git.sesse.net Git - vlc/blob - modules/codec/subtitles/subsdec.c

   1 /*****************************************************************************
   2  * subsdec.c : text subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 the VideoLAN team
   5  * $Id$
   6  *
   7  * Authors: Gildas Bazin <gbazin@videolan.org>
   8  *          Samuel Hocevar <sam@zoy.org>
   9  *          Derk-Jan Hartman <hartman at videolan dot org>
  10  *          Bernie Purcell <bitmap@videolan.org>
  11  *
  12  * This program is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * This program is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with this program; if not, write to the Free Software
  24  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  25  *****************************************************************************/
  26
  27 /*****************************************************************************
  28  * Preamble
  29  *****************************************************************************/
  30 #ifdef HAVE_CONFIG_H
  31 # include "config.h"
  32 #endif
  33
  34 #include "subsdec.h"
  35
  36 /*****************************************************************************
  37  * Local prototypes
  38  *****************************************************************************/
  39 static int  OpenDecoder   ( vlc_object_t * );
  40 static void CloseDecoder  ( vlc_object_t * );
  41
  42 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
  43 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
  44 static char           *StripTags      ( char * );
  45 static char           *CreateHtmlSubtitle ( char * );
  46
  47
  48 /*****************************************************************************
  49  * Module descriptor.
  50  *****************************************************************************/
  51 static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  52     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  53     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  54     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  55     "ISO-8859-6", "CP1256", "MacArabic", "",
  56     "ISO-8859-7", "CP1253", "MacGreek", "",
  57     "ISO-8859-8", "CP1255", "MacHebrew", "",
  58     "ISO-8859-9", "CP1254", "MacTurkish", "",
  59     "ISO-8859-13", "CP1257", "",
  60     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  61     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  62     "ISO-2022-KR", "EUC-KR", "",
  63     "MacThai", "KOI8-T", "",
  64     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  65     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  66     "Macintosh", "",
  67     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  68     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  69     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  70     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  71     "HPROMAN8", "NEXTSTEP" };
  72 /*
  73 SSA supports charset selection.
  74 The following known charsets are used:
  75
  76 0 = Ansi - Western European
  77 1 = default
  78 2 = symbol
  79 3 = invalid
  80 77 = Mac
  81 128 = Japanese (Shift JIS)
  82 129 = Hangul
  83 130 = Johab
  84 134 = GB2312 Simplified Chinese
  85 136 = Big5 Traditional Chinese
  86 161 = Greek
  87 162 = Turkish
  88 163 = Vietnamese
  89 177 = Hebrew
  90 178 = Arabic
  91 186 = Baltic
  92 204 = Russian (Cyrillic)
  93 222 = Thai
  94 238 = Eastern European
  95 254 = PC 437
  96 */
  97
  98 static int  pi_justification[] = { 0, 1, 2 };
  99 static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
 100
 101 #define ENCODING_TEXT N_("Subtitles text encoding")
 102 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 103 #define ALIGN_TEXT N_("Subtitles justification")
 104 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 105 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 106 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 107             "UTF-8 encoding within subtitles files.")
 108 #define FORMAT_TEXT N_("Formatted Subtitles")
 109 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 110  "VLC partly implements this, but you can choose to disable all formatting.")
 111
 112
 113 vlc_module_begin();
 114     set_shortname( _("Subtitles"));
 115     set_description( _("Text subtitles decoder") );
 116     set_capability( "decoder", 50 );
 117     set_callbacks( OpenDecoder, CloseDecoder );
 118     set_category( CAT_INPUT );
 119     set_subcategory( SUBCAT_INPUT_SCODEC );
 120
 121     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 122                  VLC_FALSE );
 123         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 124     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 125                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 126         change_string_list( ppsz_encodings, 0, 0 );
 127     add_bool( "subsdec-autodetect-utf8", VLC_TRUE, NULL,
 128               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, VLC_FALSE );
 129     add_bool( "subsdec-formatted", VLC_TRUE, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
 130                  VLC_FALSE );
 131 vlc_module_end();
 132
 133 /*****************************************************************************
 134  * OpenDecoder: probe the decoder and return score
 135  *****************************************************************************
 136  * Tries to launch a decoder and return score so that the interface is able
 137  * to chose.
 138  *****************************************************************************/
 139 static int OpenDecoder( vlc_object_t *p_this )
 140 {
 141     decoder_t     *p_dec = (decoder_t*)p_this;
 142     decoder_sys_t *p_sys;
 143     vlc_value_t    val;
 144
 145     switch( p_dec->fmt_in.i_codec )
 146     {
 147         case VLC_FOURCC('s','u','b','t'):
 148         case VLC_FOURCC('s','s','a',' '):
 149         case VLC_FOURCC('t','1','4','0'):
 150             break;
 151         default:
 152             return VLC_EGENERIC;
 153     }
 154
 155     p_dec->pf_decode_sub = DecodeBlock;
 156
 157     /* Allocate the memory needed to store the decoder's structure */
 158     p_dec->p_sys = p_sys = malloc( sizeof( *p_sys ) );
 159     if( p_sys == NULL )
 160     {
 161         msg_Err( p_dec, "out of memory" );
 162         return VLC_ENOMEM;
 163     }
 164
 165     /* init of p_sys */
 166     memset( p_sys, 0, sizeof( *p_sys ) );
 167     p_sys->i_align = 0;
 168     p_sys->iconv_handle = (vlc_iconv_t)-1;
 169     p_sys->b_autodetect_utf8 = VLC_FALSE;
 170     p_sys->b_ass = VLC_FALSE;
 171     p_sys->i_original_height = -1;
 172     p_sys->i_original_width = -1;
 173     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 174     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 175
 176     char *psz_charset = NULL;
 177
 178     /* First try demux-specified encoding */
 179     if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
 180         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 181     else
 182     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 183     {
 184         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 185         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 186                  p_dec->fmt_in.subs.psz_encoding ?: "not specified");
 187     }
 188
 189     /* Second, try configured encoding */
 190     if (psz_charset == NULL)
 191     {
 192         psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
 193         if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
 194         {
 195             free (psz_charset);
 196             psz_charset = NULL;
 197         }
 198
 199         msg_Dbg (p_dec, "trying configured character encoding: %s",
 200                  psz_charset ?: "not specified");
 201     }
 202
 203     /* Third, try "local" encoding with optional UTF-8 autodetection */
 204     if (psz_charset == NULL)
 205     {
 206         psz_charset = strdup (GetFallbackEncoding ());
 207         msg_Dbg (p_dec, "trying default character encoding: %s",
 208                  psz_charset ?: "not specified");
 209
 210         if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
 211         {
 212             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 213             p_sys->b_autodetect_utf8 = VLC_TRUE;
 214         }
 215     }
 216
 217     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 218     if (psz_charset == NULL)
 219     {
 220         psz_charset = strdup ("UTF-8");
 221         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 222     }
 223
 224     if ((psz_charset != NULL)
 225      && strcasecmp (psz_charset, "UTF-8")
 226      && strcasecmp (psz_charset, "utf8"))
 227     {
 228         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 229         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 230             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 231     }
 232     free (psz_charset);
 233
 234     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 235     var_Get( p_dec, "subsdec-align", &val );
 236     p_sys->i_align = val.i_int;
 237
 238     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 239      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 240     {
 241         if( p_dec->fmt_in.i_extra > 0 )
 242             ParseSSAHeader( p_dec );
 243     }
 244
 245     return VLC_SUCCESS;
 246 }
 247
 248 /****************************************************************************
 249  * DecodeBlock: the whole thing
 250  ****************************************************************************
 251  * This function must be fed with complete subtitles units.
 252  ****************************************************************************/
 253 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 254 {
 255     subpicture_t *p_spu = NULL;
 256
 257     if( !pp_block || *pp_block == NULL ) return NULL;
 258
 259     p_spu = ParseText( p_dec, *pp_block );
 260
 261     block_Release( *pp_block );
 262     *pp_block = NULL;
 263
 264     return p_spu;
 265 }
 266
 267 /*****************************************************************************
 268  * CloseDecoder: clean up the decoder
 269  *****************************************************************************/
 270 static void CloseDecoder( vlc_object_t *p_this )
 271 {
 272     decoder_t *p_dec = (decoder_t *)p_this;
 273     decoder_sys_t *p_sys = p_dec->p_sys;
 274
 275     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 276         vlc_iconv_close( p_sys->iconv_handle );
 277
 278     if( p_sys->pp_ssa_styles )
 279     {
 280         int i;
 281         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 282         {
 283             if( !p_sys->pp_ssa_styles[i] )
 284                 continue;
 285
 286             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 287             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 288             free( p_sys->pp_ssa_styles[i] );
 289         }
 290         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 291     }
 292     if( p_sys->pp_images )
 293     {
 294         int i;
 295         for( i = 0; i < p_sys->i_images; i++ )
 296         {
 297             if( !p_sys->pp_images[i] )
 298                 continue;
 299
 300             if( p_sys->pp_images[i]->p_pic )
 301                 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
 302             free( p_sys->pp_images[i]->psz_filename );
 303
 304             free( p_sys->pp_images[i] );
 305         }
 306         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 307     }
 308
 309     free( p_sys );
 310 }
 311
 312 /*****************************************************************************
 313  * ParseText: parse an text subtitle packet and send it to the video output
 314  *****************************************************************************/
 315 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 316 {
 317     decoder_sys_t *p_sys = p_dec->p_sys;
 318     subpicture_t *p_spu = NULL;
 319     char *psz_subtitle = NULL;
 320     video_format_t fmt;
 321
 322     /* We cannot display a subpicture with no date */
 323     if( p_block->i_pts == 0 )
 324     {
 325         msg_Warn( p_dec, "subtitle without a date" );
 326         return NULL;
 327     }
 328
 329     /* Check validity of packet data */
 330     /* An "empty" line containing only \0 can be used to force
 331        and ephemer picture from the screen */
 332     if( p_block->i_buffer < 1 )
 333     {
 334         msg_Warn( p_dec, "no subtitle data" );
 335         return NULL;
 336     }
 337
 338     /* Should be resiliant against bad subtitles */
 339     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 340                             p_block->i_buffer );
 341     if( psz_subtitle == NULL )
 342         return NULL;
 343
 344     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 345     {
 346         if (EnsureUTF8( psz_subtitle ) == NULL)
 347         {
 348             msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 349                      "Try manually setting a character-encoding "
 350                      "before you open the file.") );
 351         }
 352     }
 353     else
 354     {
 355
 356         if( p_sys->b_autodetect_utf8 )
 357         {
 358             if( IsUTF8( psz_subtitle ) == NULL )
 359             {
 360                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 361                          "disabling UTF-8 subtitles autodetection" );
 362                 p_sys->b_autodetect_utf8 = VLC_FALSE;
 363             }
 364         }
 365
 366         if( !p_sys->b_autodetect_utf8 )
 367         {
 368             size_t inbytes_left = strlen( psz_subtitle );
 369             size_t outbytes_left = 6 * inbytes_left;
 370             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 371             char *psz_convert_buffer_out = psz_new_subtitle;
 372             const char *psz_convert_buffer_in = psz_subtitle;
 373
 374             size_t ret = vlc_iconv( p_sys->iconv_handle,
 375                                     &psz_convert_buffer_in, &inbytes_left,
 376                                     &psz_convert_buffer_out, &outbytes_left );
 377
 378             *psz_convert_buffer_out++ = '\0';
 379             free( psz_subtitle );
 380
 381             if( ( ret == (size_t)(-1) ) || inbytes_left )
 382             {
 383                 free( psz_new_subtitle );
 384                 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 385                         "Try manually setting a character-encoding "
 386                                 "before you open the file.") );
 387                 return NULL;
 388             }
 389
 390             psz_subtitle = realloc( psz_new_subtitle,
 391                                     psz_convert_buffer_out - psz_new_subtitle );
 392         }
 393     }
 394
 395     /* Create the subpicture unit */
 396     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 397     if( !p_spu )
 398     {
 399         msg_Warn( p_dec, "can't get spu buffer" );
 400         free( psz_subtitle );
 401         return NULL;
 402     }
 403
 404     p_spu->b_pausable = VLC_TRUE;
 405
 406     /* Create a new subpicture region */
 407     memset( &fmt, 0, sizeof(video_format_t) );
 408     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 409     fmt.i_aspect = 0;
 410     fmt.i_width = fmt.i_height = 0;
 411     fmt.i_x_offset = fmt.i_y_offset = 0;
 412     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 413     if( !p_spu->p_region )
 414     {
 415         msg_Err( p_dec, "cannot allocate SPU region" );
 416         free( psz_subtitle );
 417         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 418         return NULL;
 419     }
 420
 421     /* Decode and format the subpicture unit */
 422     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 423     {
 424         /* Normal text subs, easy markup */
 425         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 426         p_spu->i_x = p_sys->i_align ? 20 : 0;
 427         p_spu->i_y = 10;
 428
 429         /* Remove formatting from string */
 430
 431         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 432         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 433         {
 434             p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
 435         }
 436
 437         p_spu->i_start = p_block->i_pts;
 438         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 439         p_spu->b_ephemer = (p_block->i_length == 0);
 440         p_spu->b_absolute = VLC_FALSE;
 441     }
 442     else
 443     {
 444         /* Decode SSA/USF strings */
 445         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 446             ParseSSAString( p_dec, psz_subtitle, p_spu );
 447
 448         p_spu->i_start = p_block->i_pts;
 449         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 450         p_spu->b_ephemer = (p_block->i_length == 0);
 451         p_spu->b_absolute = VLC_FALSE;
 452         p_spu->i_original_picture_width = p_sys->i_original_width;
 453         p_spu->i_original_picture_height = p_sys->i_original_height;
 454     }
 455     free( psz_subtitle );
 456
 457     return p_spu;
 458 }
 459
 460 char* GotoNextLine( char *psz_text )
 461 {
 462     char *p_newline = psz_text;
 463
 464     while( p_newline[0] != '\0' )
 465     {
 466         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 467         {
 468             p_newline++;
 469             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 470                 p_newline++;
 471             break;
 472         }
 473         else p_newline++;
 474     }
 475     return p_newline;
 476 }
 477
 478 /* Function now handles tags with attribute values, and tries
 479  * to deal with &' commands too. It no longer modifies the string
 480  * in place, so that the original text can be reused
 481  */
 482 static char *StripTags( char *psz_subtitle )
 483 {
 484     char *psz_text_start;
 485     char *psz_text;
 486
 487     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 488     if( !psz_text_start )
 489         return NULL;
 490
 491     while( *psz_subtitle )
 492     {
 493         if( *psz_subtitle == '<' )
 494         {
 495             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 496                 *psz_text++ = '\n';
 497
 498             psz_subtitle += strcspn( psz_subtitle, ">" );
 499         }
 500         else if( *psz_subtitle == '&' )
 501         {
 502             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 503             {
 504                 *psz_text++ = '<';
 505                 psz_subtitle += strcspn( psz_subtitle, ";" );
 506             }
 507             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 508             {
 509                 *psz_text++ = '>';
 510                 psz_subtitle += strcspn( psz_subtitle, ";" );
 511             }
 512             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 513             {
 514                 *psz_text++ = '&';
 515                 psz_subtitle += strcspn( psz_subtitle, ";" );
 516             }
 517             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 518             {
 519                 *psz_text++ = '\"';
 520                 psz_subtitle += strcspn( psz_subtitle, ";" );
 521             }
 522             else
 523             {
 524                 /* Assume it is just a normal ampersand */
 525                 *psz_text++ = '&';
 526             }
 527         }
 528         else
 529         {
 530             *psz_text++ = *psz_subtitle;
 531         }
 532
 533         psz_subtitle++;
 534     }
 535     *psz_text = '\0';
 536     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 537
 538     return psz_text_start;
 539 }
 540
 541 /* Try to respect any style tags present in the subtitle string. The main
 542  * problem here is a lack of adequate specs for the subtitle formats.
 543  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 544  * SAMI has a detailed spec, but extensive rework is needed in the demux
 545  * code to prevent all this style information being excised, as it presently
 546  * does.
 547  * That leaves the others - none of which were (I guess) originally intended
 548  * to be carrying style information. Over time people have used them that way.
 549  * In the absence of specifications from which to work, the tags supported
 550  * have been restricted to the simple set permitted by the USF DTD, ie. :
 551  *  Basic: <br>, <i>, <b>, <u>
 552  *  Extended: <font>
 553  *    Attributes: face
 554  *                family
 555  *                size
 556  *                color
 557  *                outline-color
 558  *                shadow-color
 559  *                outline-level
 560  *                shadow-level
 561  *                back-color
 562  *                alpha
 563  * There is also the further restriction that the subtitle be well-formed
 564  * as an XML entity, ie. the HTML sentence:
 565  *        <b><i>Bold and Italics</b></i>
 566  * doesn't qualify because the tags aren't nested one inside the other.
 567  * <text> tags are automatically added to the output to ensure
 568  * well-formedness.
 569  * If the text doesn't qualify for any reason, a NULL string is
 570  * returned, and the rendering engine will fall back to the
 571  * plain text version of the subtitle.
 572  */
 573 static char *CreateHtmlSubtitle( char *psz_subtitle )
 574 {
 575     char    psz_tagStack[ 100 ];
 576     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 577     char   *psz_html_start = malloc( i_buf_size );
 578
 579     psz_tagStack[ 0 ] = '\0';
 580
 581     if( psz_html_start != NULL )
 582     {
 583         char *psz_html = psz_html_start;
 584
 585         strcpy( psz_html, "<text>" );
 586         psz_html += 6;
 587
 588         while( *psz_subtitle )
 589         {
 590             if( *psz_subtitle == '\n' )
 591             {
 592                 strcpy( psz_html, "<br/>" );
 593                 psz_html += 5;
 594                 psz_subtitle++;
 595             }
 596             else if( *psz_subtitle == '<' )
 597             {
 598                 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 599                 {
 600                     strcpy( psz_html, "<br/>" );
 601                     psz_html += 5;
 602                     psz_subtitle += 5;
 603                 }
 604                 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 605                 {
 606                     strcpy( psz_html, "<b>" );
 607                     strcat( psz_tagStack, "b" );
 608                     psz_html += 3;
 609                     psz_subtitle += 3;
 610                 }
 611                 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 612                 {
 613                     strcpy( psz_html, "<i>" );
 614                     strcat( psz_tagStack, "i" );
 615                     psz_html += 3;
 616                     psz_subtitle += 3;
 617                 }
 618                 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 619                 {
 620                     strcpy( psz_html, "<u>" );
 621                     strcat( psz_tagStack, "u" );
 622                     psz_html += 3;
 623                     psz_subtitle += 3;
 624                 }
 625                 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 626                 {
 627                     const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
 628                             "color=\"", "outline-color=\"", "shadow-color=\"",
 629                             "outline-level=\"", "shadow-level=\"", "back-color=\"",
 630                             "alpha=\"", NULL };
 631
 632                     strcpy( psz_html, "<font " );
 633                     strcat( psz_tagStack, "f" );
 634                     psz_html += 6;
 635                     psz_subtitle += 6;
 636
 637                     while( *psz_subtitle != '>' )
 638                     {
 639                         int  k;
 640
 641                         for( k=0; psz_attribs[ k ]; k++ )
 642                         {
 643                             int i_len = strlen( psz_attribs[ k ] );
 644
 645                             if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
 646                             {
 647                                 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 648
 649                                 strncpy( psz_html, psz_subtitle, i_len );
 650                                 psz_html += i_len;
 651                                 psz_subtitle += i_len;
 652                                 break;
 653                             }
 654                         }
 655                         if( psz_attribs[ k ] == NULL )
 656                         {
 657                             /* Jump over unrecognised tag */
 658                             int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 659
 660                             i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 661                             psz_subtitle += i_len;
 662                         }
 663                         while (*psz_subtitle == ' ')
 664                             *psz_html++ = *psz_subtitle++;
 665                     }
 666                     *psz_html++ = *psz_subtitle++;
 667                 }
 668                 else if( !strncmp( psz_subtitle, "</", 2 ))
 669                 {
 670                     vlc_bool_t  b_match     = VLC_FALSE;
 671                     int         i_len       = strlen( psz_tagStack ) - 1;
 672                     char       *psz_lastTag = NULL;
 673
 674                     if( i_len >= 0 )
 675                     {
 676                         psz_lastTag = psz_tagStack + i_len;
 677                         i_len = 0;
 678
 679                         switch( *psz_lastTag )
 680                         {
 681                             case 'b':
 682                                 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 683                                 i_len   = 4;
 684                                 break;
 685                             case 'i':
 686                                 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 687                                 i_len   = 4;
 688                                 break;
 689                             case 'u':
 690                                 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 691                                 i_len   = 4;
 692                                 break;
 693                             case 'f':
 694                                 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 695                                 i_len   = 7;
 696                                 break;
 697                         }
 698                     }
 699                     if( ! b_match )
 700                     {
 701                         /* Not well formed -- kill everything */
 702                         free( psz_html_start );
 703                         psz_html_start = NULL;
 704                         break;
 705                     }
 706                     *psz_lastTag = '\0';
 707                     strncpy( psz_html, psz_subtitle, i_len );
 708                     psz_html += i_len;
 709                     psz_subtitle += i_len;
 710                 }
 711                 else
 712                 {
 713                     psz_subtitle += strcspn( psz_subtitle, ">" );
 714                 }
 715             }
 716             else if( *psz_subtitle == '&' )
 717             {
 718                 if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 719                 {
 720                     strcpy( psz_html, "&lt;" );
 721                     psz_html += 4;
 722                     psz_subtitle += 4;
 723                 }
 724                 else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 725                 {
 726                     strcpy( psz_html, "&gt;" );
 727                     psz_html += 4;
 728                     psz_subtitle += 4;
 729                 }
 730                 else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 731                 {
 732                     strcpy( psz_html, "&amp;" );
 733                     psz_html += 5;
 734                     psz_subtitle += 5;
 735                 }
 736                 else
 737                 {
 738                     strcpy( psz_html, "&amp;" );
 739                     psz_html += 5;
 740                     psz_subtitle++;
 741                 }
 742             }
 743             else
 744             {
 745                 *psz_html = *psz_subtitle;
 746                 if( psz_html > psz_html_start )
 747                 {
 748                     /* Check for double whitespace */
 749                     if((( *psz_html == ' ' ) ||
 750                         ( *psz_html == '\t' )) &&
 751                        (( *(psz_html-1) == ' ' ) ||
 752                         ( *(psz_html-1) == '\t' )))
 753                     {
 754                         strcpy( psz_html, NO_BREAKING_SPACE );
 755                         psz_html += strlen( NO_BREAKING_SPACE ) - 1;
 756                     }
 757                 }
 758                 psz_html++;
 759                 psz_subtitle++;
 760             }
 761
 762             if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
 763             {
 764                 int i_len = psz_html - psz_html_start;
 765
 766                 i_buf_size += 100;
 767                 psz_html_start = realloc( psz_html_start, i_buf_size );
 768                 psz_html = psz_html_start + i_len;
 769                 *psz_html = '\0';
 770             }
 771         }
 772         strcpy( psz_html, "</text>" );
 773         psz_html += 7;
 774
 775         if( psz_tagStack[ 0 ] != '\0' )
 776         {
 777             /* Not well formed -- kill everything */
 778             free( psz_html_start );
 779             psz_html_start = NULL;
 780         }
 781         else if( psz_html_start )
 782         {
 783             /* Shrink the memory requirements */
 784             psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 785         }
 786     }
 787     return psz_html_start;
 788 }