1347
|
1 /*
|
|
2 The contents of this file are subject to the Mozilla Public License
|
|
3 Version 1.1 (the "License"); you may not use this file except in
|
|
4 compliance with the License. You may obtain a copy of the License at
|
|
5 http://www.mozilla.org/MPL/
|
|
6
|
|
7 Software distributed under the License is distributed on an "AS IS"
|
|
8 basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See the
|
|
9 License for the specific language governing rights and limitations
|
|
10 under the License.
|
|
11
|
|
12 The Original Code is expat.
|
|
13
|
|
14 The Initial Developer of the Original Code is James Clark.
|
|
15 Portions created by James Clark are Copyright (C) 1998, 1999
|
|
16 James Clark. All Rights Reserved.
|
|
17
|
|
18 Contributor(s):
|
|
19
|
|
20 Alternatively, the contents of this file may be used under the terms
|
|
21 of the GNU General Public License (the "GPL"), in which case the
|
|
22 provisions of the GPL are applicable instead of those above. If you
|
|
23 wish to allow use of your version of this file only under the terms of
|
|
24 the GPL and not to allow others to use your version of this file under
|
|
25 the MPL, indicate your decision by deleting the provisions above and
|
|
26 replace them with the notice and other provisions required by the
|
|
27 GPL. If you do not delete the provisions above, a recipient may use
|
|
28 your version of this file under either the MPL or the GPL.
|
|
29 */
|
|
30
|
|
31 #include "xmldef.h"
|
|
32 #include "xmlrole.h"
|
|
33
|
|
34 /* Doesn't check:
|
|
35
|
|
36 that ,| are not mixed in a model group
|
|
37 content of literals
|
|
38
|
|
39 */
|
|
40
|
|
41 #ifndef MIN_BYTES_PER_CHAR
|
|
42 #define MIN_BYTES_PER_CHAR(enc) ((enc)->minBytesPerChar)
|
|
43 #endif
|
|
44
|
|
45 typedef int PROLOG_HANDLER(struct prolog_state *state,
|
|
46 int tok,
|
|
47 const char *ptr,
|
|
48 const char *end,
|
|
49 const ENCODING *enc);
|
|
50
|
|
51 static PROLOG_HANDLER
|
|
52 prolog0, prolog1, prolog2,
|
|
53 doctype0, doctype1, doctype2, doctype3, doctype4, doctype5,
|
|
54 internalSubset,
|
|
55 entity0, entity1, entity2, entity3, entity4, entity5, entity6,
|
|
56 entity7, entity8, entity9,
|
|
57 notation0, notation1, notation2, notation3, notation4,
|
|
58 attlist0, attlist1, attlist2, attlist3, attlist4, attlist5, attlist6,
|
|
59 attlist7, attlist8, attlist9,
|
|
60 element0, element1, element2, element3, element4, element5, element6,
|
|
61 element7,
|
|
62 declClose,
|
|
63 error;
|
|
64
|
|
65 static
|
|
66 int syntaxError(PROLOG_STATE *);
|
|
67
|
|
68 static
|
|
69 int prolog0(PROLOG_STATE *state,
|
|
70 int tok,
|
|
71 const char *ptr,
|
|
72 const char *end,
|
|
73 const ENCODING *enc)
|
|
74 {
|
|
75 switch (tok) {
|
|
76 case XML_TOK_PROLOG_S:
|
|
77 state->handler = prolog1;
|
|
78 return XML_ROLE_NONE;
|
|
79 case XML_TOK_XML_DECL:
|
|
80 state->handler = prolog1;
|
|
81 return XML_ROLE_XML_DECL;
|
|
82 case XML_TOK_PI:
|
|
83 state->handler = prolog1;
|
|
84 return XML_ROLE_NONE;
|
|
85 case XML_TOK_COMMENT:
|
|
86 state->handler = prolog1;
|
|
87 case XML_TOK_BOM:
|
|
88 return XML_ROLE_NONE;
|
|
89 case XML_TOK_DECL_OPEN:
|
|
90 if (!XmlNameMatchesAscii(enc,
|
|
91 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
92 "DOCTYPE"))
|
|
93 break;
|
|
94 state->handler = doctype0;
|
|
95 return XML_ROLE_NONE;
|
|
96 case XML_TOK_INSTANCE_START:
|
|
97 state->handler = error;
|
|
98 return XML_ROLE_INSTANCE_START;
|
|
99 }
|
|
100 return syntaxError(state);
|
|
101 }
|
|
102
|
|
103 static
|
|
104 int prolog1(PROLOG_STATE *state,
|
|
105 int tok,
|
|
106 const char *ptr,
|
|
107 const char *end,
|
|
108 const ENCODING *enc)
|
|
109 {
|
|
110 switch (tok) {
|
|
111 case XML_TOK_PROLOG_S:
|
|
112 return XML_ROLE_NONE;
|
|
113 case XML_TOK_PI:
|
|
114 case XML_TOK_COMMENT:
|
|
115 case XML_TOK_BOM:
|
|
116 return XML_ROLE_NONE;
|
|
117 case XML_TOK_DECL_OPEN:
|
|
118 if (!XmlNameMatchesAscii(enc,
|
|
119 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
120 "DOCTYPE"))
|
|
121 break;
|
|
122 state->handler = doctype0;
|
|
123 return XML_ROLE_NONE;
|
|
124 case XML_TOK_INSTANCE_START:
|
|
125 state->handler = error;
|
|
126 return XML_ROLE_INSTANCE_START;
|
|
127 }
|
|
128 return syntaxError(state);
|
|
129 }
|
|
130
|
|
131 static
|
|
132 int prolog2(PROLOG_STATE *state,
|
|
133 int tok,
|
|
134 const char *ptr,
|
|
135 const char *end,
|
|
136 const ENCODING *enc)
|
|
137 {
|
|
138 switch (tok) {
|
|
139 case XML_TOK_PROLOG_S:
|
|
140 return XML_ROLE_NONE;
|
|
141 case XML_TOK_PI:
|
|
142 case XML_TOK_COMMENT:
|
|
143 return XML_ROLE_NONE;
|
|
144 case XML_TOK_INSTANCE_START:
|
|
145 state->handler = error;
|
|
146 return XML_ROLE_INSTANCE_START;
|
|
147 }
|
|
148 return syntaxError(state);
|
|
149 }
|
|
150
|
|
151 static
|
|
152 int doctype0(PROLOG_STATE *state,
|
|
153 int tok,
|
|
154 const char *ptr,
|
|
155 const char *end,
|
|
156 const ENCODING *enc)
|
|
157 {
|
|
158 switch (tok) {
|
|
159 case XML_TOK_PROLOG_S:
|
|
160 return XML_ROLE_NONE;
|
|
161 case XML_TOK_NAME:
|
|
162 case XML_TOK_PREFIXED_NAME:
|
|
163 state->handler = doctype1;
|
|
164 return XML_ROLE_DOCTYPE_NAME;
|
|
165 }
|
|
166 return syntaxError(state);
|
|
167 }
|
|
168
|
|
169 static
|
|
170 int doctype1(PROLOG_STATE *state,
|
|
171 int tok,
|
|
172 const char *ptr,
|
|
173 const char *end,
|
|
174 const ENCODING *enc)
|
|
175 {
|
|
176 switch (tok) {
|
|
177 case XML_TOK_PROLOG_S:
|
|
178 return XML_ROLE_NONE;
|
|
179 case XML_TOK_OPEN_BRACKET:
|
|
180 state->handler = internalSubset;
|
|
181 return XML_ROLE_NONE;
|
|
182 case XML_TOK_DECL_CLOSE:
|
|
183 state->handler = prolog2;
|
|
184 return XML_ROLE_DOCTYPE_CLOSE;
|
|
185 case XML_TOK_NAME:
|
|
186 if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) {
|
|
187 state->handler = doctype3;
|
|
188 return XML_ROLE_NONE;
|
|
189 }
|
|
190 if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
|
|
191 state->handler = doctype2;
|
|
192 return XML_ROLE_NONE;
|
|
193 }
|
|
194 break;
|
|
195 }
|
|
196 return syntaxError(state);
|
|
197 }
|
|
198
|
|
199 static
|
|
200 int doctype2(PROLOG_STATE *state,
|
|
201 int tok,
|
|
202 const char *ptr,
|
|
203 const char *end,
|
|
204 const ENCODING *enc)
|
|
205 {
|
|
206 switch (tok) {
|
|
207 case XML_TOK_PROLOG_S:
|
|
208 return XML_ROLE_NONE;
|
|
209 case XML_TOK_LITERAL:
|
|
210 state->handler = doctype3;
|
|
211 return XML_ROLE_DOCTYPE_PUBLIC_ID;
|
|
212 }
|
|
213 return syntaxError(state);
|
|
214 }
|
|
215
|
|
216 static
|
|
217 int doctype3(PROLOG_STATE *state,
|
|
218 int tok,
|
|
219 const char *ptr,
|
|
220 const char *end,
|
|
221 const ENCODING *enc)
|
|
222 {
|
|
223 switch (tok) {
|
|
224 case XML_TOK_PROLOG_S:
|
|
225 return XML_ROLE_NONE;
|
|
226 case XML_TOK_LITERAL:
|
|
227 state->handler = doctype4;
|
|
228 return XML_ROLE_DOCTYPE_SYSTEM_ID;
|
|
229 }
|
|
230 return syntaxError(state);
|
|
231 }
|
|
232
|
|
233 static
|
|
234 int doctype4(PROLOG_STATE *state,
|
|
235 int tok,
|
|
236 const char *ptr,
|
|
237 const char *end,
|
|
238 const ENCODING *enc)
|
|
239 {
|
|
240 switch (tok) {
|
|
241 case XML_TOK_PROLOG_S:
|
|
242 return XML_ROLE_NONE;
|
|
243 case XML_TOK_OPEN_BRACKET:
|
|
244 state->handler = internalSubset;
|
|
245 return XML_ROLE_NONE;
|
|
246 case XML_TOK_DECL_CLOSE:
|
|
247 state->handler = prolog2;
|
|
248 return XML_ROLE_DOCTYPE_CLOSE;
|
|
249 }
|
|
250 return syntaxError(state);
|
|
251 }
|
|
252
|
|
253 static
|
|
254 int doctype5(PROLOG_STATE *state,
|
|
255 int tok,
|
|
256 const char *ptr,
|
|
257 const char *end,
|
|
258 const ENCODING *enc)
|
|
259 {
|
|
260 switch (tok) {
|
|
261 case XML_TOK_PROLOG_S:
|
|
262 return XML_ROLE_NONE;
|
|
263 case XML_TOK_DECL_CLOSE:
|
|
264 state->handler = prolog2;
|
|
265 return XML_ROLE_DOCTYPE_CLOSE;
|
|
266 }
|
|
267 return syntaxError(state);
|
|
268 }
|
|
269
|
|
270 static
|
|
271 int internalSubset(PROLOG_STATE *state,
|
|
272 int tok,
|
|
273 const char *ptr,
|
|
274 const char *end,
|
|
275 const ENCODING *enc)
|
|
276 {
|
|
277 switch (tok) {
|
|
278 case XML_TOK_PROLOG_S:
|
|
279 return XML_ROLE_NONE;
|
|
280 case XML_TOK_DECL_OPEN:
|
|
281 if (XmlNameMatchesAscii(enc,
|
|
282 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
283 "ENTITY")) {
|
|
284 state->handler = entity0;
|
|
285 return XML_ROLE_NONE;
|
|
286 }
|
|
287 if (XmlNameMatchesAscii(enc,
|
|
288 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
289 "ATTLIST")) {
|
|
290 state->handler = attlist0;
|
|
291 return XML_ROLE_NONE;
|
|
292 }
|
|
293 if (XmlNameMatchesAscii(enc,
|
|
294 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
295 "ELEMENT")) {
|
|
296 state->handler = element0;
|
|
297 return XML_ROLE_NONE;
|
|
298 }
|
|
299 if (XmlNameMatchesAscii(enc,
|
|
300 ptr + 2 * MIN_BYTES_PER_CHAR(enc),
|
|
301 "NOTATION")) {
|
|
302 state->handler = notation0;
|
|
303 return XML_ROLE_NONE;
|
|
304 }
|
|
305 break;
|
|
306 case XML_TOK_PI:
|
|
307 case XML_TOK_COMMENT:
|
|
308 return XML_ROLE_NONE;
|
|
309 case XML_TOK_PARAM_ENTITY_REF:
|
|
310 return XML_ROLE_PARAM_ENTITY_REF;
|
|
311 case XML_TOK_CLOSE_BRACKET:
|
|
312 state->handler = doctype5;
|
|
313 return XML_ROLE_NONE;
|
|
314 }
|
|
315 return syntaxError(state);
|
|
316 }
|
|
317
|
|
318 static
|
|
319 int entity0(PROLOG_STATE *state,
|
|
320 int tok,
|
|
321 const char *ptr,
|
|
322 const char *end,
|
|
323 const ENCODING *enc)
|
|
324 {
|
|
325 switch (tok) {
|
|
326 case XML_TOK_PROLOG_S:
|
|
327 return XML_ROLE_NONE;
|
|
328 case XML_TOK_PERCENT:
|
|
329 state->handler = entity1;
|
|
330 return XML_ROLE_NONE;
|
|
331 case XML_TOK_NAME:
|
|
332 state->handler = entity2;
|
|
333 return XML_ROLE_GENERAL_ENTITY_NAME;
|
|
334 }
|
|
335 return syntaxError(state);
|
|
336 }
|
|
337
|
|
338 static
|
|
339 int entity1(PROLOG_STATE *state,
|
|
340 int tok,
|
|
341 const char *ptr,
|
|
342 const char *end,
|
|
343 const ENCODING *enc)
|
|
344 {
|
|
345 switch (tok) {
|
|
346 case XML_TOK_PROLOG_S:
|
|
347 return XML_ROLE_NONE;
|
|
348 case XML_TOK_NAME:
|
|
349 state->handler = entity7;
|
|
350 return XML_ROLE_PARAM_ENTITY_NAME;
|
|
351 }
|
|
352 return syntaxError(state);
|
|
353 }
|
|
354
|
|
355 static
|
|
356 int entity2(PROLOG_STATE *state,
|
|
357 int tok,
|
|
358 const char *ptr,
|
|
359 const char *end,
|
|
360 const ENCODING *enc)
|
|
361 {
|
|
362 switch (tok) {
|
|
363 case XML_TOK_PROLOG_S:
|
|
364 return XML_ROLE_NONE;
|
|
365 case XML_TOK_NAME:
|
|
366 if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) {
|
|
367 state->handler = entity4;
|
|
368 return XML_ROLE_NONE;
|
|
369 }
|
|
370 if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
|
|
371 state->handler = entity3;
|
|
372 return XML_ROLE_NONE;
|
|
373 }
|
|
374 break;
|
|
375 case XML_TOK_LITERAL:
|
|
376 state->handler = declClose;
|
|
377 return XML_ROLE_ENTITY_VALUE;
|
|
378 }
|
|
379 return syntaxError(state);
|
|
380 }
|
|
381
|
|
382 static
|
|
383 int entity3(PROLOG_STATE *state,
|
|
384 int tok,
|
|
385 const char *ptr,
|
|
386 const char *end,
|
|
387 const ENCODING *enc)
|
|
388 {
|
|
389 switch (tok) {
|
|
390 case XML_TOK_PROLOG_S:
|
|
391 return XML_ROLE_NONE;
|
|
392 case XML_TOK_LITERAL:
|
|
393 state->handler = entity4;
|
|
394 return XML_ROLE_ENTITY_PUBLIC_ID;
|
|
395 }
|
|
396 return syntaxError(state);
|
|
397 }
|
|
398
|
|
399
|
|
400 static
|
|
401 int entity4(PROLOG_STATE *state,
|
|
402 int tok,
|
|
403 const char *ptr,
|
|
404 const char *end,
|
|
405 const ENCODING *enc)
|
|
406 {
|
|
407 switch (tok) {
|
|
408 case XML_TOK_PROLOG_S:
|
|
409 return XML_ROLE_NONE;
|
|
410 case XML_TOK_LITERAL:
|
|
411 state->handler = entity5;
|
|
412 return XML_ROLE_ENTITY_SYSTEM_ID;
|
|
413 }
|
|
414 return syntaxError(state);
|
|
415 }
|
|
416
|
|
417 static
|
|
418 int entity5(PROLOG_STATE *state,
|
|
419 int tok,
|
|
420 const char *ptr,
|
|
421 const char *end,
|
|
422 const ENCODING *enc)
|
|
423 {
|
|
424 switch (tok) {
|
|
425 case XML_TOK_PROLOG_S:
|
|
426 return XML_ROLE_NONE;
|
|
427 case XML_TOK_DECL_CLOSE:
|
|
428 state->handler = internalSubset;
|
|
429 return XML_ROLE_NONE;
|
|
430 case XML_TOK_NAME:
|
|
431 if (XmlNameMatchesAscii(enc, ptr, "NDATA")) {
|
|
432 state->handler = entity6;
|
|
433 return XML_ROLE_NONE;
|
|
434 }
|
|
435 break;
|
|
436 }
|
|
437 return syntaxError(state);
|
|
438 }
|
|
439
|
|
440 static
|
|
441 int entity6(PROLOG_STATE *state,
|
|
442 int tok,
|
|
443 const char *ptr,
|
|
444 const char *end,
|
|
445 const ENCODING *enc)
|
|
446 {
|
|
447 switch (tok) {
|
|
448 case XML_TOK_PROLOG_S:
|
|
449 return XML_ROLE_NONE;
|
|
450 case XML_TOK_NAME:
|
|
451 state->handler = declClose;
|
|
452 return XML_ROLE_ENTITY_NOTATION_NAME;
|
|
453 }
|
|
454 return syntaxError(state);
|
|
455 }
|
|
456
|
|
457 static
|
|
458 int entity7(PROLOG_STATE *state,
|
|
459 int tok,
|
|
460 const char *ptr,
|
|
461 const char *end,
|
|
462 const ENCODING *enc)
|
|
463 {
|
|
464 switch (tok) {
|
|
465 case XML_TOK_PROLOG_S:
|
|
466 return XML_ROLE_NONE;
|
|
467 case XML_TOK_NAME:
|
|
468 if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) {
|
|
469 state->handler = entity9;
|
|
470 return XML_ROLE_NONE;
|
|
471 }
|
|
472 if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
|
|
473 state->handler = entity8;
|
|
474 return XML_ROLE_NONE;
|
|
475 }
|
|
476 break;
|
|
477 case XML_TOK_LITERAL:
|
|
478 state->handler = declClose;
|
|
479 return XML_ROLE_ENTITY_VALUE;
|
|
480 }
|
|
481 return syntaxError(state);
|
|
482 }
|
|
483
|
|
484 static
|
|
485 int entity8(PROLOG_STATE *state,
|
|
486 int tok,
|
|
487 const char *ptr,
|
|
488 const char *end,
|
|
489 const ENCODING *enc)
|
|
490 {
|
|
491 switch (tok) {
|
|
492 case XML_TOK_PROLOG_S:
|
|
493 return XML_ROLE_NONE;
|
|
494 case XML_TOK_LITERAL:
|
|
495 state->handler = entity9;
|
|
496 return XML_ROLE_ENTITY_PUBLIC_ID;
|
|
497 }
|
|
498 return syntaxError(state);
|
|
499 }
|
|
500
|
|
501 static
|
|
502 int entity9(PROLOG_STATE *state,
|
|
503 int tok,
|
|
504 const char *ptr,
|
|
505 const char *end,
|
|
506 const ENCODING *enc)
|
|
507 {
|
|
508 switch (tok) {
|
|
509 case XML_TOK_PROLOG_S:
|
|
510 return XML_ROLE_NONE;
|
|
511 case XML_TOK_LITERAL:
|
|
512 state->handler = declClose;
|
|
513 return XML_ROLE_ENTITY_SYSTEM_ID;
|
|
514 }
|
|
515 return syntaxError(state);
|
|
516 }
|
|
517
|
|
518 static
|
|
519 int notation0(PROLOG_STATE *state,
|
|
520 int tok,
|
|
521 const char *ptr,
|
|
522 const char *end,
|
|
523 const ENCODING *enc)
|
|
524 {
|
|
525 switch (tok) {
|
|
526 case XML_TOK_PROLOG_S:
|
|
527 return XML_ROLE_NONE;
|
|
528 case XML_TOK_NAME:
|
|
529 state->handler = notation1;
|
|
530 return XML_ROLE_NOTATION_NAME;
|
|
531 }
|
|
532 return syntaxError(state);
|
|
533 }
|
|
534
|
|
535 static
|
|
536 int notation1(PROLOG_STATE *state,
|
|
537 int tok,
|
|
538 const char *ptr,
|
|
539 const char *end,
|
|
540 const ENCODING *enc)
|
|
541 {
|
|
542 switch (tok) {
|
|
543 case XML_TOK_PROLOG_S:
|
|
544 return XML_ROLE_NONE;
|
|
545 case XML_TOK_NAME:
|
|
546 if (XmlNameMatchesAscii(enc, ptr, "SYSTEM")) {
|
|
547 state->handler = notation3;
|
|
548 return XML_ROLE_NONE;
|
|
549 }
|
|
550 if (XmlNameMatchesAscii(enc, ptr, "PUBLIC")) {
|
|
551 state->handler = notation2;
|
|
552 return XML_ROLE_NONE;
|
|
553 }
|
|
554 break;
|
|
555 }
|
|
556 return syntaxError(state);
|
|
557 }
|
|
558
|
|
559 static
|
|
560 int notation2(PROLOG_STATE *state,
|
|
561 int tok,
|
|
562 const char *ptr,
|
|
563 const char *end,
|
|
564 const ENCODING *enc)
|
|
565 {
|
|
566 switch (tok) {
|
|
567 case XML_TOK_PROLOG_S:
|
|
568 return XML_ROLE_NONE;
|
|
569 case XML_TOK_LITERAL:
|
|
570 state->handler = notation4;
|
|
571 return XML_ROLE_NOTATION_PUBLIC_ID;
|
|
572 }
|
|
573 return syntaxError(state);
|
|
574 }
|
|
575
|
|
576 static
|
|
577 int notation3(PROLOG_STATE *state,
|
|
578 int tok,
|
|
579 const char *ptr,
|
|
580 const char *end,
|
|
581 const ENCODING *enc)
|
|
582 {
|
|
583 switch (tok) {
|
|
584 case XML_TOK_PROLOG_S:
|
|
585 return XML_ROLE_NONE;
|
|
586 case XML_TOK_LITERAL:
|
|
587 state->handler = declClose;
|
|
588 return XML_ROLE_NOTATION_SYSTEM_ID;
|
|
589 }
|
|
590 return syntaxError(state);
|
|
591 }
|
|
592
|
|
593 static
|
|
594 int notation4(PROLOG_STATE *state,
|
|
595 int tok,
|
|
596 const char *ptr,
|
|
597 const char *end,
|
|
598 const ENCODING *enc)
|
|
599 {
|
|
600 switch (tok) {
|
|
601 case XML_TOK_PROLOG_S:
|
|
602 return XML_ROLE_NONE;
|
|
603 case XML_TOK_LITERAL:
|
|
604 state->handler = declClose;
|
|
605 return XML_ROLE_NOTATION_SYSTEM_ID;
|
|
606 case XML_TOK_DECL_CLOSE:
|
|
607 state->handler = internalSubset;
|
|
608 return XML_ROLE_NOTATION_NO_SYSTEM_ID;
|
|
609 }
|
|
610 return syntaxError(state);
|
|
611 }
|
|
612
|
|
613 static
|
|
614 int attlist0(PROLOG_STATE *state,
|
|
615 int tok,
|
|
616 const char *ptr,
|
|
617 const char *end,
|
|
618 const ENCODING *enc)
|
|
619 {
|
|
620 switch (tok) {
|
|
621 case XML_TOK_PROLOG_S:
|
|
622 return XML_ROLE_NONE;
|
|
623 case XML_TOK_NAME:
|
|
624 case XML_TOK_PREFIXED_NAME:
|
|
625 state->handler = attlist1;
|
|
626 return XML_ROLE_ATTLIST_ELEMENT_NAME;
|
|
627 }
|
|
628 return syntaxError(state);
|
|
629 }
|
|
630
|
|
631 static
|
|
632 int attlist1(PROLOG_STATE *state,
|
|
633 int tok,
|
|
634 const char *ptr,
|
|
635 const char *end,
|
|
636 const ENCODING *enc)
|
|
637 {
|
|
638 switch (tok) {
|
|
639 case XML_TOK_PROLOG_S:
|
|
640 return XML_ROLE_NONE;
|
|
641 case XML_TOK_DECL_CLOSE:
|
|
642 state->handler = internalSubset;
|
|
643 return XML_ROLE_NONE;
|
|
644 case XML_TOK_NAME:
|
|
645 case XML_TOK_PREFIXED_NAME:
|
|
646 state->handler = attlist2;
|
|
647 return XML_ROLE_ATTRIBUTE_NAME;
|
|
648 }
|
|
649 return syntaxError(state);
|
|
650 }
|
|
651
|
|
652 static
|
|
653 int attlist2(PROLOG_STATE *state,
|
|
654 int tok,
|
|
655 const char *ptr,
|
|
656 const char *end,
|
|
657 const ENCODING *enc)
|
|
658 {
|
|
659 switch (tok) {
|
|
660 case XML_TOK_PROLOG_S:
|
|
661 return XML_ROLE_NONE;
|
|
662 case XML_TOK_NAME:
|
|
663 {
|
|
664 static const char *types[] = {
|
|
665 "CDATA",
|
|
666 "ID",
|
|
667 "IDREF",
|
|
668 "IDREFS",
|
|
669 "ENTITY",
|
|
670 "ENTITIES",
|
|
671 "NMTOKEN",
|
|
672 "NMTOKENS",
|
|
673 };
|
|
674 int i;
|
|
675 for (i = 0; i < (int)(sizeof(types)/sizeof(types[0])); i++)
|
|
676 if (XmlNameMatchesAscii(enc, ptr, types[i])) {
|
|
677 state->handler = attlist8;
|
|
678 return XML_ROLE_ATTRIBUTE_TYPE_CDATA + i;
|
|
679 }
|
|
680 }
|
|
681 if (XmlNameMatchesAscii(enc, ptr, "NOTATION")) {
|
|
682 state->handler = attlist5;
|
|
683 return XML_ROLE_NONE;
|
|
684 }
|
|
685 break;
|
|
686 case XML_TOK_OPEN_PAREN:
|
|
687 state->handler = attlist3;
|
|
688 return XML_ROLE_NONE;
|
|
689 }
|
|
690 return syntaxError(state);
|
|
691 }
|
|
692
|
|
693 static
|
|
694 int attlist3(PROLOG_STATE *state,
|
|
695 int tok,
|
|
696 const char *ptr,
|
|
697 const char *end,
|
|
698 const ENCODING *enc)
|
|
699 {
|
|
700 switch (tok) {
|
|
701 case XML_TOK_PROLOG_S:
|
|
702 return XML_ROLE_NONE;
|
|
703 case XML_TOK_NMTOKEN:
|
|
704 case XML_TOK_NAME:
|
|
705 case XML_TOK_PREFIXED_NAME:
|
|
706 state->handler = attlist4;
|
|
707 return XML_ROLE_ATTRIBUTE_ENUM_VALUE;
|
|
708 }
|
|
709 return syntaxError(state);
|
|
710 }
|
|
711
|
|
712 static
|
|
713 int attlist4(PROLOG_STATE *state,
|
|
714 int tok,
|
|
715 const char *ptr,
|
|
716 const char *end,
|
|
717 const ENCODING *enc)
|
|
718 {
|
|
719 switch (tok) {
|
|
720 case XML_TOK_PROLOG_S:
|
|
721 return XML_ROLE_NONE;
|
|
722 case XML_TOK_CLOSE_PAREN:
|
|
723 state->handler = attlist8;
|
|
724 return XML_ROLE_NONE;
|
|
725 case XML_TOK_OR:
|
|
726 state->handler = attlist3;
|
|
727 return XML_ROLE_NONE;
|
|
728 }
|
|
729 return syntaxError(state);
|
|
730 }
|
|
731
|
|
732 static
|
|
733 int attlist5(PROLOG_STATE *state,
|
|
734 int tok,
|
|
735 const char *ptr,
|
|
736 const char *end,
|
|
737 const ENCODING *enc)
|
|
738 {
|
|
739 switch (tok) {
|
|
740 case XML_TOK_PROLOG_S:
|
|
741 return XML_ROLE_NONE;
|
|
742 case XML_TOK_OPEN_PAREN:
|
|
743 state->handler = attlist6;
|
|
744 return XML_ROLE_NONE;
|
|
745 }
|
|
746 return syntaxError(state);
|
|
747 }
|
|
748
|
|
749
|
|
750 static
|
|
751 int attlist6(PROLOG_STATE *state,
|
|
752 int tok,
|
|
753 const char *ptr,
|
|
754 const char *end,
|
|
755 const ENCODING *enc)
|
|
756 {
|
|
757 switch (tok) {
|
|
758 case XML_TOK_PROLOG_S:
|
|
759 return XML_ROLE_NONE;
|
|
760 case XML_TOK_NAME:
|
|
761 state->handler = attlist7;
|
|
762 return XML_ROLE_ATTRIBUTE_NOTATION_VALUE;
|
|
763 }
|
|
764 return syntaxError(state);
|
|
765 }
|
|
766
|
|
767 static
|
|
768 int attlist7(PROLOG_STATE *state,
|
|
769 int tok,
|
|
770 const char *ptr,
|
|
771 const char *end,
|
|
772 const ENCODING *enc)
|
|
773 {
|
|
774 switch (tok) {
|
|
775 case XML_TOK_PROLOG_S:
|
|
776 return XML_ROLE_NONE;
|
|
777 case XML_TOK_CLOSE_PAREN:
|
|
778 state->handler = attlist8;
|
|
779 return XML_ROLE_NONE;
|
|
780 case XML_TOK_OR:
|
|
781 state->handler = attlist6;
|
|
782 return XML_ROLE_NONE;
|
|
783 }
|
|
784 return syntaxError(state);
|
|
785 }
|
|
786
|
|
787 /* default value */
|
|
788 static
|
|
789 int attlist8(PROLOG_STATE *state,
|
|
790 int tok,
|
|
791 const char *ptr,
|
|
792 const char *end,
|
|
793 const ENCODING *enc)
|
|
794 {
|
|
795 switch (tok) {
|
|
796 case XML_TOK_PROLOG_S:
|
|
797 return XML_ROLE_NONE;
|
|
798 case XML_TOK_POUND_NAME:
|
|
799 if (XmlNameMatchesAscii(enc,
|
|
800 ptr + MIN_BYTES_PER_CHAR(enc),
|
|
801 "IMPLIED")) {
|
|
802 state->handler = attlist1;
|
|
803 return XML_ROLE_IMPLIED_ATTRIBUTE_VALUE;
|
|
804 }
|
|
805 if (XmlNameMatchesAscii(enc,
|
|
806 ptr + MIN_BYTES_PER_CHAR(enc),
|
|
807 "REQUIRED")) {
|
|
808 state->handler = attlist1;
|
|
809 return XML_ROLE_REQUIRED_ATTRIBUTE_VALUE;
|
|
810 }
|
|
811 if (XmlNameMatchesAscii(enc,
|
|
812 ptr + MIN_BYTES_PER_CHAR(enc),
|
|
813 "FIXED")) {
|
|
814 state->handler = attlist9;
|
|
815 return XML_ROLE_NONE;
|
|
816 }
|
|
817 break;
|
|
818 case XML_TOK_LITERAL:
|
|
819 state->handler = attlist1;
|
|
820 return XML_ROLE_DEFAULT_ATTRIBUTE_VALUE;
|
|
821 }
|
|
822 return syntaxError(state);
|
|
823 }
|
|
824
|
|
825 static
|
|
826 int attlist9(PROLOG_STATE *state,
|
|
827 int tok,
|
|
828 const char *ptr,
|
|
829 const char *end,
|
|
830 const ENCODING *enc)
|
|
831 {
|
|
832 switch (tok) {
|
|
833 case XML_TOK_PROLOG_S:
|
|
834 return XML_ROLE_NONE;
|
|
835 case XML_TOK_LITERAL:
|
|
836 state->handler = attlist1;
|
|
837 return XML_ROLE_FIXED_ATTRIBUTE_VALUE;
|
|
838 }
|
|
839 return syntaxError(state);
|
|
840 }
|
|
841
|
|
842 static
|
|
843 int element0(PROLOG_STATE *state,
|
|
844 int tok,
|
|
845 const char *ptr,
|
|
846 const char *end,
|
|
847 const ENCODING *enc)
|
|
848 {
|
|
849 switch (tok) {
|
|
850 case XML_TOK_PROLOG_S:
|
|
851 return XML_ROLE_NONE;
|
|
852 case XML_TOK_NAME:
|
|
853 case XML_TOK_PREFIXED_NAME:
|
|
854 state->handler = element1;
|
|
855 return XML_ROLE_ELEMENT_NAME;
|
|
856 }
|
|
857 return syntaxError(state);
|
|
858 }
|
|
859
|
|
860 static
|
|
861 int element1(PROLOG_STATE *state,
|
|
862 int tok,
|
|
863 const char *ptr,
|
|
864 const char *end,
|
|
865 const ENCODING *enc)
|
|
866 {
|
|
867 switch (tok) {
|
|
868 case XML_TOK_PROLOG_S:
|
|
869 return XML_ROLE_NONE;
|
|
870 case XML_TOK_NAME:
|
|
871 if (XmlNameMatchesAscii(enc, ptr, "EMPTY")) {
|
|
872 state->handler = declClose;
|
|
873 return XML_ROLE_CONTENT_EMPTY;
|
|
874 }
|
|
875 if (XmlNameMatchesAscii(enc, ptr, "ANY")) {
|
|
876 state->handler = declClose;
|
|
877 return XML_ROLE_CONTENT_ANY;
|
|
878 }
|
|
879 break;
|
|
880 case XML_TOK_OPEN_PAREN:
|
|
881 state->handler = element2;
|
|
882 state->level = 1;
|
|
883 return XML_ROLE_GROUP_OPEN;
|
|
884 }
|
|
885 return syntaxError(state);
|
|
886 }
|
|
887
|
|
888 static
|
|
889 int element2(PROLOG_STATE *state,
|
|
890 int tok,
|
|
891 const char *ptr,
|
|
892 const char *end,
|
|
893 const ENCODING *enc)
|
|
894 {
|
|
895 switch (tok) {
|
|
896 case XML_TOK_PROLOG_S:
|
|
897 return XML_ROLE_NONE;
|
|
898 case XML_TOK_POUND_NAME:
|
|
899 if (XmlNameMatchesAscii(enc,
|
|
900 ptr + MIN_BYTES_PER_CHAR(enc),
|
|
901 "PCDATA")) {
|
|
902 state->handler = element3;
|
|
903 return XML_ROLE_CONTENT_PCDATA;
|
|
904 }
|
|
905 break;
|
|
906 case XML_TOK_OPEN_PAREN:
|
|
907 state->level = 2;
|
|
908 state->handler = element6;
|
|
909 return XML_ROLE_GROUP_OPEN;
|
|
910 case XML_TOK_NAME:
|
|
911 case XML_TOK_PREFIXED_NAME:
|
|
912 state->handler = element7;
|
|
913 return XML_ROLE_CONTENT_ELEMENT;
|
|
914 case XML_TOK_NAME_QUESTION:
|
|
915 state->handler = element7;
|
|
916 return XML_ROLE_CONTENT_ELEMENT_OPT;
|
|
917 case XML_TOK_NAME_ASTERISK:
|
|
918 state->handler = element7;
|
|
919 return XML_ROLE_CONTENT_ELEMENT_REP;
|
|
920 case XML_TOK_NAME_PLUS:
|
|
921 state->handler = element7;
|
|
922 return XML_ROLE_CONTENT_ELEMENT_PLUS;
|
|
923 }
|
|
924 return syntaxError(state);
|
|
925 }
|
|
926
|
|
927 static
|
|
928 int element3(PROLOG_STATE *state,
|
|
929 int tok,
|
|
930 const char *ptr,
|
|
931 const char *end,
|
|
932 const ENCODING *enc)
|
|
933 {
|
|
934 switch (tok) {
|
|
935 case XML_TOK_PROLOG_S:
|
|
936 return XML_ROLE_NONE;
|
|
937 case XML_TOK_CLOSE_PAREN:
|
|
938 case XML_TOK_CLOSE_PAREN_ASTERISK:
|
|
939 state->handler = declClose;
|
|
940 return XML_ROLE_GROUP_CLOSE_REP;
|
|
941 case XML_TOK_OR:
|
|
942 state->handler = element4;
|
|
943 return XML_ROLE_NONE;
|
|
944 }
|
|
945 return syntaxError(state);
|
|
946 }
|
|
947
|
|
948 static
|
|
949 int element4(PROLOG_STATE *state,
|
|
950 int tok,
|
|
951 const char *ptr,
|
|
952 const char *end,
|
|
953 const ENCODING *enc)
|
|
954 {
|
|
955 switch (tok) {
|
|
956 case XML_TOK_PROLOG_S:
|
|
957 return XML_ROLE_NONE;
|
|
958 case XML_TOK_NAME:
|
|
959 case XML_TOK_PREFIXED_NAME:
|
|
960 state->handler = element5;
|
|
961 return XML_ROLE_CONTENT_ELEMENT;
|
|
962 }
|
|
963 return syntaxError(state);
|
|
964 }
|
|
965
|
|
966 static
|
|
967 int element5(PROLOG_STATE *state,
|
|
968 int tok,
|
|
969 const char *ptr,
|
|
970 const char *end,
|
|
971 const ENCODING *enc)
|
|
972 {
|
|
973 switch (tok) {
|
|
974 case XML_TOK_PROLOG_S:
|
|
975 return XML_ROLE_NONE;
|
|
976 case XML_TOK_CLOSE_PAREN_ASTERISK:
|
|
977 state->handler = declClose;
|
|
978 return XML_ROLE_GROUP_CLOSE_REP;
|
|
979 case XML_TOK_OR:
|
|
980 state->handler = element4;
|
|
981 return XML_ROLE_NONE;
|
|
982 }
|
|
983 return syntaxError(state);
|
|
984 }
|
|
985
|
|
986 static
|
|
987 int element6(PROLOG_STATE *state,
|
|
988 int tok,
|
|
989 const char *ptr,
|
|
990 const char *end,
|
|
991 const ENCODING *enc)
|
|
992 {
|
|
993 switch (tok) {
|
|
994 case XML_TOK_PROLOG_S:
|
|
995 return XML_ROLE_NONE;
|
|
996 case XML_TOK_OPEN_PAREN:
|
|
997 state->level += 1;
|
|
998 return XML_ROLE_GROUP_OPEN;
|
|
999 case XML_TOK_NAME:
|
|
1000 case XML_TOK_PREFIXED_NAME:
|
|
1001 state->handler = element7;
|
|
1002 return XML_ROLE_CONTENT_ELEMENT;
|
|
1003 case XML_TOK_NAME_QUESTION:
|
|
1004 state->handler = element7;
|
|
1005 return XML_ROLE_CONTENT_ELEMENT_OPT;
|
|
1006 case XML_TOK_NAME_ASTERISK:
|
|
1007 state->handler = element7;
|
|
1008 return XML_ROLE_CONTENT_ELEMENT_REP;
|
|
1009 case XML_TOK_NAME_PLUS:
|
|
1010 state->handler = element7;
|
|
1011 return XML_ROLE_CONTENT_ELEMENT_PLUS;
|
|
1012 }
|
|
1013 return syntaxError(state);
|
|
1014 }
|
|
1015
|
|
1016 static
|
|
1017 int element7(PROLOG_STATE *state,
|
|
1018 int tok,
|
|
1019 const char *ptr,
|
|
1020 const char *end,
|
|
1021 const ENCODING *enc)
|
|
1022 {
|
|
1023 switch (tok) {
|
|
1024 case XML_TOK_PROLOG_S:
|
|
1025 return XML_ROLE_NONE;
|
|
1026 case XML_TOK_CLOSE_PAREN:
|
|
1027 state->level -= 1;
|
|
1028 if (state->level == 0)
|
|
1029 state->handler = declClose;
|
|
1030 return XML_ROLE_GROUP_CLOSE;
|
|
1031 case XML_TOK_CLOSE_PAREN_ASTERISK:
|
|
1032 state->level -= 1;
|
|
1033 if (state->level == 0)
|
|
1034 state->handler = declClose;
|
|
1035 return XML_ROLE_GROUP_CLOSE_REP;
|
|
1036 case XML_TOK_CLOSE_PAREN_QUESTION:
|
|
1037 state->level -= 1;
|
|
1038 if (state->level == 0)
|
|
1039 state->handler = declClose;
|
|
1040 return XML_ROLE_GROUP_CLOSE_OPT;
|
|
1041 case XML_TOK_CLOSE_PAREN_PLUS:
|
|
1042 state->level -= 1;
|
|
1043 if (state->level == 0)
|
|
1044 state->handler = declClose;
|
|
1045 return XML_ROLE_GROUP_CLOSE_PLUS;
|
|
1046 case XML_TOK_COMMA:
|
|
1047 state->handler = element6;
|
|
1048 return XML_ROLE_GROUP_SEQUENCE;
|
|
1049 case XML_TOK_OR:
|
|
1050 state->handler = element6;
|
|
1051 return XML_ROLE_GROUP_CHOICE;
|
|
1052 }
|
|
1053 return syntaxError(state);
|
|
1054 }
|
|
1055
|
|
1056 static
|
|
1057 int declClose(PROLOG_STATE *state,
|
|
1058 int tok,
|
|
1059 const char *ptr,
|
|
1060 const char *end,
|
|
1061 const ENCODING *enc)
|
|
1062 {
|
|
1063 switch (tok) {
|
|
1064 case XML_TOK_PROLOG_S:
|
|
1065 return XML_ROLE_NONE;
|
|
1066 case XML_TOK_DECL_CLOSE:
|
|
1067 state->handler = internalSubset;
|
|
1068 return XML_ROLE_NONE;
|
|
1069 }
|
|
1070 return syntaxError(state);
|
|
1071 }
|
|
1072
|
|
1073 #if 0
|
|
1074
|
|
1075 static
|
|
1076 int ignore(PROLOG_STATE *state,
|
|
1077 int tok,
|
|
1078 const char *ptr,
|
|
1079 const char *end,
|
|
1080 const ENCODING *enc)
|
|
1081 {
|
|
1082 switch (tok) {
|
|
1083 case XML_TOK_DECL_CLOSE:
|
|
1084 state->handler = internalSubset;
|
|
1085 return 0;
|
|
1086 default:
|
|
1087 return XML_ROLE_NONE;
|
|
1088 }
|
|
1089 return syntaxError(state);
|
|
1090 }
|
|
1091 #endif
|
|
1092
|
|
1093 static
|
|
1094 int error(PROLOG_STATE *state,
|
|
1095 int tok,
|
|
1096 const char *ptr,
|
|
1097 const char *end,
|
|
1098 const ENCODING *enc)
|
|
1099 {
|
|
1100 return XML_ROLE_NONE;
|
|
1101 }
|
|
1102
|
|
1103 static
|
|
1104 int syntaxError(PROLOG_STATE *state)
|
|
1105 {
|
|
1106 state->handler = error;
|
|
1107 return XML_ROLE_ERROR;
|
|
1108 }
|
|
1109
|
|
1110 void XmlPrologStateInit(PROLOG_STATE *state)
|
|
1111 {
|
|
1112 state->handler = prolog0;
|
|
1113 }
|