1 /*
2 * Summary: internals routines 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
20 #ifdef __cplusplus
21 extern "C" {
22 #endif
23
24 /**
25 * xmlParserMaxDepth:
26 *
27 * arbitrary depth limit for the XML documents that we allow to
28 * process. This is not a limitation of the parser but a safety
29 * boundary feature.
30 */
31 XMLPUBVAR unsigned int xmlParserMaxDepth;
32
33 /**
34 * XML_MAX_NAMELEN:
35 *
36 * Identifiers can be longer, but this will be more costly
37 * at runtime.
38 */
39 #define XML_MAX_NAMELEN 100
40
41 /**
42 * INPUT_CHUNK:
43 *
44 * The parser tries to always have that amount of input ready.
45 * One of the point is providing context when reporting errors.
46 */
47 #define INPUT_CHUNK 250
48
49 /************************************************************************
50 * *
51 * UNICODE version of the macros. *
52 * *
53 ************************************************************************/
54 /**
55 * IS_BYTE_CHAR:
56 * @c: an byte value (int)
57 *
58 * Macro to check the following production in the XML spec:
59 *
60 * [2] Char ::= #x9 | #xA | #xD | [#x20...]
61 * any byte character in the accepted range
62 */
63 #define IS_BYTE_CHAR(c) xmlIsChar_ch(c)
64
65 /**
66 * IS_CHAR:
67 * @c: an UNICODE value (int)
68 *
69 * Macro to check the following production in the XML spec:
70 *
71 * [2] Char ::= #x9 | #xA | #xD | [#x20-#xD7FF] | [#xE000-#xFFFD]
72 * | [#x10000-#x10FFFF]
73 * any Unicode character, excluding the surrogate blocks, FFFE, and FFFF.
74 */
75 #define IS_CHAR(c) xmlIsCharQ(c)
76
77 /**
78 * IS_CHAR_CH:
79 * @c: an xmlChar (usually an unsigned char)
80 *
81 * Behaves like IS_CHAR on single-byte value
82 */
83 #define IS_CHAR_CH(c) xmlIsChar_ch(c)
84
85 /**
86 * IS_BLANK:
87 * @c: an UNICODE value (int)
88 *
89 * Macro to check the following production in the XML spec:
90 *
91 * [3] S ::= (#x20 | #x9 | #xD | #xA)+
92 */
93 #define IS_BLANK(c) xmlIsBlankQ(c)
94
95 /**
96 * IS_BLANK_CH:
97 * @c: an xmlChar value (normally unsigned char)
98 *
99 * Behaviour same as IS_BLANK
100 */
101 #define IS_BLANK_CH(c) xmlIsBlank_ch(c)
102
103 /**
104 * IS_BASECHAR:
105 * @c: an UNICODE value (int)
106 *
107 * Macro to check the following production in the XML spec:
108 *
109 * [85] BaseChar ::= ... long list see REC ...
110 */
111 #define IS_BASECHAR(c) xmlIsBaseCharQ(c)
112
113 /**
114 * IS_DIGIT:
115 * @c: an UNICODE value (int)
116 *
117 * Macro to check the following production in the XML spec:
118 *
119 * [88] Digit ::= ... long list see REC ...
120 */
121 #define IS_DIGIT(c) xmlIsDigitQ(c)
122
123 /**
124 * IS_DIGIT_CH:
125 * @c: an xmlChar value (usually an unsigned char)
126 *
127 * Behaves like IS_DIGIT but with a single byte argument
128 */
129 #define IS_DIGIT_CH(c) xmlIsDigit_ch(c)
130
131 /**
132 * IS_COMBINING:
133 * @c: an UNICODE value (int)
134 *
135 * Macro to check the following production in the XML spec:
136 *
137 * [87] CombiningChar ::= ... long list see REC ...
138 */
139 #define IS_COMBINING(c) xmlIsCombiningQ(c)
140
141 /**
142 * IS_COMBINING_CH:
143 * @c: an xmlChar (usually an unsigned char)
144 *
145 * Always false (all combining chars > 0xff)
146 */
147 #define IS_COMBINING_CH(c) 0
148
149 /**
150 * IS_EXTENDER:
151 * @c: an UNICODE value (int)
152 *
153 * Macro to check the following production in the XML spec:
154 *
155 *
156 * [89] Extender ::= #x00B7 | #x02D0 | #x02D1 | #x0387 | #x0640 |
157 * #x0E46 | #x0EC6 | #x3005 | [#x3031-#x3035] |
158 * [#x309D-#x309E] | [#x30FC-#x30FE]
159 */
160 #define IS_EXTENDER(c) xmlIsExtenderQ(c)
161
162 /**
163 * IS_EXTENDER_CH:
164 * @c: an xmlChar value (usually an unsigned char)
165 *
166 * Behaves like IS_EXTENDER but with a single-byte argument
167 */
168 #define IS_EXTENDER_CH(c) xmlIsExtender_ch(c)
169
170 /**
171 * IS_IDEOGRAPHIC:
172 * @c: an UNICODE value (int)
173 *
174 * Macro to check the following production in the XML spec:
175 *
176 *
177 * [86] Ideographic ::= [#x4E00-#x9FA5] | #x3007 | [#x3021-#x3029]
178 */
179 #define IS_IDEOGRAPHIC(c) xmlIsIdeographicQ(c)
180
181 /**
182 * IS_LETTER:
183 * @c: an UNICODE value (int)
184 *
185 * Macro to check the following production in the XML spec:
186 *
187 *
188 * [84] Letter ::= BaseChar | Ideographic
189 */
190 #define IS_LETTER(c) (IS_BASECHAR(c) || IS_IDEOGRAPHIC(c))
191
192 /**
193 * IS_LETTER_CH:
194 * @c: an xmlChar value (normally unsigned char)
195 *
196 * Macro behaves like IS_LETTER, but only check base chars
197 *
198 */
199 #define IS_LETTER_CH(c) xmlIsBaseChar_ch(c)
200
201 /**
202 * IS_ASCII_LETTER:
203 * @c: an xmlChar value
204 *
205 * Macro to check [a-zA-Z]
206 *
207 */
208 #define IS_ASCII_LETTER(c) (((0x41 <= (c)) && ((c) <= 0x5a)) || \
209 ((0x61 <= (c)) && ((c) <= 0x7a)))
210
211 /**
212 * IS_ASCII_DIGIT:
213 * @c: an xmlChar value
214 *
215 * Macro to check [0-9]
216 *
217 */
218 #define IS_ASCII_DIGIT(c) ((0x30 <= (c)) && ((c) <= 0x39))
219
220 /**
221 * IS_PUBIDCHAR:
222 * @c: an UNICODE value (int)
223 *
224 * Macro to check the following production in the XML spec:
225 *
226 *
227 * [13] PubidChar ::= #x20 | #xD | #xA | [a-zA-Z0-9] | [-'()+,./:=?;!*#@$_%]
228 */
229 #define IS_PUBIDCHAR(c) xmlIsPubidCharQ(c)
230
231 /**
232 * IS_PUBIDCHAR_CH:
233 * @c: an xmlChar value (normally unsigned char)
234 *
235 * Same as IS_PUBIDCHAR but for single-byte value
236 */
237 #define IS_PUBIDCHAR_CH(c) xmlIsPubidChar_ch(c)
238
239 /**
240 * SKIP_EOL:
241 * @p: and UTF8 string pointer
242 *
243 * Skips the end of line chars.
244 */
245 #define SKIP_EOL(p) \
246 if (*(p) == 0x13) { p++ ; if (*(p) == 0x10) p++; } \
247 if (*(p) == 0x10) { p++ ; if (*(p) == 0x13) p++; }
248
249 /**
250 * MOVETO_ENDTAG:
251 * @p: and UTF8 string pointer
252 *
253 * Skips to the next '>' char.
254 */
255 #define MOVETO_ENDTAG(p) \
256 while ((*p) && (*(p) != '>')) (p)++
257
258 /**
259 * MOVETO_STARTTAG:
260 * @p: and UTF8 string pointer
261 *
262 * Skips to the next '<' char.
263 */
264 #define MOVETO_STARTTAG(p) \
265 while ((*p) && (*(p) != '<')) (p)++
266
267 /**
268 * Global variables used for predefined strings.
269 */
270 XMLPUBVAR const xmlChar xmlStringText[];
271 XMLPUBVAR const xmlChar xmlStringTextNoenc[];
272 XMLPUBVAR const xmlChar xmlStringComment[];
273
274 /*
275 * Function to finish the work of the macros where needed.
276 */
277 XMLPUBFUN int XMLCALL xmlIsLetter (int c);
278
279 /**
280 * Parser context.
281 */
282 XMLPUBFUN xmlParserCtxtPtr XMLCALL
283 xmlCreateFileParserCtxt (const char *filename);
284 XMLPUBFUN xmlParserCtxtPtr XMLCALL
285 xmlCreateURLParserCtxt (const char *filename,
286 int options);
287 XMLPUBFUN xmlParserCtxtPtr XMLCALL
288 xmlCreateMemoryParserCtxt(const char *buffer,
289 int size);
290 XMLPUBFUN xmlParserCtxtPtr XMLCALL
291 xmlCreateEntityParserCtxt(const xmlChar *URL,
292 const xmlChar *ID,
293 const xmlChar *base);
294 XMLPUBFUN int XMLCALL
295 xmlSwitchEncoding (xmlParserCtxtPtr ctxt,
296 xmlCharEncoding enc);
297 XMLPUBFUN int XMLCALL
298 xmlSwitchToEncoding (xmlParserCtxtPtr ctxt,
299 xmlCharEncodingHandlerPtr handler);
300 XMLPUBFUN int XMLCALL
301 xmlSwitchInputEncoding (xmlParserCtxtPtr ctxt,
302 xmlParserInputPtr input,
303 xmlCharEncodingHandlerPtr handler);
304
305 #ifdef IN_LIBXML
306 /* internal error reporting */
307 XMLPUBFUN void XMLCALL
308 __xmlErrEncoding (xmlParserCtxtPtr ctxt,
309 xmlParserErrors xmlerr,
310 const char *msg,
311 const xmlChar * str1,
312 const xmlChar * str2);
313 #endif
314
315 /**
316 * Input Streams.
317 */
318 XMLPUBFUN xmlParserInputPtr XMLCALL
319 xmlNewStringInputStream (xmlParserCtxtPtr ctxt,
320 const xmlChar *buffer);
321 XMLPUBFUN xmlParserInputPtr XMLCALL
322 xmlNewEntityInputStream (xmlParserCtxtPtr ctxt,
323 xmlEntityPtr entity);
324 XMLPUBFUN int XMLCALL
325 xmlPushInput (xmlParserCtxtPtr ctxt,
326 xmlParserInputPtr input);
327 XMLPUBFUN xmlChar XMLCALL
328 xmlPopInput (xmlParserCtxtPtr ctxt);
329 XMLPUBFUN void XMLCALL
330 xmlFreeInputStream (xmlParserInputPtr input);
331 XMLPUBFUN xmlParserInputPtr XMLCALL
332 xmlNewInputFromFile (xmlParserCtxtPtr ctxt,
333 const char *filename);
334 XMLPUBFUN xmlParserInputPtr XMLCALL
335 xmlNewInputStream (xmlParserCtxtPtr ctxt);
336
337 /**
338 * Namespaces.
339 */
340 XMLPUBFUN xmlChar * XMLCALL
341 xmlSplitQName (xmlParserCtxtPtr ctxt,
342 const xmlChar *name,
343 xmlChar **prefix);
344
345 /**
346 * Generic production rules.
347 */
348 XMLPUBFUN const xmlChar * XMLCALL
349 xmlParseName (xmlParserCtxtPtr ctxt);
350 XMLPUBFUN xmlChar * XMLCALL
351 xmlParseNmtoken (xmlParserCtxtPtr ctxt);
352 XMLPUBFUN xmlChar * XMLCALL
353 xmlParseEntityValue (xmlParserCtxtPtr ctxt,
354 xmlChar **orig);
355 XMLPUBFUN xmlChar * XMLCALL
356 xmlParseAttValue (xmlParserCtxtPtr ctxt);
357 XMLPUBFUN xmlChar * XMLCALL
358 xmlParseSystemLiteral (xmlParserCtxtPtr ctxt);
359 XMLPUBFUN xmlChar * XMLCALL
360 xmlParsePubidLiteral (xmlParserCtxtPtr ctxt);
361 XMLPUBFUN void XMLCALL
362 xmlParseCharData (xmlParserCtxtPtr ctxt,
363 int cdata);
364 XMLPUBFUN xmlChar * XMLCALL
365 xmlParseExternalID (xmlParserCtxtPtr ctxt,
366 xmlChar **publicID,
367 int strict);
368 XMLPUBFUN void XMLCALL
369 xmlParseComment (xmlParserCtxtPtr ctxt);
370 XMLPUBFUN const xmlChar * XMLCALL
371 xmlParsePITarget (xmlParserCtxtPtr ctxt);
372 XMLPUBFUN void XMLCALL
373 xmlParsePI (xmlParserCtxtPtr ctxt);
374 XMLPUBFUN void XMLCALL
375 xmlParseNotationDecl (xmlParserCtxtPtr ctxt);
376 XMLPUBFUN void XMLCALL
377 xmlParseEntityDecl (xmlParserCtxtPtr ctxt);
378 XMLPUBFUN int XMLCALL
379 xmlParseDefaultDecl (xmlParserCtxtPtr ctxt,
380 xmlChar **value);
381 XMLPUBFUN xmlEnumerationPtr XMLCALL
382 xmlParseNotationType (xmlParserCtxtPtr ctxt);
383 XMLPUBFUN xmlEnumerationPtr XMLCALL
384 xmlParseEnumerationType (xmlParserCtxtPtr ctxt);
385 XMLPUBFUN int XMLCALL
386 xmlParseEnumeratedType (xmlParserCtxtPtr ctxt,
387 xmlEnumerationPtr *tree);
388 XMLPUBFUN int XMLCALL
389 xmlParseAttributeType (xmlParserCtxtPtr ctxt,
390 xmlEnumerationPtr *tree);
391 XMLPUBFUN void XMLCALL
392 xmlParseAttributeListDecl(xmlParserCtxtPtr ctxt);
393 XMLPUBFUN xmlElementContentPtr XMLCALL
394 xmlParseElementMixedContentDecl
395 (xmlParserCtxtPtr ctxt,
396 int inputchk);
397 XMLPUBFUN xmlElementContentPtr XMLCALL
398 xmlParseElementChildrenContentDecl
399 (xmlParserCtxtPtr ctxt,
400 int inputchk);
401 XMLPUBFUN int XMLCALL
402 xmlParseElementContentDecl(xmlParserCtxtPtr ctxt,
403 const xmlChar *name,
404 xmlElementContentPtr *result);
405 XMLPUBFUN int XMLCALL
406 xmlParseElementDecl (xmlParserCtxtPtr ctxt);
407 XMLPUBFUN void XMLCALL
408 xmlParseMarkupDecl (xmlParserCtxtPtr ctxt);
409 XMLPUBFUN int XMLCALL
410 xmlParseCharRef (xmlParserCtxtPtr ctxt);
411 XMLPUBFUN xmlEntityPtr XMLCALL
412 xmlParseEntityRef (xmlParserCtxtPtr ctxt);
413 XMLPUBFUN void XMLCALL
414 xmlParseReference (xmlParserCtxtPtr ctxt);
415 XMLPUBFUN void XMLCALL
416 xmlParsePEReference (xmlParserCtxtPtr ctxt);
417 XMLPUBFUN void XMLCALL
418 xmlParseDocTypeDecl (xmlParserCtxtPtr ctxt);
419 #ifdef LIBXML_SAX1_ENABLED
420 XMLPUBFUN const xmlChar * XMLCALL
421 xmlParseAttribute (xmlParserCtxtPtr ctxt,
422 xmlChar **value);
423 XMLPUBFUN const xmlChar * XMLCALL
424 xmlParseStartTag (xmlParserCtxtPtr ctxt);
425 XMLPUBFUN void XMLCALL
426 xmlParseEndTag (xmlParserCtxtPtr ctxt);
427 #endif /* LIBXML_SAX1_ENABLED */
428 XMLPUBFUN void XMLCALL
429 xmlParseCDSect (xmlParserCtxtPtr ctxt);
430 XMLPUBFUN void XMLCALL
431 xmlParseContent (xmlParserCtxtPtr ctxt);
432 XMLPUBFUN void XMLCALL
433 xmlParseElement (xmlParserCtxtPtr ctxt);
434 XMLPUBFUN xmlChar * XMLCALL
435 xmlParseVersionNum (xmlParserCtxtPtr ctxt);
436 XMLPUBFUN xmlChar * XMLCALL
437 xmlParseVersionInfo (xmlParserCtxtPtr ctxt);
438 XMLPUBFUN xmlChar * XMLCALL
439 xmlParseEncName (xmlParserCtxtPtr ctxt);
440 XMLPUBFUN const xmlChar * XMLCALL
441 xmlParseEncodingDecl (xmlParserCtxtPtr ctxt);
442 XMLPUBFUN int XMLCALL
443 xmlParseSDDecl (xmlParserCtxtPtr ctxt);
444 XMLPUBFUN void XMLCALL
445 xmlParseXMLDecl (xmlParserCtxtPtr ctxt);
446 XMLPUBFUN void XMLCALL
447 xmlParseTextDecl (xmlParserCtxtPtr ctxt);
448 XMLPUBFUN void XMLCALL
449 xmlParseMisc (xmlParserCtxtPtr ctxt);
450 XMLPUBFUN void XMLCALL
451 xmlParseExternalSubset (xmlParserCtxtPtr ctxt,
452 const xmlChar *ExternalID,
453 const xmlChar *SystemID);
454 /**
455 * XML_SUBSTITUTE_NONE:
456 *
457 * If no entities need to be substituted.
458 */
459 #define XML_SUBSTITUTE_NONE 0
460 /**
461 * XML_SUBSTITUTE_REF:
462 *
463 * Whether general entities need to be substituted.
464 */
465 #define XML_SUBSTITUTE_REF 1
466 /**
467 * XML_SUBSTITUTE_PEREF:
468 *
469 * Whether parameter entities need to be substituted.
470 */
471 #define XML_SUBSTITUTE_PEREF 2
472 /**
473 * XML_SUBSTITUTE_BOTH:
474 *
475 * Both general and parameter entities need to be substituted.
476 */
477 #define XML_SUBSTITUTE_BOTH 3
478
479 XMLPUBFUN xmlChar * XMLCALL
480 xmlStringDecodeEntities (xmlParserCtxtPtr ctxt,
481 const xmlChar *str,
482 int what,
483 xmlChar end,
484 xmlChar end2,
485 xmlChar end3);
486 XMLPUBFUN xmlChar * XMLCALL
487 xmlStringLenDecodeEntities (xmlParserCtxtPtr ctxt,
488 const xmlChar *str,
489 int len,
490 int what,
491 xmlChar end,
492 xmlChar end2,
493 xmlChar end3);
494
495 /*
496 * Generated by MACROS on top of parser.c c.f. PUSH_AND_POP.
497 */
498 XMLPUBFUN int XMLCALL nodePush (xmlParserCtxtPtr ctxt,
499 xmlNodePtr value);
500 XMLPUBFUN xmlNodePtr XMLCALL nodePop (xmlParserCtxtPtr ctxt);
501 XMLPUBFUN int XMLCALL inputPush (xmlParserCtxtPtr ctxt,
502 xmlParserInputPtr value);
503 XMLPUBFUN xmlParserInputPtr XMLCALL inputPop (xmlParserCtxtPtr ctxt);
504 XMLPUBFUN const xmlChar * XMLCALL namePop (xmlParserCtxtPtr ctxt);
505 XMLPUBFUN int XMLCALL namePush (xmlParserCtxtPtr ctxt,
506 const xmlChar *value);
507
508 /*
509 * other commodities shared between parser.c and parserInternals.
510 */
511 XMLPUBFUN int XMLCALL xmlSkipBlankChars (xmlParserCtxtPtr ctxt);
512 XMLPUBFUN int XMLCALL xmlStringCurrentChar (xmlParserCtxtPtr ctxt,
513 const xmlChar *cur,
514 int *len);
515 XMLPUBFUN void XMLCALL xmlParserHandlePEReference(xmlParserCtxtPtr ctxt);
516 XMLPUBFUN int XMLCALL xmlCheckLanguageID (const xmlChar *lang);
517
518 /*
519 * Really core function shared with HTML parser.
520 */
521 XMLPUBFUN int XMLCALL xmlCurrentChar (xmlParserCtxtPtr ctxt,
522 int *len);
523 XMLPUBFUN int XMLCALL xmlCopyCharMultiByte (xmlChar *out,
524 int val);
525 XMLPUBFUN int XMLCALL xmlCopyChar (int len,
526 xmlChar *out,
527 int val);
528 XMLPUBFUN void XMLCALL xmlNextChar (xmlParserCtxtPtr ctxt);
529 XMLPUBFUN void XMLCALL xmlParserInputShrink (xmlParserInputPtr in);
530
531 #ifdef LIBXML_HTML_ENABLED
532 /*
533 * Actually comes from the HTML parser but launched from the init stuff.
534 */
535 XMLPUBFUN void XMLCALL htmlInitAutoClose (void);
536 XMLPUBFUN htmlParserCtxtPtr XMLCALL htmlCreateFileParserCtxt(const char *filename,
537 const char *encoding);
538 #endif
539
540 /*
541 * Specific function to keep track of entities references
542 * and used by the XSLT debugger.
543 */
544 #ifdef LIBXML_LEGACY_ENABLED
545 /**
546 * xmlEntityReferenceFunc:
547 * @ent: the entity
548 * @firstNode: the fist node in the chunk
549 * @lastNode: the last nod in the chunk
550 *
551 * Callback function used when one needs to be able to track back the
552 * provenance of a chunk of nodes inherited from an entity replacement.
553 */
554 typedef void (*xmlEntityReferenceFunc) (xmlEntityPtr ent,
555 xmlNodePtr firstNode,
556 xmlNodePtr lastNode);
557
558 XMLPUBFUN void XMLCALL xmlSetEntityReferenceFunc (xmlEntityReferenceFunc func);
559
560 XMLPUBFUN xmlChar * XMLCALL
561 xmlParseQuotedString (xmlParserCtxtPtr ctxt);
562 XMLPUBFUN void XMLCALL
563 xmlParseNamespace (xmlParserCtxtPtr ctxt);
564 XMLPUBFUN xmlChar * XMLCALL
565 xmlNamespaceParseNSDef (xmlParserCtxtPtr ctxt);
566 XMLPUBFUN xmlChar * XMLCALL
567 xmlScanName (xmlParserCtxtPtr ctxt);
568 XMLPUBFUN xmlChar * XMLCALL
569 xmlNamespaceParseNCName (xmlParserCtxtPtr ctxt);
570 XMLPUBFUN void XMLCALL xmlParserHandleReference(xmlParserCtxtPtr ctxt);
571 XMLPUBFUN xmlChar * XMLCALL
572 xmlNamespaceParseQName (xmlParserCtxtPtr ctxt,
573 xmlChar **prefix);
574 /**
575 * Entities
576 */
577 XMLPUBFUN xmlChar * XMLCALL
578 xmlDecodeEntities (xmlParserCtxtPtr ctxt,
579 int len,
580 int what,
581 xmlChar end,
582 xmlChar end2,
583 xmlChar end3);
584 XMLPUBFUN void XMLCALL
585 xmlHandleEntity (xmlParserCtxtPtr ctxt,
586 xmlEntityPtr entity);
587
588 #endif /* LIBXML_LEGACY_ENABLED */
589
590 #ifdef IN_LIBXML
591 /*
592 * internal only
593 */
594 XMLPUBFUN void XMLCALL
595 xmlErrMemory (xmlParserCtxtPtr ctxt,
596 const char *extra);
597 #endif
598
599 #ifdef __cplusplus
600 }
601 #endif
602 #endif /* __XML_PARSER_INTERNALS_H__ */