, len)); |
14394 | 0 | } |
14395 | | |
14396 | | #ifdef LIBXML_SAX1_ENABLED |
14397 | | /** |
14398 | | * xmlSAXParseDoc: |
14399 | | * @sax: the SAX handler block |
14400 | | * @cur: a pointer to an array of xmlChar |
14401 | | * @recovery: work in recovery mode, i.e. tries to read no Well Formed |
14402 | | * documents |
14403 | | * |
14404 | | * DEPRECATED: Use xmlNewSAXParserCtxt and xmlCtxtReadDoc. |
14405 | | * |
14406 | | * parse an XML in-memory document and build a tree. |
14407 | | * It use the given SAX function block to handle the parsing callback. |
14408 | | * If sax is NULL, fallback to the default DOM tree building routines. |
14409 | | * |
14410 | | * Returns the resulting document tree |
14411 | | */ |
14412 | | |
14413 | | xmlDocPtr |
14414 | 0 | xmlSAXParseDoc(xmlSAXHandlerPtr sax, const xmlChar *cur, int recovery) { |
14415 | 0 | xmlDocPtr ret; |
14416 | 0 | xmlParserCtxtPtr ctxt; |
14417 | 0 | xmlSAXHandlerPtr oldsax = NULL; |
14418 | |
|
14419 | 0 | if (cur == NULL) return(NULL); |
14420 | | |
14421 | | |
14422 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
14423 | 0 | if (ctxt == NULL) return(NULL); |
14424 | 0 | if (sax != NULL) { |
14425 | 0 | oldsax = ctxt->sax; |
14426 | 0 | ctxt->sax = sax; |
14427 | 0 | ctxt->userData = NULL; |
14428 | 0 | } |
14429 | 0 | xmlDetectSAX2(ctxt); |
14430 | |
|
14431 | 0 | xmlParseDocument(ctxt); |
14432 | 0 | if ((ctxt->wellFormed) || recovery) ret = ctxt->myDoc; |
14433 | 0 | else { |
14434 | 0 | ret = NULL; |
14435 | 0 | xmlFreeDoc(ctxt->myDoc); |
14436 | 0 | ctxt->myDoc = NULL; |
14437 | 0 | } |
14438 | 0 | if (sax != NULL) |
14439 | 0 | ctxt->sax = oldsax; |
14440 | 0 | xmlFreeParserCtxt(ctxt); |
14441 | |
|
14442 | 0 | return(ret); |
14443 | 0 | } |
14444 | | |
14445 | | /** |
14446 | | * xmlParseDoc: |
14447 | | * @cur: a pointer to an array of xmlChar |
14448 | | * |
14449 | | * DEPRECATED: Use xmlReadDoc. |
14450 | | * |
14451 | | * parse an XML in-memory document and build a tree. |
14452 | | * |
14453 | | * Returns the resulting document tree |
14454 | | */ |
14455 | | |
14456 | | xmlDocPtr |
14457 | 0 | xmlParseDoc(const xmlChar *cur) { |
14458 | 0 | return(xmlSAXParseDoc(NULL, cur, 0)); |
14459 | 0 | } |
14460 | | #endif /* LIBXML_SAX1_ENABLED */ |
14461 | | |
14462 | | #ifdef LIBXML_LEGACY_ENABLED |
14463 | | /************************************************************************ |
14464 | | * * |
14465 | | * Specific function to keep track of entities references * |
14466 | | * and used by the XSLT debugger * |
14467 | | * * |
14468 | | ************************************************************************/ |
14469 | | |
14470 | | static xmlEntityReferenceFunc xmlEntityRefFunc = NULL; |
14471 | | |
14472 | | /** |
14473 | | * xmlAddEntityReference: |
14474 | | * @ent : A valid entity |
14475 | | * @firstNode : A valid first node for children of entity |
14476 | | * @lastNode : A valid last node of children entity |
14477 | | * |
14478 | | * Notify of a reference to an entity of type XML_EXTERNAL_GENERAL_PARSED_ENTITY |
14479 | | */ |
14480 | | static void |
14481 | | xmlAddEntityReference(xmlEntityPtr ent, xmlNodePtr firstNode, |
14482 | | xmlNodePtr lastNode) |
14483 | | { |
14484 | | if (xmlEntityRefFunc != NULL) { |
14485 | | (*xmlEntityRefFunc) (ent, firstNode, lastNode); |
14486 | | } |
14487 | | } |
14488 | | |
14489 | | |
14490 | | /** |
14491 | | * xmlSetEntityReferenceFunc: |
14492 | | * @func: A valid function |
14493 | | * |
14494 | | * Set the function to call call back when a xml reference has been made |
14495 | | */ |
14496 | | void |
14497 | | xmlSetEntityReferenceFunc(xmlEntityReferenceFunc func) |
14498 | | { |
14499 | | xmlEntityRefFunc = func; |
14500 | | } |
14501 | | #endif /* LIBXML_LEGACY_ENABLED */ |
14502 | | |
14503 | | /************************************************************************ |
14504 | | * * |
14505 | | * Miscellaneous * |
14506 | | * * |
14507 | | ************************************************************************/ |
14508 | | |
14509 | | static int xmlParserInitialized = 0; |
14510 | | |
14511 | | /** |
14512 | | * xmlInitParser: |
14513 | | * |
14514 | | * Initialization function for the XML parser. |
14515 | | * This is not reentrant. Call once before processing in case of |
14516 | | * use in multithreaded programs. |
14517 | | */ |
14518 | | |
14519 | | void |
14520 | 254M | xmlInitParser(void) { |
14521 | | /* |
14522 | | * Note that the initialization code must not make memory allocations. |
14523 | | */ |
14524 | 254M | if (xmlParserInitialized != 0) |
14525 | 254M | return; |
14526 | | |
14527 | 2.96k | #ifdef LIBXML_THREAD_ENABLED |
14528 | 2.96k | __xmlGlobalInitMutexLock(); |
14529 | 2.96k | if (xmlParserInitialized == 0) { |
14530 | 2.96k | #endif |
14531 | | #if defined(_WIN32) && (!defined(LIBXML_STATIC) || defined(LIBXML_STATIC_FOR_DLL)) |
14532 | | if (xmlFree == free) |
14533 | | atexit(xmlCleanupParser); |
14534 | | #endif |
14535 | | |
14536 | 2.96k | xmlInitThreadsInternal(); |
14537 | 2.96k | xmlInitGlobalsInternal(); |
14538 | 2.96k | xmlInitMemoryInternal(); |
14539 | 2.96k | __xmlInitializeDict(); |
14540 | 2.96k | xmlInitEncodingInternal(); |
14541 | 2.96k | xmlRegisterDefaultInputCallbacks(); |
14542 | 2.96k | #ifdef LIBXML_OUTPUT_ENABLED |
14543 | 2.96k | xmlRegisterDefaultOutputCallbacks(); |
14544 | 2.96k | #endif /* LIBXML_OUTPUT_ENABLED */ |
14545 | 2.96k | #if defined(LIBXML_XPATH_ENABLED) || defined(LIBXML_SCHEMAS_ENABLED) |
14546 | 2.96k | xmlInitXPathInternal(); |
14547 | 2.96k | #endif |
14548 | 2.96k | xmlParserInitialized = 1; |
14549 | 2.96k | #ifdef LIBXML_THREAD_ENABLED |
14550 | 2.96k | } |
14551 | 2.96k | __xmlGlobalInitMutexUnlock(); |
14552 | 2.96k | #endif |
14553 | 2.96k | } |
14554 | | |
14555 | | /** |
14556 | | * xmlCleanupParser: |
14557 | | * |
14558 | | * This function name is somewhat misleading. It does not clean up |
14559 | | * parser state, it cleans up memory allocated by the library itself. |
14560 | | * It is a cleanup function for the XML library. It tries to reclaim all |
14561 | | * related global memory allocated for the library processing. |
14562 | | * It doesn't deallocate any document related memory. One should |
14563 | | * call xmlCleanupParser() only when the process has finished using |
14564 | | * the library and all XML/HTML documents built with it. |
14565 | | * See also xmlInitParser() which has the opposite function of preparing |
14566 | | * the library for operations. |
14567 | | * |
14568 | | * WARNING: if your application is multithreaded or has plugin support |
14569 | | * calling this may crash the application if another thread or |
14570 | | * a plugin is still using libxml2. It's sometimes very hard to |
14571 | | * guess if libxml2 is in use in the application, some libraries |
14572 | | * or plugins may use it without notice. In case of doubt abstain |
14573 | | * from calling this function or do it just before calling exit() |
14574 | | * to avoid leak reports from valgrind ! |
14575 | | */ |
14576 | | |
14577 | | void |
14578 | 0 | xmlCleanupParser(void) { |
14579 | 0 | if (!xmlParserInitialized) |
14580 | 0 | return; |
14581 | | |
14582 | 0 | xmlCleanupCharEncodingHandlers(); |
14583 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14584 | 0 | xmlCatalogCleanup(); |
14585 | 0 | #endif |
14586 | 0 | xmlCleanupDictInternal(); |
14587 | 0 | xmlCleanupInputCallbacks(); |
14588 | 0 | #ifdef LIBXML_OUTPUT_ENABLED |
14589 | 0 | xmlCleanupOutputCallbacks(); |
14590 | 0 | #endif |
14591 | 0 | #ifdef LIBXML_SCHEMAS_ENABLED |
14592 | 0 | xmlSchemaCleanupTypes(); |
14593 | 0 | xmlRelaxNGCleanupTypes(); |
14594 | 0 | #endif |
14595 | 0 | xmlCleanupGlobalsInternal(); |
14596 | 0 | xmlCleanupThreadsInternal(); |
14597 | 0 | xmlCleanupMemoryInternal(); |
14598 | 0 | xmlParserInitialized = 0; |
14599 | 0 | } |
14600 | | |
14601 | | #if defined(HAVE_ATTRIBUTE_DESTRUCTOR) && !defined(LIBXML_STATIC) && \ |
14602 | | !defined(_WIN32) |
14603 | | static void |
14604 | | ATTRIBUTE_DESTRUCTOR |
14605 | 0 | xmlDestructor(void) { |
14606 | | /* |
14607 | | * Calling custom deallocation functions in a destructor can cause |
14608 | | * problems, for example with Nokogiri. |
14609 | | */ |
14610 | 0 | if (xmlFree == free) |
14611 | 0 | xmlCleanupParser(); |
14612 | 0 | } |
14613 | | #endif |
14614 | | |
14615 | | /************************************************************************ |
14616 | | * * |
14617 | | * New set (2.6.0) of simpler and more flexible APIs * |
14618 | | * * |
14619 | | ************************************************************************/ |
14620 | | |
14621 | | /** |
14622 | | * DICT_FREE: |
14623 | | * @str: a string |
14624 | | * |
14625 | | * Free a string if it is not owned by the "dict" dictionary in the |
14626 | | * current scope |
14627 | | */ |
14628 | | #define DICT_FREE(str) \ |
14629 | 0 | if ((str) && ((!dict) || \ |
14630 | 0 | (xmlDictOwns(dict, (const xmlChar *)(str)) == 0))) \ |
14631 | 0 | xmlFree((char *)(str)); |
14632 | | |
14633 | | /** |
14634 | | * xmlCtxtReset: |
14635 | | * @ctxt: an XML parser context |
14636 | | * |
14637 | | * Reset a parser context |
14638 | | */ |
14639 | | void |
14640 | | xmlCtxtReset(xmlParserCtxtPtr ctxt) |
14641 | 0 | { |
14642 | 0 | xmlParserInputPtr input; |
14643 | 0 | xmlDictPtr dict; |
14644 | |
|
14645 | 0 | if (ctxt == NULL) |
14646 | 0 | return; |
14647 | | |
14648 | 0 | dict = ctxt->dict; |
14649 | |
|
14650 | 0 | while ((input = inputPop(ctxt)) != NULL) { /* Non consuming */ |
14651 | 0 | xmlFreeInputStream(input); |
14652 | 0 | } |
14653 | 0 | ctxt->inputNr = 0; |
14654 | 0 | ctxt->input = NULL; |
14655 | |
|
14656 | 0 | ctxt->spaceNr = 0; |
14657 | 0 | if (ctxt->spaceTab != NULL) { |
14658 | 0 | ctxt->spaceTab[0] = -1; |
14659 | 0 | ctxt->space = &ctxt->spaceTab[0]; |
14660 | 0 | } else { |
14661 | 0 | ctxt->space = NULL; |
14662 | 0 | } |
14663 | | |
14664 | |
|
14665 | 0 | ctxt->nodeNr = 0; |
14666 | 0 | ctxt->node = NULL; |
14667 | |
|
14668 | 0 | ctxt->nameNr = 0; |
14669 | 0 | ctxt->name = NULL; |
14670 | |
|
14671 | 0 | ctxt->nsNr = 0; |
14672 | |
|
14673 | 0 | DICT_FREE(ctxt->version); |
14674 | 0 | ctxt->version = NULL; |
14675 | 0 | DICT_FREE(ctxt->encoding); |
14676 | 0 | ctxt->encoding = NULL; |
14677 | 0 | DICT_FREE(ctxt->directory); |
14678 | 0 | ctxt->directory = NULL; |
14679 | 0 | DICT_FREE(ctxt->extSubURI); |
14680 | 0 | ctxt->extSubURI = NULL; |
14681 | 0 | DICT_FREE(ctxt->extSubSystem); |
14682 | 0 | ctxt->extSubSystem = NULL; |
14683 | 0 | if (ctxt->myDoc != NULL) |
14684 | 0 | xmlFreeDoc(ctxt->myDoc); |
14685 | 0 | ctxt->myDoc = NULL; |
14686 | |
|
14687 | 0 | ctxt->standalone = -1; |
14688 | 0 | ctxt->hasExternalSubset = 0; |
14689 | 0 | ctxt->hasPErefs = 0; |
14690 | 0 | ctxt->html = 0; |
14691 | 0 | ctxt->external = 0; |
14692 | 0 | ctxt->instate = XML_PARSER_START; |
14693 | 0 | ctxt->token = 0; |
14694 | |
|
14695 | 0 | ctxt->wellFormed = 1; |
14696 | 0 | ctxt->nsWellFormed = 1; |
14697 | 0 | ctxt->disableSAX = 0; |
14698 | 0 | ctxt->valid = 1; |
14699 | | #if 0 |
14700 | | ctxt->vctxt.userData = ctxt; |
14701 | | ctxt->vctxt.error = xmlParserValidityError; |
14702 | | ctxt->vctxt.warning = xmlParserValidityWarning; |
14703 | | #endif |
14704 | 0 | ctxt->record_info = 0; |
14705 | 0 | ctxt->checkIndex = 0; |
14706 | 0 | ctxt->endCheckState = 0; |
14707 | 0 | ctxt->inSubset = 0; |
14708 | 0 | ctxt->errNo = XML_ERR_OK; |
14709 | 0 | ctxt->depth = 0; |
14710 | 0 | ctxt->charset = XML_CHAR_ENCODING_UTF8; |
14711 | 0 | ctxt->catalogs = NULL; |
14712 | 0 | ctxt->sizeentities = 0; |
14713 | 0 | ctxt->sizeentcopy = 0; |
14714 | 0 | xmlInitNodeInfoSeq(&ctxt->node_seq); |
14715 | |
|
14716 | 0 | if (ctxt->attsDefault != NULL) { |
14717 | 0 | xmlHashFree(ctxt->attsDefault, xmlHashDefaultDeallocator); |
14718 | 0 | ctxt->attsDefault = NULL; |
14719 | 0 | } |
14720 | 0 | if (ctxt->attsSpecial != NULL) { |
14721 | 0 | xmlHashFree(ctxt->attsSpecial, NULL); |
14722 | 0 | ctxt->attsSpecial = NULL; |
14723 | 0 | } |
14724 | |
|
14725 | 0 | #ifdef LIBXML_CATALOG_ENABLED |
14726 | 0 | if (ctxt->catalogs != NULL) |
14727 | 0 | xmlCatalogFreeLocal(ctxt->catalogs); |
14728 | 0 | #endif |
14729 | 0 | ctxt->nbErrors = 0; |
14730 | 0 | ctxt->nbWarnings = 0; |
14731 | 0 | if (ctxt->lastError.code != XML_ERR_OK) |
14732 | 0 | xmlResetError(&ctxt->lastError); |
14733 | 0 | } |
14734 | | |
14735 | | /** |
14736 | | * xmlCtxtResetPush: |
14737 | | * @ctxt: an XML parser context |
14738 | | * @chunk: a pointer to an array of chars |
14739 | | * @size: number of chars in the array |
14740 | | * @filename: an optional file name or URI |
14741 | | * @encoding: the document encoding, or NULL |
14742 | | * |
14743 | | * Reset a push parser context |
14744 | | * |
14745 | | * Returns 0 in case of success and 1 in case of error |
14746 | | */ |
14747 | | int |
14748 | | xmlCtxtResetPush(xmlParserCtxtPtr ctxt, const char *chunk, |
14749 | | int size, const char *filename, const char *encoding) |
14750 | 0 | { |
14751 | 0 | xmlParserInputPtr inputStream; |
14752 | 0 | xmlParserInputBufferPtr buf; |
14753 | 0 | xmlCharEncoding enc = XML_CHAR_ENCODING_NONE; |
14754 | |
|
14755 | 0 | if (ctxt == NULL) |
14756 | 0 | return(1); |
14757 | | |
14758 | 0 | if ((encoding == NULL) && (chunk != NULL) && (size >= 4)) |
14759 | 0 | enc = xmlDetectCharEncoding((const xmlChar *) chunk, size); |
14760 | |
|
14761 | 0 | buf = xmlAllocParserInputBuffer(enc); |
14762 | 0 | if (buf == NULL) |
14763 | 0 | return(1); |
14764 | | |
14765 | 0 | if (ctxt == NULL) { |
14766 | 0 | xmlFreeParserInputBuffer(buf); |
14767 | 0 | return(1); |
14768 | 0 | } |
14769 | | |
14770 | 0 | xmlCtxtReset(ctxt); |
14771 | |
|
14772 | 0 | if (filename == NULL) { |
14773 | 0 | ctxt->directory = NULL; |
14774 | 0 | } else { |
14775 | 0 | ctxt->directory = xmlParserGetDirectory(filename); |
14776 | 0 | } |
14777 | |
|
14778 | 0 | inputStream = xmlNewInputStream(ctxt); |
14779 | 0 | if (inputStream == NULL) { |
14780 | 0 | xmlFreeParserInputBuffer(buf); |
14781 | 0 | return(1); |
14782 | 0 | } |
14783 | | |
14784 | 0 | if (filename == NULL) |
14785 | 0 | inputStream->filename = NULL; |
14786 | 0 | else |
14787 | 0 | inputStream->filename = (char *) |
14788 | 0 | xmlCanonicPath((const xmlChar *) filename); |
14789 | 0 | inputStream->buf = buf; |
14790 | 0 | xmlBufResetInput(buf->buffer, inputStream); |
14791 | |
|
14792 | 0 | inputPush(ctxt, inputStream); |
14793 | |
|
14794 | 0 | if ((size > 0) && (chunk != NULL) && (ctxt->input != NULL) && |
14795 | 0 | (ctxt->input->buf != NULL)) { |
14796 | 0 | size_t base = xmlBufGetInputBase(ctxt->input->buf->buffer, ctxt->input); |
14797 | 0 | size_t cur = ctxt->input->cur - ctxt->input->base; |
14798 | |
|
14799 | 0 | xmlParserInputBufferPush(ctxt->input->buf, size, chunk); |
14800 | |
|
14801 | 0 | xmlBufSetInputBaseCur(ctxt->input->buf->buffer, ctxt->input, base, cur); |
14802 | | #ifdef DEBUG_PUSH |
14803 | | xmlGenericError(xmlGenericErrorContext, "PP: pushed %d\n", size); |
14804 | | #endif |
14805 | 0 | } |
14806 | |
|
14807 | 0 | if (encoding != NULL) { |
14808 | 0 | xmlCharEncodingHandlerPtr hdlr; |
14809 | |
|
14810 | 0 | if (ctxt->encoding != NULL) |
14811 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
14812 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
14813 | |
|
14814 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
14815 | 0 | if (hdlr != NULL) { |
14816 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
14817 | 0 | } else { |
14818 | 0 | xmlFatalErrMsgStr(ctxt, XML_ERR_UNSUPPORTED_ENCODING, |
14819 | 0 | "Unsupported encoding %s\n", BAD_CAST encoding); |
14820 | 0 | } |
14821 | 0 | } else if (enc != XML_CHAR_ENCODING_NONE) { |
14822 | 0 | xmlSwitchEncoding(ctxt, enc); |
14823 | 0 | } |
14824 | |
|
14825 | 0 | return(0); |
14826 | 0 | } |
14827 | | |
14828 | | |
14829 | | /** |
14830 | | * xmlCtxtUseOptionsInternal: |
14831 | | * @ctxt: an XML parser context |
14832 | | * @options: a combination of xmlParserOption |
14833 | | * @encoding: the user provided encoding to use |
14834 | | * |
14835 | | * Applies the options to the parser context |
14836 | | * |
14837 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
14838 | | * in case of error. |
14839 | | */ |
14840 | | static int |
14841 | | xmlCtxtUseOptionsInternal(xmlParserCtxtPtr ctxt, int options, const char *encoding) |
14842 | 461k | { |
14843 | 461k | if (ctxt == NULL) |
14844 | 0 | return(-1); |
14845 | 461k | if (encoding != NULL) { |
14846 | 0 | if (ctxt->encoding != NULL) |
14847 | 0 | xmlFree((xmlChar *) ctxt->encoding); |
14848 | 0 | ctxt->encoding = xmlStrdup((const xmlChar *) encoding); |
14849 | 0 | } |
14850 | 461k | if (options & XML_PARSE_RECOVER) { |
14851 | 252k | ctxt->recovery = 1; |
14852 | 252k | options -= XML_PARSE_RECOVER; |
14853 | 252k | ctxt->options |= XML_PARSE_RECOVER; |
14854 | 252k | } else |
14855 | 208k | ctxt->recovery = 0; |
14856 | 461k | if (options & XML_PARSE_DTDLOAD) { |
14857 | 288k | ctxt->loadsubset = XML_DETECT_IDS; |
14858 | 288k | options -= XML_PARSE_DTDLOAD; |
14859 | 288k | ctxt->options |= XML_PARSE_DTDLOAD; |
14860 | 288k | } else |
14861 | 172k | ctxt->loadsubset = 0; |
14862 | 461k | if (options & XML_PARSE_DTDATTR) { |
14863 | 192k | ctxt->loadsubset |= XML_COMPLETE_ATTRS; |
14864 | 192k | options -= XML_PARSE_DTDATTR; |
14865 | 192k | ctxt->options |= XML_PARSE_DTDATTR; |
14866 | 192k | } |
14867 | 461k | if (options & XML_PARSE_NOENT) { |
14868 | 260k | ctxt->replaceEntities = 1; |
14869 | | /* ctxt->loadsubset |= XML_DETECT_IDS; */ |
14870 | 260k | options -= XML_PARSE_NOENT; |
14871 | 260k | ctxt->options |= XML_PARSE_NOENT; |
14872 | 260k | } else |
14873 | 200k | ctxt->replaceEntities = 0; |
14874 | 461k | if (options & XML_PARSE_PEDANTIC) { |
14875 | 59.1k | ctxt->pedantic = 1; |
14876 | 59.1k | options -= XML_PARSE_PEDANTIC; |
14877 | 59.1k | ctxt->options |= XML_PARSE_PEDANTIC; |
14878 | 59.1k | } else |
14879 | 401k | ctxt->pedantic = 0; |
14880 | 461k | if (options & XML_PARSE_NOBLANKS) { |
14881 | 183k | ctxt->keepBlanks = 0; |
14882 | 183k | ctxt->sax->ignorableWhitespace = xmlSAX2IgnorableWhitespace; |
14883 | 183k | options -= XML_PARSE_NOBLANKS; |
14884 | 183k | ctxt->options |= XML_PARSE_NOBLANKS; |
14885 | 183k | } else |
14886 | 277k | ctxt->keepBlanks = 1; |
14887 | 461k | if (options & XML_PARSE_DTDVALID) { |
14888 | 198k | ctxt->validate = 1; |
14889 | 198k | if (options & XML_PARSE_NOWARNING) |
14890 | 94.8k | ctxt->vctxt.warning = NULL; |
14891 | 198k | if (options & XML_PARSE_NOERROR) |
14892 | 157k | ctxt->vctxt.error = NULL; |
14893 | 198k | options -= XML_PARSE_DTDVALID; |
14894 | 198k | ctxt->options |= XML_PARSE_DTDVALID; |
14895 | 198k | } else |
14896 | 262k | ctxt->validate = 0; |
14897 | 461k | if (options & XML_PARSE_NOWARNING) { |
14898 | 198k | ctxt->sax->warning = NULL; |
14899 | 198k | options -= XML_PARSE_NOWARNING; |
14900 | 198k | } |
14901 | 461k | if (options & XML_PARSE_NOERROR) { |
14902 | 283k | ctxt->sax->error = NULL; |
14903 | 283k | ctxt->sax->fatalError = NULL; |
14904 | 283k | options -= XML_PARSE_NOERROR; |
14905 | 283k | } |
14906 | 461k | #ifdef LIBXML_SAX1_ENABLED |
14907 | 461k | if (options & XML_PARSE_SAX1) { |
14908 | 174k | ctxt->sax->startElement = xmlSAX2StartElement; |
14909 | 174k | ctxt->sax->endElement = xmlSAX2EndElement; |
14910 | 174k | ctxt->sax->startElementNs = NULL; |
14911 | 174k | ctxt->sax->endElementNs = NULL; |
14912 | 174k | ctxt->sax->initialized = 1; |
14913 | 174k | options -= XML_PARSE_SAX1; |
14914 | 174k | ctxt->options |= XML_PARSE_SAX1; |
14915 | 174k | } |
14916 | 461k | #endif /* LIBXML_SAX1_ENABLED */ |
14917 | 461k | if (options & XML_PARSE_NODICT) { |
14918 | 156k | ctxt->dictNames = 0; |
14919 | 156k | options -= XML_PARSE_NODICT; |
14920 | 156k | ctxt->options |= XML_PARSE_NODICT; |
14921 | 305k | } else { |
14922 | 305k | ctxt->dictNames = 1; |
14923 | 305k | } |
14924 | 461k | if (options & XML_PARSE_NOCDATA) { |
14925 | 186k | ctxt->sax->cdataBlock = NULL; |
14926 | 186k | options -= XML_PARSE_NOCDATA; |
14927 | 186k | ctxt->options |= XML_PARSE_NOCDATA; |
14928 | 186k | } |
14929 | 461k | if (options & XML_PARSE_NSCLEAN) { |
14930 | 278k | ctxt->options |= XML_PARSE_NSCLEAN; |
14931 | 278k | options -= XML_PARSE_NSCLEAN; |
14932 | 278k | } |
14933 | 461k | if (options & XML_PARSE_NONET) { |
14934 | 174k | ctxt->options |= XML_PARSE_NONET; |
14935 | 174k | options -= XML_PARSE_NONET; |
14936 | 174k | } |
14937 | 461k | if (options & XML_PARSE_COMPACT) { |
14938 | 272k | ctxt->options |= XML_PARSE_COMPACT; |
14939 | 272k | options -= XML_PARSE_COMPACT; |
14940 | 272k | } |
14941 | 461k | if (options & XML_PARSE_OLD10) { |
14942 | 160k | ctxt->options |= XML_PARSE_OLD10; |
14943 | 160k | options -= XML_PARSE_OLD10; |
14944 | 160k | } |
14945 | 461k | if (options & XML_PARSE_NOBASEFIX) { |
14946 | 191k | ctxt->options |= XML_PARSE_NOBASEFIX; |
14947 | 191k | options -= XML_PARSE_NOBASEFIX; |
14948 | 191k | } |
14949 | 461k | if (options & XML_PARSE_HUGE) { |
14950 | 168k | ctxt->options |= XML_PARSE_HUGE; |
14951 | 168k | options -= XML_PARSE_HUGE; |
14952 | 168k | if (ctxt->dict != NULL) |
14953 | 168k | xmlDictSetLimit(ctxt->dict, 0); |
14954 | 168k | } |
14955 | 461k | if (options & XML_PARSE_OLDSAX) { |
14956 | 160k | ctxt->options |= XML_PARSE_OLDSAX; |
14957 | 160k | options -= XML_PARSE_OLDSAX; |
14958 | 160k | } |
14959 | 461k | if (options & XML_PARSE_IGNORE_ENC) { |
14960 | 269k | ctxt->options |= XML_PARSE_IGNORE_ENC; |
14961 | 269k | options -= XML_PARSE_IGNORE_ENC; |
14962 | 269k | } |
14963 | 461k | if (options & XML_PARSE_BIG_LINES) { |
14964 | 191k | ctxt->options |= XML_PARSE_BIG_LINES; |
14965 | 191k | options -= XML_PARSE_BIG_LINES; |
14966 | 191k | } |
14967 | 461k | ctxt->linenumbers = 1; |
14968 | 461k | return (options); |
14969 | 461k | } |
14970 | | |
14971 | | /** |
14972 | | * xmlCtxtUseOptions: |
14973 | | * @ctxt: an XML parser context |
14974 | | * @options: a combination of xmlParserOption |
14975 | | * |
14976 | | * Applies the options to the parser context |
14977 | | * |
14978 | | * Returns 0 in case of success, the set of unknown or unimplemented options |
14979 | | * in case of error. |
14980 | | */ |
14981 | | int |
14982 | | xmlCtxtUseOptions(xmlParserCtxtPtr ctxt, int options) |
14983 | 307k | { |
14984 | 307k | return(xmlCtxtUseOptionsInternal(ctxt, options, NULL)); |
14985 | 307k | } |
14986 | | |
14987 | | /** |
14988 | | * xmlDoRead: |
14989 | | * @ctxt: an XML parser context |
14990 | | * @URL: the base URL to use for the document |
14991 | | * @encoding: the document encoding, or NULL |
14992 | | * @options: a combination of xmlParserOption |
14993 | | * @reuse: keep the context for reuse |
14994 | | * |
14995 | | * Common front-end for the xmlRead functions |
14996 | | * |
14997 | | * Returns the resulting document tree or NULL |
14998 | | */ |
14999 | | static xmlDocPtr |
15000 | | xmlDoRead(xmlParserCtxtPtr ctxt, const char *URL, const char *encoding, |
15001 | | int options, int reuse) |
15002 | 153k | { |
15003 | 153k | xmlDocPtr ret; |
15004 | | |
15005 | 153k | xmlCtxtUseOptionsInternal(ctxt, options, encoding); |
15006 | 153k | if (encoding != NULL) { |
15007 | 0 | xmlCharEncodingHandlerPtr hdlr; |
15008 | |
|
15009 | 0 | hdlr = xmlFindCharEncodingHandler(encoding); |
15010 | 0 | if (hdlr != NULL) |
15011 | 0 | xmlSwitchToEncoding(ctxt, hdlr); |
15012 | 0 | } |
15013 | 153k | if ((URL != NULL) && (ctxt->input != NULL) && |
15014 | 153k | (ctxt->input->filename == NULL)) |
15015 | 153k | ctxt->input->filename = (char *) xmlStrdup((const xmlChar *) URL); |
15016 | 153k | xmlParseDocument(ctxt); |
15017 | 153k | if ((ctxt->wellFormed) || ctxt->recovery) |
15018 | 97.3k | ret = ctxt->myDoc; |
15019 | 56.3k | else { |
15020 | 56.3k | ret = NULL; |
15021 | 56.3k | if (ctxt->myDoc != NULL) { |
15022 | 53.0k | xmlFreeDoc(ctxt->myDoc); |
15023 | 53.0k | } |
15024 | 56.3k | } |
15025 | 153k | ctxt->myDoc = NULL; |
15026 | 153k | if (!reuse) { |
15027 | 153k | xmlFreeParserCtxt(ctxt); |
15028 | 153k | } |
15029 | | |
15030 | 153k | return (ret); |
15031 | 153k | } |
15032 | | |
15033 | | /** |
15034 | | * xmlReadDoc: |
15035 | | * @cur: a pointer to a zero terminated string |
15036 | | * @URL: the base URL to use for the document |
15037 | | * @encoding: the document encoding, or NULL |
15038 | | * @options: a combination of xmlParserOption |
15039 | | * |
15040 | | * parse an XML in-memory document and build a tree. |
15041 | | * |
15042 | | * Returns the resulting document tree |
15043 | | */ |
15044 | | xmlDocPtr |
15045 | | xmlReadDoc(const xmlChar * cur, const char *URL, const char *encoding, int options) |
15046 | 0 | { |
15047 | 0 | xmlParserCtxtPtr ctxt; |
15048 | |
|
15049 | 0 | if (cur == NULL) |
15050 | 0 | return (NULL); |
15051 | 0 | xmlInitParser(); |
15052 | |
|
15053 | 0 | ctxt = xmlCreateDocParserCtxt(cur); |
15054 | 0 | if (ctxt == NULL) |
15055 | 0 | return (NULL); |
15056 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15057 | 0 | } |
15058 | | |
15059 | | /** |
15060 | | * xmlReadFile: |
15061 | | * @filename: a file or URL |
15062 | | * @encoding: the document encoding, or NULL |
15063 | | * @options: a combination of xmlParserOption |
15064 | | * |
15065 | | * parse an XML file from the filesystem or the network. |
15066 | | * |
15067 | | * Returns the resulting document tree |
15068 | | */ |
15069 | | xmlDocPtr |
15070 | | xmlReadFile(const char *filename, const char *encoding, int options) |
15071 | 0 | { |
15072 | 0 | xmlParserCtxtPtr ctxt; |
15073 | |
|
15074 | 0 | xmlInitParser(); |
15075 | 0 | ctxt = xmlCreateURLParserCtxt(filename, options); |
15076 | 0 | if (ctxt == NULL) |
15077 | 0 | return (NULL); |
15078 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 0)); |
15079 | 0 | } |
15080 | | |
15081 | | /** |
15082 | | * xmlReadMemory: |
15083 | | * @buffer: a pointer to a char array |
15084 | | * @size: the size of the array |
15085 | | * @URL: the base URL to use for the document |
15086 | | * @encoding: the document encoding, or NULL |
15087 | | * @options: a combination of xmlParserOption |
15088 | | * |
15089 | | * parse an XML in-memory document and build a tree. |
15090 | | * |
15091 | | * Returns the resulting document tree |
15092 | | */ |
15093 | | xmlDocPtr |
15094 | | xmlReadMemory(const char *buffer, int size, const char *URL, const char *encoding, int options) |
15095 | 153k | { |
15096 | 153k | xmlParserCtxtPtr ctxt; |
15097 | | |
15098 | 153k | xmlInitParser(); |
15099 | 153k | ctxt = xmlCreateMemoryParserCtxt(buffer, size); |
15100 | 153k | if (ctxt == NULL) |
15101 | 20 | return (NULL); |
15102 | 153k | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15103 | 153k | } |
15104 | | |
15105 | | /** |
15106 | | * xmlReadFd: |
15107 | | * @fd: an open file descriptor |
15108 | | * @URL: the base URL to use for the document |
15109 | | * @encoding: the document encoding, or NULL |
15110 | | * @options: a combination of xmlParserOption |
15111 | | * |
15112 | | * parse an XML from a file descriptor and build a tree. |
15113 | | * NOTE that the file descriptor will not be closed when the |
15114 | | * reader is closed or reset. |
15115 | | * |
15116 | | * Returns the resulting document tree |
15117 | | */ |
15118 | | xmlDocPtr |
15119 | | xmlReadFd(int fd, const char *URL, const char *encoding, int options) |
15120 | 0 | { |
15121 | 0 | xmlParserCtxtPtr ctxt; |
15122 | 0 | xmlParserInputBufferPtr input; |
15123 | 0 | xmlParserInputPtr stream; |
15124 | |
|
15125 | 0 | if (fd < 0) |
15126 | 0 | return (NULL); |
15127 | 0 | xmlInitParser(); |
15128 | |
|
15129 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15130 | 0 | if (input == NULL) |
15131 | 0 | return (NULL); |
15132 | 0 | input->closecallback = NULL; |
15133 | 0 | ctxt = xmlNewParserCtxt(); |
15134 | 0 | if (ctxt == NULL) { |
15135 | 0 | xmlFreeParserInputBuffer(input); |
15136 | 0 | return (NULL); |
15137 | 0 | } |
15138 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15139 | 0 | if (stream == NULL) { |
15140 | 0 | xmlFreeParserInputBuffer(input); |
15141 | 0 | xmlFreeParserCtxt(ctxt); |
15142 | 0 | return (NULL); |
15143 | 0 | } |
15144 | 0 | inputPush(ctxt, stream); |
15145 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15146 | 0 | } |
15147 | | |
15148 | | /** |
15149 | | * xmlReadIO: |
15150 | | * @ioread: an I/O read function |
15151 | | * @ioclose: an I/O close function |
15152 | | * @ioctx: an I/O handler |
15153 | | * @URL: the base URL to use for the document |
15154 | | * @encoding: the document encoding, or NULL |
15155 | | * @options: a combination of xmlParserOption |
15156 | | * |
15157 | | * parse an XML document from I/O functions and source and build a tree. |
15158 | | * |
15159 | | * Returns the resulting document tree |
15160 | | */ |
15161 | | xmlDocPtr |
15162 | | xmlReadIO(xmlInputReadCallback ioread, xmlInputCloseCallback ioclose, |
15163 | | void *ioctx, const char *URL, const char *encoding, int options) |
15164 | 0 | { |
15165 | 0 | xmlParserCtxtPtr ctxt; |
15166 | 0 | xmlParserInputBufferPtr input; |
15167 | 0 | xmlParserInputPtr stream; |
15168 | |
|
15169 | 0 | if (ioread == NULL) |
15170 | 0 | return (NULL); |
15171 | 0 | xmlInitParser(); |
15172 | |
|
15173 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15174 | 0 | XML_CHAR_ENCODING_NONE); |
15175 | 0 | if (input == NULL) { |
15176 | 0 | if (ioclose != NULL) |
15177 | 0 | ioclose(ioctx); |
15178 | 0 | return (NULL); |
15179 | 0 | } |
15180 | 0 | ctxt = xmlNewParserCtxt(); |
15181 | 0 | if (ctxt == NULL) { |
15182 | 0 | xmlFreeParserInputBuffer(input); |
15183 | 0 | return (NULL); |
15184 | 0 | } |
15185 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15186 | 0 | if (stream == NULL) { |
15187 | 0 | xmlFreeParserInputBuffer(input); |
15188 | 0 | xmlFreeParserCtxt(ctxt); |
15189 | 0 | return (NULL); |
15190 | 0 | } |
15191 | 0 | inputPush(ctxt, stream); |
15192 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 0)); |
15193 | 0 | } |
15194 | | |
15195 | | /** |
15196 | | * xmlCtxtReadDoc: |
15197 | | * @ctxt: an XML parser context |
15198 | | * @cur: a pointer to a zero terminated string |
15199 | | * @URL: the base URL to use for the document |
15200 | | * @encoding: the document encoding, or NULL |
15201 | | * @options: a combination of xmlParserOption |
15202 | | * |
15203 | | * parse an XML in-memory document and build a tree. |
15204 | | * This reuses the existing @ctxt parser context |
15205 | | * |
15206 | | * Returns the resulting document tree |
15207 | | */ |
15208 | | xmlDocPtr |
15209 | | xmlCtxtReadDoc(xmlParserCtxtPtr ctxt, const xmlChar * cur, |
15210 | | const char *URL, const char *encoding, int options) |
15211 | 0 | { |
15212 | 0 | if (cur == NULL) |
15213 | 0 | return (NULL); |
15214 | 0 | return (xmlCtxtReadMemory(ctxt, (const char *) cur, xmlStrlen(cur), URL, |
15215 | 0 | encoding, options)); |
15216 | 0 | } |
15217 | | |
15218 | | /** |
15219 | | * xmlCtxtReadFile: |
15220 | | * @ctxt: an XML parser context |
15221 | | * @filename: a file or URL |
15222 | | * @encoding: the document encoding, or NULL |
15223 | | * @options: a combination of xmlParserOption |
15224 | | * |
15225 | | * parse an XML file from the filesystem or the network. |
15226 | | * This reuses the existing @ctxt parser context |
15227 | | * |
15228 | | * Returns the resulting document tree |
15229 | | */ |
15230 | | xmlDocPtr |
15231 | | xmlCtxtReadFile(xmlParserCtxtPtr ctxt, const char *filename, |
15232 | | const char *encoding, int options) |
15233 | 0 | { |
15234 | 0 | xmlParserInputPtr stream; |
15235 | |
|
15236 | 0 | if (filename == NULL) |
15237 | 0 | return (NULL); |
15238 | 0 | if (ctxt == NULL) |
15239 | 0 | return (NULL); |
15240 | 0 | xmlInitParser(); |
15241 | |
|
15242 | 0 | xmlCtxtReset(ctxt); |
15243 | |
|
15244 | 0 | stream = xmlLoadExternalEntity(filename, NULL, ctxt); |
15245 | 0 | if (stream == NULL) { |
15246 | 0 | return (NULL); |
15247 | 0 | } |
15248 | 0 | inputPush(ctxt, stream); |
15249 | 0 | return (xmlDoRead(ctxt, NULL, encoding, options, 1)); |
15250 | 0 | } |
15251 | | |
15252 | | /** |
15253 | | * xmlCtxtReadMemory: |
15254 | | * @ctxt: an XML parser context |
15255 | | * @buffer: a pointer to a char array |
15256 | | * @size: the size of the array |
15257 | | * @URL: the base URL to use for the document |
15258 | | * @encoding: the document encoding, or NULL |
15259 | | * @options: a combination of xmlParserOption |
15260 | | * |
15261 | | * parse an XML in-memory document and build a tree. |
15262 | | * This reuses the existing @ctxt parser context |
15263 | | * |
15264 | | * Returns the resulting document tree |
15265 | | */ |
15266 | | xmlDocPtr |
15267 | | xmlCtxtReadMemory(xmlParserCtxtPtr ctxt, const char *buffer, int size, |
15268 | | const char *URL, const char *encoding, int options) |
15269 | 0 | { |
15270 | 0 | xmlParserInputBufferPtr input; |
15271 | 0 | xmlParserInputPtr stream; |
15272 | |
|
15273 | 0 | if (ctxt == NULL) |
15274 | 0 | return (NULL); |
15275 | 0 | if (buffer == NULL) |
15276 | 0 | return (NULL); |
15277 | 0 | xmlInitParser(); |
15278 | |
|
15279 | 0 | xmlCtxtReset(ctxt); |
15280 | |
|
15281 | 0 | input = xmlParserInputBufferCreateMem(buffer, size, XML_CHAR_ENCODING_NONE); |
15282 | 0 | if (input == NULL) { |
15283 | 0 | return(NULL); |
15284 | 0 | } |
15285 | | |
15286 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15287 | 0 | if (stream == NULL) { |
15288 | 0 | xmlFreeParserInputBuffer(input); |
15289 | 0 | return(NULL); |
15290 | 0 | } |
15291 | | |
15292 | 0 | inputPush(ctxt, stream); |
15293 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15294 | 0 | } |
15295 | | |
15296 | | /** |
15297 | | * xmlCtxtReadFd: |
15298 | | * @ctxt: an XML parser context |
15299 | | * @fd: an open file descriptor |
15300 | | * @URL: the base URL to use for the document |
15301 | | * @encoding: the document encoding, or NULL |
15302 | | * @options: a combination of xmlParserOption |
15303 | | * |
15304 | | * parse an XML from a file descriptor and build a tree. |
15305 | | * This reuses the existing @ctxt parser context |
15306 | | * NOTE that the file descriptor will not be closed when the |
15307 | | * reader is closed or reset. |
15308 | | * |
15309 | | * Returns the resulting document tree |
15310 | | */ |
15311 | | xmlDocPtr |
15312 | | xmlCtxtReadFd(xmlParserCtxtPtr ctxt, int fd, |
15313 | | const char *URL, const char *encoding, int options) |
15314 | 0 | { |
15315 | 0 | xmlParserInputBufferPtr input; |
15316 | 0 | xmlParserInputPtr stream; |
15317 | |
|
15318 | 0 | if (fd < 0) |
15319 | 0 | return (NULL); |
15320 | 0 | if (ctxt == NULL) |
15321 | 0 | return (NULL); |
15322 | 0 | xmlInitParser(); |
15323 | |
|
15324 | 0 | xmlCtxtReset(ctxt); |
15325 | | |
15326 | |
|
15327 | 0 | input = xmlParserInputBufferCreateFd(fd, XML_CHAR_ENCODING_NONE); |
15328 | 0 | if (input == NULL) |
15329 | 0 | return (NULL); |
15330 | 0 | input->closecallback = NULL; |
15331 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15332 | 0 | if (stream == NULL) { |
15333 | 0 | xmlFreeParserInputBuffer(input); |
15334 | 0 | return (NULL); |
15335 | 0 | } |
15336 | 0 | inputPush(ctxt, stream); |
15337 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15338 | 0 | } |
15339 | | |
15340 | | /** |
15341 | | * xmlCtxtReadIO: |
15342 | | * @ctxt: an XML parser context |
15343 | | * @ioread: an I/O read function |
15344 | | * @ioclose: an I/O close function |
15345 | | * @ioctx: an I/O handler |
15346 | | * @URL: the base URL to use for the document |
15347 | | * @encoding: the document encoding, or NULL |
15348 | | * @options: a combination of xmlParserOption |
15349 | | * |
15350 | | * parse an XML document from I/O functions and source and build a tree. |
15351 | | * This reuses the existing @ctxt parser context |
15352 | | * |
15353 | | * Returns the resulting document tree |
15354 | | */ |
15355 | | xmlDocPtr |
15356 | | xmlCtxtReadIO(xmlParserCtxtPtr ctxt, xmlInputReadCallback ioread, |
15357 | | xmlInputCloseCallback ioclose, void *ioctx, |
15358 | | const char *URL, |
15359 | | const char *encoding, int options) |
15360 | 0 | { |
15361 | 0 | xmlParserInputBufferPtr input; |
15362 | 0 | xmlParserInputPtr stream; |
15363 | |
|
15364 | 0 | if (ioread == NULL) |
15365 | 0 | return (NULL); |
15366 | 0 | if (ctxt == NULL) |
15367 | 0 | return (NULL); |
15368 | 0 | xmlInitParser(); |
15369 | |
|
15370 | 0 | xmlCtxtReset(ctxt); |
15371 | |
|
15372 | 0 | input = xmlParserInputBufferCreateIO(ioread, ioclose, ioctx, |
15373 | 0 | XML_CHAR_ENCODING_NONE); |
15374 | 0 | if (input == NULL) { |
15375 | 0 | if (ioclose != NULL) |
15376 | 0 | ioclose(ioctx); |
15377 | 0 | return (NULL); |
15378 | 0 | } |
15379 | 0 | stream = xmlNewIOInputStream(ctxt, input, XML_CHAR_ENCODING_NONE); |
15380 | 0 | if (stream == NULL) { |
15381 | 0 | xmlFreeParserInputBuffer(input); |
15382 | 0 | return (NULL); |
15383 | 0 | } |
15384 | 0 | inputPush(ctxt, stream); |
15385 | 0 | return (xmlDoRead(ctxt, URL, encoding, options, 1)); |
15386 | 0 | } |
15387 | | |