txml.c - webdump - [FORK] git://git.codemadness.org/webdump
HTML git clone git://git.z3bra.org/webdump.git
DIR Log
DIR Files
DIR Refs
DIR README
DIR LICENSE
---
txml.c (10998B)
---
1 #include <sys/types.h>
2
3 #include <ctype.h>
4 #include <errno.h>
5 #include <limits.h>
6 #include <stdio.h>
7 #include <stdlib.h>
8 #include <string.h>
9
10 #include "xml.h"
11
12 static void
13 xml_parseattrs(XMLParser *x)
14 {
15 size_t namelen = 0, valuelen;
16 int c, endsep, endname = 0, valuestart = 0;
17
18 while ((c = GETNEXT()) != EOF) {
19 if (isspace(c)) {
20 if (namelen)
21 endname = 1;
22 continue;
23 } else if (c == '?')
24 ; /* ignore */
25 else if (c == '=') {
26 x->name[namelen] = '\0';
27 valuestart = 1;
28 endname = 1;
29 } else if (namelen && ((endname && !valuestart && isalpha(c)) || (c == '>' || c == '/'))) {
30 /* attribute without value */
31 x->name[namelen] = '\0';
32 if (x->xmlattrstart)
33 x->xmlattrstart(x, x->tag, x->taglen, x->name, namelen);
34 if (x->xmlattr)
35 x->xmlattr(x, x->tag, x->taglen, x->name, namelen, "", 0);
36 if (x->xmlattrend)
37 x->xmlattrend(x, x->tag, x->taglen, x->name, namelen);
38 endname = 0;
39 x->name[0] = c;
40 namelen = 1;
41 } else if (namelen && valuestart) {
42 /* attribute with value */
43 if (x->xmlattrstart)
44 x->xmlattrstart(x, x->tag, x->taglen, x->name, namelen);
45
46 valuelen = 0;
47 if (c == '\'' || c == '"') {
48 endsep = c;
49 } else {
50 endsep = ' '; /* isspace() */
51 goto startvalue;
52 }
53
54 while ((c = GETNEXT()) != EOF) {
55 startvalue:
56 if (c == '&') { /* entities */
57 x->data[valuelen] = '\0';
58 /* call data function with data before entity if there is data */
59 if (valuelen && x->xmlattr)
60 x->xmlattr(x, x->tag, x->taglen, x->name, namelen, x->data, valuelen);
61 x->data[0] = c;
62 valuelen = 1;
63 while ((c = GETNEXT()) != EOF) {
64 if (c == endsep || (endsep == ' ' && (c == '>' || isspace(c))))
65 break;
66 if (valuelen < sizeof(x->data) - 1)
67 x->data[valuelen++] = c;
68 else {
69 /* entity too long for buffer, handle as normal data */
70 x->data[valuelen] = '\0';
71 if (x->xmlattr)
72 x->xmlattr(x, x->tag, x->taglen, x->name, namelen, x->data, valuelen);
73 x->data[0] = c;
74 valuelen = 1;
75 break;
76 }
77 if (c == ';') {
78 x->data[valuelen] = '\0';
79 if (x->xmlattrentity)
80 x->xmlattrentity(x, x->tag, x->taglen, x->name, namelen, x->data, valuelen);
81 valuelen = 0;
82 break;
83 }
84 }
85 } else if (c != endsep && !(endsep == ' ' && (c == '>' || isspace(c)))) {
86 if (valuelen < sizeof(x->data) - 1) {
87 x->data[valuelen++] = c;
88 } else {
89 x->data[valuelen] = '\0';
90 if (x->xmlattr)
91 x->xmlattr(x, x->tag, x->taglen, x->name, namelen, x->data, valuelen);
92 x->data[0] = c;
93 valuelen = 1;
94 }
95 }
96 if (c == endsep || (endsep == ' ' && (c == '>' || isspace(c)))) {
97 x->data[valuelen] = '\0';
98 if (x->xmlattr)
99 x->xmlattr(x, x->tag, x->taglen, x->name, namelen, x->data, valuelen);
100 if (x->xmlattrend)
101 x->xmlattrend(x, x->tag, x->taglen, x->name, namelen);
102 break;
103 }
104 }
105 namelen = endname = valuestart = 0;
106 } else if (namelen < sizeof(x->name) - 1) {
107 x->name[namelen++] = c;
108 }
109 if (c == '>') {
110 break;
111 } else if (c == '/') {
112 x->isshorttag = 1;
113 x->name[0] = '\0';
114 namelen = 0;
115 }
116 }
117 }
118
119 static void
120 xml_parsecomment(XMLParser *x)
121 {
122 size_t datalen = 0, i = 0;
123 int c;
124
125 if (x->xmlcommentstart)
126 x->xmlcommentstart(x);
127 while ((c = GETNEXT()) != EOF) {
128 if (c == '-' || c == '>') {
129 if (x->xmlcomment && datalen) {
130 x->data[datalen] = '\0';
131 x->xmlcomment(x, x->data, datalen);
132 datalen = 0;
133 }
134 }
135
136 if (c == '-') {
137 if (++i > 2) {
138 if (x->xmlcomment)
139 for (; i > 2; i--)
140 x->xmlcomment(x, "-", 1);
141 i = 2;
142 }
143 continue;
144 } else if (c == '>' && i == 2) {
145 if (x->xmlcommentend)
146 x->xmlcommentend(x);
147 return;
148 } else if (i) {
149 if (x->xmlcomment) {
150 for (; i > 0; i--)
151 x->xmlcomment(x, "-", 1);
152 }
153 i = 0;
154 }
155
156 if (datalen < sizeof(x->data) - 1) {
157 x->data[datalen++] = c;
158 } else {
159 x->data[datalen] = '\0';
160 if (x->xmlcomment)
161 x->xmlcomment(x, x->data, datalen);
162 x->data[0] = c;
163 datalen = 1;
164 }
165 }
166 }
167
168 static void
169 xml_parsecdata(XMLParser *x)
170 {
171 size_t datalen = 0, i = 0;
172 int c;
173
174 if (x->xmlcdatastart)
175 x->xmlcdatastart(x);
176 while ((c = GETNEXT()) != EOF) {
177 if (c == ']' || c == '>') {
178 if (x->xmlcdata && datalen) {
179 x->data[datalen] = '\0';
180 x->xmlcdata(x, x->data, datalen);
181 datalen = 0;
182 }
183 }
184
185 if (c == ']') {
186 if (++i > 2) {
187 if (x->xmlcdata)
188 for (; i > 2; i--)
189 x->xmlcdata(x, "]", 1);
190 i = 2;
191 }
192 continue;
193 } else if (c == '>' && i == 2) {
194 if (x->xmlcdataend)
195 x->xmlcdataend(x);
196 return;
197 } else if (i) {
198 if (x->xmlcdata)
199 for (; i > 0; i--)
200 x->xmlcdata(x, "]", 1);
201 i = 0;
202 }
203
204 if (datalen < sizeof(x->data) - 1) {
205 x->data[datalen++] = c;
206 } else {
207 x->data[datalen] = '\0';
208 if (x->xmlcdata)
209 x->xmlcdata(x, x->data, datalen);
210 x->data[0] = c;
211 datalen = 1;
212 }
213 }
214 }
215
216 static int
217 codepointtoutf8(long r, char *s)
218 {
219 if (r == 0) {
220 return 0; /* NUL byte */
221 } else if (r <= 0x7F) {
222 /* 1 byte: 0aaaaaaa */
223 s[0] = r;
224 return 1;
225 } else if (r <= 0x07FF) {
226 /* 2 bytes: 00000aaa aabbbbbb */
227 s[0] = 0xC0 | ((r & 0x0007C0) >> 6); /* 110aaaaa */
228 s[1] = 0x80 | (r & 0x00003F); /* 10bbbbbb */
229 return 2;
230 } else if (r <= 0xFFFF) {
231 /* 3 bytes: aaaabbbb bbcccccc */
232 s[0] = 0xE0 | ((r & 0x00F000) >> 12); /* 1110aaaa */
233 s[1] = 0x80 | ((r & 0x000FC0) >> 6); /* 10bbbbbb */
234 s[2] = 0x80 | (r & 0x00003F); /* 10cccccc */
235 return 3;
236 } else {
237 /* 4 bytes: 000aaabb bbbbcccc ccdddddd */
238 s[0] = 0xF0 | ((r & 0x1C0000) >> 18); /* 11110aaa */
239 s[1] = 0x80 | ((r & 0x03F000) >> 12); /* 10bbbbbb */
240 s[2] = 0x80 | ((r & 0x000FC0) >> 6); /* 10cccccc */
241 s[3] = 0x80 | (r & 0x00003F); /* 10dddddd */
242 return 4;
243 }
244 }
245
246 struct namedentity {
247 const char *entity;
248 long cp;
249 };
250
251 int
252 namedentitycmp(const void *v1, const void *v2)
253 {
254 struct namedentity *n1 = (struct namedentity *)v1;
255 struct namedentity *n2 = (struct namedentity *)v2;
256
257 return strcmp(n1->entity, n2->entity);
258 }
259
260 static int
261 namedentitytostr(const char *e, char *buf, size_t bufsiz)
262 {
263 static const struct namedentity entities[] = {
264 #include "namedentities.h"
265 };
266 struct namedentity find, *found;
267 size_t i;
268
269 /* buffer is too small */
270 if (bufsiz < 5)
271 return -1;
272
273 find.entity = e;
274 found = bsearch(&find, entities, sizeof(entities) / sizeof(*entities),
275 sizeof(*entities), namedentitycmp);
276 if (found) {
277 i = codepointtoutf8(found->cp, buf);
278 buf[i] = '\0';
279 return i;
280 }
281 return -1;
282 }
283
284 static int
285 numericentitytostr(const char *e, char *buf, size_t bufsiz)
286 {
287 long l;
288 int len;
289 char *end;
290
291 /* buffer is too small */
292 if (bufsiz < 5)
293 return -1;
294
295 errno = 0;
296 /* hex (16) or decimal (10) */
297 if (*e == 'x')
298 l = strtol(++e, &end, 16);
299 else
300 l = strtol(e, &end, 10);
301 /* invalid value or not a well-formed entity or invalid codepoint */
302 if (errno || e == end || *end != ';' || l < 0 || l > 0x10ffff)
303 return -1;
304 len = codepointtoutf8(l, buf);
305 buf[len] = '\0';
306
307 return len;
308 }
309
310 /* convert named- or numeric entity string to buffer string
311 * returns byte-length of string or -1 on failure. */
312 int
313 xml_entitytostr(const char *e, char *buf, size_t bufsiz)
314 {
315 /* doesn't start with & */
316 if (e[0] != '&')
317 return -1;
318 /* numeric entity */
319 if (e[1] == '#')
320 return numericentitytostr(e + 2, buf, bufsiz);
321 else /* named entity */
322 return namedentitytostr(e + 1, buf, bufsiz);
323 }
324
325 void
326 xml_parse(XMLParser *x)
327 {
328 size_t datalen, tagdatalen;
329 int c, isend;
330
331 while ((c = GETNEXT()) != EOF && c != '<')
332 ; /* skip until < */
333
334 while (c != EOF) {
335 if (c == '<') { /* parse tag */
336 if ((c = GETNEXT()) == EOF)
337 return;
338
339 if (c == '!') { /* cdata and comments */
340 for (tagdatalen = 0; (c = GETNEXT()) != EOF;) {
341 /* NOTE: sizeof(x->data) must be atleast sizeof("[CDATA[") */
342 if (tagdatalen <= sizeof("[CDATA[") - 1)
343 x->data[tagdatalen++] = c;
344 if (c == '>')
345 break;
346 else if (c == '-' && tagdatalen == sizeof("--") - 1 &&
347 (x->data[0] == '-')) {
348 xml_parsecomment(x);
349 break;
350 } else if (c == '[') {
351 if (tagdatalen == sizeof("[CDATA[") - 1 &&
352 !strncmp(x->data, "[CDATA[", tagdatalen)) {
353 xml_parsecdata(x);
354 break;
355 }
356 }
357 }
358 } else {
359 /* normal tag (open, short open, close), processing instruction. */
360 x->tag[0] = c;
361 x->taglen = 1;
362 x->isshorttag = isend = 0;
363
364 /* treat processing instruction as shorttag, don't strip "?" prefix. */
365 if (c == '?') {
366 x->isshorttag = 1;
367 } else if (c == '/') {
368 if ((c = GETNEXT()) == EOF)
369 return;
370 x->tag[0] = c;
371 isend = 1;
372 }
373
374 while ((c = GETNEXT()) != EOF) {
375 if (c == '/')
376 x->isshorttag = 1; /* short tag */
377 else if (c == '>' || isspace(c)) {
378 x->tag[x->taglen] = '\0';
379 if (isend) { /* end tag, starts with </ */
380 if (x->xmltagend)
381 x->xmltagend(x, x->tag, x->taglen, x->isshorttag);
382 x->tag[0] = '\0';
383 x->taglen = 0;
384 } else {
385 /* start tag */
386 if (x->xmltagstart)
387 x->xmltagstart(x, x->tag, x->taglen);
388 if (isspace(c))
389 xml_parseattrs(x);
390 if (x->xmltagstartparsed)
391 x->xmltagstartparsed(x, x->tag, x->taglen, x->isshorttag);
392 }
393 /* call tagend for shortform or processing instruction */
394 if (x->isshorttag) {
395 if (x->xmltagend)
396 x->xmltagend(x, x->tag, x->taglen, x->isshorttag);
397 x->tag[0] = '\0';
398 x->taglen = 0;
399 }
400 break;
401 } else if (x->taglen < sizeof(x->tag) - 1)
402 x->tag[x->taglen++] = c; /* NOTE: tag name truncation */
403 }
404 }
405 } else {
406 /* parse tag data */
407 datalen = 0;
408 if (x->xmldatastart)
409 x->xmldatastart(x);
410 while ((c = GETNEXT()) != EOF) {
411 if (c == '&') {
412 if (datalen) {
413 x->data[datalen] = '\0';
414 if (x->xmldata)
415 x->xmldata(x, x->data, datalen);
416 }
417 x->data[0] = c;
418 datalen = 1;
419 while ((c = GETNEXT()) != EOF) {
420 if (c == '<')
421 break;
422 if (datalen < sizeof(x->data) - 1)
423 x->data[datalen++] = c;
424 else {
425 /* entity too long for buffer, handle as normal data */
426 x->data[datalen] = '\0';
427 if (x->xmldata)
428 x->xmldata(x, x->data, datalen);
429 x->data[0] = c;
430 datalen = 1;
431 break;
432 }
433 if (c == ';') {
434 x->data[datalen] = '\0';
435 if (x->xmldataentity)
436 x->xmldataentity(x, x->data, datalen);
437 datalen = 0;
438 break;
439 }
440 }
441 } else if (c != '<') {
442 if (datalen < sizeof(x->data) - 1) {
443 x->data[datalen++] = c;
444 } else {
445 x->data[datalen] = '\0';
446 if (x->xmldata)
447 x->xmldata(x, x->data, datalen);
448 x->data[0] = c;
449 datalen = 1;
450 }
451 }
452 if (c == '<') {
453 x->data[datalen] = '\0';
454 if (x->xmldata && datalen)
455 x->xmldata(x, x->data, datalen);
456 if (x->xmldataend)
457 x->xmldataend(x);
458 break;
459 }
460 }
461 }
462 }
463 }