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
  31 #include "subsdec.h"
  32
  33 /*****************************************************************************
  34  * Local prototypes
  35  *****************************************************************************/
  36 static int  OpenDecoder   ( vlc_object_t * );
  37 static void CloseDecoder  ( vlc_object_t * );
  38
  39 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
  40 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
  41 static char           *StripTags      ( char * );
  42 static char           *CreateHtmlSubtitle ( char * );
  43
  44
  45 /*****************************************************************************
  46  * Module descriptor.
  47  *****************************************************************************/
  48 static const char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  49     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  50     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  51     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  52     "ISO-8859-6", "CP1256", "MacArabic", "",
  53     "ISO-8859-7", "CP1253", "MacGreek", "",
  54     "ISO-8859-8", "CP1255", "MacHebrew", "",
  55     "ISO-8859-9", "CP1254", "MacTurkish", "",
  56     "ISO-8859-13", "CP1257", "",
  57     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  58     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  59     "ISO-2022-KR", "EUC-KR", "",
  60     "MacThai", "KOI8-T", "",
  61     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  62     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  63     "Macintosh", "",
  64     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  65     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  66     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  67     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  68     "HPROMAN8", "NEXTSTEP" };
  69 /*
  70 SSA supports charset selection.
  71 The following known charsets are used:
  72
  73 0 = Ansi - Western European
  74 1 = default
  75 2 = symbol
  76 3 = invalid
  77 77 = Mac
  78 128 = Japanese (Shift JIS)
  79 129 = Hangul
  80 130 = Johab
  81 134 = GB2312 Simplified Chinese
  82 136 = Big5 Traditional Chinese
  83 161 = Greek
  84 162 = Turkish
  85 163 = Vietnamese
  86 177 = Hebrew
  87 178 = Arabic
  88 186 = Baltic
  89 204 = Russian (Cyrillic)
  90 222 = Thai
  91 238 = Eastern European
  92 254 = PC 437
  93 */
  94
  95 static int  pi_justification[] = { 0, 1, 2 };
  96 static const char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  97
  98 #define ENCODING_TEXT N_("Subtitles text encoding")
  99 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 100 #define ALIGN_TEXT N_("Subtitles justification")
 101 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 102 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 103 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 104             "UTF-8 encoding within subtitles files.")
 105 #define FORMAT_TEXT N_("Formatted Subtitles")
 106 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 107  "VLC partly implements this, but you can choose to disable all formatting.")
 108
 109
 110 vlc_module_begin();
 111     set_shortname( _("Subtitles"));
 112     set_description( _("Text subtitles decoder") );
 113     set_capability( "decoder", 50 );
 114     set_callbacks( OpenDecoder, CloseDecoder );
 115     set_category( CAT_INPUT );
 116     set_subcategory( SUBCAT_INPUT_SCODEC );
 117
 118     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 119                  VLC_FALSE );
 120         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 121     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 122                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 123         change_string_list( ppsz_encodings, 0, 0 );
 124     add_bool( "subsdec-autodetect-utf8", VLC_TRUE, NULL,
 125               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, VLC_FALSE );
 126     add_bool( "subsdec-formatted", VLC_TRUE, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
 127                  VLC_FALSE );
 128 vlc_module_end();
 129
 130 /*****************************************************************************
 131  * OpenDecoder: probe the decoder and return score
 132  *****************************************************************************
 133  * Tries to launch a decoder and return score so that the interface is able
 134  * to chose.
 135  *****************************************************************************/
 136 static int OpenDecoder( vlc_object_t *p_this )
 137 {
 138     decoder_t     *p_dec = (decoder_t*)p_this;
 139     decoder_sys_t *p_sys;
 140     vlc_value_t    val;
 141
 142     switch( p_dec->fmt_in.i_codec )
 143     {
 144         case VLC_FOURCC('s','u','b','t'):
 145         case VLC_FOURCC('s','s','a',' '):
 146         case VLC_FOURCC('t','1','4','0'):
 147             break;
 148         default:
 149             return VLC_EGENERIC;
 150     }
 151
 152     p_dec->pf_decode_sub = DecodeBlock;
 153
 154     /* Allocate the memory needed to store the decoder's structure */
 155     p_dec->p_sys = p_sys = malloc( sizeof( *p_sys ) );
 156     if( p_sys == NULL )
 157     {
 158         msg_Err( p_dec, "out of memory" );
 159         return VLC_ENOMEM;
 160     }
 161
 162     /* init of p_sys */
 163     memset( p_sys, 0, sizeof( *p_sys ) );
 164     p_sys->i_align = 0;
 165     p_sys->iconv_handle = (vlc_iconv_t)-1;
 166     p_sys->b_autodetect_utf8 = VLC_FALSE;
 167     p_sys->b_ass = VLC_FALSE;
 168     p_sys->i_original_height = -1;
 169     p_sys->i_original_width = -1;
 170     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 171     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 172
 173     char *psz_charset = NULL;
 174
 175     /* First try demux-specified encoding */
 176     if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
 177         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 178     else
 179     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 180     {
 181         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 182         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 183                  p_dec->fmt_in.subs.psz_encoding ?: "not specified");
 184     }
 185
 186     /* Second, try configured encoding */
 187     if (psz_charset == NULL)
 188     {
 189         psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
 190         if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
 191         {
 192             free (psz_charset);
 193             psz_charset = NULL;
 194         }
 195
 196         msg_Dbg (p_dec, "trying configured character encoding: %s",
 197                  psz_charset ?: "not specified");
 198     }
 199
 200     /* Third, try "local" encoding with optional UTF-8 autodetection */
 201     if (psz_charset == NULL)
 202     {
 203         psz_charset = strdup (GetFallbackEncoding ());
 204         msg_Dbg (p_dec, "trying default character encoding: %s",
 205                  psz_charset ?: "not specified");
 206
 207         if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
 208         {
 209             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 210             p_sys->b_autodetect_utf8 = VLC_TRUE;
 211         }
 212     }
 213
 214     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 215     if (psz_charset == NULL)
 216     {
 217         psz_charset = strdup ("UTF-8");
 218         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 219     }
 220
 221     if ((psz_charset != NULL)
 222      && strcasecmp (psz_charset, "UTF-8")
 223      && strcasecmp (psz_charset, "utf8"))
 224     {
 225         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 226         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 227             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 228     }
 229     free (psz_charset);
 230
 231     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 232     var_Get( p_dec, "subsdec-align", &val );
 233     p_sys->i_align = val.i_int;
 234
 235     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 236      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 237     {
 238         if( p_dec->fmt_in.i_extra > 0 )
 239             ParseSSAHeader( p_dec );
 240     }
 241
 242     return VLC_SUCCESS;
 243 }
 244
 245 /****************************************************************************
 246  * DecodeBlock: the whole thing
 247  ****************************************************************************
 248  * This function must be fed with complete subtitles units.
 249  ****************************************************************************/
 250 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 251 {
 252     subpicture_t *p_spu = NULL;
 253
 254     if( !pp_block || *pp_block == NULL ) return NULL;
 255
 256     p_spu = ParseText( p_dec, *pp_block );
 257
 258     block_Release( *pp_block );
 259     *pp_block = NULL;
 260
 261     return p_spu;
 262 }
 263
 264 /*****************************************************************************
 265  * CloseDecoder: clean up the decoder
 266  *****************************************************************************/
 267 static void CloseDecoder( vlc_object_t *p_this )
 268 {
 269     decoder_t *p_dec = (decoder_t *)p_this;
 270     decoder_sys_t *p_sys = p_dec->p_sys;
 271
 272     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 273         vlc_iconv_close( p_sys->iconv_handle );
 274
 275     if( p_sys->pp_ssa_styles )
 276     {
 277         int i;
 278         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 279         {
 280             if( !p_sys->pp_ssa_styles[i] )
 281                 continue;
 282
 283             if( p_sys->pp_ssa_styles[i]->psz_stylename )
 284                 free( p_sys->pp_ssa_styles[i]->psz_stylename );
 285             if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname )
 286                 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 287             if( p_sys->pp_ssa_styles[i] )
 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             if( p_sys->pp_images[i]->psz_filename )
 303                 free( p_sys->pp_images[i]->psz_filename );
 304
 305             free( p_sys->pp_images[i] );
 306         }
 307         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 308     }
 309
 310     free( p_sys );
 311 }
 312
 313 /*****************************************************************************
 314  * ParseText: parse an text subtitle packet and send it to the video output
 315  *****************************************************************************/
 316 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 317 {
 318     decoder_sys_t *p_sys = p_dec->p_sys;
 319     subpicture_t *p_spu = NULL;
 320     char *psz_subtitle = NULL;
 321     video_format_t fmt;
 322
 323     /* We cannot display a subpicture with no date */
 324     if( p_block->i_pts == 0 )
 325     {
 326         msg_Warn( p_dec, "subtitle without a date" );
 327         return NULL;
 328     }
 329
 330     /* Check validity of packet data */
 331     /* An "empty" line containing only \0 can be used to force
 332        and ephemer picture from the screen */
 333     if( p_block->i_buffer < 1 )
 334     {
 335         msg_Warn( p_dec, "no subtitle data" );
 336         return NULL;
 337     }
 338
 339     /* Should be resiliant against bad subtitles */
 340     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 341                             p_block->i_buffer );
 342     if( psz_subtitle == NULL )
 343         return NULL;
 344
 345     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 346     {
 347         if (EnsureUTF8( psz_subtitle ) == NULL)
 348         {
 349             msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 350                      "Try manually setting a character-encoding "
 351                      "before you open the file.") );
 352         }
 353     }
 354     else
 355     {
 356
 357         if( p_sys->b_autodetect_utf8 )
 358         {
 359             if( IsUTF8( psz_subtitle ) == NULL )
 360             {
 361                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 362                          "disabling UTF-8 subtitles autodetection" );
 363                 p_sys->b_autodetect_utf8 = VLC_FALSE;
 364             }
 365         }
 366
 367         if( !p_sys->b_autodetect_utf8 )
 368         {
 369             size_t inbytes_left = strlen( psz_subtitle );
 370             size_t outbytes_left = 6 * inbytes_left;
 371             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 372             char *psz_convert_buffer_out = psz_new_subtitle;
 373             const char *psz_convert_buffer_in = psz_subtitle;
 374
 375             size_t ret = vlc_iconv( p_sys->iconv_handle,
 376                                     &psz_convert_buffer_in, &inbytes_left,
 377                                     &psz_convert_buffer_out, &outbytes_left );
 378
 379             *psz_convert_buffer_out++ = '\0';
 380             free( psz_subtitle );
 381
 382             if( ( ret == (size_t)(-1) ) || inbytes_left )
 383             {
 384                 free( psz_new_subtitle );
 385                 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 386                         "Try manually setting a character-encoding "
 387                                 "before you open the file.") );
 388                 return NULL;
 389             }
 390
 391             psz_subtitle = realloc( psz_new_subtitle,
 392                                     psz_convert_buffer_out - psz_new_subtitle );
 393         }
 394     }
 395
 396     /* Create the subpicture unit */
 397     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 398     if( !p_spu )
 399     {
 400         msg_Warn( p_dec, "can't get spu buffer" );
 401         if( psz_subtitle ) free( psz_subtitle );
 402         return NULL;
 403     }
 404
 405     p_spu->b_pausable = VLC_TRUE;
 406
 407     /* Create a new subpicture region */
 408     memset( &fmt, 0, sizeof(video_format_t) );
 409     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 410     fmt.i_aspect = 0;
 411     fmt.i_width = fmt.i_height = 0;
 412     fmt.i_x_offset = fmt.i_y_offset = 0;
 413     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 414     if( !p_spu->p_region )
 415     {
 416         msg_Err( p_dec, "cannot allocate SPU region" );
 417         if( psz_subtitle ) free( psz_subtitle );
 418         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 419         return NULL;
 420     }
 421
 422     /* Decode and format the subpicture unit */
 423     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 424     {
 425         /* Normal text subs, easy markup */
 426         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 427         p_spu->i_x = p_sys->i_align ? 20 : 0;
 428         p_spu->i_y = 10;
 429
 430         /* Remove formatting from string */
 431
 432         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 433         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 434         {
 435             p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
 436         }
 437
 438         p_spu->i_start = p_block->i_pts;
 439         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 440         p_spu->b_ephemer = (p_block->i_length == 0);
 441         p_spu->b_absolute = VLC_FALSE;
 442     }
 443     else
 444     {
 445         /* Decode SSA/USF strings */
 446         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 447             ParseSSAString( p_dec, psz_subtitle, p_spu );
 448
 449         p_spu->i_start = p_block->i_pts;
 450         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 451         p_spu->b_ephemer = (p_block->i_length == 0);
 452         p_spu->b_absolute = VLC_FALSE;
 453         p_spu->i_original_picture_width = p_sys->i_original_width;
 454         p_spu->i_original_picture_height = p_sys->i_original_height;
 455     }
 456     if( psz_subtitle ) free( psz_subtitle );
 457
 458     return p_spu;
 459 }
 460
 461 char* GotoNextLine( char *psz_text )
 462 {
 463     char *p_newline = psz_text;
 464
 465     while( p_newline[0] != '\0' )
 466     {
 467         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 468         {
 469             p_newline++;
 470             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 471                 p_newline++;
 472             break;
 473         }
 474         else p_newline++;
 475     }
 476     return p_newline;
 477 }
 478
 479 /* Function now handles tags with attribute values, and tries
 480  * to deal with &' commands too. It no longer modifies the string
 481  * in place, so that the original text can be reused
 482  */
 483 static char *StripTags( char *psz_subtitle )
 484 {
 485     char *psz_text_start;
 486     char *psz_text;
 487
 488     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 489     if( !psz_text_start )
 490         return NULL;
 491
 492     while( *psz_subtitle )
 493     {
 494         if( *psz_subtitle == '<' )
 495         {
 496             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 497                 *psz_text++ = '\n';
 498
 499             psz_subtitle += strcspn( psz_subtitle, ">" );
 500         }
 501         else if( *psz_subtitle == '&' )
 502         {
 503             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 504             {
 505                 *psz_text++ = '<';
 506                 psz_subtitle += strcspn( psz_subtitle, ";" );
 507             }
 508             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 509             {
 510                 *psz_text++ = '>';
 511                 psz_subtitle += strcspn( psz_subtitle, ";" );
 512             }
 513             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 514             {
 515                 *psz_text++ = '&';
 516                 psz_subtitle += strcspn( psz_subtitle, ";" );
 517             }
 518             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 519             {
 520                 *psz_text++ = '\"';
 521                 psz_subtitle += strcspn( psz_subtitle, ";" );
 522             }
 523             else
 524             {
 525                 /* Assume it is just a normal ampersand */
 526                 *psz_text++ = '&';
 527             }
 528         }
 529         else
 530         {
 531             *psz_text++ = *psz_subtitle;
 532         }
 533
 534         psz_subtitle++;
 535     }
 536     *psz_text = '\0';
 537     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 538
 539     return psz_text_start;
 540 }
 541
 542 /* Try to respect any style tags present in the subtitle string. The main
 543  * problem here is a lack of adequate specs for the subtitle formats.
 544  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 545  * SAMI has a detailed spec, but extensive rework is needed in the demux
 546  * code to prevent all this style information being excised, as it presently
 547  * does.
 548  * That leaves the others - none of which were (I guess) originally intended
 549  * to be carrying style information. Over time people have used them that way.
 550  * In the absence of specifications from which to work, the tags supported
 551  * have been restricted to the simple set permitted by the USF DTD, ie. :
 552  *  Basic: <br>, <i>, <b>, <u>
 553  *  Extended: <font>
 554  *    Attributes: face
 555  *                family
 556  *                size
 557  *                color
 558  *                outline-color
 559  *                shadow-color
 560  *                outline-level
 561  *                shadow-level
 562  *                back-color
 563  *                alpha
 564  * There is also the further restriction that the subtitle be well-formed
 565  * as an XML entity, ie. the HTML sentence:
 566  *        <b><i>Bold and Italics</b></i>
 567  * doesn't qualify because the tags aren't nested one inside the other.
 568  * <text> tags are automatically added to the output to ensure
 569  * well-formedness.
 570  * If the text doesn't qualify for any reason, a NULL string is
 571  * returned, and the rendering engine will fall back to the
 572  * plain text version of the subtitle.
 573  */
 574 static char *CreateHtmlSubtitle( char *psz_subtitle )
 575 {
 576     char    psz_tagStack[ 100 ];
 577     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 578     char   *psz_html_start = malloc( i_buf_size );
 579
 580     psz_tagStack[ 0 ] = '\0';
 581
 582     if( psz_html_start != NULL )
 583     {
 584         char *psz_html = psz_html_start;
 585
 586         strcpy( psz_html, "<text>" );
 587         psz_html += 6;
 588
 589         while( *psz_subtitle )
 590         {
 591             if( *psz_subtitle == '\n' )
 592             {
 593                 strcpy( psz_html, "<br/>" );
 594                 psz_html += 5;
 595                 psz_subtitle++;
 596             }
 597             else if( *psz_subtitle == '<' )
 598             {
 599                 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 600                 {
 601                     strcpy( psz_html, "<br/>" );
 602                     psz_html += 5;
 603                     psz_subtitle += 5;
 604                 }
 605                 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 606                 {
 607                     strcpy( psz_html, "<b>" );
 608                     strcat( psz_tagStack, "b" );
 609                     psz_html += 3;
 610                     psz_subtitle += 3;
 611                 }
 612                 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 613                 {
 614                     strcpy( psz_html, "<i>" );
 615                     strcat( psz_tagStack, "i" );
 616                     psz_html += 3;
 617                     psz_subtitle += 3;
 618                 }
 619                 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 620                 {
 621                     strcpy( psz_html, "<u>" );
 622                     strcat( psz_tagStack, "u" );
 623                     psz_html += 3;
 624                     psz_subtitle += 3;
 625                 }
 626                 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 627                 {
 628                     const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
 629                             "color=\"", "outline-color=\"", "shadow-color=\"",
 630                             "outline-level=\"", "shadow-level=\"", "back-color=\"",
 631                             "alpha=\"", NULL };
 632
 633                     strcpy( psz_html, "<font " );
 634                     strcat( psz_tagStack, "f" );
 635                     psz_html += 6;
 636                     psz_subtitle += 6;
 637
 638                     while( *psz_subtitle != '>' )
 639                     {
 640                         int  k;
 641
 642                         for( k=0; psz_attribs[ k ]; k++ )
 643                         {
 644                             int i_len = strlen( psz_attribs[ k ] );
 645
 646                             if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
 647                             {
 648                                 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 649
 650                                 strncpy( psz_html, psz_subtitle, i_len );
 651                                 psz_html += i_len;
 652                                 psz_subtitle += i_len;
 653                                 break;
 654                             }
 655                         }
 656                         if( psz_attribs[ k ] == NULL )
 657                         {
 658                             /* Jump over unrecognised tag */
 659                             int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 660
 661                             i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 662                             psz_subtitle += i_len;
 663                         }
 664                         while (*psz_subtitle == ' ')
 665                             *psz_html++ = *psz_subtitle++;
 666                     }
 667                     *psz_html++ = *psz_subtitle++;
 668                 }
 669                 else if( !strncmp( psz_subtitle, "</", 2 ))
 670                 {
 671                     vlc_bool_t  b_match     = VLC_FALSE;
 672                     int         i_len       = strlen( psz_tagStack ) - 1;
 673                     char       *psz_lastTag = NULL;
 674
 675                     if( i_len >= 0 )
 676                     {
 677                         psz_lastTag = psz_tagStack + i_len;
 678                         i_len = 0;
 679
 680                         switch( *psz_lastTag )
 681                         {
 682                             case 'b':
 683                                 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 684                                 i_len   = 4;
 685                                 break;
 686                             case 'i':
 687                                 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 688                                 i_len   = 4;
 689                                 break;
 690                             case 'u':
 691                                 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 692                                 i_len   = 4;
 693                                 break;
 694                             case 'f':
 695                                 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 696                                 i_len   = 7;
 697                                 break;
 698                         }
 699                     }
 700                     if( ! b_match )
 701                     {
 702                         /* Not well formed -- kill everything */
 703                         free( psz_html_start );
 704                         psz_html_start = NULL;
 705                         break;
 706                     }
 707                     *psz_lastTag = '\0';
 708                     strncpy( psz_html, psz_subtitle, i_len );
 709                     psz_html += i_len;
 710                     psz_subtitle += i_len;
 711                 }
 712                 else
 713                 {
 714                     psz_subtitle += strcspn( psz_subtitle, ">" );
 715                 }
 716             }
 717             else if( *psz_subtitle == '&' )
 718             {
 719                 if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 720                 {
 721                     strcpy( psz_html, "&lt;" );
 722                     psz_html += 4;
 723                     psz_subtitle += 4;
 724                 }
 725                 else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 726                 {
 727                     strcpy( psz_html, "&gt;" );
 728                     psz_html += 4;
 729                     psz_subtitle += 4;
 730                 }
 731                 else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 732                 {
 733                     strcpy( psz_html, "&amp;" );
 734                     psz_html += 5;
 735                     psz_subtitle += 5;
 736                 }
 737                 else
 738                 {
 739                     strcpy( psz_html, "&amp;" );
 740                     psz_html += 5;
 741                     psz_subtitle++;
 742                 }
 743             }
 744             else
 745             {
 746                 *psz_html = *psz_subtitle;
 747                 if( psz_html > psz_html_start )
 748                 {
 749                     /* Check for double whitespace */
 750                     if((( *psz_html == ' ' ) ||
 751                         ( *psz_html == '\t' )) &&
 752                        (( *(psz_html-1) == ' ' ) ||
 753                         ( *(psz_html-1) == '\t' )))
 754                     {
 755                         strcpy( psz_html, NO_BREAKING_SPACE );
 756                         psz_html += strlen( NO_BREAKING_SPACE ) - 1;
 757                     }
 758                 }
 759                 psz_html++;
 760                 psz_subtitle++;
 761             }
 762
 763             if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
 764             {
 765                 int i_len = psz_html - psz_html_start;
 766
 767                 i_buf_size += 100;
 768                 psz_html_start = realloc( psz_html_start, i_buf_size );
 769                 psz_html = psz_html_start + i_len;
 770                 *psz_html = '\0';
 771             }
 772         }
 773         strcpy( psz_html, "</text>" );
 774         psz_html += 7;
 775
 776         if( psz_tagStack[ 0 ] != '\0' )
 777         {
 778             /* Not well formed -- kill everything */
 779             free( psz_html_start );
 780             psz_html_start = NULL;
 781         }
 782         else if( psz_html_start )
 783         {
 784             /* Shrink the memory requirements */
 785             psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 786         }
 787     }
 788     return psz_html_start;
 789 }