2 * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
3 * Copyright (C) 1999-2006 Hiroyuki Yamamoto and the Sylpheed-Claws team
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.
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.
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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
29 #define SC_HTMLBUFSIZE 8192
30 #define HR_STR "------------------------------------------------"
32 typedef struct _SC_HTMLSymbol SC_HTMLSymbol;
40 static SC_HTMLSymbol symbol_list[] = {
55 static SC_HTMLSymbol ascii_symbol_list[] = {
118 static GHashTable *default_symbol_table;
120 static SC_HTMLState sc_html_read_line (SC_HTMLParser *parser);
121 static void sc_html_append_char (SC_HTMLParser *parser,
123 static void sc_html_append_str (SC_HTMLParser *parser,
126 static SC_HTMLState sc_html_parse_tag (SC_HTMLParser *parser);
127 static void sc_html_parse_special (SC_HTMLParser *parser);
128 static void sc_html_get_parenthesis (SC_HTMLParser *parser,
133 SC_HTMLParser *sc_html_parser_new(FILE *fp, CodeConverter *conv)
135 SC_HTMLParser *parser;
137 g_return_val_if_fail(fp != NULL, NULL);
138 g_return_val_if_fail(conv != NULL, NULL);
140 parser = g_new0(SC_HTMLParser, 1);
143 parser->str = g_string_new(NULL);
144 parser->buf = g_string_new(NULL);
145 parser->bufp = parser->buf->str;
146 parser->state = SC_HTML_NORMAL;
148 parser->newline = TRUE;
149 parser->empty_line = TRUE;
150 parser->space = FALSE;
153 #define SYMBOL_TABLE_ADD(table, list) \
157 for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
158 g_hash_table_insert(table, list[i].key, list[i].val); \
161 if (!default_symbol_table) {
162 default_symbol_table =
163 g_hash_table_new(g_str_hash, g_str_equal);
164 SYMBOL_TABLE_ADD(default_symbol_table, symbol_list);
165 SYMBOL_TABLE_ADD(default_symbol_table, ascii_symbol_list);
168 #undef SYMBOL_TABLE_ADD
170 parser->symbol_table = default_symbol_table;
175 void sc_html_parser_destroy(SC_HTMLParser *parser)
177 g_string_free(parser->str, TRUE);
178 g_string_free(parser->buf, TRUE);
179 g_free(parser->href);
183 gchar *sc_html_parse(SC_HTMLParser *parser)
185 parser->state = SC_HTML_NORMAL;
186 g_string_truncate(parser->str, 0);
188 if (*parser->bufp == '\0') {
189 g_string_truncate(parser->buf, 0);
190 parser->bufp = parser->buf->str;
191 if (sc_html_read_line(parser) == SC_HTML_EOF)
195 while (*parser->bufp != '\0') {
196 switch (*parser->bufp) {
199 st = sc_html_parse_tag(parser);
200 /* when we see an href, we need to flush the str
201 * buffer. Then collect all the chars until we
202 * see the end anchor tag
204 if (SC_HTML_HREF_BEG == st || SC_HTML_HREF == st)
205 return parser->str->str;
209 sc_html_parse_special(parser);
215 if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
219 if (!parser->newline)
220 parser->space = TRUE;
227 sc_html_append_char(parser, *parser->bufp++);
231 return parser->str->str;
234 static SC_HTMLState sc_html_read_line(SC_HTMLParser *parser)
236 gchar buf[SC_HTMLBUFSIZE];
237 gchar buf2[SC_HTMLBUFSIZE];
240 if (fgets(buf, sizeof(buf), parser->fp) == NULL) {
241 parser->state = SC_HTML_EOF;
245 if (conv_convert(parser->conv, buf2, sizeof(buf2), buf) < 0) {
246 index = parser->bufp - parser->buf->str;
248 conv_utf8todisp(buf2, sizeof(buf2), buf);
249 g_string_append(parser->buf, buf2);
251 parser->bufp = parser->buf->str + index;
253 return SC_HTML_CONV_FAILED;
256 index = parser->bufp - parser->buf->str;
258 g_string_append(parser->buf, buf2);
260 parser->bufp = parser->buf->str + index;
262 return SC_HTML_NORMAL;
265 static void sc_html_append_char(SC_HTMLParser *parser, gchar ch)
267 GString *str = parser->str;
269 if (!parser->pre && parser->space) {
270 g_string_append_c(str, ' ');
271 parser->space = FALSE;
274 g_string_append_c(str, ch);
276 parser->empty_line = FALSE;
278 parser->newline = TRUE;
279 if (str->len > 1 && str->str[str->len - 2] == '\n')
280 parser->empty_line = TRUE;
282 parser->newline = FALSE;
285 static void sc_html_append_str(SC_HTMLParser *parser, const gchar *str, gint len)
287 GString *string = parser->str;
289 if (!parser->pre && parser->space) {
290 g_string_append_c(string, ' ');
291 parser->space = FALSE;
294 if (len == 0) return;
296 g_string_append(string, str);
299 Xstrndup_a(s, str, len, return);
300 g_string_append(string, s);
303 parser->empty_line = FALSE;
304 if (string->len > 0 && string->str[string->len - 1] == '\n') {
305 parser->newline = TRUE;
306 if (string->len > 1 && string->str[string->len - 2] == '\n')
307 parser->empty_line = TRUE;
309 parser->newline = FALSE;
312 static SC_HTMLTag *sc_html_get_tag(const gchar *str)
318 g_return_val_if_fail(str != NULL, NULL);
320 if (*str == '\0' || *str == '!') return NULL;
322 Xstrdup_a(tmp, str, return NULL);
324 tag = g_new0(SC_HTMLTag, 1);
326 for (tmpp = tmp; *tmpp != '\0' && !g_ascii_isspace(*tmpp); tmpp++)
331 tag->name = g_strdup(tmp);
336 tag->name = g_strdup(tmp);
339 while (*tmpp != '\0') {
346 while (g_ascii_isspace(*tmpp)) tmpp++;
349 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp) &&
352 if (*tmpp != '\0' && *tmpp != '=') {
354 while (g_ascii_isspace(*tmpp)) tmpp++;
359 while (g_ascii_isspace(*tmpp)) tmpp++;
361 if (*tmpp == '"' || *tmpp == '\'') {
366 if ((p = strchr(attr_value, quote)) == NULL) {
367 g_warning("sc_html_get_tag(): syntax error in tag: '%s'\n", str);
372 while (g_ascii_isspace(*tmpp)) tmpp++;
376 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp)) tmpp++;
383 g_strchomp(attr_name);
384 g_strdown(attr_name);
385 attr = g_new(SC_HTMLAttr, 1);
386 attr->name = g_strdup(attr_name);
387 attr->value = g_strdup(attr_value);
388 tag->attr = g_list_append(tag->attr, attr);
394 static void sc_html_free_tag(SC_HTMLTag *tag)
399 while (tag->attr != NULL) {
400 SC_HTMLAttr *attr = (SC_HTMLAttr *)tag->attr->data;
404 tag->attr = g_list_remove(tag->attr, tag->attr->data);
409 static SC_HTMLState sc_html_parse_tag(SC_HTMLParser *parser)
411 gchar buf[SC_HTMLBUFSIZE];
414 sc_html_get_parenthesis(parser, buf, sizeof(buf));
416 tag = sc_html_get_tag(buf);
418 parser->state = SC_HTML_UNKNOWN;
419 if (!tag) return SC_HTML_UNKNOWN;
421 if (!strcmp(tag->name, "br")) {
422 parser->space = FALSE;
423 sc_html_append_char(parser, '\n');
424 parser->state = SC_HTML_BR;
425 } else if (!strcmp(tag->name, "a")) {
427 for (cur = tag->attr; cur != NULL; cur = cur->next) {
428 if (cur->data && !strcmp(((SC_HTMLAttr *)cur->data)->name, "href")) {
429 g_free(parser->href);
430 parser->href = g_strdup(((SC_HTMLAttr *)cur->data)->value);
431 parser->state = SC_HTML_HREF_BEG;
435 } else if (!strcmp(tag->name, "/a")) {
436 parser->state = SC_HTML_HREF;
437 } else if (!strcmp(tag->name, "p")) {
438 parser->space = FALSE;
439 if (!parser->empty_line) {
440 parser->space = FALSE;
441 if (!parser->newline) sc_html_append_char(parser, '\n');
442 sc_html_append_char(parser, '\n');
444 parser->state = SC_HTML_PAR;
445 } else if (!strcmp(tag->name, "pre")) {
447 parser->state = SC_HTML_PRE;
448 } else if (!strcmp(tag->name, "/pre")) {
450 parser->state = SC_HTML_NORMAL;
451 } else if (!strcmp(tag->name, "hr")) {
452 if (!parser->newline) {
453 parser->space = FALSE;
454 sc_html_append_char(parser, '\n');
456 sc_html_append_str(parser, HR_STR "\n", -1);
457 parser->state = SC_HTML_HR;
458 } else if (!strcmp(tag->name, "div") ||
459 !strcmp(tag->name, "ul") ||
460 !strcmp(tag->name, "li") ||
461 !strcmp(tag->name, "table") ||
462 !strcmp(tag->name, "tr") ||
463 (tag->name[0] == 'h' && g_ascii_isdigit(tag->name[1]))) {
464 if (!parser->newline) {
465 parser->space = FALSE;
466 sc_html_append_char(parser, '\n');
468 parser->state = SC_HTML_NORMAL;
469 } else if (!strcmp(tag->name, "/table") ||
470 (tag->name[0] == '/' &&
471 tag->name[1] == 'h' &&
472 g_ascii_isdigit(tag->name[1]))) {
473 if (!parser->empty_line) {
474 parser->space = FALSE;
475 if (!parser->newline) sc_html_append_char(parser, '\n');
476 sc_html_append_char(parser, '\n');
478 parser->state = SC_HTML_NORMAL;
479 } else if (!strcmp(tag->name, "/div") ||
480 !strcmp(tag->name, "/ul") ||
481 !strcmp(tag->name, "/li")) {
482 if (!parser->newline) {
483 parser->space = FALSE;
484 sc_html_append_char(parser, '\n');
486 parser->state = SC_HTML_NORMAL;
489 sc_html_free_tag(tag);
491 return parser->state;
494 static void sc_html_parse_special(SC_HTMLParser *parser)
496 gchar symbol_name[9];
500 parser->state = SC_HTML_UNKNOWN;
501 g_return_if_fail(*parser->bufp == '&');
504 for (n = 0; parser->bufp[n] != '\0' && parser->bufp[n] != ';'; n++)
506 if (n > 7 || parser->bufp[n] != ';') {
507 /* output literal `&' */
508 sc_html_append_char(parser, *parser->bufp++);
509 parser->state = SC_HTML_NORMAL;
512 strncpy2(symbol_name, parser->bufp, n + 2);
513 parser->bufp += n + 1;
515 if ((val = g_hash_table_lookup(parser->symbol_table, symbol_name))
517 sc_html_append_str(parser, val, -1);
518 parser->state = SC_HTML_NORMAL;
520 } else if (symbol_name[1] == '#' && g_ascii_isdigit(symbol_name[2])) {
523 ch = atoi(symbol_name + 2);
524 if ((ch > 0 && ch <= 127) ||
525 (ch >= 128 && ch <= 255 &&
526 parser->conv->charset == C_ISO_8859_1)) {
527 sc_html_append_char(parser, ch);
528 parser->state = SC_HTML_NORMAL;
533 /* http://www.w3schools.com/html/html_entitiesref.asp */
534 case 96: /* backtick */
537 case 338: /* capital ligature OE Œ */
540 case 339: /* small ligature OE œ */
543 case 352: /* capital S w/caron Š */
544 case 353: /* small S w/caron š */
545 case 376: /* cap Y w/ diaeres Ÿ */
547 case 710: /* circumflex accent ˆ */
550 case 732: /* small tilde ˜ */
553 case 8194: /* en space   */
554 case 8195: /* em space   */
555 case 8201: /* thin space   */
558 case 8204: /* zero width non-joiner ‌ */
559 case 8205: /* zero width joiner ‍ */
560 case 8206: /* l-t-r mark ‎ */
561 case 8207: /* r-t-l mark &rlm */
563 case 8211: /* en dash – */
566 case 8212: /* em dash — */
569 case 8216: /* l single quot mark ‘ */
570 case 8217: /* r single quot mark ’ */
573 case 8218: /* single low-9 quot ‚ */
576 case 8220: /* l double quot mark “ */
577 case 8221: /* r double quot mark ” */
580 case 8222: /* double low-9 quot „ */
583 case 8224: /* dagger † */
584 case 8225: /* double dagger ‡ */
586 case 8230: /* horizontal ellipsis … */
589 case 8240: /* per mile ‰ */
592 case 8249: /* l-pointing angle quot ‹ */
595 case 8250: /* r-pointing angle quot › */
598 case 8364: /* euro € */
601 case 8482: /* trademark ™ */
608 sc_html_append_str(parser, symb, -1);
609 parser->state = SC_HTML_NORMAL;
615 sc_html_append_str(parser, symbol_name, -1);
618 static void sc_html_get_parenthesis(SC_HTMLParser *parser, gchar *buf, gint len)
623 g_return_if_fail(*parser->bufp == '<');
625 /* ignore comment / CSS / script stuff */
626 if (!strncmp(parser->bufp, "<!--", 4)) {
628 while ((p = strstr(parser->bufp, "-->")) == NULL)
629 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
630 parser->bufp = p + 3;
633 if (!g_ascii_strncasecmp(parser->bufp, "<style", 6)) {
635 while ((p = strcasestr(parser->bufp, "</style>")) == NULL)
636 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
637 parser->bufp = p + 8;
640 if (!g_ascii_strncasecmp(parser->bufp, "<script", 7)) {
642 while ((p = strcasestr(parser->bufp, "</script>")) == NULL)
643 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
644 parser->bufp = p + 9;
649 while ((p = strchr(parser->bufp, '>')) == NULL)
650 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
652 strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
654 parser->bufp = p + 1;