huwcampbell.com
Reactos
1/*
2 * Summary: internals routines and limits exported by the parser.
3 * Description: this module exports a number of internal parsing routines
4 * they are not really all intended for applications but
5 * can prove useful doing low level processing.
6 *
7 * Copy: See Copyright for the status of this software.
8 *
9 * Author: Daniel Veillard
10 */
11
12#ifndef __XML_PARSER_INTERNALS_H__
13#define __XML_PARSER_INTERNALS_H__
14
15#include <libxml/xmlversion.h>
16#include <libxml/parser.h>
17#include <libxml/HTMLparser.h>
18#include <libxml/chvalid.h>
19#include <libxml/SAX2.h>
20
21#ifdef __cplusplus
22extern "C" {
23#endif
24
25/**
26 * xmlParserMaxDepth:
27 *
28 * arbitrary depth limit for the XML documents that we allow to
29 * process. This is not a limitation of the parser but a safety
30 * boundary feature, use XML_PARSE_HUGE option to override it.
31 */
32XMLPUBVAR unsigned int xmlParserMaxDepth;
33
34/**
35 * XML_MAX_TEXT_LENGTH:
36 *
37 * Maximum size allowed for a single text node when building a tree.
38 * This is not a limitation of the parser but a safety boundary feature,
39 * use XML_PARSE_HUGE option to override it.
40 * Introduced in 2.9.0
41 */
42#define XML_MAX_TEXT_LENGTH 10000000
43
44/**
45 * XML_MAX_HUGE_LENGTH:
46 *
47 * Maximum size allowed when XML_PARSE_HUGE is set.
48 */
49#define XML_MAX_HUGE_LENGTH 1000000000
50
51/**
52 * XML_MAX_NAME_LENGTH:
53 *
54 * Maximum size allowed for a markup identifier.
55 * This is not a limitation of the parser but a safety boundary feature,
56 * use XML_PARSE_HUGE option to override it.
57 * Note that with the use of parsing dictionaries overriding the limit
58 * may result in more runtime memory usage in face of "unfriendly' content
59 * Introduced in 2.9.0
60 */
61#define XML_MAX_NAME_LENGTH 50000
62
63/**
64 * XML_MAX_DICTIONARY_LIMIT:
65 *
66 * Maximum size allowed by the parser for a dictionary by default
67 * This is not a limitation of the parser but a safety boundary feature,
68 * use XML_PARSE_HUGE option to override it.
69 * Introduced in 2.9.0
70 */
71#define XML_MAX_DICTIONARY_LIMIT 10000000
72
73/**
74 * XML_MAX_LOOKUP_LIMIT:
75 *
76 * Maximum size allowed by the parser for ahead lookup
77 * This is an upper boundary enforced by the parser to avoid bad
78 * behaviour on "unfriendly' content
79 * Introduced in 2.9.0
80 */
81#define XML_MAX_LOOKUP_LIMIT 10000000
82
83/**
84 * XML_MAX_NAMELEN:
85 *
86 * Identifiers can be longer, but this will be more costly
87 * at runtime.
88 */
89#define XML_MAX_NAMELEN 100
90
91/**
92 * INPUT_CHUNK:
93 *
94 * The parser tries to always have that amount of input ready.
95 * One of the point is providing context when reporting errors.
96 */
97#define INPUT_CHUNK 250
98
99/************************************************************************
100 * *
101 * UNICODE version of the macros. *
102 * *
103 ************************************************************************/
104/**
105 * IS_BYTE_CHAR:
106 * @c: an byte value (int)
107 *
108 * Macro to check the following production in the XML spec:
109 *
110 * [2] Char ::= #x9 | #xA | #xD | [#x20...]
111 * any byte character in the accepted range
112 */
113#define IS_BYTE_CHAR(c) xmlIsChar_ch(c)
114
115/**
116 * IS_CHAR:
117 * @c: an UNICODE value (int)
118 *
119 * Macro to check the following production in the XML spec:
120 *
121 * [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
122 * | [#x10000-#x10FFFF]
123 * any Unicode character, excluding the surrogate blocks, FFFE, and FFFF.
124 */
125#define IS_CHAR(c) xmlIsCharQ(c)
126
127/**
128 * IS_CHAR_CH:
129 * @c: an xmlChar (usually an unsigned char)
130 *
131 * Behaves like IS_CHAR on single-byte value
132 */
133#define IS_CHAR_CH(c) xmlIsChar_ch(c)
134
135/**
136 * IS_BLANK:
137 * @c: an UNICODE value (int)
138 *
139 * Macro to check the following production in the XML spec:
140 *
141 * [3] S ::= (#x20 | #x9 | #xD | #xA)+
142 */
143#define IS_BLANK(c) xmlIsBlankQ(c)
144
145/**
146 * IS_BLANK_CH:
147 * @c: an xmlChar value (normally unsigned char)
148 *
149 * Behaviour same as IS_BLANK
150 */
151#define IS_BLANK_CH(c) xmlIsBlank_ch(c)
152
153/**
154 * IS_BASECHAR:
155 * @c: an UNICODE value (int)
156 *
157 * Macro to check the following production in the XML spec:
158 *
159 * [85] BaseChar ::= ... long list see REC ...
160 */
161#define IS_BASECHAR(c) xmlIsBaseCharQ(c)
162
163/**
164 * IS_DIGIT:
165 * @c: an UNICODE value (int)
166 *
167 * Macro to check the following production in the XML spec:
168 *
169 * [88] Digit ::= ... long list see REC ...
170 */
171#define IS_DIGIT(c) xmlIsDigitQ(c)
172
173/**
174 * IS_DIGIT_CH:
175 * @c: an xmlChar value (usually an unsigned char)
176 *
177 * Behaves like IS_DIGIT but with a single byte argument
178 */
179#define IS_DIGIT_CH(c) xmlIsDigit_ch(c)
180
181/**
182 * IS_COMBINING:
183 * @c: an UNICODE value (int)
184 *
185 * Macro to check the following production in the XML spec:
186 *
187 * [87] CombiningChar ::= ... long list see REC ...
188 */
189#define IS_COMBINING(c) xmlIsCombiningQ(c)
190
191/**
192 * IS_COMBINING_CH:
193 * @c: an xmlChar (usually an unsigned char)
194 *
195 * Always false (all combining chars > 0xff)
196 */
197#define IS_COMBINING_CH(c) 0
198
199/**
200 * IS_EXTENDER:
201 * @c: an UNICODE value (int)
202 *
203 * Macro to check the following production in the XML spec:
204 *
205 *
206 * [89] Extender ::= #x00B7 | #x02D0 | #x02D1 | #x0387 | #x0640 |
207 * #x0E46 | #x0EC6 | #x3005 | [#x3031-#x3035] |
208 * [#x309D-#x309E] | [#x30FC-#x30FE]
209 */
210#define IS_EXTENDER(c) xmlIsExtenderQ(c)
211
212/**
213 * IS_EXTENDER_CH:
214 * @c: an xmlChar value (usually an unsigned char)
215 *
216 * Behaves like IS_EXTENDER but with a single-byte argument
217 */
218#define IS_EXTENDER_CH(c) xmlIsExtender_ch(c)
219
220/**
221 * IS_IDEOGRAPHIC:
222 * @c: an UNICODE value (int)
223 *
224 * Macro to check the following production in the XML spec:
225 *
226 *
227 * [86] Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
228 */
229#define IS_IDEOGRAPHIC(c) xmlIsIdeographicQ(c)
230
231/**
232 * IS_LETTER:
233 * @c: an UNICODE value (int)
234 *
235 * Macro to check the following production in the XML spec:
236 *
237 *
238 * [84] Letter ::= BaseChar | Ideographic
239 */
240#define IS_LETTER(c) (IS_BASECHAR(c) || IS_IDEOGRAPHIC(c))
241
242/**
243 * IS_LETTER_CH:
244 * @c: an xmlChar value (normally unsigned char)
245 *
246 * Macro behaves like IS_LETTER, but only check base chars
247 *
248 */
249#define IS_LETTER_CH(c) xmlIsBaseChar_ch(c)
250
251/**
252 * IS_ASCII_LETTER:
253 * @c: an xmlChar value
254 *
255 * Macro to check [a-zA-Z]
256 *
257 */
258#define IS_ASCII_LETTER(c) (((0x41 <= (c)) && ((c) <= 0x5a)) || \
259 ((0x61 <= (c)) && ((c) <= 0x7a)))
260
261/**
262 * IS_ASCII_DIGIT:
263 * @c: an xmlChar value
264 *
265 * Macro to check [0-9]
266 *
267 */
268#define IS_ASCII_DIGIT(c) ((0x30 <= (c)) && ((c) <= 0x39))
269
270/**
271 * IS_PUBIDCHAR:
272 * @c: an UNICODE value (int)
273 *
274 * Macro to check the following production in the XML spec:
275 *
276 *
277 * [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%]
278 */
279#define IS_PUBIDCHAR(c) xmlIsPubidCharQ(c)
280
281/**
282 * IS_PUBIDCHAR_CH:
283 * @c: an xmlChar value (normally unsigned char)
284 *
285 * Same as IS_PUBIDCHAR but for single-byte value
286 */
287#define IS_PUBIDCHAR_CH(c) xmlIsPubidChar_ch(c)
288
289/**
290 * Global variables used for predefined strings.
291 */
292XMLPUBVAR const xmlChar xmlStringText[];
293XMLPUBVAR const xmlChar xmlStringTextNoenc[];
294XMLPUBVAR const xmlChar xmlStringComment[];
295
296/*
297 * Function to finish the work of the macros where needed.
298 */
299XMLPUBFUN int xmlIsLetter (int c);
300
301/**
302 * Parser context.
303 */
304XMLPUBFUN xmlParserCtxtPtr
305 xmlCreateFileParserCtxt (const char *filename);
306XMLPUBFUN xmlParserCtxtPtr
307 xmlCreateURLParserCtxt (const char *filename,
308 int options);
309XMLPUBFUN xmlParserCtxtPtr
310 xmlCreateMemoryParserCtxt(const char *buffer,
311 int size);
312XMLPUBFUN xmlParserCtxtPtr
313 xmlCreateEntityParserCtxt(const xmlChar *URL,
314 const xmlChar *ID,
315 const xmlChar *base);
316XMLPUBFUN int
317 xmlSwitchEncoding (xmlParserCtxtPtr ctxt,
318 xmlCharEncoding enc);
319XMLPUBFUN int
320 xmlSwitchToEncoding (xmlParserCtxtPtr ctxt,
321 xmlCharEncodingHandlerPtr handler);
322XML_DEPRECATED
323XMLPUBFUN int
324 xmlSwitchInputEncoding (xmlParserCtxtPtr ctxt,
325 xmlParserInputPtr input,
326 xmlCharEncodingHandlerPtr handler);
327
328/**
329 * Input Streams.
330 */
331XMLPUBFUN xmlParserInputPtr
332 xmlNewStringInputStream (xmlParserCtxtPtr ctxt,
333 const xmlChar *buffer);
334XML_DEPRECATED
335XMLPUBFUN xmlParserInputPtr
336 xmlNewEntityInputStream (xmlParserCtxtPtr ctxt,
337 xmlEntityPtr entity);
338XMLPUBFUN int
339 xmlPushInput (xmlParserCtxtPtr ctxt,
340 xmlParserInputPtr input);
341XMLPUBFUN xmlChar
342 xmlPopInput (xmlParserCtxtPtr ctxt);
343XMLPUBFUN void
344 xmlFreeInputStream (xmlParserInputPtr input);
345XMLPUBFUN xmlParserInputPtr
346 xmlNewInputFromFile (xmlParserCtxtPtr ctxt,
347 const char *filename);
348XMLPUBFUN xmlParserInputPtr
349 xmlNewInputStream (xmlParserCtxtPtr ctxt);
350
351/**
352 * Namespaces.
353 */
354XMLPUBFUN xmlChar *
355 xmlSplitQName (xmlParserCtxtPtr ctxt,
356 const xmlChar *name,
357 xmlChar **prefix);
358
359/**
360 * Generic production rules.
361 */
362XML_DEPRECATED
363XMLPUBFUN const xmlChar *
364 xmlParseName (xmlParserCtxtPtr ctxt);
365XML_DEPRECATED
366XMLPUBFUN xmlChar *
367 xmlParseNmtoken (xmlParserCtxtPtr ctxt);
368XML_DEPRECATED
369XMLPUBFUN xmlChar *
370 xmlParseEntityValue (xmlParserCtxtPtr ctxt,
371 xmlChar **orig);
372XML_DEPRECATED
373XMLPUBFUN xmlChar *
374 xmlParseAttValue (xmlParserCtxtPtr ctxt);
375XML_DEPRECATED
376XMLPUBFUN xmlChar *
377 xmlParseSystemLiteral (xmlParserCtxtPtr ctxt);
378XML_DEPRECATED
379XMLPUBFUN xmlChar *
380 xmlParsePubidLiteral (xmlParserCtxtPtr ctxt);
381XML_DEPRECATED
382XMLPUBFUN void
383 xmlParseCharData (xmlParserCtxtPtr ctxt,
384 int cdata);
385XML_DEPRECATED
386XMLPUBFUN xmlChar *
387 xmlParseExternalID (xmlParserCtxtPtr ctxt,
388 xmlChar **publicID,
389 int strict);
390XML_DEPRECATED
391XMLPUBFUN void
392 xmlParseComment (xmlParserCtxtPtr ctxt);
393XML_DEPRECATED
394XMLPUBFUN const xmlChar *
395 xmlParsePITarget (xmlParserCtxtPtr ctxt);
396XML_DEPRECATED
397XMLPUBFUN void
398 xmlParsePI (xmlParserCtxtPtr ctxt);
399XML_DEPRECATED
400XMLPUBFUN void
401 xmlParseNotationDecl (xmlParserCtxtPtr ctxt);
402XML_DEPRECATED
403XMLPUBFUN void
404 xmlParseEntityDecl (xmlParserCtxtPtr ctxt);
405XML_DEPRECATED
406XMLPUBFUN int
407 xmlParseDefaultDecl (xmlParserCtxtPtr ctxt,
408 xmlChar **value);
409XML_DEPRECATED
410XMLPUBFUN xmlEnumerationPtr
411 xmlParseNotationType (xmlParserCtxtPtr ctxt);
412XML_DEPRECATED
413XMLPUBFUN xmlEnumerationPtr
414 xmlParseEnumerationType (xmlParserCtxtPtr ctxt);
415XML_DEPRECATED
416XMLPUBFUN int
417 xmlParseEnumeratedType (xmlParserCtxtPtr ctxt,
418 xmlEnumerationPtr *tree);
419XML_DEPRECATED
420XMLPUBFUN int
421 xmlParseAttributeType (xmlParserCtxtPtr ctxt,
422 xmlEnumerationPtr *tree);
423XML_DEPRECATED
424XMLPUBFUN void
425 xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt);
426XML_DEPRECATED
427XMLPUBFUN xmlElementContentPtr
428 xmlParseElementMixedContentDecl
429 (xmlParserCtxtPtr ctxt,
430 int inputchk);
431XML_DEPRECATED
432XMLPUBFUN xmlElementContentPtr
433 xmlParseElementChildrenContentDecl
434 (xmlParserCtxtPtr ctxt,
435 int inputchk);
436XML_DEPRECATED
437XMLPUBFUN int
438 xmlParseElementContentDecl(xmlParserCtxtPtr ctxt,
439 const xmlChar *name,
440 xmlElementContentPtr *result);
441XML_DEPRECATED
442XMLPUBFUN int
443 xmlParseElementDecl (xmlParserCtxtPtr ctxt);
444XML_DEPRECATED
445XMLPUBFUN void
446 xmlParseMarkupDecl (xmlParserCtxtPtr ctxt);
447XML_DEPRECATED
448XMLPUBFUN int
449 xmlParseCharRef (xmlParserCtxtPtr ctxt);
450XML_DEPRECATED
451XMLPUBFUN xmlEntityPtr
452 xmlParseEntityRef (xmlParserCtxtPtr ctxt);
453XML_DEPRECATED
454XMLPUBFUN void
455 xmlParseReference (xmlParserCtxtPtr ctxt);
456XML_DEPRECATED
457XMLPUBFUN void
458 xmlParsePEReference (xmlParserCtxtPtr ctxt);
459XML_DEPRECATED
460XMLPUBFUN void
461 xmlParseDocTypeDecl (xmlParserCtxtPtr ctxt);
462#ifdef LIBXML_SAX1_ENABLED
463XML_DEPRECATED
464XMLPUBFUN const xmlChar *
465 xmlParseAttribute (xmlParserCtxtPtr ctxt,
466 xmlChar **value);
467XML_DEPRECATED
468XMLPUBFUN const xmlChar *
469 xmlParseStartTag (xmlParserCtxtPtr ctxt);
470XML_DEPRECATED
471XMLPUBFUN void
472 xmlParseEndTag (xmlParserCtxtPtr ctxt);
473#endif /* LIBXML_SAX1_ENABLED */
474XML_DEPRECATED
475XMLPUBFUN void
476 xmlParseCDSect (xmlParserCtxtPtr ctxt);
477XMLPUBFUN void
478 xmlParseContent (xmlParserCtxtPtr ctxt);
479XML_DEPRECATED
480XMLPUBFUN void
481 xmlParseElement (xmlParserCtxtPtr ctxt);
482XML_DEPRECATED
483XMLPUBFUN xmlChar *
484 xmlParseVersionNum (xmlParserCtxtPtr ctxt);
485XML_DEPRECATED
486XMLPUBFUN xmlChar *
487 xmlParseVersionInfo (xmlParserCtxtPtr ctxt);
488XML_DEPRECATED
489XMLPUBFUN xmlChar *
490 xmlParseEncName (xmlParserCtxtPtr ctxt);
491XML_DEPRECATED
492XMLPUBFUN const xmlChar *
493 xmlParseEncodingDecl (xmlParserCtxtPtr ctxt);
494XML_DEPRECATED
495XMLPUBFUN int
496 xmlParseSDDecl (xmlParserCtxtPtr ctxt);
497XML_DEPRECATED
498XMLPUBFUN void
499 xmlParseXMLDecl (xmlParserCtxtPtr ctxt);
500XML_DEPRECATED
501XMLPUBFUN void
502 xmlParseTextDecl (xmlParserCtxtPtr ctxt);
503XML_DEPRECATED
504XMLPUBFUN void
505 xmlParseMisc (xmlParserCtxtPtr ctxt);
506XMLPUBFUN void
507 xmlParseExternalSubset (xmlParserCtxtPtr ctxt,
508 const xmlChar *ExternalID,
509 const xmlChar *SystemID);
510/**
511 * XML_SUBSTITUTE_NONE:
512 *
513 * If no entities need to be substituted.
514 */
515#define XML_SUBSTITUTE_NONE 0
516/**
517 * XML_SUBSTITUTE_REF:
518 *
519 * Whether general entities need to be substituted.
520 */
521#define XML_SUBSTITUTE_REF 1
522/**
523 * XML_SUBSTITUTE_PEREF:
524 *
525 * Whether parameter entities need to be substituted.
526 */
527#define XML_SUBSTITUTE_PEREF 2
528/**
529 * XML_SUBSTITUTE_BOTH:
530 *
531 * Both general and parameter entities need to be substituted.
532 */
533#define XML_SUBSTITUTE_BOTH 3
534
535XML_DEPRECATED
536XMLPUBFUN xmlChar *
537 xmlStringDecodeEntities (xmlParserCtxtPtr ctxt,
538 const xmlChar *str,
539 int what,
540 xmlChar end,
541 xmlChar end2,
542 xmlChar end3);
543XML_DEPRECATED
544XMLPUBFUN xmlChar *
545 xmlStringLenDecodeEntities (xmlParserCtxtPtr ctxt,
546 const xmlChar *str,
547 int len,
548 int what,
549 xmlChar end,
550 xmlChar end2,
551 xmlChar end3);
552
553/*
554 * Generated by MACROS on top of parser.c c.f. PUSH_AND_POP.
555 */
556XML_DEPRECATED
557XMLPUBFUN int nodePush (xmlParserCtxtPtr ctxt,
558 xmlNodePtr value);
559XML_DEPRECATED
560XMLPUBFUN xmlNodePtr nodePop (xmlParserCtxtPtr ctxt);
561XMLPUBFUN int inputPush (xmlParserCtxtPtr ctxt,
562 xmlParserInputPtr value);
563XMLPUBFUN xmlParserInputPtr inputPop (xmlParserCtxtPtr ctxt);
564XML_DEPRECATED
565XMLPUBFUN const xmlChar * namePop (xmlParserCtxtPtr ctxt);
566XML_DEPRECATED
567XMLPUBFUN int namePush (xmlParserCtxtPtr ctxt,
568 const xmlChar *value);
569
570/*
571 * other commodities shared between parser.c and parserInternals.
572 */
573XML_DEPRECATED
574XMLPUBFUN int xmlSkipBlankChars (xmlParserCtxtPtr ctxt);
575XML_DEPRECATED
576XMLPUBFUN int xmlStringCurrentChar (xmlParserCtxtPtr ctxt,
577 const xmlChar *cur,
578 int *len);
579XML_DEPRECATED
580XMLPUBFUN void xmlParserHandlePEReference(xmlParserCtxtPtr ctxt);
581XML_DEPRECATED
582XMLPUBFUN int xmlCheckLanguageID (const xmlChar *lang);
583
584/*
585 * Really core function shared with HTML parser.
586 */
587XML_DEPRECATED
588XMLPUBFUN int xmlCurrentChar (xmlParserCtxtPtr ctxt,
589 int *len);
590XMLPUBFUN int xmlCopyCharMultiByte (xmlChar *out,
591 int val);
592XMLPUBFUN int xmlCopyChar (int len,
593 xmlChar *out,
594 int val);
595XML_DEPRECATED
596XMLPUBFUN void xmlNextChar (xmlParserCtxtPtr ctxt);
597XML_DEPRECATED
598XMLPUBFUN void xmlParserInputShrink (xmlParserInputPtr in);
599
600/*
601 * Specific function to keep track of entities references
602 * and used by the XSLT debugger.
603 */
604#ifdef LIBXML_LEGACY_ENABLED
605/**
606 * xmlEntityReferenceFunc:
607 * @ent: the entity
608 * @firstNode: the fist node in the chunk
609 * @lastNode: the last nod in the chunk
610 *
611 * Callback function used when one needs to be able to track back the
612 * provenance of a chunk of nodes inherited from an entity replacement.
613 */
614typedef void (*xmlEntityReferenceFunc) (xmlEntityPtr ent,
615 xmlNodePtr firstNode,
616 xmlNodePtr lastNode);
617
618XML_DEPRECATED
619XMLPUBFUN void xmlSetEntityReferenceFunc (xmlEntityReferenceFunc func);
620
621XML_DEPRECATED
622XMLPUBFUN xmlChar *
623 xmlParseQuotedString (xmlParserCtxtPtr ctxt);
624XML_DEPRECATED
625XMLPUBFUN void
626 xmlParseNamespace (xmlParserCtxtPtr ctxt);
627XML_DEPRECATED
628XMLPUBFUN xmlChar *
629 xmlNamespaceParseNSDef (xmlParserCtxtPtr ctxt);
630XML_DEPRECATED
631XMLPUBFUN xmlChar *
632 xmlScanName (xmlParserCtxtPtr ctxt);
633XML_DEPRECATED
634XMLPUBFUN xmlChar *
635 xmlNamespaceParseNCName (xmlParserCtxtPtr ctxt);
636XML_DEPRECATED
637XMLPUBFUN void xmlParserHandleReference(xmlParserCtxtPtr ctxt);
638XML_DEPRECATED
639XMLPUBFUN xmlChar *
640 xmlNamespaceParseQName (xmlParserCtxtPtr ctxt,
641 xmlChar **prefix);
642/**
643 * Entities
644 */
645XML_DEPRECATED
646XMLPUBFUN xmlChar *
647 xmlDecodeEntities (xmlParserCtxtPtr ctxt,
648 int len,
649 int what,
650 xmlChar end,
651 xmlChar end2,
652 xmlChar end3);
653XML_DEPRECATED
654XMLPUBFUN void
655 xmlHandleEntity (xmlParserCtxtPtr ctxt,
656 xmlEntityPtr entity);
657
658#endif /* LIBXML_LEGACY_ENABLED */
659
660#ifdef __cplusplus
661}
662#endif
663#endif /* __XML_PARSER_INTERNALS_H__ */