Statistics
| Revision:

root / src / html.c @ 1

History | View | Annotate | Download (17.6 KB)

1
/*
2
 * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
3
 * Copyright (C) 1999-2003 Hiroyuki Yamamoto
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 2 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, write to the Free Software
17
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
18
 */
19

    
20
#include <glib.h>
21
#include <stdio.h>
22
#include <string.h>
23
#include <ctype.h>
24

    
25
#include "html.h"
26
#include "codeconv.h"
27
#include "utils.h"
28

    
29
#define HTMLBUFSIZE        8192
30
#define HR_STR                "------------------------------------------------"
31

    
32
typedef struct _HTMLSymbol        HTMLSymbol;
33

    
34
struct _HTMLSymbol
35
{
36
        gchar *const key;
37
        gchar *const val;
38
};
39

    
40
static HTMLSymbol symbol_list[] = {
41
        {"&lt;"    , "<"},
42
        {"&gt;"    , ">"},
43
        {"&amp;"   , "&"},
44
        {"&quot;"  , "\""},
45
        {"&nbsp;"  , " "},
46
        {"&trade;" , "(TM)"},
47

    
48
        {"&#153;", "(TM)"},
49
};
50

    
51
static HTMLSymbol ascii_symbol_list[] = {
52
        {"&iexcl;" , "^!"},
53
        {"&brvbar;", "|"},
54
        {"&copy;"  , "(C)"},
55
        {"&laquo;" , "<<"},
56
        {"&reg;"   , "(R)"},
57

    
58
        {"&sup2;"  , "^2"},
59
        {"&sup3;"  , "^3"},
60
        {"&acute;" , "'"},
61
        {"&cedil;" , ","},
62
        {"&sup1;"  , "^1"},
63
        {"&raquo;" , ">>"},
64
        {"&frac14;", "1/4"},
65
        {"&frac12;", "1/2"},
66
        {"&frac34;", "3/4"},
67
        {"&iquest;", "^?"},
68

    
69
        {"&Agrave;", "A`"},
70
        {"&Aacute;", "A'"},
71
        {"&Acirc;" , "A^"},
72
        {"&Atilde;", "A~"},
73
        {"&AElig;" , "AE"},
74
        {"&Egrave;", "E`"},
75
        {"&Eacute;", "E'"},
76
        {"&Ecirc;" , "E^"},
77
        {"&Igrave;", "I`"},
78
        {"&Iacute;", "I'"},
79
        {"&Icirc;" , "I^"},
80

    
81
        {"&Ntilde;", "N~"},
82
        {"&Ograve;", "O`"},
83
        {"&Oacute;", "O'"},
84
        {"&Ocirc;" , "O^"},
85
        {"&Otilde;", "O~"},
86
        {"&Ugrave;", "U`"},
87
        {"&Uacute;", "U'"},
88
        {"&Ucirc;" , "U^"},
89
        {"&Yacute;", "Y'"},
90

    
91
        {"&agrave;", "a`"},
92
        {"&aacute;", "a'"},
93
        {"&acirc;" , "a^"},
94
        {"&atilde;", "a~"},
95
        {"&aelig;" , "ae"},
96
        {"&egrave;", "e`"},
97
        {"&eacute;", "e'"},
98
        {"&ecirc;" , "e^"},
99
        {"&igrave;", "i`"},
100
        {"&iacute;", "i'"},
101
        {"&icirc;" , "i^"},
102

    
103
        {"&ntilde;", "n~"},
104
        {"&ograve;", "o`"},
105
        {"&oacute;", "o'"},
106
        {"&ocirc;" , "o^"},
107
        {"&otilde;", "o~"},
108
        {"&ugrave;", "u`"},
109
        {"&uacute;", "u'"},
110
        {"&ucirc;" , "u^"},
111
        {"&yacute;", "y'"},
112
};
113

    
114
static HTMLSymbol eucjp_symbol_list[] = {
115
        {"&iexcl;" , "^!"},
116
        {"&cent;"  , "\xa1\xf1"},
117
        {"&pound;" , "\xa1\xf2"},
118
        {"&yen;"   , "\xa1\xef"},
119
        {"&brvbar;", "|"},
120
        {"&sect;"  , "\xa1\xf8"},
121
        {"&uml;"   , "\xa1\xaf"},
122
        {"&copy;"  , "(C)"},
123
        {"&laquo;" , "<<"},
124
        {"&reg;"   , "(R)"},
125

    
126
        {"&deg;"   , "\xa1\xeb"},
127
        {"&plusmn;", "\xa1\xde"},
128
        {"&sup2;"  , "^2"},
129
        {"&sup3;"  , "^3"},
130
        {"&acute;" , "'"},
131
        {"&micro;" , "\xa6\xcc"},
132
        {"&para;"  , "\xa2\xf9"},
133
        {"&middot;", "\xa1\xa6"},
134
        {"&cedil;" , ","},
135
        {"&sup1;"  , "^1"},
136
        {"&raquo;" , ">>"},
137
        {"&frac14;", "1/4"},
138
        {"&frac12;", "1/2"},
139
        {"&frac34;", "3/4"},
140
        {"&iquest;", "^?"},
141

    
142
        {"&Agrave;", "A`"},
143
        {"&Aacute;", "A'"},
144
        {"&Acirc;" , "A^"},
145
        {"&Atilde;", "A~"},
146
        {"&Auml;"  , "A\xa1\xaf"},
147
        {"&Aring;" , "A\xa1\xeb"},
148
        {"&AElig;" , "AE"},
149
        {"&Egrave;", "E`"},
150
        {"&Eacute;", "E'"},
151
        {"&Ecirc;" , "E^"},
152
        {"&Euml;"  , "E\xa1\xaf"},
153
        {"&Igrave;", "I`"},
154
        {"&Iacute;", "I'"},
155
        {"&Icirc;" , "I^"},
156
        {"&Iuml;"  , "I\xa1\xaf"},
157

    
158
        {"&Ntilde;", "N~"},
159
        {"&Ograve;", "O`"},
160
        {"&Oacute;", "O'"},
161
        {"&Ocirc;" , "O^"},
162
        {"&Otilde;", "O~"},
163
        {"&Ouml;"  , "O\xa1\xaf"},
164
        {"&times;" , "\xa1\xdf"},
165
        {"&Ugrave;", "U`"},
166
        {"&Uacute;", "U'"},
167
        {"&Ucirc;" , "U^"},
168
        {"&Uuml;"  , "U\xa1\xaf"},
169
        {"&Yacute;", "Y'"},
170

    
171
        {"&agrave;", "a`"},
172
        {"&aacute;", "a'"},
173
        {"&acirc;" , "a^"},
174
        {"&atilde;", "a~"},
175
        {"&auml;"  , "a\xa1\xaf"},
176
        {"&aring;" , "a\xa1\xeb"},
177
        {"&aelig;" , "ae"},
178
        {"&egrave;", "e`"},
179
        {"&eacute;", "e'"},
180
        {"&ecirc;" , "e^"},
181
        {"&euml;"  , "e\xa1\xaf"},
182
        {"&igrave;", "i`"},
183
        {"&iacute;", "i'"},
184
        {"&icirc;" , "i^"},
185
        {"&iuml;"  , "i\xa1\xaf"},
186

    
187
        {"&eth;"   , "\xa2\xdf"},
188
        {"&ntilde;", "n~"},
189
        {"&ograve;", "o`"},
190
        {"&oacute;", "o'"},
191
        {"&ocirc;" , "o^"},
192
        {"&otilde;", "o~"},
193
        {"&ouml;"  , "o\xa1\xaf"},
194
        {"&divide;", "\xa1\xe0"},
195
        {"&ugrave;", "u`"},
196
        {"&uacute;", "u'"},
197
        {"&ucirc;" , "u^"},
198
        {"&uuml;"  , "u\xa1\xaf"},
199
        {"&yacute;", "y'"},
200
        {"&yuml;"  , "y\xa1\xaf"},
201
};
202

    
203
static HTMLSymbol latin_symbol_list[] = {
204
        {"&iexcl;" , "\xa1"},
205
        {"&cent;"  , "\xa2"},
206
        {"&pound;" , "\xa3"},
207
        {"&curren;", "\xa4"},
208
        {"&yen;"   , "\xa5"},
209
        {"&brvbar;", "\xa6"},
210
        {"&sect;"  , "\xa7"},
211
        {"&uml;"   , "\xa8"},
212
        {"&copy;"  , "\xa9"},
213
        {"&ordf;"  , "\xaa"},
214
        {"&laquo;" , "\xab"},
215
        {"&not;"   , "\xac"},
216
        {"&shy;"   , "\xad"},
217
        {"&reg;"   , "\xae"},
218
        {"&macr;"  , "\xaf"},
219

    
220
        {"&deg;"   , "\xb0"},
221
        {"&plusmn;", "\xb1"},
222
        {"&sup2;"  , "\xb2"},
223
        {"&sup3;"  , "\xb3"},
224
        {"&acute;" , "\xb4"},
225
        {"&micro;" , "\xb5"},
226
        {"&para;"  , "\xb6"},
227
        {"&middot;", "\xb7"},
228
        {"&cedil;" , "\xb8"},
229
        {"&sup1;"  , "\xb9"},
230
        {"&ordm;"  , "\xba"},
231
        {"&raquo;" , "\xbb"},
232
        {"&frac14;", "\xbc"},
233
        {"&frac12;", "\xbd"},
234
        {"&frac34;", "\xbe"},
235
        {"&iquest;", "\xbf"},
236

    
237
        {"&Agrave;", "\xc0"},
238
        {"&Aacute;", "\xc1"},
239
        {"&Acirc;" , "\xc2"},
240
        {"&Atilde;", "\xc3"},
241
        {"&Auml;"  , "\xc4"},
242
        {"&Aring;" , "\xc5"},
243
        {"&AElig;" , "\xc6"},
244
        {"&Ccedil;", "\xc7"},
245
        {"&Egrave;", "\xc8"},
246
        {"&Eacute;", "\xc9"},
247
        {"&Ecirc;" , "\xca"},
248
        {"&Euml;"  , "\xcb"},
249
        {"&Igrave;", "\xcc"},
250
        {"&Iacute;", "\xcd"},
251
        {"&Icirc;" , "\xce"},
252
        {"&Iuml;"  , "\xcf"},
253

    
254
        {"&ETH;"   , "\xd0"},
255
        {"&Ntilde;", "\xd1"},
256
        {"&Ograve;", "\xd2"},
257
        {"&Oacute;", "\xd3"},
258
        {"&Ocirc;" , "\xd4"},
259
        {"&Otilde;", "\xd5"},
260
        {"&Ouml;"  , "\xd6"},
261
        {"&times;" , "\xd7"},
262
        {"&Oslash;", "\xd8"},
263
        {"&Ugrave;", "\xd9"},
264
        {"&Uacute;", "\xda"},
265
        {"&Ucirc;" , "\xdb"},
266
        {"&Uuml;"  , "\xdc"},
267
        {"&Yacute;", "\xdd"},
268
        {"&THORN;" , "\xde"},
269
        {"&szlig;" , "\xdf"},
270

    
271
        {"&agrave;", "\xe0"},
272
        {"&aacute;", "\xe1"},
273
        {"&acirc;" , "\xe2"},
274
        {"&atilde;", "\xe3"},
275
        {"&auml;"  , "\xe4"},
276
        {"&aring;" , "\xe5"},
277
        {"&aelig;" , "\xe6"},
278
        {"&ccedil;", "\xe7"},
279
        {"&egrave;", "\xe8"},
280
        {"&eacute;", "\xe9"},
281
        {"&ecirc;" , "\xea"},
282
        {"&euml;"  , "\xeb"},
283
        {"&igrave;", "\xec"},
284
        {"&iacute;", "\xed"},
285
        {"&icirc;" , "\xee"},
286
        {"&iuml;"  , "\xef"},
287

    
288
        {"&eth;"   , "\xf0"},
289
        {"&ntilde;", "\xf1"},
290
        {"&ograve;", "\xf2"},
291
        {"&oacute;", "\xf3"},
292
        {"&ocirc;" , "\xf4"},
293
        {"&otilde;", "\xf5"},
294
        {"&ouml;"  , "\xf6"},
295
        {"&divide;", "\xf7"},
296
        {"&oslash;", "\xf8"},
297
        {"&ugrave;", "\xf9"},
298
        {"&uacute;", "\xfa"},
299
        {"&ucirc;" , "\xfb"},
300
        {"&uuml;"  , "\xfc"},
301
        {"&yacute;", "\xfd"},
302
        {"&thorn;" , "\xfe"},
303
        {"&yuml;"  , "\xff"},
304
};
305

    
306
static GHashTable *default_symbol_table;
307
static GHashTable *eucjp_symbol_table;
308
static GHashTable *latin_symbol_table;
309

    
310
static HTMLState html_read_line                (HTMLParser        *parser);
311
static void html_append_char                (HTMLParser        *parser,
312
                                         gchar                 ch);
313
static void html_append_str                (HTMLParser        *parser,
314
                                         const gchar        *str,
315
                                         gint                 len);
316
static HTMLState html_parse_tag                (HTMLParser        *parser);
317
static void html_parse_special                (HTMLParser        *parser);
318
static void html_get_parenthesis        (HTMLParser        *parser,
319
                                         gchar                *buf,
320
                                         gint                 len);
321

    
322

    
323
HTMLParser *html_parser_new(FILE *fp, CodeConverter *conv)
324
{
325
        HTMLParser *parser;
326

    
327
        g_return_val_if_fail(fp != NULL, NULL);
328
        g_return_val_if_fail(conv != NULL, NULL);
329

    
330
        parser = g_new0(HTMLParser, 1);
331
        parser->fp = fp;
332
        parser->conv = conv;
333
        parser->str = g_string_new(NULL);
334
        parser->buf = g_string_new(NULL);
335
        parser->bufp = parser->buf->str;
336
        parser->state = HTML_NORMAL;
337
        parser->href = NULL;
338
        parser->newline = TRUE;
339
        parser->empty_line = TRUE;
340
        parser->space = FALSE;
341
        parser->pre = FALSE;
342

    
343
#define SYMBOL_TABLE_ADD(table, list) \
344
{ \
345
        gint i; \
346
 \
347
        for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
348
                g_hash_table_insert(table, list[i].key, list[i].val); \
349
}
350

    
351
        if (!default_symbol_table) {
352
                default_symbol_table =
353
                        g_hash_table_new(g_str_hash, g_str_equal);
354
                SYMBOL_TABLE_ADD(default_symbol_table, symbol_list);
355
                SYMBOL_TABLE_ADD(default_symbol_table, ascii_symbol_list);
356
        }
357
        if (!eucjp_symbol_table) {
358
                eucjp_symbol_table =
359
                        g_hash_table_new(g_str_hash, g_str_equal);
360
                SYMBOL_TABLE_ADD(eucjp_symbol_table, symbol_list);
361
                SYMBOL_TABLE_ADD(eucjp_symbol_table, eucjp_symbol_list);
362
        }
363
        if (!latin_symbol_table) {
364
                latin_symbol_table =
365
                        g_hash_table_new(g_str_hash, g_str_equal);
366
                SYMBOL_TABLE_ADD(latin_symbol_table, symbol_list);
367
                SYMBOL_TABLE_ADD(latin_symbol_table, latin_symbol_list);
368
        }
369

    
370
#undef SYMBOL_TABLE_ADD
371

    
372
        if (conv->charset == C_ISO_8859_1)
373
                parser->symbol_table = latin_symbol_table;
374
        else if ((conv->charset == C_ISO_2022_JP   ||
375
                  conv->charset == C_ISO_2022_JP_2 ||
376
                  conv->charset == C_EUC_JP        ||
377
                  conv->charset == C_SHIFT_JIS) &&
378
                 conv_get_locale_charset() == C_EUC_JP)
379
                parser->symbol_table = eucjp_symbol_table;
380
        else
381
                parser->symbol_table = default_symbol_table;
382

    
383
        return parser;
384
}
385

    
386
void html_parser_destroy(HTMLParser *parser)
387
{
388
        g_string_free(parser->str, TRUE);
389
        g_string_free(parser->buf, TRUE);
390
        g_free(parser->href);
391
        g_free(parser);
392
}
393

    
394
gchar *html_parse(HTMLParser *parser)
395
{
396
        parser->state = HTML_NORMAL;
397
        g_string_truncate(parser->str, 0);
398

    
399
        if (*parser->bufp == '\0') {
400
                g_string_truncate(parser->buf, 0);
401
                parser->bufp = parser->buf->str;
402
                if (html_read_line(parser) == HTML_EOF)
403
                        return NULL;
404
        }
405

    
406
        while (*parser->bufp != '\0') {
407
                switch (*parser->bufp) {
408
                case '<':
409
                        if (parser->str->len == 0)
410
                                html_parse_tag(parser);
411
                        else
412
                                return parser->str->str;
413
                        break;
414
                case '&':
415
                        html_parse_special(parser);
416
                        break;
417
                case ' ':
418
                case '\t':
419
                case '\r':
420
                case '\n':
421
                        if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
422
                                parser->bufp++;
423

    
424
                        if (!parser->pre) {
425
                                if (!parser->newline)
426
                                        parser->space = TRUE;
427

    
428
                                parser->bufp++;
429
                                break;
430
                        }
431
                        /* fallthrough */
432
                default:
433
                        html_append_char(parser, *parser->bufp++);
434
                }
435
        }
436

    
437
        return parser->str->str;
438
}
439

    
440
static HTMLState html_read_line(HTMLParser *parser)
441
{
442
        gchar buf[HTMLBUFSIZE];
443
        gchar buf2[HTMLBUFSIZE];
444
        gint index;
445

    
446
        if (fgets(buf, sizeof(buf), parser->fp) == NULL) {
447
                parser->state = HTML_EOF;
448
                return HTML_EOF;
449
        }
450

    
451
        if (conv_convert(parser->conv, buf2, sizeof(buf2), buf) < 0) {
452
                index = parser->bufp - parser->buf->str;
453

    
454
                conv_localetodisp(buf2, sizeof(buf2), buf);
455
                g_string_append(parser->buf, buf2);
456

    
457
                parser->bufp = parser->buf->str + index;
458

    
459
                return HTML_CONV_FAILED;
460
        }
461

    
462
        index = parser->bufp - parser->buf->str;
463

    
464
        g_string_append(parser->buf, buf2);
465

    
466
        parser->bufp = parser->buf->str + index;
467

    
468
        return HTML_NORMAL;
469
}
470

    
471
static void html_append_char(HTMLParser *parser, gchar ch)
472
{
473
        GString *str = parser->str;
474

    
475
        if (!parser->pre && parser->space) {
476
                g_string_append_c(str, ' ');
477
                parser->space = FALSE;
478
        }
479

    
480
        g_string_append_c(str, ch);
481

    
482
        parser->empty_line = FALSE;
483
        if (ch == '\n') {
484
                parser->newline = TRUE;
485
                if (str->len > 1 && str->str[str->len - 2] == '\n')
486
                        parser->empty_line = TRUE;
487
        } else
488
                parser->newline = FALSE;
489
}
490

    
491
static void html_append_str(HTMLParser *parser, const gchar *str, gint len)
492
{
493
        GString *string = parser->str;
494

    
495
        if (!parser->pre && parser->space) {
496
                g_string_append_c(string, ' ');
497
                parser->space = FALSE;
498
        }
499

    
500
        if (len == 0) return;
501
        if (len < 0)
502
                g_string_append(string, str);
503
        else {
504
                gchar *s;
505
                Xstrndup_a(s, str, len, return);
506
                g_string_append(string, s);
507
        }
508

    
509
        parser->empty_line = FALSE;
510
        if (string->len > 0 && string->str[string->len - 1] == '\n') {
511
                parser->newline = TRUE;
512
                if (string->len > 1 && string->str[string->len - 2] == '\n')
513
                        parser->empty_line = TRUE;
514
        } else
515
                parser->newline = FALSE;
516
}
517

    
518
static HTMLTag *html_get_tag(const gchar *str)
519
{
520
        HTMLTag *tag;
521
        gchar *tmp;
522
        guchar *tmpp;
523

    
524
        g_return_val_if_fail(str != NULL, NULL);
525

    
526
        if (*str == '\0' || *str == '!') return NULL;
527

    
528
        Xstrdup_a(tmp, str, return NULL);
529

    
530
        tag = g_new0(HTMLTag, 1);
531

    
532
        for (tmpp = tmp; *tmpp != '\0' && !isspace(*tmpp); tmpp++)
533
                ;
534

    
535
        if (*tmpp == '\0') {
536
                g_strdown(tmp);
537
                tag->name = g_strdup(tmp);
538
                return tag;
539
        } else {
540
                *tmpp++ = '\0';
541
                g_strdown(tmp);
542
                tag->name = g_strdup(tmp);
543
        }
544

    
545
        while (*tmpp != '\0') {
546
                HTMLAttr *attr;
547
                gchar *attr_name;
548
                gchar *attr_value;
549
                gchar *p;
550
                gchar quote;
551

    
552
                while (isspace(*tmpp)) tmpp++;
553
                attr_name = tmpp;
554

    
555
                while (*tmpp != '\0' && !isspace(*tmpp) && *tmpp != '=') tmpp++;
556
                if (*tmpp != '\0' && *tmpp != '=') {
557
                        *tmpp++ = '\0';
558
                        while (isspace(*tmpp)) tmpp++;
559
                }
560

    
561
                if (*tmpp == '=') {
562
                        *tmpp++ = '\0';
563
                        while (isspace(*tmpp)) tmpp++;
564

    
565
                        if (*tmpp == '"' || *tmpp == '\'') {
566
                                /* name="value" */
567
                                quote = *tmpp;
568
                                tmpp++;
569
                                attr_value = tmpp;
570
                                if ((p = strchr(attr_value, quote)) == NULL) {
571
                                        g_warning("html_get_tag(): syntax error in tag: '%s'\n", str);
572
                                        return tag;
573
                                }
574
                                tmpp = p;
575
                                *tmpp++ = '\0';
576
                                while (isspace(*tmpp)) tmpp++;
577
                        } else {
578
                                /* name=value */
579
                                attr_value = tmpp;
580
                                while (*tmpp != '\0' && !isspace(*tmpp)) tmpp++;
581
                                if (*tmpp != '\0')
582
                                        *tmpp++ = '\0';
583
                        }
584
                } else
585
                        attr_value = "";
586

    
587
                g_strchomp(attr_name);
588
                g_strdown(attr_name);
589
                attr = g_new(HTMLAttr, 1);
590
                attr->name = g_strdup(attr_name);
591
                attr->value = g_strdup(attr_value);
592
                tag->attr = g_list_append(tag->attr, attr);
593
        }
594

    
595
        return tag;
596
}
597

    
598
static void html_free_tag(HTMLTag *tag)
599
{
600
        if (!tag) return;
601

    
602
        g_free(tag->name);
603
        while (tag->attr != NULL) {
604
                HTMLAttr *attr = (HTMLAttr *)tag->attr->data;
605
                g_free(attr->name);
606
                g_free(attr->value);
607
                g_free(attr);
608
                tag->attr = g_list_remove(tag->attr, tag->attr->data);
609
        }
610
        g_free(tag);
611
}
612

    
613
static HTMLState html_parse_tag(HTMLParser *parser)
614
{
615
        gchar buf[HTMLBUFSIZE];
616
        HTMLTag *tag;
617

    
618
        html_get_parenthesis(parser, buf, sizeof(buf));
619

    
620
        tag = html_get_tag(buf);
621

    
622
        parser->state = HTML_UNKNOWN;
623
        if (!tag) return HTML_UNKNOWN;
624

    
625
        if (!strcmp(tag->name, "br")) {
626
                parser->space = FALSE;
627
                html_append_char(parser, '\n');
628
                parser->state = HTML_BR;
629
        } else if (!strcmp(tag->name, "a")) {
630
                if (tag->attr && tag->attr->data &&
631
                    !strcmp(((HTMLAttr *)tag->attr->data)->name, "href")) {
632
                        g_free(parser->href);
633
                        parser->href =
634
                                g_strdup(((HTMLAttr *)tag->attr->data)->value);
635
                        parser->state = HTML_HREF;
636
                }
637
        } else if (!strcmp(tag->name, "/a")) {
638
                g_free(parser->href);
639
                parser->href = NULL;
640
                parser->state = HTML_NORMAL;
641
        } else if (!strcmp(tag->name, "p")) {
642
                parser->space = FALSE;
643
                if (!parser->empty_line) {
644
                        parser->space = FALSE;
645
                        if (!parser->newline) html_append_char(parser, '\n');
646
                        html_append_char(parser, '\n');
647
                }
648
                parser->state = HTML_PAR;
649
        } else if (!strcmp(tag->name, "pre")) {
650
                parser->pre = TRUE;
651
                parser->state = HTML_PRE;
652
        } else if (!strcmp(tag->name, "/pre")) {
653
                parser->pre = FALSE;
654
                parser->state = HTML_NORMAL;
655
        } else if (!strcmp(tag->name, "hr")) {
656
                if (!parser->newline) {
657
                        parser->space = FALSE;
658
                        html_append_char(parser, '\n');
659
                }
660
                html_append_str(parser, HR_STR "\n", -1);
661
                parser->state = HTML_HR;
662
        } else if (!strcmp(tag->name, "div")    ||
663
                   !strcmp(tag->name, "ul")     ||
664
                   !strcmp(tag->name, "li")     ||
665
                   !strcmp(tag->name, "table")  ||
666
                   !strcmp(tag->name, "tr")     ||
667
                   (tag->name[0] == 'h' && isdigit((guchar)tag->name[1]))) {
668
                if (!parser->newline) {
669
                        parser->space = FALSE;
670
                        html_append_char(parser, '\n');
671
                }
672
                parser->state = HTML_NORMAL;
673
        } else if (!strcmp(tag->name, "/table") ||
674
                   (tag->name[0] == '/' &&
675
                    tag->name[1] == 'h' &&
676
                    isdigit((guchar)tag->name[1]))) {
677
                if (!parser->empty_line) {
678
                        parser->space = FALSE;
679
                        if (!parser->newline) html_append_char(parser, '\n');
680
                        html_append_char(parser, '\n');
681
                }
682
                parser->state = HTML_NORMAL;
683
        } else if (!strcmp(tag->name, "/div")   ||
684
                   !strcmp(tag->name, "/ul")    ||
685
                   !strcmp(tag->name, "/li")) {
686
                if (!parser->newline) {
687
                        parser->space = FALSE;
688
                        html_append_char(parser, '\n');
689
                }
690
                parser->state = HTML_NORMAL;
691
                        }
692

    
693
        html_free_tag(tag);
694

    
695
        return parser->state;
696
}
697

    
698
static void html_parse_special(HTMLParser *parser)
699
{
700
        gchar symbol_name[9];
701
        gint n;
702
        const gchar *val;
703

    
704
        parser->state = HTML_UNKNOWN;
705
        g_return_if_fail(*parser->bufp == '&');
706

    
707
        /* &foo; */
708
        for (n = 0; parser->bufp[n] != '\0' && parser->bufp[n] != ';'; n++)
709
                ;
710
        if (n > 7 || parser->bufp[n] != ';') {
711
                /* output literal `&' */
712
                html_append_char(parser, *parser->bufp++);
713
                parser->state = HTML_NORMAL;
714
                return;
715
        }
716
        strncpy2(symbol_name, parser->bufp, n + 2);
717
        parser->bufp += n + 1;
718

    
719
        if ((val = g_hash_table_lookup(parser->symbol_table, symbol_name))
720
            != NULL) {
721
                html_append_str(parser, val, -1);
722
                parser->state = HTML_NORMAL;
723
                return;
724
        } else if (symbol_name[1] == '#' && isdigit((guchar)symbol_name[2])) {
725
                gint ch;
726

    
727
                ch = atoi(symbol_name + 2);
728
                if ((ch > 0 && ch <= 127) ||
729
                    (ch >= 128 && ch <= 255 &&
730
                     parser->conv->charset == C_ISO_8859_1)) {
731
                        html_append_char(parser, ch);
732
                        parser->state = HTML_NORMAL;
733
                        return;
734
                }
735
        }
736

    
737
        html_append_str(parser, symbol_name, -1);
738
}
739

    
740
static void html_get_parenthesis(HTMLParser *parser, gchar *buf, gint len)
741
{
742
        gchar *p;
743

    
744
        buf[0] = '\0';
745
        g_return_if_fail(*parser->bufp == '<');
746

    
747
        /* ignore comment / CSS / script stuff */
748
        if (!strncmp(parser->bufp, "<!--", 4)) {
749
                parser->bufp += 4;
750
                while ((p = strstr(parser->bufp, "-->")) == NULL)
751
                        if (html_read_line(parser) == HTML_EOF) return;
752
                parser->bufp = p + 3;
753
                return;
754
        }
755
        if (!g_strncasecmp(parser->bufp, "<style", 6)) {
756
                parser->bufp += 6;
757
                while ((p = strcasestr(parser->bufp, "</style>")) == NULL)
758
                        if (html_read_line(parser) == HTML_EOF) return;
759
                parser->bufp = p + 8;
760
                return;
761
        }
762
        if (!g_strncasecmp(parser->bufp, "<script", 7)) {
763
                parser->bufp += 7;
764
                while ((p = strcasestr(parser->bufp, "</script>")) == NULL)
765
                        if (html_read_line(parser) == HTML_EOF) return;
766
                parser->bufp = p + 9;
767
                return;
768
        }
769

    
770
        parser->bufp++;
771
        while ((p = strchr(parser->bufp, '>')) == NULL)
772
                if (html_read_line(parser) == HTML_EOF) return;
773

    
774
        strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
775
        g_strstrip(buf);
776
        parser->bufp = p + 1;
777
}