src/html.c

   1 /*
   2  * Sylpheed -- a GTK+ based, lightweight, and fast e-mail client
   3  * Copyright (C) 1999-2006 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         {"&#153;", "(TM)"},
  52         {"&hellip;", "..."},
  53 };
  54
  55 static SC_HTMLSymbol ascii_symbol_list[] = {
  56         {"&iexcl;" , "\302\241"},
  57         {"&brvbar;", "\302\246"},
  58         {"&copy;"  , "\302\251"},
  59         {"&laquo;" , "\302\253"},
  60         {"&reg;"   , "\302\256"},
  61
  62         {"&sup2;"  , "\302\262"},
  63         {"&sup3;"  , "\302\263"},
  64         {"&acute;" , "\302\264"},
  65         {"&cedil;" , "\302\270"},
  66         {"&sup1;"  , "\302\271"},
  67         {"&raquo;" , "\302\273"},
  68         {"&frac14;", "\302\274"},
  69         {"&frac12;", "\302\275"},
  70         {"&frac34;", "\302\276"},
  71         {"&iquest;", "\302\277"},
  72
  73         {"&Agrave;", "\303\200"},
  74         {"&Aacute;", "\303\201"},
  75         {"&Acirc;" , "\303\202"},
  76         {"&Atilde;", "\303\203"},
  77         {"&AElig;" , "\303\206"},
  78         {"&Egrave;", "\303\210"},
  79         {"&Eacute;", "\303\211"},
  80         {"&Ecirc;" , "\303\212"},
  81         {"&Igrave;", "\303\214"},
  82         {"&Iacute;", "\303\215"},
  83         {"&Icirc;" , "\303\216"},
  84
  85         {"&Ntilde;", "\303\221"},
  86         {"&Ograve;", "\303\222"},
  87         {"&Oacute;", "\303\223"},
  88         {"&Ocirc;" , "\303\224"},
  89         {"&Otilde;", "\303\225"},
  90         {"&Ugrave;", "\303\231"},
  91         {"&Uacute;", "\303\232"},
  92         {"&Ucirc;" , "\303\233"},
  93         {"&Yacute;", "\303\235"},
  94
  95         {"&agrave;", "\303\240"},
  96         {"&aacute;", "\303\241"},
  97         {"&acirc;" , "\303\242"},
  98         {"&atilde;", "\303\243"},
  99         {"&aelig;" , "\303\246"},
 100         {"&egrave;", "\303\250"},
 101         {"&eacute;", "\303\251"},
 102         {"&ecirc;" , "\303\252"},
 103         {"&igrave;", "\303\254"},
 104         {"&iacute;", "\303\255"},
 105         {"&icirc;" , "\303\256"},
 106
 107         {"&ntilde;", "\303\261"},
 108         {"&ograve;", "\303\262"},
 109         {"&oacute;", "\303\263"},
 110         {"&ocirc;" , "\303\264"},
 111         {"&otilde;", "\303\265"},
 112         {"&ugrave;", "\303\271"},
 113         {"&uacute;", "\303\272"},
 114         {"&ucirc;" , "\303\273"},
 115         {"&yacute;", "\303\275"},
 116 };
 117
 118 static GHashTable *default_symbol_table;
 119
 120 static SC_HTMLState sc_html_read_line   (SC_HTMLParser  *parser);
 121 static void sc_html_append_char                 (SC_HTMLParser  *parser,
 122                                          gchar           ch);
 123 static void sc_html_append_str                  (SC_HTMLParser  *parser,
 124                                          const gchar    *str,
 125                                          gint            len);
 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,
 129                                          gchar          *buf,
 130                                          gint            len);
 131
 132
 133 SC_HTMLParser *sc_html_parser_new(FILE *fp, CodeConverter *conv)
 134 {
 135         SC_HTMLParser *parser;
 136
 137         g_return_val_if_fail(fp != NULL, NULL);
 138         g_return_val_if_fail(conv != NULL, NULL);
 139
 140         parser = g_new0(SC_HTMLParser, 1);
 141         parser->fp = fp;
 142         parser->conv = conv;
 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;
 147         parser->href = NULL;
 148         parser->newline = TRUE;
 149         parser->empty_line = TRUE;
 150         parser->space = FALSE;
 151         parser->pre = FALSE;
 152
 153 #define SYMBOL_TABLE_ADD(table, list) \
 154 { \
 155         gint i; \
 156  \
 157         for (i = 0; i < sizeof(list) / sizeof(list[0]); i++) \
 158                 g_hash_table_insert(table, list[i].key, list[i].val); \
 159 }
 160
 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);
 166         }
 167
 168 #undef SYMBOL_TABLE_ADD
 169
 170         parser->symbol_table = default_symbol_table;
 171
 172         return parser;
 173 }
 174
 175 void sc_html_parser_destroy(SC_HTMLParser *parser)
 176 {
 177         g_string_free(parser->str, TRUE);
 178         g_string_free(parser->buf, TRUE);
 179         g_free(parser->href);
 180         g_free(parser);
 181 }
 182
 183 gchar *sc_html_parse(SC_HTMLParser *parser)
 184 {
 185         parser->state = SC_HTML_NORMAL;
 186         g_string_truncate(parser->str, 0);
 187
 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)
 192                         return NULL;
 193         }
 194
 195         while (*parser->bufp != '\0') {
 196                 switch (*parser->bufp) {
 197                 case '<': {
 198                         SC_HTMLState st;
 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
 203                          */
 204                         if (SC_HTML_HREF_BEG == st || SC_HTML_HREF == st)
 205                                 return parser->str->str;
 206                         }
 207                         break;
 208                 case '&':
 209                         sc_html_parse_special(parser);
 210                         break;
 211                 case ' ':
 212                 case '\t':
 213                 case '\r':
 214                 case '\n':
 215                         if (parser->bufp[0] == '\r' && parser->bufp[1] == '\n')
 216                                 parser->bufp++;
 217
 218                         if (!parser->pre) {
 219                                 if (!parser->newline)
 220                                         parser->space = TRUE;
 221
 222                                 parser->bufp++;
 223                                 break;
 224                         }
 225                         /* fallthrough */
 226                 default:
 227                         sc_html_append_char(parser, *parser->bufp++);
 228                 }
 229         }
 230
 231         return parser->str->str;
 232 }
 233
 234 static SC_HTMLState sc_html_read_line(SC_HTMLParser *parser)
 235 {
 236         gchar buf[SC_HTMLBUFSIZE];
 237         gchar buf2[SC_HTMLBUFSIZE];
 238         gint index;
 239
 240         if (fgets(buf, sizeof(buf), parser->fp) == NULL) {
 241                 parser->state = SC_HTML_EOF;
 242                 return SC_HTML_EOF;
 243         }
 244
 245         if (conv_convert(parser->conv, buf2, sizeof(buf2), buf) < 0) {
 246                 index = parser->bufp - parser->buf->str;
 247
 248                 conv_utf8todisp(buf2, sizeof(buf2), buf);
 249                 g_string_append(parser->buf, buf2);
 250
 251                 parser->bufp = parser->buf->str + index;
 252
 253                 return SC_HTML_CONV_FAILED;
 254         }
 255
 256         index = parser->bufp - parser->buf->str;
 257
 258         g_string_append(parser->buf, buf2);
 259
 260         parser->bufp = parser->buf->str + index;
 261
 262         return SC_HTML_NORMAL;
 263 }
 264
 265 static void sc_html_append_char(SC_HTMLParser *parser, gchar ch)
 266 {
 267         GString *str = parser->str;
 268
 269         if (!parser->pre && parser->space) {
 270                 g_string_append_c(str, ' ');
 271                 parser->space = FALSE;
 272         }
 273
 274         g_string_append_c(str, ch);
 275
 276         parser->empty_line = FALSE;
 277         if (ch == '\n') {
 278                 parser->newline = TRUE;
 279                 if (str->len > 1 && str->str[str->len - 2] == '\n')
 280                         parser->empty_line = TRUE;
 281         } else
 282                 parser->newline = FALSE;
 283 }
 284
 285 static void sc_html_append_str(SC_HTMLParser *parser, const gchar *str, gint len)
 286 {
 287         GString *string = parser->str;
 288
 289         if (!parser->pre && parser->space) {
 290                 g_string_append_c(string, ' ');
 291                 parser->space = FALSE;
 292         }
 293
 294         if (len == 0) return;
 295         if (len < 0)
 296                 g_string_append(string, str);
 297         else {
 298                 gchar *s;
 299                 Xstrndup_a(s, str, len, return);
 300                 g_string_append(string, s);
 301         }
 302
 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;
 308         } else
 309                 parser->newline = FALSE;
 310 }
 311
 312 static SC_HTMLTag *sc_html_get_tag(const gchar *str)
 313 {
 314         SC_HTMLTag *tag;
 315         gchar *tmp;
 316         guchar *tmpp;
 317
 318         g_return_val_if_fail(str != NULL, NULL);
 319
 320         if (*str == '\0' || *str == '!') return NULL;
 321
 322         Xstrdup_a(tmp, str, return NULL);
 323
 324         tag = g_new0(SC_HTMLTag, 1);
 325
 326         for (tmpp = tmp; *tmpp != '\0' && !g_ascii_isspace(*tmpp); tmpp++)
 327                 ;
 328
 329         if (*tmpp == '\0') {
 330                 g_strdown(tmp);
 331                 tag->name = g_strdup(tmp);
 332                 return tag;
 333         } else {
 334                 *tmpp++ = '\0';
 335                 g_strdown(tmp);
 336                 tag->name = g_strdup(tmp);
 337         }
 338
 339         while (*tmpp != '\0') {
 340                 SC_HTMLAttr *attr;
 341                 gchar *attr_name;
 342                 gchar *attr_value;
 343                 gchar *p;
 344                 gchar quote;
 345
 346                 while (g_ascii_isspace(*tmpp)) tmpp++;
 347                 attr_name = tmpp;
 348
 349                 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp) &&
 350                        *tmpp != '=')
 351                         tmpp++;
 352                 if (*tmpp != '\0' && *tmpp != '=') {
 353                         *tmpp++ = '\0';
 354                         while (g_ascii_isspace(*tmpp)) tmpp++;
 355                 }
 356
 357                 if (*tmpp == '=') {
 358                         *tmpp++ = '\0';
 359                         while (g_ascii_isspace(*tmpp)) tmpp++;
 360
 361                         if (*tmpp == '"' || *tmpp == '\'') {
 362                                 /* name="value" */
 363                                 quote = *tmpp;
 364                                 tmpp++;
 365                                 attr_value = tmpp;
 366                                 if ((p = strchr(attr_value, quote)) == NULL) {
 367                                         g_warning("sc_html_get_tag(): syntax error in tag: '%s'\n", str);
 368                                         return tag;
 369                                 }
 370                                 tmpp = p;
 371                                 *tmpp++ = '\0';
 372                                 while (g_ascii_isspace(*tmpp)) tmpp++;
 373                         } else {
 374                                 /* name=value */
 375                                 attr_value = tmpp;
 376                                 while (*tmpp != '\0' && !g_ascii_isspace(*tmpp)) tmpp++;
 377                                 if (*tmpp != '\0')
 378                                         *tmpp++ = '\0';
 379                         }
 380                 } else
 381                         attr_value = "";
 382
 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);
 389         }
 390
 391         return tag;
 392 }
 393
 394 static void sc_html_free_tag(SC_HTMLTag *tag)
 395 {
 396         if (!tag) return;
 397
 398         g_free(tag->name);
 399         while (tag->attr != NULL) {
 400                 SC_HTMLAttr *attr = (SC_HTMLAttr *)tag->attr->data;
 401                 g_free(attr->name);
 402                 g_free(attr->value);
 403                 g_free(attr);
 404                 tag->attr = g_list_remove(tag->attr, tag->attr->data);
 405         }
 406         g_free(tag);
 407 }
 408
 409 static SC_HTMLState sc_html_parse_tag(SC_HTMLParser *parser)
 410 {
 411         gchar buf[SC_HTMLBUFSIZE];
 412         SC_HTMLTag *tag;
 413
 414         sc_html_get_parenthesis(parser, buf, sizeof(buf));
 415
 416         tag = sc_html_get_tag(buf);
 417
 418         parser->state = SC_HTML_UNKNOWN;
 419         if (!tag) return SC_HTML_UNKNOWN;
 420
 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")) {
 426                 GList *cur;
 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;
 432                                 break;
 433                         }
 434                 }
 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');
 443                 }
 444                 parser->state = SC_HTML_PAR;
 445         } else if (!strcmp(tag->name, "pre")) {
 446                 parser->pre = TRUE;
 447                 parser->state = SC_HTML_PRE;
 448         } else if (!strcmp(tag->name, "/pre")) {
 449                 parser->pre = FALSE;
 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');
 455                 }
 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');
 467                 }
 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');
 477                 }
 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');
 485                 }
 486                 parser->state = SC_HTML_NORMAL;
 487                         }
 488
 489         sc_html_free_tag(tag);
 490
 491         return parser->state;
 492 }
 493
 494 static void sc_html_parse_special(SC_HTMLParser *parser)
 495 {
 496         gchar symbol_name[9];
 497         gint n;
 498         const gchar *val;
 499
 500         parser->state = SC_HTML_UNKNOWN;
 501         g_return_if_fail(*parser->bufp == '&');
 502
 503         /* &foo; */
 504         for (n = 0; parser->bufp[n] != '\0' && parser->bufp[n] != ';'; n++)
 505                 ;
 506         if (n > 7 || parser->bufp[n] != ';') {
 507                 /* output literal `&' */
 508                 sc_html_append_char(parser, *parser->bufp++);
 509                 parser->state = SC_HTML_NORMAL;
 510                 return;
 511         }
 512         strncpy2(symbol_name, parser->bufp, n + 2);
 513         parser->bufp += n + 1;
 514
 515         if ((val = g_hash_table_lookup(parser->symbol_table, symbol_name))
 516             != NULL) {
 517                 sc_html_append_str(parser, val, -1);
 518                 parser->state = SC_HTML_NORMAL;
 519                 return;
 520         } else if (symbol_name[1] == '#' && g_ascii_isdigit(symbol_name[2])) {
 521                 gint ch;
 522
 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;
 529                         return;
 530                 } else {
 531                         char *symb = NULL;
 532                         switch (ch) {
 533                         /* http://www.w3schools.com/html/html_entitiesref.asp */
 534                         case 96:        /* backtick  */
 535                                 symb = "`";
 536                                 break;
 537                         case 338:       /* capital ligature OE  &OElig;  */
 538                                 symb = "OE";
 539                                 break;
 540                         case 339:       /* small ligature OE    &oelig;  */
 541                                 symb = "oe";
 542                                 break;
 543                         case 352:       /* capital S w/caron    &Scaron; */
 544                         case 353:       /* small S w/caron      &scaron; */
 545                         case 376:       /* cap Y w/ diaeres     &Yuml;   */
 546                                 break;
 547                         case 710:       /* circumflex accent    &circ;   */
 548                                 symb = "^";
 549                                 break;
 550                         case 732:       /* small tilde          &tilde;  */
 551                                 symb = "~";
 552                                 break;
 553                         case 8194:      /* en space             &ensp;   */
 554                         case 8195:      /* em space             &emsp;   */
 555                         case 8201:      /* thin space           &thinsp; */
 556                                 symb = " ";
 557                                 break;
 558                         case 8204:      /* zero width non-joiner &zwnj;  */
 559                         case 8205:      /* zero width joiner    &zwj;   */
 560                         case 8206:      /* l-t-r mark           &lrm;   */
 561                         case 8207:      /* r-t-l mark           &rlm     */
 562                                 break;
 563                         case 8211:      /* en dash              &ndash;  */
 564                                 symb = "-";
 565                                 break;
 566                         case 8212:      /* em dash              &mdash;  */
 567                                 symb = "--";
 568                                 break;
 569                         case 8216:      /* l single quot mark   &lsquo;  */
 570                         case 8217:      /* r single quot mark   &rsquo;  */
 571                                 symb = "'";
 572                                 break;
 573                         case 8218:      /* single low-9 quot    &sbquo;  */
 574                                 symb = ",";
 575                                 break;
 576                         case 8220:      /* l double quot mark   &ldquo;  */
 577                         case 8221:      /* r double quot mark   &rdquo;  */
 578                                 symb = "\"";
 579                                 break;
 580                         case 8222:      /* double low-9 quot    &bdquo;  */
 581                                 symb = ",,";
 582                                 break;
 583                         case 8224:      /* dagger               &dagger; */
 584                         case 8225:      /* double dagger        &Dagger; */
 585                                 break;
 586                         case 8230:      /* horizontal ellipsis  &hellip; */
 587                                 symb = "...";
 588                                 break;
 589                         case 8240:      /* per mile             &permil; */
 590                                 symb = "\%o";
 591                                 break;
 592                         case 8249:      /* l-pointing angle quot &lsaquo; */
 593                                 symb = "<";
 594                                 break;
 595                         case 8250:      /* r-pointing angle quot &rsaquo; */
 596                                 symb = ">";
 597                                 break;
 598                         case 8364:      /* euro                 &euro;   */
 599                                 symb = "EUR";
 600                                 break;
 601                         case 8482:      /* trademark            &trade;  */
 602                                 symb  = "(TM)";
 603                                 break;
 604                         default:
 605                                 break;
 606                         }
 607                         if (symb) {
 608                                 sc_html_append_str(parser, symb, -1);
 609                                 parser->state = SC_HTML_NORMAL;
 610                                 return;
 611                         }
 612                 }
 613         }
 614
 615         sc_html_append_str(parser, symbol_name, -1);
 616 }
 617
 618 static void sc_html_get_parenthesis(SC_HTMLParser *parser, gchar *buf, gint len)
 619 {
 620         gchar *p;
 621
 622         buf[0] = '\0';
 623         g_return_if_fail(*parser->bufp == '<');
 624
 625         /* ignore comment / CSS / script stuff */
 626         if (!strncmp(parser->bufp, "<!--", 4)) {
 627                 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;
 631                 return;
 632         }
 633         if (!g_ascii_strncasecmp(parser->bufp, "<style", 6)) {
 634                 parser->bufp += 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;
 638                 return;
 639         }
 640         if (!g_ascii_strncasecmp(parser->bufp, "<script", 7)) {
 641                 parser->bufp += 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;
 645                 return;
 646         }
 647
 648         parser->bufp++;
 649         while ((p = strchr(parser->bufp, '>')) == NULL)
 650                 if (sc_html_read_line(parser) == SC_HTML_EOF) return;
 651
 652         strncpy2(buf, parser->bufp, MIN(p - parser->bufp + 1, len));
 653         g_strstrip(buf);
 654         parser->bufp = p + 1;
 655 }