1 |
|
---|
2 | /*
|
---|
3 | *sourcefile xmltok_impl.c
|
---|
4 | * part of the expat implementation. See xmlparse.c.
|
---|
5 | *
|
---|
6 | * NOTE: This file must not be compiled directly. It is
|
---|
7 | * #include'd from xmltok.c several times.
|
---|
8 | */
|
---|
9 |
|
---|
10 | /*
|
---|
11 | * Copyright (C) 2001 Ulrich Mller.
|
---|
12 | * Copyright (c) 1998, 1999, 2000 Thai Open Source Software Center Ltd.
|
---|
13 | * and Clark Cooper.
|
---|
14 | *
|
---|
15 | * Permission is hereby granted, free of charge, to any person obtaining
|
---|
16 | * a copy of this software and associated documentation files (the
|
---|
17 | * "Software"), to deal in the Software without restriction, including
|
---|
18 | * without limitation the rights to use, copy, modify, merge, publish,
|
---|
19 | * distribute, sublicense, and/or sell copies of the Software, and to
|
---|
20 | * permit persons to whom the Software is furnished to do so, subject to
|
---|
21 | * the following conditions:
|
---|
22 | *
|
---|
23 | * The above copyright notice and this permission notice shall be included
|
---|
24 | * in all copies or substantial portions of the Software.
|
---|
25 | *
|
---|
26 | * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
|
---|
27 | * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
|
---|
28 | * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
|
---|
29 | * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
|
---|
30 | * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
|
---|
31 | * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
|
---|
32 | * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
|
---|
33 | */
|
---|
34 |
|
---|
35 | #ifndef IS_INVALID_CHAR
|
---|
36 | #define IS_INVALID_CHAR(enc, ptr, n) (0)
|
---|
37 | #endif
|
---|
38 |
|
---|
39 | #define INVALID_LEAD_CASE(n, ptr, nextTokPtr) \
|
---|
40 | case BT_LEAD ## n: \
|
---|
41 | if (end - ptr < n) \
|
---|
42 | return XML_TOK_PARTIAL_CHAR; \
|
---|
43 | if (IS_INVALID_CHAR(enc, ptr, n)) { \
|
---|
44 | *(nextTokPtr) = (ptr); \
|
---|
45 | return XML_TOK_INVALID; \
|
---|
46 | } \
|
---|
47 | ptr += n; \
|
---|
48 | break;
|
---|
49 |
|
---|
50 | #define INVALID_CASES(ptr, nextTokPtr) \
|
---|
51 | INVALID_LEAD_CASE(2, ptr, nextTokPtr) \
|
---|
52 | INVALID_LEAD_CASE(3, ptr, nextTokPtr) \
|
---|
53 | INVALID_LEAD_CASE(4, ptr, nextTokPtr) \
|
---|
54 | case BT_NONXML: \
|
---|
55 | case BT_MALFORM: \
|
---|
56 | case BT_TRAIL: \
|
---|
57 | *(nextTokPtr) = (ptr); \
|
---|
58 | return XML_TOK_INVALID;
|
---|
59 |
|
---|
60 | #define CHECK_NAME_CASE(n, enc, ptr, end, nextTokPtr) \
|
---|
61 | case BT_LEAD ## n: \
|
---|
62 | if (end - ptr < n) \
|
---|
63 | return XML_TOK_PARTIAL_CHAR; \
|
---|
64 | if (!IS_NAME_CHAR(enc, ptr, n)) { \
|
---|
65 | *nextTokPtr = ptr; \
|
---|
66 | return XML_TOK_INVALID; \
|
---|
67 | } \
|
---|
68 | ptr += n; \
|
---|
69 | break;
|
---|
70 |
|
---|
71 | #define CHECK_NAME_CASES(enc, ptr, end, nextTokPtr) \
|
---|
72 | case BT_NONASCII: \
|
---|
73 | if (!IS_NAME_CHAR_MINBPC(enc, ptr)) { \
|
---|
74 | *nextTokPtr = ptr; \
|
---|
75 | return XML_TOK_INVALID; \
|
---|
76 | } \
|
---|
77 | case BT_NMSTRT: \
|
---|
78 | case BT_HEX: \
|
---|
79 | case BT_DIGIT: \
|
---|
80 | case BT_NAME: \
|
---|
81 | case BT_MINUS: \
|
---|
82 | ptr += MINBPC(enc); \
|
---|
83 | break; \
|
---|
84 | CHECK_NAME_CASE(2, enc, ptr, end, nextTokPtr) \
|
---|
85 | CHECK_NAME_CASE(3, enc, ptr, end, nextTokPtr) \
|
---|
86 | CHECK_NAME_CASE(4, enc, ptr, end, nextTokPtr)
|
---|
87 |
|
---|
88 | #define CHECK_NMSTRT_CASE(n, enc, ptr, end, nextTokPtr) \
|
---|
89 | case BT_LEAD ## n: \
|
---|
90 | if (end - ptr < n) \
|
---|
91 | return XML_TOK_PARTIAL_CHAR; \
|
---|
92 | if (!IS_NMSTRT_CHAR(enc, ptr, n)) { \
|
---|
93 | *nextTokPtr = ptr; \
|
---|
94 | return XML_TOK_INVALID; \
|
---|
95 | } \
|
---|
96 | ptr += n; \
|
---|
97 | break;
|
---|
98 |
|
---|
99 | #define CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr) \
|
---|
100 | case BT_NONASCII: \
|
---|
101 | if (!IS_NMSTRT_CHAR_MINBPC(enc, ptr)) { \
|
---|
102 | *nextTokPtr = ptr; \
|
---|
103 | return XML_TOK_INVALID; \
|
---|
104 | } \
|
---|
105 | case BT_NMSTRT: \
|
---|
106 | case BT_HEX: \
|
---|
107 | ptr += MINBPC(enc); \
|
---|
108 | break; \
|
---|
109 | CHECK_NMSTRT_CASE(2, enc, ptr, end, nextTokPtr) \
|
---|
110 | CHECK_NMSTRT_CASE(3, enc, ptr, end, nextTokPtr) \
|
---|
111 | CHECK_NMSTRT_CASE(4, enc, ptr, end, nextTokPtr)
|
---|
112 |
|
---|
113 | #ifndef PREFIX
|
---|
114 | #define PREFIX(ident) ident
|
---|
115 | #endif
|
---|
116 |
|
---|
117 | /* ptr points to character following "<!-" */
|
---|
118 |
|
---|
119 | static int EXPATENTRY PREFIX(scanComment)(const ENCODING *enc,
|
---|
120 | const char *ptr,
|
---|
121 | const char *end,
|
---|
122 | const char **nextTokPtr)
|
---|
123 | {
|
---|
124 | if (ptr != end) {
|
---|
125 | if (!CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
|
---|
126 | *nextTokPtr = ptr;
|
---|
127 | return XML_TOK_INVALID;
|
---|
128 | }
|
---|
129 | ptr += MINBPC(enc);
|
---|
130 | while (ptr != end) {
|
---|
131 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
132 | INVALID_CASES(ptr, nextTokPtr)
|
---|
133 | case BT_MINUS:
|
---|
134 | if ((ptr += MINBPC(enc)) == end)
|
---|
135 | return XML_TOK_PARTIAL;
|
---|
136 | if (CHAR_MATCHES(enc, ptr, ASCII_MINUS)) {
|
---|
137 | if ((ptr += MINBPC(enc)) == end)
|
---|
138 | return XML_TOK_PARTIAL;
|
---|
139 | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
140 | *nextTokPtr = ptr;
|
---|
141 | return XML_TOK_INVALID;
|
---|
142 | }
|
---|
143 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
144 | return XML_TOK_COMMENT;
|
---|
145 | }
|
---|
146 | break;
|
---|
147 | default:
|
---|
148 | ptr += MINBPC(enc);
|
---|
149 | break;
|
---|
150 | }
|
---|
151 | }
|
---|
152 | }
|
---|
153 | return XML_TOK_PARTIAL;
|
---|
154 | }
|
---|
155 |
|
---|
156 | /* ptr points to character following "<!" */
|
---|
157 |
|
---|
158 | static int EXPATENTRY PREFIX(scanDecl)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
159 | const char **nextTokPtr)
|
---|
160 | {
|
---|
161 | if (ptr == end)
|
---|
162 | return XML_TOK_PARTIAL;
|
---|
163 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
164 | case BT_MINUS:
|
---|
165 | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
166 | case BT_LSQB:
|
---|
167 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
168 | return XML_TOK_COND_SECT_OPEN;
|
---|
169 | case BT_NMSTRT:
|
---|
170 | case BT_HEX:
|
---|
171 | ptr += MINBPC(enc);
|
---|
172 | break;
|
---|
173 | default:
|
---|
174 | *nextTokPtr = ptr;
|
---|
175 | return XML_TOK_INVALID;
|
---|
176 | }
|
---|
177 | while (ptr != end) {
|
---|
178 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
179 | case BT_PERCNT:
|
---|
180 | if (ptr + MINBPC(enc) == end)
|
---|
181 | return XML_TOK_PARTIAL;
|
---|
182 | /* don't allow <!ENTITY% foo "whatever"> */
|
---|
183 | switch (BYTE_TYPE(enc, ptr + MINBPC(enc))) {
|
---|
184 | case BT_S: case BT_CR: case BT_LF: case BT_PERCNT:
|
---|
185 | *nextTokPtr = ptr;
|
---|
186 | return XML_TOK_INVALID;
|
---|
187 | }
|
---|
188 | /* fall through */
|
---|
189 | case BT_S: case BT_CR: case BT_LF:
|
---|
190 | *nextTokPtr = ptr;
|
---|
191 | return XML_TOK_DECL_OPEN;
|
---|
192 | case BT_NMSTRT:
|
---|
193 | case BT_HEX:
|
---|
194 | ptr += MINBPC(enc);
|
---|
195 | break;
|
---|
196 | default:
|
---|
197 | *nextTokPtr = ptr;
|
---|
198 | return XML_TOK_INVALID;
|
---|
199 | }
|
---|
200 | }
|
---|
201 | return XML_TOK_PARTIAL;
|
---|
202 | }
|
---|
203 |
|
---|
204 | static int EXPATENTRY PREFIX(checkPiTarget)(const ENCODING *enc, const char *ptr, const char *end, int *tokPtr)
|
---|
205 | {
|
---|
206 | int upper = 0;
|
---|
207 | *tokPtr = XML_TOK_PI;
|
---|
208 | if (end - ptr != MINBPC(enc)*3)
|
---|
209 | return 1;
|
---|
210 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
211 | case ASCII_x:
|
---|
212 | break;
|
---|
213 | case ASCII_X:
|
---|
214 | upper = 1;
|
---|
215 | break;
|
---|
216 | default:
|
---|
217 | return 1;
|
---|
218 | }
|
---|
219 | ptr += MINBPC(enc);
|
---|
220 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
221 | case ASCII_m:
|
---|
222 | break;
|
---|
223 | case ASCII_M:
|
---|
224 | upper = 1;
|
---|
225 | break;
|
---|
226 | default:
|
---|
227 | return 1;
|
---|
228 | }
|
---|
229 | ptr += MINBPC(enc);
|
---|
230 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
231 | case ASCII_l:
|
---|
232 | break;
|
---|
233 | case ASCII_L:
|
---|
234 | upper = 1;
|
---|
235 | break;
|
---|
236 | default:
|
---|
237 | return 1;
|
---|
238 | }
|
---|
239 | if (upper)
|
---|
240 | return 0;
|
---|
241 | *tokPtr = XML_TOK_XML_DECL;
|
---|
242 | return 1;
|
---|
243 | }
|
---|
244 |
|
---|
245 | /* ptr points to character following "<?" */
|
---|
246 |
|
---|
247 | static int EXPATENTRY PREFIX(scanPi)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
248 | const char **nextTokPtr)
|
---|
249 | {
|
---|
250 | int tok;
|
---|
251 | const char *target = ptr;
|
---|
252 | if (ptr == end)
|
---|
253 | return XML_TOK_PARTIAL;
|
---|
254 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
255 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
256 | default:
|
---|
257 | *nextTokPtr = ptr;
|
---|
258 | return XML_TOK_INVALID;
|
---|
259 | }
|
---|
260 | while (ptr != end) {
|
---|
261 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
262 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
263 | case BT_S: case BT_CR: case BT_LF:
|
---|
264 | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
|
---|
265 | *nextTokPtr = ptr;
|
---|
266 | return XML_TOK_INVALID;
|
---|
267 | }
|
---|
268 | ptr += MINBPC(enc);
|
---|
269 | while (ptr != end) {
|
---|
270 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
271 | INVALID_CASES(ptr, nextTokPtr)
|
---|
272 | case BT_QUEST:
|
---|
273 | ptr += MINBPC(enc);
|
---|
274 | if (ptr == end)
|
---|
275 | return XML_TOK_PARTIAL;
|
---|
276 | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
277 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
278 | return tok;
|
---|
279 | }
|
---|
280 | break;
|
---|
281 | default:
|
---|
282 | ptr += MINBPC(enc);
|
---|
283 | break;
|
---|
284 | }
|
---|
285 | }
|
---|
286 | return XML_TOK_PARTIAL;
|
---|
287 | case BT_QUEST:
|
---|
288 | if (!PREFIX(checkPiTarget)(enc, target, ptr, &tok)) {
|
---|
289 | *nextTokPtr = ptr;
|
---|
290 | return XML_TOK_INVALID;
|
---|
291 | }
|
---|
292 | ptr += MINBPC(enc);
|
---|
293 | if (ptr == end)
|
---|
294 | return XML_TOK_PARTIAL;
|
---|
295 | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
296 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
297 | return tok;
|
---|
298 | }
|
---|
299 | /* fall through */
|
---|
300 | default:
|
---|
301 | *nextTokPtr = ptr;
|
---|
302 | return XML_TOK_INVALID;
|
---|
303 | }
|
---|
304 | }
|
---|
305 | return XML_TOK_PARTIAL;
|
---|
306 | }
|
---|
307 |
|
---|
308 |
|
---|
309 | static int EXPATENTRY PREFIX(scanCdataSection)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
310 | const char **nextTokPtr)
|
---|
311 | {
|
---|
312 | static const char CDATA_LSQB[] = { ASCII_C, ASCII_D, ASCII_A, ASCII_T, ASCII_A, ASCII_LSQB };
|
---|
313 | int i;
|
---|
314 | /* CDATA[ */
|
---|
315 | if (end - ptr < 6 * MINBPC(enc))
|
---|
316 | return XML_TOK_PARTIAL;
|
---|
317 | for (i = 0; i < 6; i++, ptr += MINBPC(enc)) {
|
---|
318 | if (!CHAR_MATCHES(enc, ptr, CDATA_LSQB[i])) {
|
---|
319 | *nextTokPtr = ptr;
|
---|
320 | return XML_TOK_INVALID;
|
---|
321 | }
|
---|
322 | }
|
---|
323 | *nextTokPtr = ptr;
|
---|
324 | return XML_TOK_CDATA_SECT_OPEN;
|
---|
325 | }
|
---|
326 |
|
---|
327 | static int EXPATENTRY PREFIX(cdataSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
328 | const char **nextTokPtr)
|
---|
329 | {
|
---|
330 | if (ptr == end)
|
---|
331 | return XML_TOK_NONE;
|
---|
332 | if (MINBPC(enc) > 1) {
|
---|
333 | size_t n = end - ptr;
|
---|
334 | if (n & (MINBPC(enc) - 1)) {
|
---|
335 | n &= ~(MINBPC(enc) - 1);
|
---|
336 | if (n == 0)
|
---|
337 | return XML_TOK_PARTIAL;
|
---|
338 | end = ptr + n;
|
---|
339 | }
|
---|
340 | }
|
---|
341 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
342 | case BT_RSQB:
|
---|
343 | ptr += MINBPC(enc);
|
---|
344 | if (ptr == end)
|
---|
345 | return XML_TOK_PARTIAL;
|
---|
346 | if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
|
---|
347 | break;
|
---|
348 | ptr += MINBPC(enc);
|
---|
349 | if (ptr == end)
|
---|
350 | return XML_TOK_PARTIAL;
|
---|
351 | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
352 | ptr -= MINBPC(enc);
|
---|
353 | break;
|
---|
354 | }
|
---|
355 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
356 | return XML_TOK_CDATA_SECT_CLOSE;
|
---|
357 | case BT_CR:
|
---|
358 | ptr += MINBPC(enc);
|
---|
359 | if (ptr == end)
|
---|
360 | return XML_TOK_PARTIAL;
|
---|
361 | if (BYTE_TYPE(enc, ptr) == BT_LF)
|
---|
362 | ptr += MINBPC(enc);
|
---|
363 | *nextTokPtr = ptr;
|
---|
364 | return XML_TOK_DATA_NEWLINE;
|
---|
365 | case BT_LF:
|
---|
366 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
367 | return XML_TOK_DATA_NEWLINE;
|
---|
368 | INVALID_CASES(ptr, nextTokPtr)
|
---|
369 | default:
|
---|
370 | ptr += MINBPC(enc);
|
---|
371 | break;
|
---|
372 | }
|
---|
373 | while (ptr != end) {
|
---|
374 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
375 | #define LEAD_CASE(n) \
|
---|
376 | case BT_LEAD ## n: \
|
---|
377 | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
|
---|
378 | *nextTokPtr = ptr; \
|
---|
379 | return XML_TOK_DATA_CHARS; \
|
---|
380 | } \
|
---|
381 | ptr += n; \
|
---|
382 | break;
|
---|
383 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
384 | #undef LEAD_CASE
|
---|
385 | case BT_NONXML:
|
---|
386 | case BT_MALFORM:
|
---|
387 | case BT_TRAIL:
|
---|
388 | case BT_CR:
|
---|
389 | case BT_LF:
|
---|
390 | case BT_RSQB:
|
---|
391 | *nextTokPtr = ptr;
|
---|
392 | return XML_TOK_DATA_CHARS;
|
---|
393 | default:
|
---|
394 | ptr += MINBPC(enc);
|
---|
395 | break;
|
---|
396 | }
|
---|
397 | }
|
---|
398 | *nextTokPtr = ptr;
|
---|
399 | return XML_TOK_DATA_CHARS;
|
---|
400 | }
|
---|
401 |
|
---|
402 | /* ptr points to character following "</" */
|
---|
403 |
|
---|
404 | static int EXPATENTRY PREFIX(scanEndTag)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
405 | const char **nextTokPtr)
|
---|
406 | {
|
---|
407 | if (ptr == end)
|
---|
408 | return XML_TOK_PARTIAL;
|
---|
409 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
410 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
411 | default:
|
---|
412 | *nextTokPtr = ptr;
|
---|
413 | return XML_TOK_INVALID;
|
---|
414 | }
|
---|
415 | while (ptr != end) {
|
---|
416 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
417 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
418 | case BT_S: case BT_CR: case BT_LF:
|
---|
419 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
|
---|
420 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
421 | case BT_S: case BT_CR: case BT_LF:
|
---|
422 | break;
|
---|
423 | case BT_GT:
|
---|
424 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
425 | return XML_TOK_END_TAG;
|
---|
426 | default:
|
---|
427 | *nextTokPtr = ptr;
|
---|
428 | return XML_TOK_INVALID;
|
---|
429 | }
|
---|
430 | }
|
---|
431 | return XML_TOK_PARTIAL;
|
---|
432 | #ifdef XML_NS
|
---|
433 | case BT_COLON:
|
---|
434 | /* no need to check qname syntax here, since end-tag must match exactly */
|
---|
435 | ptr += MINBPC(enc);
|
---|
436 | break;
|
---|
437 | #endif
|
---|
438 | case BT_GT:
|
---|
439 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
440 | return XML_TOK_END_TAG;
|
---|
441 | default:
|
---|
442 | *nextTokPtr = ptr;
|
---|
443 | return XML_TOK_INVALID;
|
---|
444 | }
|
---|
445 | }
|
---|
446 | return XML_TOK_PARTIAL;
|
---|
447 | }
|
---|
448 |
|
---|
449 | /* ptr points to character following "&#X" */
|
---|
450 |
|
---|
451 | static int EXPATENTRY PREFIX(scanHexCharRef)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
452 | const char **nextTokPtr)
|
---|
453 | {
|
---|
454 | if (ptr != end) {
|
---|
455 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
456 | case BT_DIGIT:
|
---|
457 | case BT_HEX:
|
---|
458 | break;
|
---|
459 | default:
|
---|
460 | *nextTokPtr = ptr;
|
---|
461 | return XML_TOK_INVALID;
|
---|
462 | }
|
---|
463 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
|
---|
464 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
465 | case BT_DIGIT:
|
---|
466 | case BT_HEX:
|
---|
467 | break;
|
---|
468 | case BT_SEMI:
|
---|
469 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
470 | return XML_TOK_CHAR_REF;
|
---|
471 | default:
|
---|
472 | *nextTokPtr = ptr;
|
---|
473 | return XML_TOK_INVALID;
|
---|
474 | }
|
---|
475 | }
|
---|
476 | }
|
---|
477 | return XML_TOK_PARTIAL;
|
---|
478 | }
|
---|
479 |
|
---|
480 | /* ptr points to character following "&#" */
|
---|
481 |
|
---|
482 | static int EXPATENTRY PREFIX(scanCharRef)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
483 | const char **nextTokPtr)
|
---|
484 | {
|
---|
485 | if (ptr != end) {
|
---|
486 | if (CHAR_MATCHES(enc, ptr, ASCII_x))
|
---|
487 | return PREFIX(scanHexCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
488 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
489 | case BT_DIGIT:
|
---|
490 | break;
|
---|
491 | default:
|
---|
492 | *nextTokPtr = ptr;
|
---|
493 | return XML_TOK_INVALID;
|
---|
494 | }
|
---|
495 | for (ptr += MINBPC(enc); ptr != end; ptr += MINBPC(enc)) {
|
---|
496 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
497 | case BT_DIGIT:
|
---|
498 | break;
|
---|
499 | case BT_SEMI:
|
---|
500 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
501 | return XML_TOK_CHAR_REF;
|
---|
502 | default:
|
---|
503 | *nextTokPtr = ptr;
|
---|
504 | return XML_TOK_INVALID;
|
---|
505 | }
|
---|
506 | }
|
---|
507 | }
|
---|
508 | return XML_TOK_PARTIAL;
|
---|
509 | }
|
---|
510 |
|
---|
511 | /* ptr points to character following "&" */
|
---|
512 |
|
---|
513 | static int EXPATENTRY PREFIX(scanRef)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
514 | const char **nextTokPtr)
|
---|
515 | {
|
---|
516 | if (ptr == end)
|
---|
517 | return XML_TOK_PARTIAL;
|
---|
518 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
519 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
520 | case BT_NUM:
|
---|
521 | return PREFIX(scanCharRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
522 | default:
|
---|
523 | *nextTokPtr = ptr;
|
---|
524 | return XML_TOK_INVALID;
|
---|
525 | }
|
---|
526 | while (ptr != end) {
|
---|
527 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
528 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
529 | case BT_SEMI:
|
---|
530 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
531 | return XML_TOK_ENTITY_REF;
|
---|
532 | default:
|
---|
533 | *nextTokPtr = ptr;
|
---|
534 | return XML_TOK_INVALID;
|
---|
535 | }
|
---|
536 | }
|
---|
537 | return XML_TOK_PARTIAL;
|
---|
538 | }
|
---|
539 |
|
---|
540 | /* ptr points to character following first character of attribute name */
|
---|
541 |
|
---|
542 | static int EXPATENTRY PREFIX(scanAtts)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
543 | const char **nextTokPtr)
|
---|
544 | {
|
---|
545 | #ifdef XML_NS
|
---|
546 | int hadColon = 0;
|
---|
547 | #endif
|
---|
548 | while (ptr != end) {
|
---|
549 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
550 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
551 | #ifdef XML_NS
|
---|
552 | case BT_COLON:
|
---|
553 | if (hadColon) {
|
---|
554 | *nextTokPtr = ptr;
|
---|
555 | return XML_TOK_INVALID;
|
---|
556 | }
|
---|
557 | hadColon = 1;
|
---|
558 | ptr += MINBPC(enc);
|
---|
559 | if (ptr == end)
|
---|
560 | return XML_TOK_PARTIAL;
|
---|
561 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
562 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
563 | default:
|
---|
564 | *nextTokPtr = ptr;
|
---|
565 | return XML_TOK_INVALID;
|
---|
566 | }
|
---|
567 | break;
|
---|
568 | #endif
|
---|
569 | case BT_S: case BT_CR: case BT_LF:
|
---|
570 | for (;;) {
|
---|
571 | int t;
|
---|
572 |
|
---|
573 | ptr += MINBPC(enc);
|
---|
574 | if (ptr == end)
|
---|
575 | return XML_TOK_PARTIAL;
|
---|
576 | t = BYTE_TYPE(enc, ptr);
|
---|
577 | if (t == BT_EQUALS)
|
---|
578 | break;
|
---|
579 | switch (t) {
|
---|
580 | case BT_S:
|
---|
581 | case BT_LF:
|
---|
582 | case BT_CR:
|
---|
583 | break;
|
---|
584 | default:
|
---|
585 | *nextTokPtr = ptr;
|
---|
586 | return XML_TOK_INVALID;
|
---|
587 | }
|
---|
588 | }
|
---|
589 | /* fall through */
|
---|
590 | case BT_EQUALS:
|
---|
591 | {
|
---|
592 | int open;
|
---|
593 | #ifdef XML_NS
|
---|
594 | hadColon = 0;
|
---|
595 | #endif
|
---|
596 | for (;;) {
|
---|
597 |
|
---|
598 | ptr += MINBPC(enc);
|
---|
599 | if (ptr == end)
|
---|
600 | return XML_TOK_PARTIAL;
|
---|
601 | open = BYTE_TYPE(enc, ptr);
|
---|
602 | if (open == BT_QUOT || open == BT_APOS)
|
---|
603 | break;
|
---|
604 | switch (open) {
|
---|
605 | case BT_S:
|
---|
606 | case BT_LF:
|
---|
607 | case BT_CR:
|
---|
608 | break;
|
---|
609 | default:
|
---|
610 | *nextTokPtr = ptr;
|
---|
611 | return XML_TOK_INVALID;
|
---|
612 | }
|
---|
613 | }
|
---|
614 | ptr += MINBPC(enc);
|
---|
615 | /* in attribute value */
|
---|
616 | for (;;) {
|
---|
617 | int t;
|
---|
618 | if (ptr == end)
|
---|
619 | return XML_TOK_PARTIAL;
|
---|
620 | t = BYTE_TYPE(enc, ptr);
|
---|
621 | if (t == open)
|
---|
622 | break;
|
---|
623 | switch (t) {
|
---|
624 | INVALID_CASES(ptr, nextTokPtr)
|
---|
625 | case BT_AMP:
|
---|
626 | {
|
---|
627 | int tok = PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, &ptr);
|
---|
628 | if (tok <= 0) {
|
---|
629 | if (tok == XML_TOK_INVALID)
|
---|
630 | *nextTokPtr = ptr;
|
---|
631 | return tok;
|
---|
632 | }
|
---|
633 | break;
|
---|
634 | }
|
---|
635 | case BT_LT:
|
---|
636 | *nextTokPtr = ptr;
|
---|
637 | return XML_TOK_INVALID;
|
---|
638 | default:
|
---|
639 | ptr += MINBPC(enc);
|
---|
640 | break;
|
---|
641 | }
|
---|
642 | }
|
---|
643 | ptr += MINBPC(enc);
|
---|
644 | if (ptr == end)
|
---|
645 | return XML_TOK_PARTIAL;
|
---|
646 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
647 | case BT_S:
|
---|
648 | case BT_CR:
|
---|
649 | case BT_LF:
|
---|
650 | break;
|
---|
651 | case BT_SOL:
|
---|
652 | goto sol;
|
---|
653 | case BT_GT:
|
---|
654 | goto gt;
|
---|
655 | default:
|
---|
656 | *nextTokPtr = ptr;
|
---|
657 | return XML_TOK_INVALID;
|
---|
658 | }
|
---|
659 | /* ptr points to closing quote */
|
---|
660 | for (;;) {
|
---|
661 | ptr += MINBPC(enc);
|
---|
662 | if (ptr == end)
|
---|
663 | return XML_TOK_PARTIAL;
|
---|
664 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
665 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
666 | case BT_S: case BT_CR: case BT_LF:
|
---|
667 | continue;
|
---|
668 | case BT_GT:
|
---|
669 | gt:
|
---|
670 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
671 | return XML_TOK_START_TAG_WITH_ATTS;
|
---|
672 | case BT_SOL:
|
---|
673 | sol:
|
---|
674 | ptr += MINBPC(enc);
|
---|
675 | if (ptr == end)
|
---|
676 | return XML_TOK_PARTIAL;
|
---|
677 | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
678 | *nextTokPtr = ptr;
|
---|
679 | return XML_TOK_INVALID;
|
---|
680 | }
|
---|
681 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
682 | return XML_TOK_EMPTY_ELEMENT_WITH_ATTS;
|
---|
683 | default:
|
---|
684 | *nextTokPtr = ptr;
|
---|
685 | return XML_TOK_INVALID;
|
---|
686 | }
|
---|
687 | break;
|
---|
688 | }
|
---|
689 | break;
|
---|
690 | }
|
---|
691 | default:
|
---|
692 | *nextTokPtr = ptr;
|
---|
693 | return XML_TOK_INVALID;
|
---|
694 | }
|
---|
695 | }
|
---|
696 | return XML_TOK_PARTIAL;
|
---|
697 | }
|
---|
698 |
|
---|
699 | /* ptr points to character following "<" */
|
---|
700 |
|
---|
701 | static int EXPATENTRY PREFIX(scanLt)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
702 | const char **nextTokPtr)
|
---|
703 | {
|
---|
704 | #ifdef XML_NS
|
---|
705 | int hadColon;
|
---|
706 | #endif
|
---|
707 | if (ptr == end)
|
---|
708 | return XML_TOK_PARTIAL;
|
---|
709 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
710 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
711 | case BT_EXCL:
|
---|
712 | if ((ptr += MINBPC(enc)) == end)
|
---|
713 | return XML_TOK_PARTIAL;
|
---|
714 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
715 | case BT_MINUS:
|
---|
716 | return PREFIX(scanComment)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
717 | case BT_LSQB:
|
---|
718 | return PREFIX(scanCdataSection)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
719 | }
|
---|
720 | *nextTokPtr = ptr;
|
---|
721 | return XML_TOK_INVALID;
|
---|
722 | case BT_QUEST:
|
---|
723 | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
724 | case BT_SOL:
|
---|
725 | return PREFIX(scanEndTag)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
726 | default:
|
---|
727 | *nextTokPtr = ptr;
|
---|
728 | return XML_TOK_INVALID;
|
---|
729 | }
|
---|
730 | #ifdef XML_NS
|
---|
731 | hadColon = 0;
|
---|
732 | #endif
|
---|
733 | /* we have a start-tag */
|
---|
734 | while (ptr != end) {
|
---|
735 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
736 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
737 | #ifdef XML_NS
|
---|
738 | case BT_COLON:
|
---|
739 | if (hadColon) {
|
---|
740 | *nextTokPtr = ptr;
|
---|
741 | return XML_TOK_INVALID;
|
---|
742 | }
|
---|
743 | hadColon = 1;
|
---|
744 | ptr += MINBPC(enc);
|
---|
745 | if (ptr == end)
|
---|
746 | return XML_TOK_PARTIAL;
|
---|
747 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
748 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
749 | default:
|
---|
750 | *nextTokPtr = ptr;
|
---|
751 | return XML_TOK_INVALID;
|
---|
752 | }
|
---|
753 | break;
|
---|
754 | #endif
|
---|
755 | case BT_S: case BT_CR: case BT_LF:
|
---|
756 | {
|
---|
757 | ptr += MINBPC(enc);
|
---|
758 | while (ptr != end) {
|
---|
759 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
760 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
761 | case BT_GT:
|
---|
762 | goto gt;
|
---|
763 | case BT_SOL:
|
---|
764 | goto sol;
|
---|
765 | case BT_S: case BT_CR: case BT_LF:
|
---|
766 | ptr += MINBPC(enc);
|
---|
767 | continue;
|
---|
768 | default:
|
---|
769 | *nextTokPtr = ptr;
|
---|
770 | return XML_TOK_INVALID;
|
---|
771 | }
|
---|
772 | return PREFIX(scanAtts)(enc, ptr, end, nextTokPtr);
|
---|
773 | }
|
---|
774 | return XML_TOK_PARTIAL;
|
---|
775 | }
|
---|
776 | case BT_GT:
|
---|
777 | gt:
|
---|
778 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
779 | return XML_TOK_START_TAG_NO_ATTS;
|
---|
780 | case BT_SOL:
|
---|
781 | sol:
|
---|
782 | ptr += MINBPC(enc);
|
---|
783 | if (ptr == end)
|
---|
784 | return XML_TOK_PARTIAL;
|
---|
785 | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
786 | *nextTokPtr = ptr;
|
---|
787 | return XML_TOK_INVALID;
|
---|
788 | }
|
---|
789 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
790 | return XML_TOK_EMPTY_ELEMENT_NO_ATTS;
|
---|
791 | default:
|
---|
792 | *nextTokPtr = ptr;
|
---|
793 | return XML_TOK_INVALID;
|
---|
794 | }
|
---|
795 | }
|
---|
796 | return XML_TOK_PARTIAL;
|
---|
797 | }
|
---|
798 |
|
---|
799 | static int EXPATENTRY PREFIX(contentTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
800 | const char **nextTokPtr)
|
---|
801 | {
|
---|
802 | if (ptr == end)
|
---|
803 | return XML_TOK_NONE;
|
---|
804 | if (MINBPC(enc) > 1) {
|
---|
805 | size_t n = end - ptr;
|
---|
806 | if (n & (MINBPC(enc) - 1)) {
|
---|
807 | n &= ~(MINBPC(enc) - 1);
|
---|
808 | if (n == 0)
|
---|
809 | return XML_TOK_PARTIAL;
|
---|
810 | end = ptr + n;
|
---|
811 | }
|
---|
812 | }
|
---|
813 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
814 | case BT_LT:
|
---|
815 | return PREFIX(scanLt)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
816 | case BT_AMP:
|
---|
817 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
818 | case BT_CR:
|
---|
819 | ptr += MINBPC(enc);
|
---|
820 | if (ptr == end)
|
---|
821 | return XML_TOK_TRAILING_CR;
|
---|
822 | if (BYTE_TYPE(enc, ptr) == BT_LF)
|
---|
823 | ptr += MINBPC(enc);
|
---|
824 | *nextTokPtr = ptr;
|
---|
825 | return XML_TOK_DATA_NEWLINE;
|
---|
826 | case BT_LF:
|
---|
827 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
828 | return XML_TOK_DATA_NEWLINE;
|
---|
829 | case BT_RSQB:
|
---|
830 | ptr += MINBPC(enc);
|
---|
831 | if (ptr == end)
|
---|
832 | return XML_TOK_TRAILING_RSQB;
|
---|
833 | if (!CHAR_MATCHES(enc, ptr, ASCII_RSQB))
|
---|
834 | break;
|
---|
835 | ptr += MINBPC(enc);
|
---|
836 | if (ptr == end)
|
---|
837 | return XML_TOK_TRAILING_RSQB;
|
---|
838 | if (!CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
839 | ptr -= MINBPC(enc);
|
---|
840 | break;
|
---|
841 | }
|
---|
842 | *nextTokPtr = ptr;
|
---|
843 | return XML_TOK_INVALID;
|
---|
844 | INVALID_CASES(ptr, nextTokPtr)
|
---|
845 | default:
|
---|
846 | ptr += MINBPC(enc);
|
---|
847 | break;
|
---|
848 | }
|
---|
849 | while (ptr != end) {
|
---|
850 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
851 | #define LEAD_CASE(n) \
|
---|
852 | case BT_LEAD ## n: \
|
---|
853 | if (end - ptr < n || IS_INVALID_CHAR(enc, ptr, n)) { \
|
---|
854 | *nextTokPtr = ptr; \
|
---|
855 | return XML_TOK_DATA_CHARS; \
|
---|
856 | } \
|
---|
857 | ptr += n; \
|
---|
858 | break;
|
---|
859 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
860 | #undef LEAD_CASE
|
---|
861 | case BT_RSQB:
|
---|
862 | if (ptr + MINBPC(enc) != end) {
|
---|
863 | if (!CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_RSQB)) {
|
---|
864 | ptr += MINBPC(enc);
|
---|
865 | break;
|
---|
866 | }
|
---|
867 | if (ptr + 2*MINBPC(enc) != end) {
|
---|
868 | if (!CHAR_MATCHES(enc, ptr + 2*MINBPC(enc), ASCII_GT)) {
|
---|
869 | ptr += MINBPC(enc);
|
---|
870 | break;
|
---|
871 | }
|
---|
872 | *nextTokPtr = ptr + 2*MINBPC(enc);
|
---|
873 | return XML_TOK_INVALID;
|
---|
874 | }
|
---|
875 | }
|
---|
876 | /* fall through */
|
---|
877 | case BT_AMP:
|
---|
878 | case BT_LT:
|
---|
879 | case BT_NONXML:
|
---|
880 | case BT_MALFORM:
|
---|
881 | case BT_TRAIL:
|
---|
882 | case BT_CR:
|
---|
883 | case BT_LF:
|
---|
884 | *nextTokPtr = ptr;
|
---|
885 | return XML_TOK_DATA_CHARS;
|
---|
886 | default:
|
---|
887 | ptr += MINBPC(enc);
|
---|
888 | break;
|
---|
889 | }
|
---|
890 | }
|
---|
891 | *nextTokPtr = ptr;
|
---|
892 | return XML_TOK_DATA_CHARS;
|
---|
893 | }
|
---|
894 |
|
---|
895 | /* ptr points to character following "%" */
|
---|
896 |
|
---|
897 | static int EXPATENTRY PREFIX(scanPercent)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
898 | const char **nextTokPtr)
|
---|
899 | {
|
---|
900 | if (ptr == end)
|
---|
901 | return XML_TOK_PARTIAL;
|
---|
902 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
903 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
904 | case BT_S: case BT_LF: case BT_CR: case BT_PERCNT:
|
---|
905 | *nextTokPtr = ptr;
|
---|
906 | return XML_TOK_PERCENT;
|
---|
907 | default:
|
---|
908 | *nextTokPtr = ptr;
|
---|
909 | return XML_TOK_INVALID;
|
---|
910 | }
|
---|
911 | while (ptr != end) {
|
---|
912 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
913 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
914 | case BT_SEMI:
|
---|
915 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
916 | return XML_TOK_PARAM_ENTITY_REF;
|
---|
917 | default:
|
---|
918 | *nextTokPtr = ptr;
|
---|
919 | return XML_TOK_INVALID;
|
---|
920 | }
|
---|
921 | }
|
---|
922 | return XML_TOK_PARTIAL;
|
---|
923 | }
|
---|
924 |
|
---|
925 | static int EXPATENTRY PREFIX(scanPoundName)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
926 | const char **nextTokPtr)
|
---|
927 | {
|
---|
928 | if (ptr == end)
|
---|
929 | return XML_TOK_PARTIAL;
|
---|
930 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
931 | CHECK_NMSTRT_CASES(enc, ptr, end, nextTokPtr)
|
---|
932 | default:
|
---|
933 | *nextTokPtr = ptr;
|
---|
934 | return XML_TOK_INVALID;
|
---|
935 | }
|
---|
936 | while (ptr != end) {
|
---|
937 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
938 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
939 | case BT_CR: case BT_LF: case BT_S:
|
---|
940 | case BT_RPAR: case BT_GT: case BT_PERCNT: case BT_VERBAR:
|
---|
941 | *nextTokPtr = ptr;
|
---|
942 | return XML_TOK_POUND_NAME;
|
---|
943 | default:
|
---|
944 | *nextTokPtr = ptr;
|
---|
945 | return XML_TOK_INVALID;
|
---|
946 | }
|
---|
947 | }
|
---|
948 | return -XML_TOK_POUND_NAME;
|
---|
949 | }
|
---|
950 |
|
---|
951 | static int EXPATENTRY PREFIX(scanLit)(int open, const ENCODING *enc,
|
---|
952 | const char *ptr, const char *end,
|
---|
953 | const char **nextTokPtr)
|
---|
954 | {
|
---|
955 | while (ptr != end) {
|
---|
956 | int t = BYTE_TYPE(enc, ptr);
|
---|
957 | switch (t) {
|
---|
958 | INVALID_CASES(ptr, nextTokPtr)
|
---|
959 | case BT_QUOT:
|
---|
960 | case BT_APOS:
|
---|
961 | ptr += MINBPC(enc);
|
---|
962 | if (t != open)
|
---|
963 | break;
|
---|
964 | if (ptr == end)
|
---|
965 | return -XML_TOK_LITERAL;
|
---|
966 | *nextTokPtr = ptr;
|
---|
967 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
968 | case BT_S: case BT_CR: case BT_LF:
|
---|
969 | case BT_GT: case BT_PERCNT: case BT_LSQB:
|
---|
970 | return XML_TOK_LITERAL;
|
---|
971 | default:
|
---|
972 | return XML_TOK_INVALID;
|
---|
973 | }
|
---|
974 | default:
|
---|
975 | ptr += MINBPC(enc);
|
---|
976 | break;
|
---|
977 | }
|
---|
978 | }
|
---|
979 | return XML_TOK_PARTIAL;
|
---|
980 | }
|
---|
981 |
|
---|
982 | static int EXPATENTRY PREFIX(prologTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
983 | const char **nextTokPtr)
|
---|
984 | {
|
---|
985 | int tok;
|
---|
986 | if (ptr == end)
|
---|
987 | return XML_TOK_NONE;
|
---|
988 | if (MINBPC(enc) > 1) {
|
---|
989 | size_t n = end - ptr;
|
---|
990 | if (n & (MINBPC(enc) - 1)) {
|
---|
991 | n &= ~(MINBPC(enc) - 1);
|
---|
992 | if (n == 0)
|
---|
993 | return XML_TOK_PARTIAL;
|
---|
994 | end = ptr + n;
|
---|
995 | }
|
---|
996 | }
|
---|
997 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
998 | case BT_QUOT:
|
---|
999 | return PREFIX(scanLit)(BT_QUOT, enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1000 | case BT_APOS:
|
---|
1001 | return PREFIX(scanLit)(BT_APOS, enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1002 | case BT_LT:
|
---|
1003 | {
|
---|
1004 | ptr += MINBPC(enc);
|
---|
1005 | if (ptr == end)
|
---|
1006 | return XML_TOK_PARTIAL;
|
---|
1007 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1008 | case BT_EXCL:
|
---|
1009 | return PREFIX(scanDecl)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1010 | case BT_QUEST:
|
---|
1011 | return PREFIX(scanPi)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1012 | case BT_NMSTRT:
|
---|
1013 | case BT_HEX:
|
---|
1014 | case BT_NONASCII:
|
---|
1015 | case BT_LEAD2:
|
---|
1016 | case BT_LEAD3:
|
---|
1017 | case BT_LEAD4:
|
---|
1018 | *nextTokPtr = ptr - MINBPC(enc);
|
---|
1019 | return XML_TOK_INSTANCE_START;
|
---|
1020 | }
|
---|
1021 | *nextTokPtr = ptr;
|
---|
1022 | return XML_TOK_INVALID;
|
---|
1023 | }
|
---|
1024 | case BT_CR:
|
---|
1025 | if (ptr + MINBPC(enc) == end)
|
---|
1026 | return -XML_TOK_PROLOG_S;
|
---|
1027 | /* fall through */
|
---|
1028 | case BT_S: case BT_LF:
|
---|
1029 | for (;;) {
|
---|
1030 | ptr += MINBPC(enc);
|
---|
1031 | if (ptr == end)
|
---|
1032 | break;
|
---|
1033 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1034 | case BT_S: case BT_LF:
|
---|
1035 | break;
|
---|
1036 | case BT_CR:
|
---|
1037 | /* don't split CR/LF pair */
|
---|
1038 | if (ptr + MINBPC(enc) != end)
|
---|
1039 | break;
|
---|
1040 | /* fall through */
|
---|
1041 | default:
|
---|
1042 | *nextTokPtr = ptr;
|
---|
1043 | return XML_TOK_PROLOG_S;
|
---|
1044 | }
|
---|
1045 | }
|
---|
1046 | *nextTokPtr = ptr;
|
---|
1047 | return XML_TOK_PROLOG_S;
|
---|
1048 | case BT_PERCNT:
|
---|
1049 | return PREFIX(scanPercent)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1050 | case BT_COMMA:
|
---|
1051 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1052 | return XML_TOK_COMMA;
|
---|
1053 | case BT_LSQB:
|
---|
1054 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1055 | return XML_TOK_OPEN_BRACKET;
|
---|
1056 | case BT_RSQB:
|
---|
1057 | ptr += MINBPC(enc);
|
---|
1058 | if (ptr == end)
|
---|
1059 | return -XML_TOK_CLOSE_BRACKET;
|
---|
1060 | if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
|
---|
1061 | if (ptr + MINBPC(enc) == end)
|
---|
1062 | return XML_TOK_PARTIAL;
|
---|
1063 | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_GT)) {
|
---|
1064 | *nextTokPtr = ptr + 2*MINBPC(enc);
|
---|
1065 | return XML_TOK_COND_SECT_CLOSE;
|
---|
1066 | }
|
---|
1067 | }
|
---|
1068 | *nextTokPtr = ptr;
|
---|
1069 | return XML_TOK_CLOSE_BRACKET;
|
---|
1070 | case BT_LPAR:
|
---|
1071 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1072 | return XML_TOK_OPEN_PAREN;
|
---|
1073 | case BT_RPAR:
|
---|
1074 | ptr += MINBPC(enc);
|
---|
1075 | if (ptr == end)
|
---|
1076 | return -XML_TOK_CLOSE_PAREN;
|
---|
1077 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1078 | case BT_AST:
|
---|
1079 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1080 | return XML_TOK_CLOSE_PAREN_ASTERISK;
|
---|
1081 | case BT_QUEST:
|
---|
1082 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1083 | return XML_TOK_CLOSE_PAREN_QUESTION;
|
---|
1084 | case BT_PLUS:
|
---|
1085 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1086 | return XML_TOK_CLOSE_PAREN_PLUS;
|
---|
1087 | case BT_CR: case BT_LF: case BT_S:
|
---|
1088 | case BT_GT: case BT_COMMA: case BT_VERBAR:
|
---|
1089 | case BT_RPAR:
|
---|
1090 | *nextTokPtr = ptr;
|
---|
1091 | return XML_TOK_CLOSE_PAREN;
|
---|
1092 | }
|
---|
1093 | *nextTokPtr = ptr;
|
---|
1094 | return XML_TOK_INVALID;
|
---|
1095 | case BT_VERBAR:
|
---|
1096 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1097 | return XML_TOK_OR;
|
---|
1098 | case BT_GT:
|
---|
1099 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1100 | return XML_TOK_DECL_CLOSE;
|
---|
1101 | case BT_NUM:
|
---|
1102 | return PREFIX(scanPoundName)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1103 | #define LEAD_CASE(n) \
|
---|
1104 | case BT_LEAD ## n: \
|
---|
1105 | if (end - ptr < n) \
|
---|
1106 | return XML_TOK_PARTIAL_CHAR; \
|
---|
1107 | if (IS_NMSTRT_CHAR(enc, ptr, n)) { \
|
---|
1108 | ptr += n; \
|
---|
1109 | tok = XML_TOK_NAME; \
|
---|
1110 | break; \
|
---|
1111 | } \
|
---|
1112 | if (IS_NAME_CHAR(enc, ptr, n)) { \
|
---|
1113 | ptr += n; \
|
---|
1114 | tok = XML_TOK_NMTOKEN; \
|
---|
1115 | break; \
|
---|
1116 | } \
|
---|
1117 | *nextTokPtr = ptr; \
|
---|
1118 | return XML_TOK_INVALID;
|
---|
1119 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1120 | #undef LEAD_CASE
|
---|
1121 | case BT_NMSTRT:
|
---|
1122 | case BT_HEX:
|
---|
1123 | tok = XML_TOK_NAME;
|
---|
1124 | ptr += MINBPC(enc);
|
---|
1125 | break;
|
---|
1126 | case BT_DIGIT:
|
---|
1127 | case BT_NAME:
|
---|
1128 | case BT_MINUS:
|
---|
1129 | #ifdef XML_NS
|
---|
1130 | case BT_COLON:
|
---|
1131 | #endif
|
---|
1132 | tok = XML_TOK_NMTOKEN;
|
---|
1133 | ptr += MINBPC(enc);
|
---|
1134 | break;
|
---|
1135 | case BT_NONASCII:
|
---|
1136 | if (IS_NMSTRT_CHAR_MINBPC(enc, ptr)) {
|
---|
1137 | ptr += MINBPC(enc);
|
---|
1138 | tok = XML_TOK_NAME;
|
---|
1139 | break;
|
---|
1140 | }
|
---|
1141 | if (IS_NAME_CHAR_MINBPC(enc, ptr)) {
|
---|
1142 | ptr += MINBPC(enc);
|
---|
1143 | tok = XML_TOK_NMTOKEN;
|
---|
1144 | break;
|
---|
1145 | }
|
---|
1146 | /* fall through */
|
---|
1147 | default:
|
---|
1148 | *nextTokPtr = ptr;
|
---|
1149 | return XML_TOK_INVALID;
|
---|
1150 | }
|
---|
1151 | while (ptr != end) {
|
---|
1152 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1153 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
1154 | case BT_GT: case BT_RPAR: case BT_COMMA:
|
---|
1155 | case BT_VERBAR: case BT_LSQB: case BT_PERCNT:
|
---|
1156 | case BT_S: case BT_CR: case BT_LF:
|
---|
1157 | *nextTokPtr = ptr;
|
---|
1158 | return tok;
|
---|
1159 | #ifdef XML_NS
|
---|
1160 | case BT_COLON:
|
---|
1161 | ptr += MINBPC(enc);
|
---|
1162 | switch (tok) {
|
---|
1163 | case XML_TOK_NAME:
|
---|
1164 | if (ptr == end)
|
---|
1165 | return XML_TOK_PARTIAL;
|
---|
1166 | tok = XML_TOK_PREFIXED_NAME;
|
---|
1167 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1168 | CHECK_NAME_CASES(enc, ptr, end, nextTokPtr)
|
---|
1169 | default:
|
---|
1170 | tok = XML_TOK_NMTOKEN;
|
---|
1171 | break;
|
---|
1172 | }
|
---|
1173 | break;
|
---|
1174 | case XML_TOK_PREFIXED_NAME:
|
---|
1175 | tok = XML_TOK_NMTOKEN;
|
---|
1176 | break;
|
---|
1177 | }
|
---|
1178 | break;
|
---|
1179 | #endif
|
---|
1180 | case BT_PLUS:
|
---|
1181 | if (tok == XML_TOK_NMTOKEN) {
|
---|
1182 | *nextTokPtr = ptr;
|
---|
1183 | return XML_TOK_INVALID;
|
---|
1184 | }
|
---|
1185 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1186 | return XML_TOK_NAME_PLUS;
|
---|
1187 | case BT_AST:
|
---|
1188 | if (tok == XML_TOK_NMTOKEN) {
|
---|
1189 | *nextTokPtr = ptr;
|
---|
1190 | return XML_TOK_INVALID;
|
---|
1191 | }
|
---|
1192 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1193 | return XML_TOK_NAME_ASTERISK;
|
---|
1194 | case BT_QUEST:
|
---|
1195 | if (tok == XML_TOK_NMTOKEN) {
|
---|
1196 | *nextTokPtr = ptr;
|
---|
1197 | return XML_TOK_INVALID;
|
---|
1198 | }
|
---|
1199 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1200 | return XML_TOK_NAME_QUESTION;
|
---|
1201 | default:
|
---|
1202 | *nextTokPtr = ptr;
|
---|
1203 | return XML_TOK_INVALID;
|
---|
1204 | }
|
---|
1205 | }
|
---|
1206 | return -tok;
|
---|
1207 | }
|
---|
1208 |
|
---|
1209 | static int EXPATENTRY PREFIX(attributeValueTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
1210 | const char **nextTokPtr)
|
---|
1211 | {
|
---|
1212 | const char *start;
|
---|
1213 | if (ptr == end)
|
---|
1214 | return XML_TOK_NONE;
|
---|
1215 | start = ptr;
|
---|
1216 | while (ptr != end) {
|
---|
1217 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1218 | #define LEAD_CASE(n) \
|
---|
1219 | case BT_LEAD ## n: ptr += n; break;
|
---|
1220 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1221 | #undef LEAD_CASE
|
---|
1222 | case BT_AMP:
|
---|
1223 | if (ptr == start)
|
---|
1224 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1225 | *nextTokPtr = ptr;
|
---|
1226 | return XML_TOK_DATA_CHARS;
|
---|
1227 | case BT_LT:
|
---|
1228 | /* this is for inside entity references */
|
---|
1229 | *nextTokPtr = ptr;
|
---|
1230 | return XML_TOK_INVALID;
|
---|
1231 | case BT_LF:
|
---|
1232 | if (ptr == start) {
|
---|
1233 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1234 | return XML_TOK_DATA_NEWLINE;
|
---|
1235 | }
|
---|
1236 | *nextTokPtr = ptr;
|
---|
1237 | return XML_TOK_DATA_CHARS;
|
---|
1238 | case BT_CR:
|
---|
1239 | if (ptr == start) {
|
---|
1240 | ptr += MINBPC(enc);
|
---|
1241 | if (ptr == end)
|
---|
1242 | return XML_TOK_TRAILING_CR;
|
---|
1243 | if (BYTE_TYPE(enc, ptr) == BT_LF)
|
---|
1244 | ptr += MINBPC(enc);
|
---|
1245 | *nextTokPtr = ptr;
|
---|
1246 | return XML_TOK_DATA_NEWLINE;
|
---|
1247 | }
|
---|
1248 | *nextTokPtr = ptr;
|
---|
1249 | return XML_TOK_DATA_CHARS;
|
---|
1250 | case BT_S:
|
---|
1251 | if (ptr == start) {
|
---|
1252 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1253 | return XML_TOK_ATTRIBUTE_VALUE_S;
|
---|
1254 | }
|
---|
1255 | *nextTokPtr = ptr;
|
---|
1256 | return XML_TOK_DATA_CHARS;
|
---|
1257 | default:
|
---|
1258 | ptr += MINBPC(enc);
|
---|
1259 | break;
|
---|
1260 | }
|
---|
1261 | }
|
---|
1262 | *nextTokPtr = ptr;
|
---|
1263 | return XML_TOK_DATA_CHARS;
|
---|
1264 | }
|
---|
1265 |
|
---|
1266 | static int EXPATENTRY PREFIX(entityValueTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
1267 | const char **nextTokPtr)
|
---|
1268 | {
|
---|
1269 | const char *start;
|
---|
1270 | if (ptr == end)
|
---|
1271 | return XML_TOK_NONE;
|
---|
1272 | start = ptr;
|
---|
1273 | while (ptr != end) {
|
---|
1274 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1275 | #define LEAD_CASE(n) \
|
---|
1276 | case BT_LEAD ## n: ptr += n; break;
|
---|
1277 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1278 | #undef LEAD_CASE
|
---|
1279 | case BT_AMP:
|
---|
1280 | if (ptr == start)
|
---|
1281 | return PREFIX(scanRef)(enc, ptr + MINBPC(enc), end, nextTokPtr);
|
---|
1282 | *nextTokPtr = ptr;
|
---|
1283 | return XML_TOK_DATA_CHARS;
|
---|
1284 | case BT_PERCNT:
|
---|
1285 | if (ptr == start) {
|
---|
1286 | int tok = PREFIX(scanPercent)(enc, ptr + MINBPC(enc),
|
---|
1287 | end, nextTokPtr);
|
---|
1288 | return (tok == XML_TOK_PERCENT) ? XML_TOK_INVALID : tok;
|
---|
1289 | }
|
---|
1290 | *nextTokPtr = ptr;
|
---|
1291 | return XML_TOK_DATA_CHARS;
|
---|
1292 | case BT_LF:
|
---|
1293 | if (ptr == start) {
|
---|
1294 | *nextTokPtr = ptr + MINBPC(enc);
|
---|
1295 | return XML_TOK_DATA_NEWLINE;
|
---|
1296 | }
|
---|
1297 | *nextTokPtr = ptr;
|
---|
1298 | return XML_TOK_DATA_CHARS;
|
---|
1299 | case BT_CR:
|
---|
1300 | if (ptr == start) {
|
---|
1301 | ptr += MINBPC(enc);
|
---|
1302 | if (ptr == end)
|
---|
1303 | return XML_TOK_TRAILING_CR;
|
---|
1304 | if (BYTE_TYPE(enc, ptr) == BT_LF)
|
---|
1305 | ptr += MINBPC(enc);
|
---|
1306 | *nextTokPtr = ptr;
|
---|
1307 | return XML_TOK_DATA_NEWLINE;
|
---|
1308 | }
|
---|
1309 | *nextTokPtr = ptr;
|
---|
1310 | return XML_TOK_DATA_CHARS;
|
---|
1311 | default:
|
---|
1312 | ptr += MINBPC(enc);
|
---|
1313 | break;
|
---|
1314 | }
|
---|
1315 | }
|
---|
1316 | *nextTokPtr = ptr;
|
---|
1317 | return XML_TOK_DATA_CHARS;
|
---|
1318 | }
|
---|
1319 |
|
---|
1320 | #ifdef XML_DTD
|
---|
1321 |
|
---|
1322 | static int EXPATENTRY PREFIX(ignoreSectionTok)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
1323 | const char **nextTokPtr)
|
---|
1324 | {
|
---|
1325 | int level = 0;
|
---|
1326 | if (MINBPC(enc) > 1) {
|
---|
1327 | size_t n = end - ptr;
|
---|
1328 | if (n & (MINBPC(enc) - 1)) {
|
---|
1329 | n &= ~(MINBPC(enc) - 1);
|
---|
1330 | end = ptr + n;
|
---|
1331 | }
|
---|
1332 | }
|
---|
1333 | while (ptr != end) {
|
---|
1334 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1335 | INVALID_CASES(ptr, nextTokPtr)
|
---|
1336 | case BT_LT:
|
---|
1337 | if ((ptr += MINBPC(enc)) == end)
|
---|
1338 | return XML_TOK_PARTIAL;
|
---|
1339 | if (CHAR_MATCHES(enc, ptr, ASCII_EXCL)) {
|
---|
1340 | if ((ptr += MINBPC(enc)) == end)
|
---|
1341 | return XML_TOK_PARTIAL;
|
---|
1342 | if (CHAR_MATCHES(enc, ptr, ASCII_LSQB)) {
|
---|
1343 | ++level;
|
---|
1344 | ptr += MINBPC(enc);
|
---|
1345 | }
|
---|
1346 | }
|
---|
1347 | break;
|
---|
1348 | case BT_RSQB:
|
---|
1349 | if ((ptr += MINBPC(enc)) == end)
|
---|
1350 | return XML_TOK_PARTIAL;
|
---|
1351 | if (CHAR_MATCHES(enc, ptr, ASCII_RSQB)) {
|
---|
1352 | if ((ptr += MINBPC(enc)) == end)
|
---|
1353 | return XML_TOK_PARTIAL;
|
---|
1354 | if (CHAR_MATCHES(enc, ptr, ASCII_GT)) {
|
---|
1355 | ptr += MINBPC(enc);
|
---|
1356 | if (level == 0) {
|
---|
1357 | *nextTokPtr = ptr;
|
---|
1358 | return XML_TOK_IGNORE_SECT;
|
---|
1359 | }
|
---|
1360 | --level;
|
---|
1361 | }
|
---|
1362 | }
|
---|
1363 | break;
|
---|
1364 | default:
|
---|
1365 | ptr += MINBPC(enc);
|
---|
1366 | break;
|
---|
1367 | }
|
---|
1368 | }
|
---|
1369 | return XML_TOK_PARTIAL;
|
---|
1370 | }
|
---|
1371 |
|
---|
1372 | #endif /* XML_DTD */
|
---|
1373 |
|
---|
1374 | static int EXPATENTRY PREFIX(isPublicId)(const ENCODING *enc, const char *ptr, const char *end,
|
---|
1375 | const char **badPtr)
|
---|
1376 | {
|
---|
1377 | ptr += MINBPC(enc);
|
---|
1378 | end -= MINBPC(enc);
|
---|
1379 | for (; ptr != end; ptr += MINBPC(enc)) {
|
---|
1380 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1381 | case BT_DIGIT:
|
---|
1382 | case BT_HEX:
|
---|
1383 | case BT_MINUS:
|
---|
1384 | case BT_APOS:
|
---|
1385 | case BT_LPAR:
|
---|
1386 | case BT_RPAR:
|
---|
1387 | case BT_PLUS:
|
---|
1388 | case BT_COMMA:
|
---|
1389 | case BT_SOL:
|
---|
1390 | case BT_EQUALS:
|
---|
1391 | case BT_QUEST:
|
---|
1392 | case BT_CR:
|
---|
1393 | case BT_LF:
|
---|
1394 | case BT_SEMI:
|
---|
1395 | case BT_EXCL:
|
---|
1396 | case BT_AST:
|
---|
1397 | case BT_PERCNT:
|
---|
1398 | case BT_NUM:
|
---|
1399 | #ifdef XML_NS
|
---|
1400 | case BT_COLON:
|
---|
1401 | #endif
|
---|
1402 | break;
|
---|
1403 | case BT_S:
|
---|
1404 | if (CHAR_MATCHES(enc, ptr, ASCII_TAB)) {
|
---|
1405 | *badPtr = ptr;
|
---|
1406 | return 0;
|
---|
1407 | }
|
---|
1408 | break;
|
---|
1409 | case BT_NAME:
|
---|
1410 | case BT_NMSTRT:
|
---|
1411 | if (!(BYTE_TO_ASCII(enc, ptr) & ~0x7f))
|
---|
1412 | break;
|
---|
1413 | default:
|
---|
1414 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
1415 | case 0x24: /* $ */
|
---|
1416 | case 0x40: /* @ */
|
---|
1417 | break;
|
---|
1418 | default:
|
---|
1419 | *badPtr = ptr;
|
---|
1420 | return 0;
|
---|
1421 | }
|
---|
1422 | break;
|
---|
1423 | }
|
---|
1424 | }
|
---|
1425 | return 1;
|
---|
1426 | }
|
---|
1427 |
|
---|
1428 | /* This must only be called for a well-formed start-tag or empty element tag.
|
---|
1429 | Returns the number of attributes. Pointers to the first attsMax attributes
|
---|
1430 | are stored in atts. */
|
---|
1431 |
|
---|
1432 | static int EXPATENTRY PREFIX(getAtts)(const ENCODING *enc, const char *ptr,
|
---|
1433 | int attsMax, ATTRIBUTE *atts)
|
---|
1434 | {
|
---|
1435 | enum { other, inName, inValue } state = inName;
|
---|
1436 | int nAtts = 0;
|
---|
1437 | int open = 0; /* defined when state == inValue;
|
---|
1438 | initialization just to shut up compilers */
|
---|
1439 |
|
---|
1440 | for (ptr += MINBPC(enc);; ptr += MINBPC(enc)) {
|
---|
1441 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1442 | #define START_NAME \
|
---|
1443 | if (state == other) { \
|
---|
1444 | if (nAtts < attsMax) { \
|
---|
1445 | atts[nAtts].name = ptr; \
|
---|
1446 | atts[nAtts].normalized = 1; \
|
---|
1447 | } \
|
---|
1448 | state = inName; \
|
---|
1449 | }
|
---|
1450 | #define LEAD_CASE(n) \
|
---|
1451 | case BT_LEAD ## n: START_NAME ptr += (n - MINBPC(enc)); break;
|
---|
1452 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1453 | #undef LEAD_CASE
|
---|
1454 | case BT_NONASCII:
|
---|
1455 | case BT_NMSTRT:
|
---|
1456 | case BT_HEX:
|
---|
1457 | START_NAME
|
---|
1458 | break;
|
---|
1459 | #undef START_NAME
|
---|
1460 | case BT_QUOT:
|
---|
1461 | if (state != inValue) {
|
---|
1462 | if (nAtts < attsMax)
|
---|
1463 | atts[nAtts].valuePtr = ptr + MINBPC(enc);
|
---|
1464 | state = inValue;
|
---|
1465 | open = BT_QUOT;
|
---|
1466 | }
|
---|
1467 | else if (open == BT_QUOT) {
|
---|
1468 | state = other;
|
---|
1469 | if (nAtts < attsMax)
|
---|
1470 | atts[nAtts].valueEnd = ptr;
|
---|
1471 | nAtts++;
|
---|
1472 | }
|
---|
1473 | break;
|
---|
1474 | case BT_APOS:
|
---|
1475 | if (state != inValue) {
|
---|
1476 | if (nAtts < attsMax)
|
---|
1477 | atts[nAtts].valuePtr = ptr + MINBPC(enc);
|
---|
1478 | state = inValue;
|
---|
1479 | open = BT_APOS;
|
---|
1480 | }
|
---|
1481 | else if (open == BT_APOS) {
|
---|
1482 | state = other;
|
---|
1483 | if (nAtts < attsMax)
|
---|
1484 | atts[nAtts].valueEnd = ptr;
|
---|
1485 | nAtts++;
|
---|
1486 | }
|
---|
1487 | break;
|
---|
1488 | case BT_AMP:
|
---|
1489 | if (nAtts < attsMax)
|
---|
1490 | atts[nAtts].normalized = 0;
|
---|
1491 | break;
|
---|
1492 | case BT_S:
|
---|
1493 | if (state == inName)
|
---|
1494 | state = other;
|
---|
1495 | else if (state == inValue
|
---|
1496 | && nAtts < attsMax
|
---|
1497 | && atts[nAtts].normalized
|
---|
1498 | && (ptr == atts[nAtts].valuePtr
|
---|
1499 | || BYTE_TO_ASCII(enc, ptr) != ASCII_SPACE
|
---|
1500 | || BYTE_TO_ASCII(enc, ptr + MINBPC(enc)) == ASCII_SPACE
|
---|
1501 | || BYTE_TYPE(enc, ptr + MINBPC(enc)) == open))
|
---|
1502 | atts[nAtts].normalized = 0;
|
---|
1503 | break;
|
---|
1504 | case BT_CR: case BT_LF:
|
---|
1505 | /* This case ensures that the first attribute name is counted
|
---|
1506 | Apart from that we could just change state on the quote. */
|
---|
1507 | if (state == inName)
|
---|
1508 | state = other;
|
---|
1509 | else if (state == inValue && nAtts < attsMax)
|
---|
1510 | atts[nAtts].normalized = 0;
|
---|
1511 | break;
|
---|
1512 | case BT_GT:
|
---|
1513 | case BT_SOL:
|
---|
1514 | if (state != inValue)
|
---|
1515 | return nAtts;
|
---|
1516 | break;
|
---|
1517 | default:
|
---|
1518 | break;
|
---|
1519 | }
|
---|
1520 | }
|
---|
1521 | /* not reached */
|
---|
1522 | }
|
---|
1523 |
|
---|
1524 | static int EXPATENTRY PREFIX(charRefNumber)(const ENCODING *enc, const char *ptr)
|
---|
1525 | {
|
---|
1526 | int result = 0;
|
---|
1527 | /* skip &# */
|
---|
1528 | ptr += 2*MINBPC(enc);
|
---|
1529 | if (CHAR_MATCHES(enc, ptr, ASCII_x)) {
|
---|
1530 | for (ptr += MINBPC(enc); !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
|
---|
1531 | int c = BYTE_TO_ASCII(enc, ptr);
|
---|
1532 | switch (c) {
|
---|
1533 | case ASCII_0: case ASCII_1: case ASCII_2: case ASCII_3: case ASCII_4:
|
---|
1534 | case ASCII_5: case ASCII_6: case ASCII_7: case ASCII_8: case ASCII_9:
|
---|
1535 | result <<= 4;
|
---|
1536 | result |= (c - ASCII_0);
|
---|
1537 | break;
|
---|
1538 | case ASCII_A: case ASCII_B: case ASCII_C: case ASCII_D: case ASCII_E: case ASCII_F:
|
---|
1539 | result <<= 4;
|
---|
1540 | result += 10 + (c - ASCII_A);
|
---|
1541 | break;
|
---|
1542 | case ASCII_a: case ASCII_b: case ASCII_c: case ASCII_d: case ASCII_e: case ASCII_f:
|
---|
1543 | result <<= 4;
|
---|
1544 | result += 10 + (c - ASCII_a);
|
---|
1545 | break;
|
---|
1546 | }
|
---|
1547 | if (result >= 0x110000)
|
---|
1548 | return -1;
|
---|
1549 | }
|
---|
1550 | }
|
---|
1551 | else {
|
---|
1552 | for (; !CHAR_MATCHES(enc, ptr, ASCII_SEMI); ptr += MINBPC(enc)) {
|
---|
1553 | int c = BYTE_TO_ASCII(enc, ptr);
|
---|
1554 | result *= 10;
|
---|
1555 | result += (c - ASCII_0);
|
---|
1556 | if (result >= 0x110000)
|
---|
1557 | return -1;
|
---|
1558 | }
|
---|
1559 | }
|
---|
1560 | return checkCharRefNumber(result);
|
---|
1561 | }
|
---|
1562 |
|
---|
1563 | static int EXPATENTRY PREFIX(predefinedEntityName)(const ENCODING *enc, const char *ptr, const char *end)
|
---|
1564 | {
|
---|
1565 | switch ((end - ptr)/MINBPC(enc)) {
|
---|
1566 | case 2:
|
---|
1567 | if (CHAR_MATCHES(enc, ptr + MINBPC(enc), ASCII_t)) {
|
---|
1568 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
1569 | case ASCII_l:
|
---|
1570 | return ASCII_LT;
|
---|
1571 | case ASCII_g:
|
---|
1572 | return ASCII_GT;
|
---|
1573 | }
|
---|
1574 | }
|
---|
1575 | break;
|
---|
1576 | case 3:
|
---|
1577 | if (CHAR_MATCHES(enc, ptr, ASCII_a)) {
|
---|
1578 | ptr += MINBPC(enc);
|
---|
1579 | if (CHAR_MATCHES(enc, ptr, ASCII_m)) {
|
---|
1580 | ptr += MINBPC(enc);
|
---|
1581 | if (CHAR_MATCHES(enc, ptr, ASCII_p))
|
---|
1582 | return ASCII_AMP;
|
---|
1583 | }
|
---|
1584 | }
|
---|
1585 | break;
|
---|
1586 | case 4:
|
---|
1587 | switch (BYTE_TO_ASCII(enc, ptr)) {
|
---|
1588 | case ASCII_q:
|
---|
1589 | ptr += MINBPC(enc);
|
---|
1590 | if (CHAR_MATCHES(enc, ptr, ASCII_u)) {
|
---|
1591 | ptr += MINBPC(enc);
|
---|
1592 | if (CHAR_MATCHES(enc, ptr, ASCII_o)) {
|
---|
1593 | ptr += MINBPC(enc);
|
---|
1594 | if (CHAR_MATCHES(enc, ptr, ASCII_t))
|
---|
1595 | return ASCII_QUOT;
|
---|
1596 | }
|
---|
1597 | }
|
---|
1598 | break;
|
---|
1599 | case ASCII_a:
|
---|
1600 | ptr += MINBPC(enc);
|
---|
1601 | if (CHAR_MATCHES(enc, ptr, ASCII_p)) {
|
---|
1602 | ptr += MINBPC(enc);
|
---|
1603 | if (CHAR_MATCHES(enc, ptr, ASCII_o)) {
|
---|
1604 | ptr += MINBPC(enc);
|
---|
1605 | if (CHAR_MATCHES(enc, ptr, ASCII_s))
|
---|
1606 | return ASCII_APOS;
|
---|
1607 | }
|
---|
1608 | }
|
---|
1609 | break;
|
---|
1610 | }
|
---|
1611 | }
|
---|
1612 | return 0;
|
---|
1613 | }
|
---|
1614 |
|
---|
1615 | static int EXPATENTRY PREFIX(sameName)(const ENCODING *enc, const char *ptr1, const char *ptr2)
|
---|
1616 | {
|
---|
1617 | for (;;) {
|
---|
1618 | switch (BYTE_TYPE(enc, ptr1)) {
|
---|
1619 | #define LEAD_CASE(n) \
|
---|
1620 | case BT_LEAD ## n: \
|
---|
1621 | if (*ptr1++ != *ptr2++) \
|
---|
1622 | return 0;
|
---|
1623 | LEAD_CASE(4) LEAD_CASE(3) LEAD_CASE(2)
|
---|
1624 | #undef LEAD_CASE
|
---|
1625 | /* fall through */
|
---|
1626 | if (*ptr1++ != *ptr2++)
|
---|
1627 | return 0;
|
---|
1628 | break;
|
---|
1629 | case BT_NONASCII:
|
---|
1630 | case BT_NMSTRT:
|
---|
1631 | #ifdef XML_NS
|
---|
1632 | case BT_COLON:
|
---|
1633 | #endif
|
---|
1634 | case BT_HEX:
|
---|
1635 | case BT_DIGIT:
|
---|
1636 | case BT_NAME:
|
---|
1637 | case BT_MINUS:
|
---|
1638 | if (*ptr2++ != *ptr1++)
|
---|
1639 | return 0;
|
---|
1640 | if (MINBPC(enc) > 1) {
|
---|
1641 | if (*ptr2++ != *ptr1++)
|
---|
1642 | return 0;
|
---|
1643 | if (MINBPC(enc) > 2) {
|
---|
1644 | if (*ptr2++ != *ptr1++)
|
---|
1645 | return 0;
|
---|
1646 | if (MINBPC(enc) > 3) {
|
---|
1647 | if (*ptr2++ != *ptr1++)
|
---|
1648 | return 0;
|
---|
1649 | }
|
---|
1650 | }
|
---|
1651 | }
|
---|
1652 | break;
|
---|
1653 | default:
|
---|
1654 | if (MINBPC(enc) == 1 && *ptr1 == *ptr2)
|
---|
1655 | return 1;
|
---|
1656 | switch (BYTE_TYPE(enc, ptr2)) {
|
---|
1657 | case BT_LEAD2:
|
---|
1658 | case BT_LEAD3:
|
---|
1659 | case BT_LEAD4:
|
---|
1660 | case BT_NONASCII:
|
---|
1661 | case BT_NMSTRT:
|
---|
1662 | #ifdef XML_NS
|
---|
1663 | case BT_COLON:
|
---|
1664 | #endif
|
---|
1665 | case BT_HEX:
|
---|
1666 | case BT_DIGIT:
|
---|
1667 | case BT_NAME:
|
---|
1668 | case BT_MINUS:
|
---|
1669 | return 0;
|
---|
1670 | default:
|
---|
1671 | return 1;
|
---|
1672 | }
|
---|
1673 | }
|
---|
1674 | }
|
---|
1675 | /* not reached */
|
---|
1676 | }
|
---|
1677 |
|
---|
1678 | static int EXPATENTRY PREFIX(nameMatchesAscii)(const ENCODING *enc, const char *ptr1,
|
---|
1679 | const char *end1, const char *ptr2)
|
---|
1680 | {
|
---|
1681 | for (; *ptr2; ptr1 += MINBPC(enc), ptr2++) {
|
---|
1682 | if (ptr1 == end1)
|
---|
1683 | return 0;
|
---|
1684 | if (!CHAR_MATCHES(enc, ptr1, *ptr2))
|
---|
1685 | return 0;
|
---|
1686 | }
|
---|
1687 | return ptr1 == end1;
|
---|
1688 | }
|
---|
1689 |
|
---|
1690 | static int EXPATENTRY PREFIX(nameLength)(const ENCODING *enc, const char *ptr)
|
---|
1691 | {
|
---|
1692 | const char *start = ptr;
|
---|
1693 | for (;;) {
|
---|
1694 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1695 | #define LEAD_CASE(n) \
|
---|
1696 | case BT_LEAD ## n: ptr += n; break;
|
---|
1697 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1698 | #undef LEAD_CASE
|
---|
1699 | case BT_NONASCII:
|
---|
1700 | case BT_NMSTRT:
|
---|
1701 | #ifdef XML_NS
|
---|
1702 | case BT_COLON:
|
---|
1703 | #endif
|
---|
1704 | case BT_HEX:
|
---|
1705 | case BT_DIGIT:
|
---|
1706 | case BT_NAME:
|
---|
1707 | case BT_MINUS:
|
---|
1708 | ptr += MINBPC(enc);
|
---|
1709 | break;
|
---|
1710 | default:
|
---|
1711 | return ptr - start;
|
---|
1712 | }
|
---|
1713 | }
|
---|
1714 | }
|
---|
1715 |
|
---|
1716 | static const char* EXPATENTRY PREFIX(skipS)(const ENCODING *enc, const char *ptr)
|
---|
1717 | {
|
---|
1718 | for (;;) {
|
---|
1719 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1720 | case BT_LF:
|
---|
1721 | case BT_CR:
|
---|
1722 | case BT_S:
|
---|
1723 | ptr += MINBPC(enc);
|
---|
1724 | break;
|
---|
1725 | default:
|
---|
1726 | return ptr;
|
---|
1727 | }
|
---|
1728 | }
|
---|
1729 | }
|
---|
1730 |
|
---|
1731 | static void EXPATENTRY PREFIX(updatePosition)(const ENCODING *enc,
|
---|
1732 | const char *ptr,
|
---|
1733 | const char *end,
|
---|
1734 | POSITION *pos)
|
---|
1735 | {
|
---|
1736 | while (ptr != end) {
|
---|
1737 | switch (BYTE_TYPE(enc, ptr)) {
|
---|
1738 | #define LEAD_CASE(n) \
|
---|
1739 | case BT_LEAD ## n: \
|
---|
1740 | ptr += n; \
|
---|
1741 | break;
|
---|
1742 | LEAD_CASE(2) LEAD_CASE(3) LEAD_CASE(4)
|
---|
1743 | #undef LEAD_CASE
|
---|
1744 | case BT_LF:
|
---|
1745 | pos->columnNumber = (unsigned)-1;
|
---|
1746 | pos->lineNumber++;
|
---|
1747 | ptr += MINBPC(enc);
|
---|
1748 | break;
|
---|
1749 | case BT_CR:
|
---|
1750 | pos->lineNumber++;
|
---|
1751 | ptr += MINBPC(enc);
|
---|
1752 | if (ptr != end && BYTE_TYPE(enc, ptr) == BT_LF)
|
---|
1753 | ptr += MINBPC(enc);
|
---|
1754 | pos->columnNumber = (unsigned)-1;
|
---|
1755 | break;
|
---|
1756 | default:
|
---|
1757 | ptr += MINBPC(enc);
|
---|
1758 | break;
|
---|
1759 | }
|
---|
1760 | pos->columnNumber++;
|
---|
1761 | }
|
---|
1762 | }
|
---|
1763 |
|
---|
1764 | #undef DO_LEAD_CASE
|
---|
1765 | #undef MULTIBYTE_CASES
|
---|
1766 | #undef INVALID_CASES
|
---|
1767 | #undef CHECK_NAME_CASE
|
---|
1768 | #undef CHECK_NAME_CASES
|
---|
1769 | #undef CHECK_NMSTRT_CASE
|
---|
1770 | #undef CHECK_NMSTRT_CASES
|
---|