src/html.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 1999-2007 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 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., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, 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 SC_HTMLBUFSIZE  8192
  30 #define HR_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         {"&lt;"    , "<"},
  42         {"&gt;"    , ">"},
  43         {"&amp;"   , "&"},
  44         {"&quot;"  , "\""},
  45         {"&lsquo;",  "'"},
  46         {"&rsquo;",  "'"},
  47         {"&ldquo;",  "\""},
  48         {"&rdquo;",  "\""},
  49         {"&nbsp;"  , " "},
  50         {"&trade;" , "(TM)"},
  51         {"&hellip;", "..."},
  52         {"&bull;", "*"},
  53         {"&ndash;", "-"},
  54         {"&mdash;", "--"},
  55 };
  56
  57 static SC_HTMLSymbol ascii_symbol_list[] = {
  58         {"&iexcl;" , "\302\241"},
  59         {"&brvbar;", "\302\246"},
  60         {"&copy;"  , "\302\251"},
  61         {"&laquo;" , "\302\253"},
  62         {"&reg;"   , "\302\256"},
  63
  64         {"&sup2;"  , "\302\262"},
  65         {"&sup3;"  , "\302\263"},
  66         {"&acute;" , "\302\264"},
  67         {"&cedil;" , "\302\270"},
  68         {"&sup1;"  , "\302\271"},
  69         {"&raquo;" , "\302\273"},
  70         {"&frac14;", "\302\274"},
  71         {"&frac12;", "\302\275"},
  72         {"&frac34;", "\302\276"},
  73         {"&iquest;", "\302\277"},
  74
  75         {"&Agrave;", "\303\200"},
  76         {"&Aacute;", "\303\201"},
  77         {"&Acirc;" , "\303\202"},
  78         {"&Atilde;", "\303\203"},
  79         {"&AElig;" , "\303\206"},
  80         {"&Egrave;", "\303\210"},
  81         {"&Eacute;", "\303\211"},
  82         {"&Ecirc;" , "\303\212"},
  83         {"&Igrave;", "\303\214"},
  84         {"&Iacute;", "\303\215"},
  85         {"&Icirc;" , "\303\216"},
  86
  87         {"&Ntilde;", "\303\221"},
  88         {"&Ograve;", "\303\222"},
  89         {"&Oacute;", "\303\223"},
  90         {"&Ocirc;" , "\303\224"},
  91         {"&Otilde;", "\303\225"},
  92         {"&Ugrave;", "\303\231"},
  93         {"&Uacute;", "\303\232"},
  94         {"&Ucirc;" , "\303\233"},
  95         {"&Yacute;", "\303\235"},
  96
  97         {"&agrave;", "\303\240"},
  98         {"&aacute;", "\303\241"},
  99         {"&acirc;" , "\303\242"},
 100         {"&atilde;", "\303\243"},
 101         {"&aelig;" , "\303\246"},
 102         {"&egrave;", "\303\250"},
 103         {"&eacute;", "\303\251"},
 104         {"&ecirc;" , "\303\252"},
 105         {"&igrave;", "\303\254"},
 106         {"&iacute;", "\303\255"},
 107         {"&icirc;" , "\303\256"},
 108
 109         {"&ntilde;", "\303\261"},
 110         {"&ograve;", "\303\262"},
 111         {"&oacute;", "\303\263"},
 112         {"&ocirc;" , "\303\264"},
 113         {"&otilde;", "\303\265"},
 114         {"&ugrave;", "\303\271"},
 115         {"&uacute;", "\303\272"},
 116         {"&ucirc;" , "\303\273"},
 117         {"&yacute;", "\303\275"},
 118 };
 119
 120 static GHashTable *default_symbol_table;
 121
 122 static SC_HTMLState sc_html_read_line   (SC_HTMLParser  *parser);
 123 static void sc_html_append_char                 (SC_HTMLParser  *parser,
 124                                          gchar           ch);
 125 static void sc_html_append_str                  (SC_HTMLParser  *parser,
 126                                          const gchar    *str,
 127                                          gint            len);
 128 static SC_HTMLState sc_html_parse_tag   (SC_HTMLParser  *parser);
 129 static void sc_html_parse_special               (SC_HTMLParser  *parser);
 130 static void sc_html_get_parenthesis             (SC_HTMLParser  *parser,
 131                                          gchar          *buf,
 132                                          gint            len);
 133
 134
 135 SC_HTMLParser *sc_html_parser_new(FILE *fp, CodeConverter *conv)
 136 {
 137         SC_HTMLParser *parser;
 138
 139         g_return_val_if_fail(fp != NULL, NULL);
 140         g_return_val_if_fail(conv != NULL, NULL);
 141
 142         parser = g_new0(SC_HTMLParser, 1);
 143         parser->fp = fp;
 144         parser->conv = conv;
 145         parser->str = g_string_new(NULL);
 146         parser->buf = g_string_new(NULL);
 147         parser->bufp = parser->buf->str;
 148         parser->state = SC_HTML_NORMAL;
 149         parser->href = NULL;
 150         parser->newline = TRUE;
 151         parser->empty_line = TRUE;
 152         parser->space = FALSE;
 153         parser->pre = FALSE;
 154
 155 #define SYMBOL_TABLE_ADD(table, list) \
 156 { \
 157         gint i; \
 158  \
 159         for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
 160                 g_hash_table_insert(table, list[i].key, list[i].val); \
 161 }
 162
 163         if (!default_symbol_table) {
 164                 default_symbol_table =
 165                         g_hash_table_new(g_str_hash, g_str_equal);
 166                 SYMBOL_TABLE_ADD(default_symbol_table, symbol_list);
 167                 SYMBOL_TABLE_ADD(default_symbol_table, ascii_symbol_list);
 168         }
 169
 170 #undef SYMBOL_TABLE_ADD
 171
 172         parser->symbol_table = default_symbol_table;
 173
 174         return parser;
 175 }
 176
 177 void sc_html_parser_destroy(SC_HTMLParser *parser)
 178 {
 179         g_string_free(parser->str, TRUE);
 180         g_string_free(parser->buf, TRUE);
 181         g_free(parser->href);
 182         g_free(parser);
 183 }
 184
 185 gchar *sc_html_parse(SC_HTMLParser *parser)
 186 {
 187         parser->state = SC_HTML_NORMAL;
 188         g_string_truncate(parser->str, 0);
 189
 190         if (*parser->bufp == '\0') {
 191                 g_string_truncate(parser->buf, 0);
 192                 parser->bufp = parser->buf->str;
 193                 if (sc_html_read_line(parser) == SC_HTML_EOF)
 194                         return NULL;
 195         }
 196
 197         while (*parser->bufp != '\0') {
 198                 switch (*parser->bufp) {
 199                 case '<': {
 200                         SC_HTMLState st;
 201                         st = sc_html_parse_tag(parser);
 202                         /* when we see an href, we need to flush the str
 203                          * buffer.  Then collect all the chars until we
 204                          * see the end anchor tag
 205                          */
 206                         if (SC_HTML_HREF_BEG == st || SC_HTML_HREF == st)
 207                                 return parser->str->str;
 208                         }
 209                         break;
 210                 case '&':
 211                         sc_html_parse_special(parser);
 212                         break;
 213                 case ' ':
 214                 case '\t':
 215                 case '\r':
 216                 case '\n':
 217                         if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
 218                                 parser->bufp++;
 219
 220                         if (!parser->pre) {
 221                                 if (!parser->newline)
 222                                         parser->space = TRUE;
 223
 224                                 parser->bufp++;
 225                                 break;
 226                         }
 227                         /* fallthrough */
 228                 default:
 229                         sc_html_append_char(parser, *parser->bufp++);
 230                 }
 231         }
 232
 233         return parser->str->str;
 234 }
 235
 236 static SC_HTMLState sc_html_read_line(SC_HTMLParser *parser)
 237 {
 238         gchar buf[SC_HTMLBUFSIZE];
 239         gchar buf2[SC_HTMLBUFSIZE];
 240         gint index;
 241
 242         if (fgets(buf, sizeof(buf), parser->fp) == NULL) {
 243                 parser->state = SC_HTML_EOF;
 244                 return SC_HTML_EOF;
 245         }
 246
 247         if (conv_convert(parser->conv, buf2, sizeof(buf2), buf) < 0) {
 248                 index = parser->bufp - parser->buf->str;
 249
 250                 conv_utf8todisp(buf2, sizeof(buf2), buf);
 251                 g_string_append(parser->buf, buf2);
 252
 253                 parser->bufp = parser->buf->str + index;
 254
 255                 return SC_HTML_CONV_FAILED;
 256         }
 257
 258         index = parser->bufp - parser->buf->str;
 259
 260         g_string_append(parser->buf, buf2);
 261
 262         parser->bufp = parser->buf->str + index;
 263
 264         return SC_HTML_NORMAL;
 265 }
 266
 267 static void sc_html_append_char(SC_HTMLParser *parser, gchar ch)
 268 {
 269         GString *str = parser->str;
 270
 271         if (!parser->pre && parser->space) {
 272                 g_string_append_c(str, ' ');
 273                 parser->space = FALSE;
 274         }
 275
 276         g_string_append_c(str, ch);
 277
 278         parser->empty_line = FALSE;
 279         if (ch == '\n') {
 280                 parser->newline = TRUE;
 281                 if (str->len > 1 && str->str[str->len - 2] == '\n')
 282                         parser->empty_line = TRUE;
 283         } else
 284                 parser->newline = FALSE;
 285 }
 286
 287 static void sc_html_append_str(SC_HTMLParser *parser, const gchar *str, gint len)
 288 {
 289         GString *string = parser->str;
 290
 291         if (!parser->pre && parser->space) {
 292                 g_string_append_c(string, ' ');
 293                 parser->space = FALSE;
 294         }
 295
 296         if (len == 0) return;
 297         if (len < 0)
 298                 g_string_append(string, str);
 299         else {
 300                 gchar *s;
 301                 Xstrndup_a(s, str, len, return);
 302                 g_string_append(string, s);
 303         }
 304
 305         parser->empty_line = FALSE;
 306         if (string->len > 0 && string->str[string->len - 1] == '\n') {
 307                 parser->newline = TRUE;
 308                 if (string->len > 1 && string->str[string->len - 2] == '\n')
 309                         parser->empty_line = TRUE;
 310         } else
 311                 parser->newline = FALSE;
 312 }
 313
 314 static SC_HTMLTag *sc_html_get_tag(const gchar *str)
 315 {
 316         SC_HTMLTag *tag;
 317         gchar *tmp;
 318         guchar *tmpp;
 319
 320         g_return_val_if_fail(str != NULL, NULL);
 321
 322         if (*str == '\0' || *str == '!') return NULL;
 323
 324         Xstrdup_a(tmp, str, return NULL);
 325
 326         tag = g_new0(SC_HTMLTag, 1);
 327
 328         for (tmpp = tmp; *tmpp != '\0' && !g_ascii_isspace(*tmpp); tmpp++)
 329                 ;
 330
 331         if (*tmpp == '\0') {
 332                 g_strdown(tmp);
 333                 tag->name = g_strdup(tmp);
 334                 return tag;
 335         } else {
 336                 *tmpp++ = '\0';
 337                 g_strdown(tmp);
 338                 tag->name = g_strdup(tmp);
 339         }
 340
 341         while (*tmpp != '\0') {
 342                 SC_HTMLAttr *attr;
 343                 gchar *attr_name;
 344                 gchar *attr_value;
 345                 gchar *p;
 346                 gchar quote;
 347
 348                 while (g_ascii_isspace(*tmpp)) tmpp++;
 349                 attr_name = tmpp;
 350
 351                 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp) &&
 352                        *tmpp != '=')
 353                         tmpp++;
 354                 if (*tmpp != '\0' && *tmpp != '=') {
 355                         *tmpp++ = '\0';
 356                         while (g_ascii_isspace(*tmpp)) tmpp++;
 357                 }
 358
 359                 if (*tmpp == '=') {
 360                         *tmpp++ = '\0';
 361                         while (g_ascii_isspace(*tmpp)) tmpp++;
 362
 363                         if (*tmpp == '"' || *tmpp == '\'') {
 364                                 /* name="value" */
 365                                 quote = *tmpp;
 366                                 tmpp++;
 367                                 attr_value = tmpp;
 368                                 if ((p = strchr(attr_value, quote)) == NULL) {
 369                                         g_warning("sc_html_get_tag(): syntax error in tag: '%s'\n", str);
 370                                         return tag;
 371                                 }
 372                                 tmpp = p;
 373                                 *tmpp++ = '\0';
 374                                 while (g_ascii_isspace(*tmpp)) tmpp++;
 375                         } else {
 376                                 /* name=value */
 377                                 attr_value = tmpp;
 378                                 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp)) tmpp++;
 379                                 if (*tmpp != '\0')
 380                                         *tmpp++ = '\0';
 381                         }
 382                 } else
 383                         attr_value = "";
 384
 385                 g_strchomp(attr_name);
 386                 g_strdown(attr_name);
 387                 attr = g_new(SC_HTMLAttr, 1);
 388                 attr->name = g_strdup(attr_name);
 389                 attr->value = g_strdup(attr_value);
 390                 tag->attr = g_list_append(tag->attr, attr);
 391         }
 392
 393         return tag;
 394 }
 395
 396 static void sc_html_free_tag(SC_HTMLTag *tag)
 397 {
 398         if (!tag) return;
 399
 400         g_free(tag->name);
 401         while (tag->attr != NULL) {
 402                 SC_HTMLAttr *attr = (SC_HTMLAttr *)tag->attr->data;
 403                 g_free(attr->name);
 404                 g_free(attr->value);
 405                 g_free(attr);
 406                 tag->attr = g_list_remove(tag->attr, tag->attr->data);
 407         }
 408         g_free(tag);
 409 }
 410
 411 static SC_HTMLState sc_html_parse_tag(SC_HTMLParser *parser)
 412 {
 413         gchar buf[SC_HTMLBUFSIZE];
 414         SC_HTMLTag *tag;
 415
 416         sc_html_get_parenthesis(parser, buf, sizeof(buf));
 417
 418         tag = sc_html_get_tag(buf);
 419
 420         parser->state = SC_HTML_UNKNOWN;
 421         if (!tag) return SC_HTML_UNKNOWN;
 422
 423         if (!strcmp(tag->name, "br")) {
 424                 parser->space = FALSE;
 425                 sc_html_append_char(parser, '\n');
 426                 parser->state = SC_HTML_BR;
 427         } else if (!strcmp(tag->name, "a")) {
 428                 GList *cur;
 429                 for (cur = tag->attr; cur != NULL; cur = cur->next) {
 430                         if (cur->data && !strcmp(((SC_HTMLAttr *)cur->data)->name, "href")) {
 431                                 g_free(parser->href);
 432                                 parser->href = g_strdup(((SC_HTMLAttr *)cur->data)->value);
 433                                 parser->state = SC_HTML_HREF_BEG;
 434                                 break;
 435                         }
 436                 }
 437         } else if (!strcmp(tag->name, "/a")) {
 438                 parser->state = SC_HTML_HREF;
 439         } else if (!strcmp(tag->name, "p")) {
 440                 parser->space = FALSE;
 441                 if (!parser->empty_line) {
 442                         parser->space = FALSE;
 443                         if (!parser->newline) sc_html_append_char(parser, '\n');
 444                         sc_html_append_char(parser, '\n');
 445                 }
 446                 parser->state = SC_HTML_PAR;
 447         } else if (!strcmp(tag->name, "pre")) {
 448                 parser->pre = TRUE;
 449                 parser->state = SC_HTML_PRE;
 450         } else if (!strcmp(tag->name, "/pre")) {
 451                 parser->pre = FALSE;
 452                 parser->state = SC_HTML_NORMAL;
 453         } else if (!strcmp(tag->name, "hr")) {
 454                 if (!parser->newline) {
 455                         parser->space = FALSE;
 456                         sc_html_append_char(parser, '\n');
 457                 }
 458                 sc_html_append_str(parser, HR_STR "\n", -1);
 459                 parser->state = SC_HTML_HR;
 460         } else if (!strcmp(tag->name, "div")    ||
 461                    !strcmp(tag->name, "ul")     ||
 462                    !strcmp(tag->name, "li")     ||
 463                    !strcmp(tag->name, "table")  ||
 464                    !strcmp(tag->name, "tr")     ||
 465                    (tag->name[0] == 'h' && g_ascii_isdigit(tag->name[1]))) {
 466                 if (!parser->newline) {
 467                         parser->space = FALSE;
 468                         sc_html_append_char(parser, '\n');
 469                 }
 470                 parser->state = SC_HTML_NORMAL;
 471         } else if (!strcmp(tag->name, "/table") ||
 472                    (tag->name[0] == '/' &&
 473                     tag->name[1] == 'h' &&
 474                     g_ascii_isdigit(tag->name[1]))) {
 475                 if (!parser->empty_line) {
 476                         parser->space = FALSE;
 477                         if (!parser->newline) sc_html_append_char(parser, '\n');
 478                         sc_html_append_char(parser, '\n');
 479                 }
 480                 parser->state = SC_HTML_NORMAL;
 481         } else if (!strcmp(tag->name, "/div")   ||
 482                    !strcmp(tag->name, "/ul")    ||
 483                    !strcmp(tag->name, "/li")) {
 484                 if (!parser->newline) {
 485                         parser->space = FALSE;
 486                         sc_html_append_char(parser, '\n');
 487                 }
 488                 parser->state = SC_HTML_NORMAL;
 489                         }
 490
 491         sc_html_free_tag(tag);
 492
 493         return parser->state;
 494 }
 495
 496 static void sc_html_parse_special(SC_HTMLParser *parser)
 497 {
 498         gchar symbol_name[9];
 499         gint n;
 500         const gchar *val;
 501
 502         parser->state = SC_HTML_UNKNOWN;
 503         g_return_if_fail(*parser->bufp == '&');
 504
 505         /* &foo; */
 506         for (n = 0; parser->bufp[n] != '\0' && parser->bufp[n] != ';'; n++)
 507                 ;
 508         if (n > 7 || parser->bufp[n] != ';') {
 509                 /* output literal `&' */
 510                 sc_html_append_char(parser, *parser->bufp++);
 511                 parser->state = SC_HTML_NORMAL;
 512                 return;
 513         }
 514         strncpy2(symbol_name, parser->bufp, n + 2);
 515         parser->bufp += n + 1;
 516
 517         if ((val = g_hash_table_lookup(parser->symbol_table, symbol_name))
 518             != NULL) {
 519                 sc_html_append_str(parser, val, -1);
 520                 parser->state = SC_HTML_NORMAL;
 521                 return;
 522         } else if (symbol_name[1] == '#' && g_ascii_isdigit(symbol_name[2])) {
 523                 gint ch;
 524
 525                 ch = atoi(symbol_name + 2);
 526                 if ((ch > 0 && ch <= 127) ||
 527                     (ch >= 128 && ch <= 255 &&
 528                      parser->conv->charset == C_ISO_8859_1)) {
 529                         sc_html_append_char(parser, ch);
 530                         parser->state = SC_HTML_NORMAL;
 531                         return;
 532                 } else {
 533                         char *symb = NULL;
 534                         switch (ch) {
 535                         /* http://www.w3schools.com/html/html_entitiesref.asp */
 536                         case 96:        /* backtick  */
 537                                 symb = "`";
 538                                 break;
 539                         case 153:       /* trademark */
 540                                 symb = "(TM)";
 541                                 break;
 542                         case 338:       /* capital ligature OE  &OElig;  */
 543                                 symb = "OE";
 544                                 break;
 545                         case 339:       /* small ligature OE    &oelig;  */
 546                                 symb = "oe";
 547                                 break;
 548                         case 352:       /* capital S w/caron    &Scaron; */
 549                         case 353:       /* small S w/caron      &scaron; */
 550                         case 376:       /* cap Y w/ diaeres     &Yuml;   */
 551                                 break;
 552                         case 710:       /* circumflex accent    &circ;   */
 553                                 symb = "^";
 554                                 break;
 555                         case 732:       /* small tilde          &tilde;  */
 556                                 symb = "~";
 557                                 break;
 558                         case 8194:      /* en space             &ensp;   */
 559                         case 8195:      /* em space             &emsp;   */
 560                         case 8201:      /* thin space           &thinsp; */
 561                                 symb = " ";
 562                                 break;
 563                         case 8204:      /* zero width non-joiner &zwnj;  */
 564                         case 8205:      /* zero width joiner    &zwj;   */
 565                         case 8206:      /* l-t-r mark           &lrm;   */
 566                         case 8207:      /* r-t-l mark           &rlm     */
 567                                 break;
 568                         case 8211:      /* en dash              &ndash;  */
 569                                 symb = "-";
 570                                 break;
 571                         case 8212:      /* em dash              &mdash;  */
 572                                 symb = "--";
 573                                 break;
 574                         case 8216:      /* l single quot mark   &lsquo;  */
 575                         case 8217:      /* r single quot mark   &rsquo;  */
 576                                 symb = "'";
 577                                 break;
 578                         case 8218:      /* single low-9 quot    &sbquo;  */
 579                                 symb = ",";
 580                                 break;
 581                         case 8220:      /* l double quot mark   &ldquo;  */
 582                         case 8221:      /* r double quot mark   &rdquo;  */
 583                                 symb = "\"";
 584                                 break;
 585                         case 8222:      /* double low-9 quot    &bdquo;  */
 586                                 symb = ",,";
 587                                 break;
 588                         case 8224:      /* dagger               &dagger; */
 589                         case 8225:      /* double dagger        &Dagger; */
 590                                 break;
 591                         case 8226:      /* bullet       &bull;  */
 592                                 symb = "*";
 593                                 break;
 594                         case 8230:      /* horizontal ellipsis  &hellip; */
 595                                 symb = "...";
 596                                 break;
 597                         case 8240:      /* per mile             &permil; */
 598                                 symb = "\%o";
 599                                 break;
 600                         case 8249:      /* l-pointing angle quot &lsaquo; */
 601                                 symb = "<";
 602                                 break;
 603                         case 8250:      /* r-pointing angle quot &rsaquo; */
 604                                 symb = ">";
 605                                 break;
 606                         case 8364:      /* euro                 &euro;   */
 607                                 symb = "EUR";
 608                                 break;
 609                         case 8482:      /* trademark            &trade;  */
 610                                 symb  = "(TM)";
 611                                 break;
 612                         default:
 613                                 break;
 614                         }
 615                         if (symb) {
 616                                 sc_html_append_str(parser, symb, -1);
 617                                 parser->state = SC_HTML_NORMAL;
 618                                 return;
 619                         }
 620                 }
 621         }
 622
 623         sc_html_append_str(parser, symbol_name, -1);
 624 }
 625
 626 static void sc_html_get_parenthesis(SC_HTMLParser *parser, gchar *buf, gint len)
 627 {
 628         gchar *p;
 629
 630         buf[0] = '\0';
 631         g_return_if_fail(*parser->bufp == '<');
 632
 633         /* ignore comment / CSS / script stuff */
 634         if (!strncmp(parser->bufp, "<!--", 4)) {
 635                 parser->bufp += 4;
 636                 while ((p = strstr(parser->bufp, "-->")) == NULL)
 637                         if (sc_html_read_line(parser) == SC_HTML_EOF) return;
 638                 parser->bufp = p + 3;
 639                 return;
 640         }
 641         if (!g_ascii_strncasecmp(parser->bufp, "<style", 6)) {
 642                 parser->bufp += 6;
 643                 while ((p = strcasestr(parser->bufp, "</style>")) == NULL)
 644                         if (sc_html_read_line(parser) == SC_HTML_EOF) return;
 645                 parser->bufp = p + 8;
 646                 return;
 647         }
 648         if (!g_ascii_strncasecmp(parser->bufp, "<script", 7)) {
 649                 parser->bufp += 7;
 650                 while ((p = strcasestr(parser->bufp, "</script>")) == NULL)
 651                         if (sc_html_read_line(parser) == SC_HTML_EOF) return;
 652                 parser->bufp = p + 9;
 653                 return;
 654         }
 655
 656         parser->bufp++;
 657         while ((p = strchr(parser->bufp, '>')) == NULL)
 658                 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
 659
 660         strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
 661         g_strstrip(buf);
 662         parser->bufp = p + 1;
 663 }