?  search:
summary | log | graphiclog1 | graphiclog2 | commit | commitdiff | tree | refs | edit | fork
blame | history | raw | HEAD
Fix bug #3574: Template addressing
[claws.git] / src / html.c
blob815f4464663d8f0c238d3a2e6503e9e49a9a63e8
1 /*
2 * Claws Mail -- a GTK+ based, lightweight, and fast e-mail client
3 * Copyright (C) 1999-2016 Hiroyuki Yamamoto and the Claws Mail team
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 3 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program. If not, see <http://www.gnu.org/licenses/>.
17 */
18
19 #include <glib.h>
20 #include <stdio.h>
21 #include <string.h>
22 #include <ctype.h>
23
24 #include "html.h"
25 #include "codeconv.h"
26 #include "utils.h"
27
28 #define SC_HTMLBUFSIZE 8192
29 #define HR_STR "────────────────────────────────────────────────"
30 #define LI_STR "• "
31
32 typedef struct _SC_HTMLSymbol SC_HTMLSymbol;
33
34 struct _SC_HTMLSymbol
35 {
36 gchar *const key;
37 gchar *const val;
38 };
39
40 static SC_HTMLSymbol symbol_list[] = {
41 {"&#34;", "\42"},
42 {"&#38;", "\46"},
43 {"&#39;", "\47"},
44 {"&#60;", "\74"},
45 {"&#62;", "\76"},
46 {"&#034;", "\42"},
47 {"&#038;", "\46"},
48 {"&#039;", "\47"},
49 {"&#060;", "\74"},
50 {"&#062;", "\76"},
51 {"&#146;", "\47"},
52 {"&#153;", "\342\204\242"},
53 {"&#160;", "\40"},
54 {"&#161;", "\302\241"},
55 {"&#162;", "\302\242"},
56 {"&#163;", "\302\243"},
57 {"&#164;", "\302\244"},
58 {"&#165;", "\302\245"},
59 {"&#166;", "\302\246"},
60 {"&#167;", "\302\247"},
61 {"&#168;", "\302\250"},
62 {"&#169;", "\302\251"},
63 {"&#170;", "\302\252"},
64 {"&#171;", "\302\253"},
65 {"&#172;", "\302\254"},
66 {"&#173;", "\302\255"},
67 {"&#174;", "\302\256"},
68 {"&#175;", "\302\257"},
69 {"&#176;", "\302\260"},
70 {"&#177;", "\302\261"},
71 {"&#178;", "\302\262"},
72 {"&#179;", "\302\263"},
73 {"&#180;", "\302\264"},
74 {"&#181;", "\302\265"},
75 {"&#182;", "\302\266"},
76 {"&#183;", "\302\267"},
77 {"&#184;", "\302\270"},
78 {"&#185;", "\302\271"},
79 {"&#186;", "\302\272"},
80 {"&#187;", "\302\273"},
81 {"&#188;", "\302\274"},
82 {"&#189;", "\302\275"},
83 {"&#190;", "\302\276"},
84 {"&#191;", "\302\277"},
85 {"&#192;", "\303\200"},
86 {"&#193;", "\303\201"},
87 {"&#194;", "\303\202"},
88 {"&#195;", "\303\203"},
89 {"&#196;", "\303\204"},
90 {"&#197;", "\303\205"},
91 {"&#198;", "\303\206"},
92 {"&#199;", "\303\207"},
93 {"&#200;", "\303\210"},
94 {"&#201;", "\303\211"},
95 {"&#202;", "\303\212"},
96 {"&#203;", "\303\213"},
97 {"&#204;", "\303\214"},
98 {"&#205;", "\303\215"},
99 {"&#206;", "\303\216"},
100 {"&#207;", "\303\217"},
101 {"&#208;", "\303\220"},
102 {"&#209;", "\303\221"},
103 {"&#210;", "\303\222"},
104 {"&#211;", "\303\223"},
105 {"&#212;", "\303\224"},
106 {"&#213;", "\303\225"},
107 {"&#214;", "\303\226"},
108 {"&#215;", "\303\227"},
109 {"&#216;", "\303\230"},
110 {"&#217;", "\303\231"},
111 {"&#218;", "\303\232"},
112 {"&#219;", "\303\233"},
113 {"&#220;", "\303\234"},
114 {"&#221;", "\303\235"},
115 {"&#222;", "\303\236"},
116 {"&#223;", "\303\237"},
117 {"&#224;", "\303\240"},
118 {"&#225;", "\303\241"},
119 {"&#226;", "\303\242"},
120 {"&#227;", "\303\243"},
121 {"&#228;", "\303\244"},
122 {"&#229;", "\303\245"},
123 {"&#230;", "\303\246"},
124 {"&#231;", "\303\247"},
125 {"&#232;", "\303\250"},
126 {"&#233;", "\303\251"},
127 {"&#234;", "\303\252"},
128 {"&#235;", "\303\253"},
129 {"&#236;", "\303\254"},
130 {"&#237;", "\303\255"},
131 {"&#238;", "\303\256"},
132 {"&#239;", "\303\257"},
133 {"&#240;", "\303\260"},
134 {"&#241;", "\303\261"},
135 {"&#242;", "\303\262"},
136 {"&#243;", "\303\263"},
137 {"&#244;", "\303\264"},
138 {"&#245;", "\303\265"},
139 {"&#246;", "\303\266"},
140 {"&#247;", "\303\267"},
141 {"&#248;", "\303\270"},
142 {"&#249;", "\303\271"},
143 {"&#250;", "\303\272"},
144 {"&#251;", "\303\273"},
145 {"&#252;", "\303\274"},
146 {"&#253;", "\303\275"},
147 {"&#254;", "\303\276"},
148 {"&#255;", "\303\277"},
149 {"&#338;", "\305\222"},
150 {"&#339;", "\305\223"},
151 {"&#352;", "\305\240"},
152 {"&#353;", "\305\241"},
153 {"&#376;", "\305\270"},
154 {"&#710;", "\313\206"},
155 {"&#732;", "\313\234"},
156 {"&#8194;", "\342\200\202"},
157 {"&#8195;", "\342\200\203"},
158 {"&#8201;", "\342\200\211"},
159 {"&#8211;", "\342\200\223"},
160 {"&#8212;", "\342\200\224"},
161 {"&#8216;", "\342\200\230"},
162 {"&#8217;", "\342\200\231"},
163 {"&#8218;", "\342\200\232"},
164 {"&#8220;", "\342\200\234"},
165 {"&#8221;", "\342\200\235"},
166 {"&#8222;", "\342\200\236"},
167 {"&#8224;", "\342\200\240"},
168 {"&#8225;", "\342\200\241"},
169 {"&#8226;", "\342\200\242"},
170 {"&#8230;", "\342\200\246"},
171 {"&#8240;", "\342\200\260"},
172 {"&#8249;", "\342\200\271"},
173 {"&#8250;", "\342\200\272"},
174 {"&#8364;", "\342\202\254"},
175 {"&#8482;", "\342\204\242"},
176 {"&quot;", "\42"},
177 {"&amp;", "\46"},
178 {"&apos;", "\47"},
179 {"&lt;", "\74"},
180 {"&gt;", "\76"},
181 {"&squot;", "\47"},
182 {"&nbsp;", "\40"},
183 {"&iexcl;", "\302\241"},
184 {"&cent;", "\302\242"},
185 {"&pound;", "\302\243"},
186 {"&curren;", "\302\244"},
187 {"&yen;", "\302\245"},
188 {"&brvbar;", "\302\246"},
189 {"&sect;", "\302\247"},
190 {"&uml;", "\302\250"},
191 {"&copy;", "\302\251"},
192 {"&ordf;", "\302\252"},
193 {"&laquo;", "\302\253"},
194 {"&not;", "\302\254"},
195 {"&shy;", "\302\255"},
196 {"&reg;", "\302\256"},
197 {"&macr;", "\302\257"},
198 {"&deg;", "\302\260"},
199 {"&plusmn;", "\302\261"},
200 {"&sup2;", "\302\262"},
201 {"&sup3;", "\302\263"},
202 {"&acute;", "\302\264"},
203 {"&micro;", "\302\265"},
204 {"&para;", "\302\266"},
205 {"&middot;", "\302\267"},
206 {"&cedil;", "\302\270"},
207 {"&sup1;", "\302\271"},
208 {"&ordm;", "\302\272"},
209 {"&raquo;", "\302\273"},
210 {"&frac14;", "\302\274"},
211 {"&frac12;", "\302\275"},
212 {"&frac34;", "\302\276"},
213 {"&iquest;", "\302\277"},
214 {"&Agrave;", "\303\200"},
215 {"&Aacute;", "\303\201"},
216 {"&Acirc;", "\303\202"},
217 {"&Atilde;", "\303\203"},
218 {"&Auml;", "\303\204"},
219 {"&Aring;", "\303\205"},
220 {"&AElig;", "\303\206"},
221 {"&Ccedil;", "\303\207"},
222 {"&Egrave;", "\303\210"},
223 {"&Eacute;", "\303\211"},
224 {"&Ecirc;", "\303\212"},
225 {"&Euml;", "\303\213"},
226 {"&Igrave;", "\303\214"},
227 {"&Iacute;", "\303\215"},
228 {"&Icirc;", "\303\216"},
229 {"&Iuml;", "\303\217"},
230 {"&ETH;", "\303\220"},
231 {"&Ntilde;", "\303\221"},
232 {"&Ograve;", "\303\222"},
233 {"&Oacute;", "\303\223"},
234 {"&Ocirc;", "\303\224"},
235 {"&Otilde;", "\303\225"},
236 {"&Ouml;", "\303\226"},
237 {"&times;", "\303\227"},
238 {"&Oslash;", "\303\230"},
239 {"&Ugrave;", "\303\231"},
240 {"&Uacute;", "\303\232"},
241 {"&Ucirc;", "\303\233"},
242 {"&Uuml;", "\303\234"},
243 {"&Yacute;", "\303\235"},
244 {"&THORN;", "\303\236"},
245 {"&szlig;", "\303\237"},
246 {"&agrave;", "\303\240"},
247 {"&aacute;", "\303\241"},
248 {"&acirc;", "\303\242"},
249 {"&atilde;", "\303\243"},
250 {"&auml;", "\303\244"},
251 {"&aring;", "\303\245"},
252 {"&aelig;", "\303\246"},
253 {"&ccedil;", "\303\247"},
254 {"&egrave;", "\303\250"},
255 {"&eacute;", "\303\251"},
256 {"&ecirc;", "\303\252"},
257 {"&euml;", "\303\253"},
258 {"&igrave;", "\303\254"},
259 {"&iacute;", "\303\255"},
260 {"&icirc;", "\303\256"},
261 {"&iuml;", "\303\257"},
262 {"&eth;", "\303\260"},
263 {"&ntilde;", "\303\261"},
264 {"&ograve;", "\303\262"},
265 {"&oacute;", "\303\263"},
266 {"&ocirc;", "\303\264"},
267 {"&otilde;", "\303\265"},
268 {"&ouml;", "\303\266"},
269 {"&divide;", "\303\267"},
270 {"&oslash;", "\303\270"},
271 {"&ugrave;", "\303\271"},
272 {"&uacute;", "\303\272"},
273 {"&ucirc;", "\303\273"},
274 {"&uuml;", "\303\274"},
275 {"&yacute;", "\303\275"},
276 {"&thorn;", "\303\276"},
277 {"&yuml;", "\303\277"},
278 {"&OElig;", "\305\222"},
279 {"&oelig;", "\305\223"},
280 {"&Scaron;", "\305\240"},
281 {"&scaron;", "\305\241"},
282 {"&Yuml;", "\305\270"},
283 {"&circ;", "\313\206"},
284 {"&tilde;", "\313\234"},
285 {"&ensp;", "\342\200\202"},
286 {"&emsp;", "\342\200\203"},
287 {"&thinsp;", "\342\200\211"},
288 {"&ndash;", "\342\200\223"},
289 {"&mdash;", "\342\200\224"},
290 {"&lsquo;", "\342\200\230"},
291 {"&rsquo;", "\342\200\231"},
292 {"&sbquo;", "\342\200\232"},
293 {"&ldquo;", "\342\200\234"},
294 {"&rdquo;", "\342\200\235"},
295 {"&bdquo;", "\342\200\236"},
296 {"&dagger;", "\342\200\240"},
297 {"&Dagger;", "\342\200\241"},
298 {"&bull;", "\342\200\242"},
299 {"&hellip;", "\342\200\246"},
300 {"&permil;", "\342\200\260"},
301 {"&lsaquo;", "\342\200\271"},
302 {"&rsaquo;", "\342\200\272"},
303 {"&euro;", "\342\202\254"},
304 {"&trade;", "\342\204\242"}
305 };
306
307 typedef struct _SC_HTMLAltSymbol SC_HTMLAltSymbol;
308
309 struct _SC_HTMLAltSymbol
310 {
311 gint key;
312 gchar *const val;
313 };
314
315 static GHashTable *default_symbol_table;
316
317 static SC_HTMLState sc_html_read_line (SC_HTMLParser *parser);
318 static void sc_html_append_char (SC_HTMLParser *parser,
319 gchar ch);
320 static void sc_html_append_str (SC_HTMLParser *parser,
321 const gchar *str,
322 gint len);
323 static SC_HTMLState sc_html_parse_tag (SC_HTMLParser *parser);
324 static void sc_html_parse_special (SC_HTMLParser *parser);
325 static void sc_html_get_parenthesis (SC_HTMLParser *parser,
326 gchar *buf,
327 gint len);
328
329
330 SC_HTMLParser *sc_html_parser_new(FILE *fp, CodeConverter *conv)
331 {
332 SC_HTMLParser *parser;
333
334 cm_return_val_if_fail(fp != NULL, NULL);
335 cm_return_val_if_fail(conv != NULL, NULL);
336
337 parser = g_new0(SC_HTMLParser, 1);
338 parser->fp = fp;
339 parser->conv = conv;
340 parser->str = g_string_new(NULL);
341 parser->buf = g_string_new(NULL);
342 parser->bufp = parser->buf->str;
343 parser->state = SC_HTML_NORMAL;
344 parser->href = NULL;
345 parser->newline = TRUE;
346 parser->empty_line = TRUE;
347 parser->space = FALSE;
348 parser->pre = FALSE;
349 parser->indent = 0;
350
351 #define SYMBOL_TABLE_ADD(table, list) \
352 { \
353 gint i; \
354 \
355 for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
356 g_hash_table_insert(table, list[i].key, list[i].val); \
357 }
358 #define SYMBOL_TABLE_REF_ADD(table, list) \
359 { \
360 gint i; \
361 \
362 for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
363 g_hash_table_insert(table, &list[i].key, list[i].val); \
364 }
365
366 if (!default_symbol_table) {
367 default_symbol_table =
368 g_hash_table_new(g_str_hash, g_str_equal);
369 SYMBOL_TABLE_ADD(default_symbol_table, symbol_list);
370 }
371
372 #undef SYMBOL_TABLE_ADD
373 #undef SYMBOL_TABLE_REF_ADD
374
375 parser->symbol_table = default_symbol_table;
376
377 return parser;
378 }
379
380 void sc_html_parser_destroy(SC_HTMLParser *parser)
381 {
382 g_string_free(parser->str, TRUE);
383 g_string_free(parser->buf, TRUE);
384 g_free(parser->href);
385 g_free(parser);
386 }
387
388 gchar *sc_html_parse(SC_HTMLParser *parser)
389 {
390 parser->state = SC_HTML_NORMAL;
391 g_string_truncate(parser->str, 0);
392
393 if (*parser->bufp == '\0') {
394 g_string_truncate(parser->buf, 0);
395 parser->bufp = parser->buf->str;
396 if (sc_html_read_line(parser) == SC_HTML_EOF)
397 return NULL;
398 }
399
400 while (*parser->bufp != '\0') {
401 switch (*parser->bufp) {
402 case '<': {
403 SC_HTMLState st;
404 st = sc_html_parse_tag(parser);
405 /* when we see an href, we need to flush the str
406 * buffer. Then collect all the chars until we
407 * see the end anchor tag
408 */
409 if (SC_HTML_HREF_BEG == st || SC_HTML_HREF == st)
410 return parser->str->str;
411 }
412 break;
413 case '&':
414 sc_html_parse_special(parser);
415 break;
416 case ' ':
417 case '\t':
418 case '\r':
419 case '\n':
420 if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
421 parser->bufp++;
422
423 if (!parser->pre) {
424 if (!parser->newline)
425 parser->space = TRUE;
426
427 parser->bufp++;
428 break;
429 }
430 /* fallthrough */
431 default:
432 sc_html_append_char(parser, *parser->bufp++);
433 }
434 }
435
436 return parser->str->str;
437 }
438
439 static SC_HTMLState sc_html_read_line(SC_HTMLParser *parser)
440 {
441 gchar buf[SC_HTMLBUFSIZE];
442 gchar buf2[SC_HTMLBUFSIZE];
443 gint index;
444 gint n;
445
446 if (parser->fp == NULL)
447 return SC_HTML_EOF;
448
449 n = fread(buf, 1, sizeof(buf) - 1, parser->fp);
450 if (n == 0) {
451 parser->state = SC_HTML_EOF;
452 return SC_HTML_EOF;
453 } else
454 buf[n] = '\0';
455
456 if (conv_convert(parser->conv, buf2, sizeof(buf2), buf) < 0) {
457 index = parser->bufp - parser->buf->str;
458
459 conv_utf8todisp(buf2, sizeof(buf2), buf);
460 g_string_append(parser->buf, buf2);
461
462 parser->bufp = parser->buf->str + index;
463
464 return SC_HTML_CONV_FAILED;
465 }
466
467 index = parser->bufp - parser->buf->str;
468
469 g_string_append(parser->buf, buf2);
470
471 parser->bufp = parser->buf->str + index;
472
473 return SC_HTML_NORMAL;
474 }
475
476 static void sc_html_append_char(SC_HTMLParser *parser, gchar ch)
477 {
478 GString *str = parser->str;
479
480 if (!parser->pre && parser->space) {
481 g_string_append_c(str, ' ');
482 parser->space = FALSE;
483 }
484
485 g_string_append_c(str, ch);
486
487 parser->empty_line = FALSE;
488 if (ch == '\n') {
489 parser->newline = TRUE;
490 if (str->len > 1 && str->str[str->len - 2] == '\n')
491 parser->empty_line = TRUE;
492 if (parser->indent > 0) {
493 gint i, n = parser->indent;
494 for (i = 0; i < n; i++)
495 g_string_append_c(str, '>');
496 g_string_append_c(str, ' ');
497 }
498 } else
499 parser->newline = FALSE;
500 }
501
502 static void sc_html_append_str(SC_HTMLParser *parser, const gchar *str, gint len)
503 {
504 GString *string = parser->str;
505
506 if (!parser->pre && parser->space) {
507 g_string_append_c(string, ' ');
508 parser->space = FALSE;
509 }
510
511 if (len == 0) return;
512 if (len < 0)
513 g_string_append(string, str);
514 else {
515 gchar *s;
516 Xstrndup_a(s, str, len, return);
517 g_string_append(string, s);
518 }
519
520 parser->empty_line = FALSE;
521 if (string->len > 0 && string->str[string->len - 1] == '\n') {
522 parser->newline = TRUE;
523 if (string->len > 1 && string->str[string->len - 2] == '\n')
524 parser->empty_line = TRUE;
525 } else
526 parser->newline = FALSE;
527 }
528
529 static SC_HTMLTag *sc_html_get_tag(const gchar *str)
530 {
531 SC_HTMLTag *tag;
532 gchar *tmp;
533 guchar *tmpp;
534
535 cm_return_val_if_fail(str != NULL, NULL);
536
537 if (*str == '\0' || *str == '!') return NULL;
538
539 Xstrdup_a(tmp, str, return NULL);
540
541 tag = g_new0(SC_HTMLTag, 1);
542
543 for (tmpp = tmp; *tmpp != '\0' && !g_ascii_isspace(*tmpp); tmpp++)
544 ;
545
546 if (*tmpp == '\0') {
547 tag->name = g_utf8_strdown(tmp, -1);
548 return tag;
549 } else {
550 *tmpp++ = '\0';
551 tag->name = g_utf8_strdown(tmp, -1);
552 }
553
554 while (*tmpp != '\0') {
555 SC_HTMLAttr *attr;
556 gchar *attr_name;
557 gchar *attr_value;
558 gchar *p;
559 gchar quote;
560
561 while (g_ascii_isspace(*tmpp)) tmpp++;
562 attr_name = tmpp;
563
564 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp) &&
565 *tmpp != '=')
566 tmpp++;
567 if (*tmpp != '\0' && *tmpp != '=') {
568 *tmpp++ = '\0';
569 while (g_ascii_isspace(*tmpp)) tmpp++;
570 }
571
572 if (*tmpp == '=') {
573 *tmpp++ = '\0';
574 while (g_ascii_isspace(*tmpp)) tmpp++;
575
576 if (*tmpp == '"' || *tmpp == '\'') {
577 /* name="value" */
578 quote = *tmpp;
579 tmpp++;
580 attr_value = tmpp;
581 if ((p = strchr(attr_value, quote)) == NULL) {
582 g_warning("sc_html_get_tag(): syntax error in tag: '%s'", str);
583 return tag;
584 }
585 tmpp = p;
586 *tmpp++ = '\0';
587 while (g_ascii_isspace(*tmpp)) tmpp++;
588 } else {
589 /* name=value */
590 attr_value = tmpp;
591 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp)) tmpp++;
592 if (*tmpp != '\0')
593 *tmpp++ = '\0';
594 }
595 } else
596 attr_value = "";
597
598 g_strchomp(attr_name);
599 attr = g_new(SC_HTMLAttr, 1);
600 attr->name = g_utf8_strdown(attr_name, -1);
601 attr->value = g_strdup(attr_value);
602 tag->attr = g_list_append(tag->attr, attr);
603 }
604
605 return tag;
606 }
607
608 static void sc_html_free_tag(SC_HTMLTag *tag)
609 {
610 if (!tag) return;
611
612 g_free(tag->name);
613 while (tag->attr != NULL) {
614 SC_HTMLAttr *attr = (SC_HTMLAttr *)tag->attr->data;
615 g_free(attr->name);
616 g_free(attr->value);
617 g_free(attr);
618 tag->attr = g_list_remove(tag->attr, tag->attr->data);
619 }
620 g_free(tag);
621 }
622
623 static void decode_href(SC_HTMLParser *parser)
624 {
625 gchar *tmp;
626 SC_HTMLParser *tparser = g_new0(SC_HTMLParser, 1);
627
628 tparser->str = g_string_new(NULL);
629 tparser->buf = g_string_new(parser->href);
630 tparser->bufp = tparser->buf->str;
631 tparser->symbol_table = default_symbol_table;
632
633 tmp = sc_html_parse(tparser);
634
635 g_free(parser->href);
636 parser->href = g_strdup(tmp);
637
638 sc_html_parser_destroy(tparser);
639 }
640
641 static SC_HTMLState sc_html_parse_tag(SC_HTMLParser *parser)
642 {
643 gchar buf[SC_HTMLBUFSIZE];
644 SC_HTMLTag *tag;
645
646 sc_html_get_parenthesis(parser, buf, sizeof(buf));
647
648 tag = sc_html_get_tag(buf);
649
650 parser->state = SC_HTML_UNKNOWN;
651 if (!tag) return SC_HTML_UNKNOWN;
652
653 if (!strcmp(tag->name, "br") || !strcmp(tag->name, "br/")) {
654 parser->space = FALSE;
655 sc_html_append_char(parser, '\n');
656 parser->state = SC_HTML_BR;
657 } else if (!strcmp(tag->name, "a")) {
658 GList *cur;
659 parser->href = NULL;
660 for (cur = tag->attr; cur != NULL; cur = cur->next) {
661 if (cur->data && !strcmp(((SC_HTMLAttr *)cur->data)->name, "href")) {
662 g_free(parser->href);
663 parser->href = g_strdup(((SC_HTMLAttr *)cur->data)->value);
664 decode_href(parser);
665 parser->state = SC_HTML_HREF_BEG;
666 break;
667 }
668 }
669 if (parser->href == NULL)
670 parser->href = g_strdup("");
671 parser->state = SC_HTML_HREF_BEG;
672 } else if (!strcmp(tag->name, "/a")) {
673 parser->state = SC_HTML_HREF;
674 } else if (!strcmp(tag->name, "p")) {
675 parser->space = FALSE;
676 if (!parser->empty_line) {
677 parser->space = FALSE;
678 if (!parser->newline) sc_html_append_char(parser, '\n');
679 sc_html_append_char(parser, '\n');
680 }
681 parser->state = SC_HTML_PAR;
682 } else if (!strcmp(tag->name, "pre")) {
683 parser->pre = TRUE;
684 parser->state = SC_HTML_PRE;
685 } else if (!strcmp(tag->name, "/pre")) {
686 parser->pre = FALSE;
687 parser->state = SC_HTML_NORMAL;
688 } else if (!strcmp(tag->name, "hr")) {
689 if (!parser->newline) {
690 parser->space = FALSE;
691 sc_html_append_char(parser, '\n');
692 }
693 sc_html_append_str(parser, HR_STR, -1);
694 sc_html_append_char(parser, '\n');
695 parser->state = SC_HTML_HR;
696 } else if (!strcmp(tag->name, "div") ||
697 !strcmp(tag->name, "ul") ||
698 !strcmp(tag->name, "li") ||
699 !strcmp(tag->name, "table") ||
700 !strcmp(tag->name, "dd") ||
701 !strcmp(tag->name, "tr") ||
702 (tag->name[0] == 'h' && g_ascii_isdigit(tag->name[1]))) {
703 if (!parser->newline) {
704 parser->space = FALSE;
705 sc_html_append_char(parser, '\n');
706 }
707 if (!strcmp(tag->name, "li")) {
708 sc_html_append_str(parser, LI_STR, -1);
709 }
710 parser->state = SC_HTML_NORMAL;
711 } else if (!strcmp(tag->name, "blockquote")) {
712 parser->state = SC_HTML_NORMAL;
713 parser->indent++;
714 } else if (!strcmp(tag->name, "/blockquote")) {
715 parser->state = SC_HTML_NORMAL;
716 parser->indent--;
717 } else if (!strcmp(tag->name, "/table") ||
718 (tag->name[0] == '/' &&
719 tag->name[1] == 'h' &&
720 g_ascii_isdigit(tag->name[1]))) {
721 if (!parser->empty_line) {
722 parser->space = FALSE;
723 if (!parser->newline) sc_html_append_char(parser, '\n');
724 sc_html_append_char(parser, '\n');
725 }
726 parser->state = SC_HTML_NORMAL;
727 } else if (!strcmp(tag->name, "/div") ||
728 !strcmp(tag->name, "/ul") ||
729 !strcmp(tag->name, "/li")) {
730 if (!parser->newline) {
731 parser->space = FALSE;
732 sc_html_append_char(parser, '\n');
733 }
734 parser->state = SC_HTML_NORMAL;
735 }
736
737 sc_html_free_tag(tag);
738
739 return parser->state;
740 }
741
742 static void sc_html_parse_special(SC_HTMLParser *parser)
743 {
744 gchar symbol_name[9];
745 gint n;
746 const gchar *val;
747
748 parser->state = SC_HTML_UNKNOWN;
749 cm_return_if_fail(*parser->bufp == '&');
750
751 /* &foo; */
752 for (n = 0; parser->bufp[n] != '\0' && parser->bufp[n] != ';'; n++)
753 ;
754 if (n > 7 || parser->bufp[n] != ';') {
755 /* output literal `&' */
756 sc_html_append_char(parser, *parser->bufp++);
757 parser->state = SC_HTML_NORMAL;
758 return;
759 }
760 strncpy2(symbol_name, parser->bufp, n + 2);
761 parser->bufp += n + 1;
762
763 if ((val = g_hash_table_lookup(parser->symbol_table, symbol_name))
764 != NULL) {
765 sc_html_append_str(parser, val, -1);
766 parser->state = SC_HTML_NORMAL;
767 return;
768 }
769
770 sc_html_append_str(parser, symbol_name, -1);
771 }
772
773 static gchar *sc_html_find_tag(SC_HTMLParser *parser, const gchar *tag)
774 {
775 gchar *cur = parser->bufp;
776 gint len = strlen(tag);
777
778 if (cur == NULL)
779 return NULL;
780
781 while ((cur = strstr(cur, "<")) != NULL) {
782 if (!g_ascii_strncasecmp(cur, tag, len))
783 return cur;
784 cur += 2;
785 }
786 return NULL;
787 }
788
789 static void sc_html_get_parenthesis(SC_HTMLParser *parser, gchar *buf, gint len)
790 {
791 gchar *p;
792
793 buf[0] = '\0';
794 cm_return_if_fail(*parser->bufp == '<');
795
796 /* ignore comment / CSS / script stuff */
797 if (!strncmp(parser->bufp, "<!--", 4)) {
798 parser->bufp += 4;
799 while ((p = strstr(parser->bufp, "-->")) == NULL)
800 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
801 parser->bufp = p + 3;
802 return;
803 }
804 if (!g_ascii_strncasecmp(parser->bufp, "<style", 6)) {
805 parser->bufp += 6;
806 while ((p = sc_html_find_tag(parser, "</style>")) == NULL)
807 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
808 parser->bufp = p + 8;
809 return;
810 }
811 if (!g_ascii_strncasecmp(parser->bufp, "<script", 7)) {
812 parser->bufp += 7;
813 while ((p = sc_html_find_tag(parser, "</script>")) == NULL)
814 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
815 parser->bufp = p + 9;
816 return;
817 }
818
819 parser->bufp++;
820 while ((p = strchr(parser->bufp, '>')) == NULL)
821 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
822
823 strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
824 g_strstrip(buf);
825 parser->bufp = p + 1;
826 }