swfdec/swfdec_text_field_movie_html.c

   1 /* Swfdec
   2  * Copyright (C) 2007 Pekka Lampila <pekka.lampila@iki.fi>
   3  *
   4  * This library is free software; you can redistribute it and/or
   5  * modify it under the terms of the GNU Lesser General Public
   6  * License as published by the Free Software Foundation; either
   7  * version 2.1 of the License, or (at your option) any later version.
   8  *
   9  * This library is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  12  * Lesser General Public License for more details.
  13  *
  14  * You should have received a copy of the GNU Lesser General Public
  15  * License along with this library; if not, write to the Free Software
  16  * Foundation, Inc., 51 Franklin Street, Fifth Floor,
  17  * Boston, MA  02110-1301  USA
  18  */
  19
  20 #ifdef HAVE_CONFIG_H
  21 #include "config.h"
  22 #endif
  23
  24 #include <stdlib.h>
  25 #include <string.h>
  26
  27 #include "swfdec_text_field_movie.h"
  28 #include "swfdec_as_strings.h"
  29 #include "swfdec_style_sheet.h"
  30 #include "swfdec_xml.h"
  31 #include "swfdec_debug.h"
  32
  33 /*
  34  * Parsing
  35  */
  36 typedef struct {
  37   const char *          name;
  38   int                   name_length;
  39   guint                 index;
  40   guint                 end_index;
  41   SwfdecTextFormat *    format;
  42 } ParserTag;
  43
  44 typedef struct {
  45   SwfdecAsContext *     cx;
  46   gboolean              multiline;
  47   gboolean              condense_white;
  48   SwfdecStyleSheet *    style_sheet;
  49   SwfdecTextBuffer *    text;
  50   GSList *              tags_open;
  51   GSList *              tags_closed;
  52 } ParserData;
  53
  54 static void
  55 swfdec_text_field_movie_html_parse_close_tag (ParserData *data, ParserTag *tag,
  56     gboolean end)
  57 {
  58   g_return_if_fail (data != NULL);
  59   g_return_if_fail (tag != NULL);
  60
  61   if (data->multiline && !end &&
  62       ((tag->name_length == 1 && !g_strncasecmp (tag->name, "p", 1)) ||
  63        (tag->name_length == 2 && !g_strncasecmp (tag->name, "li", 2))))
  64   {
  65     GSList *iter;
  66
  67     for (iter = data->tags_closed; iter != NULL; iter = iter->next) {
  68       ParserTag *f = iter->data;
  69       if (f->end_index < tag->index)
  70         break;
  71       if (f->name_length == 4 && !g_strncasecmp (f->name, "font", 4)) {
  72         ParserTag *n = g_new0 (ParserTag, 1);
  73         n->name = f->name;
  74         n->name_length = f->name_length;
  75         n->index = swfdec_text_buffer_get_length (data->text);
  76         n->end_index = n->index + 1;
  77         if (f->format != NULL) {
  78           n->format = swfdec_text_format_copy (f->format);
  79         } else {
  80           n->format = NULL;
  81         }
  82         data->tags_closed = g_slist_prepend (data->tags_closed, n);
  83         break;
  84       }
  85     }
  86     swfdec_text_buffer_append_text (data->text, "\n");
  87   }
  88
  89   tag->end_index = swfdec_text_buffer_get_length (data->text);
  90
  91   data->tags_open = g_slist_remove (data->tags_open, tag);
  92   data->tags_closed = g_slist_prepend (data->tags_closed, tag);
  93 }
  94
  95 static const char *
  96 swfdec_text_field_movie_html_parse_comment (ParserData *data, const char *p)
  97 {
  98   const char *end;
  99
 100   g_return_val_if_fail (data != NULL, NULL);
 101   g_return_val_if_fail (p != NULL, NULL);
 102   g_return_val_if_fail (strncmp (p, "<!--", strlen ("<!--")) == 0, NULL);
 103
 104   end = strstr (p + strlen ("<!--"), "-->");
 105   if (end != NULL)
 106     end += strlen("-->");
 107
 108   // return NULL if no end found
 109   return end;
 110 }
 111
 112 static void
 113 swfdec_text_field_movie_html_tag_set_attribute (ParserData *data,
 114     ParserTag *tag, const char *name, int name_length, const char *value,
 115     int value_length)
 116 {
 117   SwfdecAsValue val;
 118   SwfdecAsObject *object;
 119
 120   g_return_if_fail (data != NULL);
 121   g_return_if_fail (tag != NULL);
 122   g_return_if_fail (name != NULL);
 123   g_return_if_fail (name_length >= 0);
 124   g_return_if_fail (value != NULL);
 125   g_return_if_fail (value_length >= 0);
 126
 127   if (!tag->format)
 128     return;
 129
 130   object = SWFDEC_AS_OBJECT (tag->format);
 131   SWFDEC_AS_VALUE_SET_STRING (&val, swfdec_as_context_give_string (
 132         swfdec_gc_object_get_context (object), g_strndup (value, value_length)));
 133
 134   if (tag->name_length == 10 && !g_strncasecmp (tag->name, "textformat", 10))
 135   {
 136     if (name_length == 10 && !g_strncasecmp (name, "leftmargin", 10))
 137     {
 138       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_leftMargin, &val);
 139     }
 140     else if (name_length == 11 && !g_strncasecmp (name, "rightmargin", 11))
 141     {
 142       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_rightMargin, &val);
 143     }
 144     else if (name_length == 6 && !g_strncasecmp (name, "indent", 6))
 145     {
 146       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_indent, &val);
 147     }
 148     else if (name_length == 11 && !g_strncasecmp (name, "blockindent", 11))
 149     {
 150       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_blockIndent, &val);
 151     }
 152     else if (name_length == 8 && !g_strncasecmp (name, "tabstops", 8))
 153     {
 154       // FIXME
 155       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_tabStops, &val);
 156     }
 157   }
 158   else if (tag->name_length == 1 && !g_strncasecmp (tag->name, "p", 1))
 159   {
 160     if (name_length == 5 && !g_strncasecmp (name, "align", 5))
 161     {
 162       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_align, &val);
 163     }
 164   }
 165   else if (tag->name_length == 4 && !g_strncasecmp (tag->name, "font", 4))
 166   {
 167     if (name_length == 4 && !g_strncasecmp (name, "face", 4))
 168     {
 169       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_font, &val);
 170     }
 171     else if (name_length == 4 && !g_strncasecmp (name, "size", 4))
 172     {
 173       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_size, &val);
 174     }
 175     else if (name_length == 5 && !g_strncasecmp (name, "color", 5))
 176     {
 177       SwfdecAsValue val_number;
 178
 179       if (value_length != 7 || *value != '#') {
 180         SWFDEC_AS_VALUE_SET_NUMBER (&val_number, 0);
 181       } else {
 182         int number;
 183         char *tail;
 184
 185         number = g_ascii_strtoll (value + 1, &tail, 16);
 186         if (tail != value + 7)
 187           number = 0;
 188         SWFDEC_AS_VALUE_SET_NUMBER (&val_number, number);
 189       }
 190
 191       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_color, &val_number);
 192     }
 193     else if (name_length == 13 && !g_strncasecmp (name, "letterspacing", 13))
 194     {
 195       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_letterSpacing,
 196           &val);
 197     }
 198     // special case: Don't parse kerning
 199   }
 200   else if (tag->name_length == 1 && !g_strncasecmp (tag->name, "a", 1))
 201   {
 202     if (name_length == 4 && !g_strncasecmp (name, "href", 4))
 203     {
 204       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_url, &val);
 205     }
 206     else if (name_length == 6 && !g_strncasecmp (name, "target", 6))
 207     {
 208       swfdec_as_object_set_variable (object, SWFDEC_AS_STR_target, &val);
 209     }
 210   }
 211
 212   if (data->style_sheet &&
 213       ((tag->name_length == 2 && !g_strncasecmp (tag->name, "li", 2)) ||
 214       (tag->name_length == 4 && !g_strncasecmp (tag->name, "span", 4)) ||
 215       (tag->name_length == 1 && !g_strncasecmp (tag->name, "p", 1))))
 216   {
 217     if (name_length == 5 && !g_strncasecmp (name, "class", 5)) {
 218       SwfdecTextFormat *format = swfdec_style_sheet_get_class_format (
 219           data->style_sheet, swfdec_as_context_give_string (data->cx,
 220               g_strndup (value, value_length)));
 221       if (format != NULL)
 222         swfdec_text_format_add (tag->format, format);
 223     }
 224   }
 225 }
 226
 227 static const char *
 228 swfdec_text_field_movie_html_parse_attribute (ParserData *data, ParserTag *tag,
 229     const char *p)
 230 {
 231   const char *end, *name, *value;
 232   int name_length, value_length;
 233
 234   g_return_val_if_fail (data != NULL, NULL);
 235   g_return_val_if_fail (tag != NULL, NULL);
 236   g_return_val_if_fail ((*p != '>' && *p != '\0'), NULL);
 237
 238   end = p + strcspn (p, "=> \r\n\t");
 239   if (end - p <= 0)
 240     return NULL; // Correct?
 241
 242   name = p;
 243   name_length = end - p;
 244
 245   p = end + strspn (end, " \r\n\t");
 246   if (*p != '=')
 247     return NULL; // FIXME: Correct?
 248   p = p + 1;
 249   p = p + strspn (p, " \r\n\t");
 250
 251   if (*p != '"' && *p != '\'')
 252     return NULL; // FIXME: Correct?
 253
 254   end = p + 1;
 255   do {
 256     end = strchr (end, *p);
 257   } while (end != NULL && *(end - 1) == '\\');
 258
 259   if (end == NULL)
 260     return NULL; // FIXME: Correct?
 261
 262   value = p + 1;
 263   value_length = end - (p + 1);
 264
 265   if (tag != NULL) {
 266     swfdec_text_field_movie_html_tag_set_attribute (data, tag, name,
 267         name_length, value, value_length);
 268   }
 269
 270   g_return_val_if_fail (end + 1 > p, NULL);
 271
 272   return end + 1;
 273 }
 274
 275 static const char *
 276 swfdec_text_field_movie_html_parse_tag (ParserData *data, const char *p)
 277 {
 278   ParserTag *tag;
 279   const char *name, *end;
 280   int name_length;
 281   gboolean close;
 282
 283   g_return_val_if_fail (data != NULL, NULL);
 284   g_return_val_if_fail (p != NULL, NULL);
 285   g_return_val_if_fail (*p == '<', NULL);
 286
 287   p++;
 288
 289   // closing tag or opening tag?
 290   if (*p == '/') {
 291     close = TRUE;
 292     p++;
 293   } else {
 294     close = FALSE;
 295   }
 296
 297   // find the end of the name
 298   end = p + strcspn (p, "> \r\n\t");
 299
 300   if (*end == '\0')
 301     return NULL;
 302
 303   // don't count trailing / as part of the name if it's followed by >
 304   // we still act like it's a normal opening tag even if it has /
 305   if (*end == '>' && *(end - 1) == '/')
 306     end = end - 1;
 307
 308   if (end == p) // empty name
 309     return NULL;
 310
 311   name = p;
 312   name_length = end - p;
 313
 314   if (close)
 315   {
 316     if (name_length == 1 && !g_strncasecmp (name, "p", 1)) {
 317       GSList *iter, *found;
 318
 319       found = NULL;
 320       iter = data->tags_open;
 321       while (iter != NULL) {
 322         tag = iter->data;
 323         if (tag->name_length == 1 && !g_strncasecmp (tag->name, "p", 1))
 324           found = iter;
 325         iter = iter->next;
 326       }
 327
 328       if (found != NULL) {
 329         iter = data->tags_open;
 330         while (iter != found) {
 331           tag = iter->data;
 332           iter = iter->next;
 333           if ((tag->name_length == 2 && !g_strncasecmp (tag->name, "li", 2)) ||
 334               (tag->name_length == 10 &&
 335                !g_strncasecmp (tag->name, "textformat", 10)))
 336             continue;
 337           swfdec_text_field_movie_html_parse_close_tag (data, tag, TRUE);
 338         }
 339         if (data->multiline)
 340           swfdec_text_buffer_append_text (data->text, "\n");
 341         swfdec_text_field_movie_html_parse_close_tag (data, found->data,
 342             TRUE);
 343       }
 344     } else {
 345       if (data->tags_open != NULL) {
 346         tag = data->tags_open->data;
 347         if (name_length == tag->name_length &&
 348             !g_strncasecmp (name, tag->name, name_length))
 349           swfdec_text_field_movie_html_parse_close_tag (data, tag, FALSE);
 350       }
 351     }
 352
 353     end = strchr (end, '>');
 354     if (end != NULL)
 355       end += 1;
 356   }
 357   else
 358   {
 359     SwfdecAsObject *object;
 360     SwfdecAsValue val;
 361
 362     if (name_length == 3 && !g_strncasecmp (name, "tab", 3))
 363       swfdec_text_buffer_append_text (data->text, "\t");
 364
 365     if (data->multiline) {
 366       if (name_length == 2 && !g_strncasecmp (name, "br", 2))
 367       {
 368         swfdec_text_buffer_append_text (data->text, "\n");
 369       }
 370       else if (name_length == 2 && !g_strncasecmp (name, "li", 1))
 371       {
 372         gsize length = swfdec_text_buffer_get_length (data->text);
 373         const char *s = swfdec_text_buffer_get_text (data->text);
 374         if (length > 0 && s[length-1] != '\n' && s[length-1] != '\r')
 375           swfdec_text_buffer_append_text (data->text, "\n");
 376       }
 377     }
 378
 379     tag = g_new0 (ParserTag, 1);
 380     tag->name = name;
 381     tag->name_length = name_length;
 382     tag->format = SWFDEC_TEXT_FORMAT (swfdec_text_format_new (data->cx));
 383     tag->index = swfdec_text_buffer_get_length (data->text);
 384
 385     data->tags_open = g_slist_prepend (data->tags_open, tag);
 386
 387     // set format based on tag
 388     if (tag->format != NULL) {
 389       object = SWFDEC_AS_OBJECT (tag->format);
 390       SWFDEC_AS_VALUE_SET_BOOLEAN (&val, TRUE);
 391
 392       if (tag->name_length == 2 && !g_strncasecmp (tag->name, "li", 2)) {
 393         swfdec_as_object_set_variable (object, SWFDEC_AS_STR_bullet, &val);
 394       } else if (tag->name_length == 1 && !g_strncasecmp (tag->name, "b", 1)) {
 395         swfdec_as_object_set_variable (object, SWFDEC_AS_STR_bold, &val);
 396       } else if (tag->name_length == 1 && !g_strncasecmp (tag->name, "i", 1)) {
 397         swfdec_as_object_set_variable (object, SWFDEC_AS_STR_italic, &val);
 398       } else if (tag->name_length == 1 && !g_strncasecmp (tag->name, "u", 1)) {
 399         swfdec_as_object_set_variable (object, SWFDEC_AS_STR_underline, &val);
 400       }
 401       else if (tag->name_length == 3 && !g_strncasecmp (tag->name, "img", 3))
 402       {
 403         SWFDEC_FIXME ("IMG tag support for TextField's HTML input missing");
 404       }
 405     }
 406
 407     if (data->style_sheet &&
 408         ((tag->name_length == 2 && !g_strncasecmp (tag->name, "li", 2)) ||
 409         (tag->name_length == 1 && !g_strncasecmp (tag->name, "p", 1)))) {
 410       SwfdecTextFormat *format = swfdec_style_sheet_get_tag_format (
 411           data->style_sheet, swfdec_as_context_give_string (data->cx,
 412               g_strndup (tag->name, tag->name_length)));
 413       if (format != NULL)
 414         swfdec_text_format_add (tag->format, format);
 415     }
 416
 417     // parse attributes
 418     end = end + strspn (end, " \r\n\t");
 419     while (*end != '\0' && *end != '>' && (*end != '/' || *(end + 1) != '>')) {
 420       end = swfdec_text_field_movie_html_parse_attribute (data, tag, end);
 421       if (end == NULL)
 422         break;
 423       end = end + strspn (end, " \r\n\t");
 424     }
 425     if (end != NULL) {
 426       if (*end == '/')
 427         end += 1;
 428       if (*end == '>')
 429         end += 1;
 430     }
 431   }
 432
 433   return end;
 434 }
 435
 436 static const char *
 437 swfdec_text_field_movie_html_parse_text (ParserData *data, const char *p)
 438 {
 439   const char *end;
 440   char *unescaped;
 441
 442   g_return_val_if_fail (data != NULL, NULL);
 443   g_return_val_if_fail (p != NULL, NULL);
 444   g_return_val_if_fail (*p != '\0' && *p != '<', NULL);
 445
 446   // condense the whitespace with previous text
 447   // also skip any whitespace that would be preceding first actual text
 448   if (data->condense_white && data->cx->version >= 8) {
 449     gsize length = swfdec_text_buffer_get_length (data->text);
 450     const char *s = swfdec_text_buffer_get_text (data->text);
 451     if (length == 0 || g_ascii_isspace (s[length - 1]))
 452       p += strspn (p, " \n\r\t");
 453   }
 454
 455   // if it's only whitespace, don't add it
 456   if (data->cx->version < 8) {
 457     end = p + strspn (p, " \n\r\t");
 458     if (*end == '\0' || *end == '<')
 459       return end;
 460   }
 461
 462   // get the text
 463   // if condense_white: all whitespace blocks are converted to a single space
 464   while (*p != '\0' && *p != '<') {
 465     if (data->condense_white) {
 466       end = p + strcspn (p, "< \n\r\t");
 467     } else {
 468       end = strchr (p, '<');
 469       if (end == NULL)
 470         end = strchr (p, '\0');
 471     }
 472
 473     unescaped = swfdec_xml_unescape_len (data->cx, p, end - p, TRUE);
 474     swfdec_text_buffer_append_text (data->text, unescaped);
 475     g_free (unescaped);
 476
 477     if (data->condense_white && g_ascii_isspace (*end)) {
 478       swfdec_text_buffer_append_text (data->text, " ");
 479       p = end + strspn (end, " \n\r\t");
 480     } else {
 481       p = end;
 482     }
 483   }
 484
 485   return p;
 486 }
 487
 488 void
 489 swfdec_text_field_movie_html_parse (SwfdecTextFieldMovie *text, const char *str)
 490 {
 491   ParserData data;
 492   const char *p;
 493
 494   g_return_if_fail (SWFDEC_IS_TEXT_FIELD_MOVIE (text));
 495   g_return_if_fail (str != NULL);
 496
 497   data.cx = swfdec_gc_object_get_context (text);
 498   data.multiline = (data.cx->version < 7 || text->multiline);
 499   data.condense_white = text->condense_white;
 500   if (text->style_sheet != NULL && SWFDEC_IS_STYLESHEET (text->style_sheet)) {
 501     data.style_sheet = SWFDEC_STYLESHEET (text->style_sheet);
 502   } else {
 503     data.style_sheet = NULL;
 504   }
 505   data.text = text->text;
 506   data.tags_open = NULL;
 507   data.tags_closed = NULL;
 508
 509   p = str;
 510   while (p != NULL && *p != '\0') {
 511     if (*p == '<') {
 512       if (strncmp (p + 1, "!--", strlen ("!--")) == 0) {
 513         p = swfdec_text_field_movie_html_parse_comment (&data, p);
 514       } else {
 515         p = swfdec_text_field_movie_html_parse_tag (&data, p);
 516       }
 517     } else {
 518       p = swfdec_text_field_movie_html_parse_text (&data, p);
 519     }
 520   }
 521
 522   // close remaining tags
 523   while (data.tags_open != NULL) {
 524     /* yes, this really appends to the default format */
 525     swfdec_text_buffer_set_default_attributes (text->text,
 526         &((ParserTag *)data.tags_open->data)->format->attr,
 527         ((ParserTag *)data.tags_open->data)->format->values_set);
 528     swfdec_text_field_movie_html_parse_close_tag (&data,
 529         (ParserTag *)data.tags_open->data, TRUE);
 530   }
 531
 532   // add parsed styles
 533   while (data.tags_closed != NULL) {
 534     ParserTag *tag = (ParserTag *)data.tags_closed->data;
 535
 536     if (tag->index != tag->end_index && tag->format != NULL) {
 537       swfdec_text_buffer_set_attributes (text->text, tag->index,
 538           tag->end_index - tag->index, &tag->format->attr, tag->format->values_set);
 539     }
 540
 541     g_free (tag);
 542     data.tags_closed = g_slist_remove (data.tags_closed, tag);
 543   }
 544 }
 545
 546 /*
 547  * Generating
 548  */
 549 static const char *
 550 swfdec_text_field_movie_html_text_align_to_string (SwfdecTextAlign align)
 551 {
 552   switch (align) {
 553     case SWFDEC_TEXT_ALIGN_LEFT:
 554       return "LEFT";
 555     case SWFDEC_TEXT_ALIGN_RIGHT:
 556       return "RIGHT";
 557     case SWFDEC_TEXT_ALIGN_CENTER:
 558       return "CENTER";
 559     case SWFDEC_TEXT_ALIGN_JUSTIFY:
 560       return "JUSTIFY";
 561     default:
 562       g_assert_not_reached ();
 563       return "";
 564   }
 565 }
 566
 567 /*
 568  * Order of tags:
 569  * TEXTFORMAT / P or LI / FONT / A / B / I / U
 570  *
 571  * Order of attributes:
 572  * TEXTFORMAT:
 573  * LEFTMARGIN / RIGHTMARGIN / INDENT / LEADING / BLOCKINDENT / TABSTOPS
 574  * P: ALIGN
 575  * LI: none
 576  * FONT: FACE / SIZE / COLOR / LETTERSPACING / KERNING
 577  * A: HREF / TARGET
 578  * B: none
 579  * I: none
 580  * U: none
 581  */
 582 static GString *
 583 swfdec_text_field_movie_html_text_append_paragraph (SwfdecTextFieldMovie *text,
 584     GString *string, guint start_index, guint end_index)
 585 {
 586   const SwfdecTextAttributes *attr, *attr_prev, *attr_font;
 587   SwfdecTextBufferIter *iter;
 588   GSList *fonts, *iter_font;
 589   guint index_, index_prev;
 590   gboolean textformat, bullet, font = FALSE;
 591   char *escaped;
 592
 593   g_return_val_if_fail (SWFDEC_IS_TEXT_FIELD_MOVIE (text), string);
 594   g_return_val_if_fail (string != NULL, string);
 595   g_return_val_if_fail (start_index <= end_index, string);
 596
 597   iter = swfdec_text_buffer_get_iter (text->text, start_index);
 598   index_ = start_index;
 599   attr = swfdec_text_buffer_iter_get_attributes (text->text, iter);
 600
 601   if (attr->left_margin != 0 || attr->right_margin != 0 ||
 602       attr->indent != 0 || attr->leading != 0 ||
 603       attr->block_indent != 0 ||
 604       attr->n_tab_stops > 0)
 605   {
 606     string = g_string_append (string, "<TEXTFORMAT");
 607     if (attr->left_margin) {
 608       g_string_append_printf (string, " LEFTMARGIN=\"%i\"",
 609           attr->left_margin);
 610     }
 611     if (attr->right_margin) {
 612       g_string_append_printf (string, " RIGHTMARGIN=\"%i\"",
 613           attr->right_margin);
 614     }
 615     if (attr->indent)
 616       g_string_append_printf (string, " INDENT=\"%i\"", attr->indent);
 617     if (attr->leading)
 618       g_string_append_printf (string, " LEADING=\"%i\"", attr->leading);
 619     if (attr->block_indent) {
 620       g_string_append_printf (string, " BLOCKINDENT=\"%i\"",
 621           attr->block_indent);
 622     }
 623     if (attr->n_tab_stops > 0) {
 624       guint i;
 625       g_string_append (string, " TABSTOPS=\"\"");
 626       for (i = 0; i < attr->n_tab_stops; i++) {
 627         g_string_append_printf (string, "%d,", attr->tab_stops[i]);
 628       }
 629       string->str[string->len - 1] = '\"';
 630     }
 631     string = g_string_append (string, ">");
 632
 633     textformat = TRUE;
 634   }
 635   else
 636   {
 637     textformat = FALSE;
 638   }
 639
 640   if (attr->bullet) {
 641     string = g_string_append (string, "<LI>");
 642     bullet = TRUE;
 643   } else {
 644     g_string_append_printf (string, "<P ALIGN=\"%s\">",
 645         swfdec_text_field_movie_html_text_align_to_string (attr->align));
 646     bullet = FALSE;
 647   }
 648
 649   // note we don't escape attr->font, even thought it can have evil chars
 650   g_string_append_printf (string, "<FONT FACE=\"%s\" SIZE=\"%i\" COLOR=\"#%06X\" LETTERSPACING=\"%i\" KERNING=\"%i\">",
 651       attr->font, attr->size, attr->color, (int)attr->letter_spacing,
 652       (attr->kerning ? 1 : 0));
 653   fonts = g_slist_prepend (NULL, (gpointer) attr);
 654
 655   if (attr->url != SWFDEC_AS_STR_EMPTY)
 656     g_string_append_printf (string, "<A HREF=\"%s\" TARGET=\"%s\">",
 657         attr->url, attr->target);
 658   if (attr->bold)
 659     string = g_string_append (string, "<B>");
 660   if (attr->italic)
 661     string = g_string_append (string, "<I>");
 662   if (attr->underline)
 663     string = g_string_append (string, "<U>");
 664
 665   // special case: use <= instead of < to add some extra markup
 666   for (iter = swfdec_text_buffer_iter_next (text->text, iter);
 667       iter != NULL && swfdec_text_buffer_iter_get_start (text->text, iter) <= end_index;
 668       iter = swfdec_text_buffer_iter_next (text->text, iter))
 669   {
 670     index_prev = index_;
 671     attr_prev = attr;
 672     index_ = swfdec_text_buffer_iter_get_start (text->text, iter);
 673     attr = swfdec_text_buffer_iter_get_attributes (text->text, iter);
 674
 675     escaped = swfdec_xml_escape_len (swfdec_text_buffer_get_text (text->text) + index_prev,
 676         index_ - index_prev);
 677     string = g_string_append (string, escaped);
 678     g_free (escaped);
 679     escaped = NULL;
 680
 681     // Figure out what tags need to be rewritten
 682     if (attr->font != attr_prev->font ||
 683         attr->size != attr_prev->size ||
 684         attr->color != attr_prev->color ||
 685         (int)attr->letter_spacing != (int)attr_prev->letter_spacing ||
 686         attr->kerning != attr_prev->kerning) {
 687       font = TRUE;
 688     } else if (attr->url == attr_prev->url &&
 689         attr->target == attr_prev->target &&
 690         attr->bold == attr_prev->bold &&
 691         attr->italic == attr_prev->italic &&
 692         attr->underline == attr_prev->underline) {
 693       continue;
 694     }
 695
 696     // Close tags
 697     for (iter_font = fonts; iter_font != NULL; iter_font = iter_font->next)
 698     {
 699       attr_font = iter_font->data;
 700       if (attr->font == attr_font->font &&
 701         attr->size == attr_font->size &&
 702         attr->color == attr_font->color &&
 703         (int)attr->letter_spacing == (int)attr_font->letter_spacing &&
 704         attr->kerning == attr_font->kerning) {
 705         break;
 706       }
 707     }
 708     if (attr_prev->underline)
 709       string = g_string_append (string, "</U>");
 710     if (attr_prev->italic)
 711       string = g_string_append (string, "</I>");
 712     if (attr_prev->bold)
 713       string = g_string_append (string, "</B>");
 714     if (attr_prev->url != SWFDEC_AS_STR_EMPTY)
 715       string = g_string_append (string, "</A>");
 716     if (iter_font != NULL) {
 717       while (fonts != iter_font) {
 718         string = g_string_append (string, "</FONT>");
 719         fonts = g_slist_remove (fonts, fonts->data);
 720       }
 721     }
 722
 723     // Open tags
 724     attr_font = fonts->data;
 725     if (font && (attr->font != attr_font->font ||
 726          attr->size != attr_font->size ||
 727          attr->color != attr_font->color ||
 728          (int)attr->letter_spacing != (int)attr_font->letter_spacing ||
 729          attr->kerning != attr_font->kerning))
 730     {
 731       fonts = g_slist_prepend (fonts, (gpointer) attr);
 732
 733       string = g_string_append (string, "<FONT");
 734       // note we don't escape attr->font, even thought it can have evil chars
 735       if (attr->font != attr_font->font)
 736         g_string_append_printf (string, " FACE=\"%s\"", attr->font);
 737       if (attr->size != attr_font->size)
 738         g_string_append_printf (string, " SIZE=\"%i\"", attr->size);
 739       if (attr->color != attr_font->color)
 740         g_string_append_printf (string, " COLOR=\"#%06X\"", attr->color);
 741       if ((int)attr->letter_spacing != (int)attr_font->letter_spacing) {
 742         g_string_append_printf (string, " LETTERSPACING=\"%i\"",
 743             (int)attr->letter_spacing);
 744       }
 745       if (attr->kerning != attr_font->kerning) {
 746         g_string_append_printf (string, " KERNING=\"%i\"",
 747             (attr->kerning ? 1 : 0));
 748       }
 749       string = g_string_append (string, ">");
 750     }
 751     if (attr->url != SWFDEC_AS_STR_EMPTY) {
 752       g_string_append_printf (string, "<A HREF=\"%s\" TARGET=\"%s\">",
 753           attr->url, attr->target);
 754     }
 755     if (attr->bold)
 756       string = g_string_append (string, "<B>");
 757     if (attr->italic)
 758       string = g_string_append (string, "<I>");
 759     if (attr->underline)
 760       string = g_string_append (string, "<U>");
 761   }
 762
 763   escaped = swfdec_xml_escape_len (swfdec_text_buffer_get_text (text->text) + index_,
 764       end_index - index_);
 765   string = g_string_append (string, escaped);
 766   g_free (escaped);
 767
 768   if (attr->underline)
 769     string = g_string_append (string, "</U>");
 770   if (attr->italic)
 771     string = g_string_append (string, "</I>");
 772   if (attr->bold)
 773     string = g_string_append (string, "</B>");
 774   if (attr->url != SWFDEC_AS_STR_EMPTY)
 775     string = g_string_append (string, "</A>");
 776   for (iter_font = fonts; iter_font != NULL; iter_font = iter_font->next)
 777     string = g_string_append (string, "</FONT>");
 778   g_slist_free (fonts);
 779   if (bullet) {
 780     string = g_string_append (string, "</LI>");
 781   } else {
 782     string = g_string_append (string, "</P>");
 783   }
 784   if (textformat)
 785     string = g_string_append (string, "</TEXTFORMAT>");
 786
 787   return string;
 788 }
 789
 790 const char *
 791 swfdec_text_field_movie_get_html_text (SwfdecTextFieldMovie *text)
 792 {
 793   const char *p, *end, *start;
 794   GString *string;
 795
 796   g_return_val_if_fail (SWFDEC_IS_TEXT_FIELD_MOVIE (text),
 797       SWFDEC_AS_STR_EMPTY);
 798
 799   string = g_string_new ("");
 800   p = start = swfdec_text_buffer_get_text (text->text);
 801
 802   while (*p != '\0') {
 803     end = strpbrk (p, "\r\n");
 804     if (end == NULL)
 805       end = strchr (p, '\0');
 806
 807     string = swfdec_text_field_movie_html_text_append_paragraph (text, string,
 808         p - start, end - start);
 809
 810     p = end;
 811     if (*p != '\0') p++;
 812   }
 813
 814   return swfdec_as_context_give_string (swfdec_gc_object_get_context (text),
 815       g_string_free (string, FALSE));
 816 }