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             if( p_sys->pp_ssa_styles[i]->psz_stylename )
 287                 free( p_sys->pp_ssa_styles[i]->psz_stylename );
 288             if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname )
 289                 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 290             if( p_sys->pp_ssa_styles[i] )
 291                 free( p_sys->pp_ssa_styles[i] );
 292         }
 293         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 294     }
 295     if( p_sys->pp_images )
 296     {
 297         int i;
 298         for( i = 0; i < p_sys->i_images; i++ )
 299         {
 300             if( !p_sys->pp_images[i] )
 301                 continue;
 302
 303             if( p_sys->pp_images[i]->p_pic )
 304                 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
 305             if( p_sys->pp_images[i]->psz_filename )
 306                 free( p_sys->pp_images[i]->psz_filename );
 307
 308             free( p_sys->pp_images[i] );
 309         }
 310         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 311     }
 312
 313     free( p_sys );
 314 }
 315
 316 /*****************************************************************************
 317  * ParseText: parse an text subtitle packet and send it to the video output
 318  *****************************************************************************/
 319 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 320 {
 321     decoder_sys_t *p_sys = p_dec->p_sys;
 322     subpicture_t *p_spu = NULL;
 323     char *psz_subtitle = NULL;
 324     video_format_t fmt;
 325
 326     /* We cannot display a subpicture with no date */
 327     if( p_block->i_pts == 0 )
 328     {
 329         msg_Warn( p_dec, "subtitle without a date" );
 330         return NULL;
 331     }
 332
 333     /* Check validity of packet data */
 334     /* An "empty" line containing only \0 can be used to force
 335        and ephemer picture from the screen */
 336     if( p_block->i_buffer < 1 )
 337     {
 338         msg_Warn( p_dec, "no subtitle data" );
 339         return NULL;
 340     }
 341
 342     /* Should be resiliant against bad subtitles */
 343     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 344                             p_block->i_buffer );
 345     if( psz_subtitle == NULL )
 346         return NULL;
 347
 348     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 349     {
 350         if (EnsureUTF8( psz_subtitle ) == NULL)
 351         {
 352             msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 353                      "Try manually setting a character-encoding "
 354                      "before you open the file.") );
 355         }
 356     }
 357     else
 358     {
 359
 360         if( p_sys->b_autodetect_utf8 )
 361         {
 362             if( IsUTF8( psz_subtitle ) == NULL )
 363             {
 364                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 365                          "disabling UTF-8 subtitles autodetection" );
 366                 p_sys->b_autodetect_utf8 = VLC_FALSE;
 367             }
 368         }
 369
 370         if( !p_sys->b_autodetect_utf8 )
 371         {
 372             size_t inbytes_left = strlen( psz_subtitle );
 373             size_t outbytes_left = 6 * inbytes_left;
 374             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 375             char *psz_convert_buffer_out = psz_new_subtitle;
 376             const char *psz_convert_buffer_in = psz_subtitle;
 377
 378             size_t ret = vlc_iconv( p_sys->iconv_handle,
 379                                     &psz_convert_buffer_in, &inbytes_left,
 380                                     &psz_convert_buffer_out, &outbytes_left );
 381
 382             *psz_convert_buffer_out++ = '\0';
 383             free( psz_subtitle );
 384
 385             if( ( ret == (size_t)(-1) ) || inbytes_left )
 386             {
 387                 free( psz_new_subtitle );
 388                 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 389                         "Try manually setting a character-encoding "
 390                                 "before you open the file.") );
 391                 return NULL;
 392             }
 393
 394             psz_subtitle = realloc( psz_new_subtitle,
 395                                     psz_convert_buffer_out - psz_new_subtitle );
 396         }
 397     }
 398
 399     /* Create the subpicture unit */
 400     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 401     if( !p_spu )
 402     {
 403         msg_Warn( p_dec, "can't get spu buffer" );
 404         if( psz_subtitle ) free( psz_subtitle );
 405         return NULL;
 406     }
 407
 408     p_spu->b_pausable = VLC_TRUE;
 409
 410     /* Create a new subpicture region */
 411     memset( &fmt, 0, sizeof(video_format_t) );
 412     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 413     fmt.i_aspect = 0;
 414     fmt.i_width = fmt.i_height = 0;
 415     fmt.i_x_offset = fmt.i_y_offset = 0;
 416     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 417     if( !p_spu->p_region )
 418     {
 419         msg_Err( p_dec, "cannot allocate SPU region" );
 420         if( psz_subtitle ) free( psz_subtitle );
 421         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 422         return NULL;
 423     }
 424
 425     /* Decode and format the subpicture unit */
 426     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 427     {
 428         /* Normal text subs, easy markup */
 429         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 430         p_spu->i_x = p_sys->i_align ? 20 : 0;
 431         p_spu->i_y = 10;
 432
 433         /* Remove formatting from string */
 434
 435         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 436         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 437         {
 438             p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
 439         }
 440
 441         p_spu->i_start = p_block->i_pts;
 442         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 443         p_spu->b_ephemer = (p_block->i_length == 0);
 444         p_spu->b_absolute = VLC_FALSE;
 445     }
 446     else
 447     {
 448         /* Decode SSA/USF strings */
 449         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 450             ParseSSAString( p_dec, psz_subtitle, p_spu );
 451
 452         p_spu->i_start = p_block->i_pts;
 453         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 454         p_spu->b_ephemer = (p_block->i_length == 0);
 455         p_spu->b_absolute = VLC_FALSE;
 456         p_spu->i_original_picture_width = p_sys->i_original_width;
 457         p_spu->i_original_picture_height = p_sys->i_original_height;
 458     }
 459     if( psz_subtitle ) free( psz_subtitle );
 460
 461     return p_spu;
 462 }
 463
 464 char* GotoNextLine( char *psz_text )
 465 {
 466     char *p_newline = psz_text;
 467
 468     while( p_newline[0] != '\0' )
 469     {
 470         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 471         {
 472             p_newline++;
 473             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 474                 p_newline++;
 475             break;
 476         }
 477         else p_newline++;
 478     }
 479     return p_newline;
 480 }
 481
 482 /* Function now handles tags with attribute values, and tries
 483  * to deal with &' commands too. It no longer modifies the string
 484  * in place, so that the original text can be reused
 485  */
 486 static char *StripTags( char *psz_subtitle )
 487 {
 488     char *psz_text_start;
 489     char *psz_text;
 490
 491     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 492     if( !psz_text_start )
 493         return NULL;
 494
 495     while( *psz_subtitle )
 496     {
 497         if( *psz_subtitle == '<' )
 498         {
 499             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 500                 *psz_text++ = '\n';
 501
 502             psz_subtitle += strcspn( psz_subtitle, ">" );
 503         }
 504         else if( *psz_subtitle == '&' )
 505         {
 506             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 507             {
 508                 *psz_text++ = '<';
 509                 psz_subtitle += strcspn( psz_subtitle, ";" );
 510             }
 511             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 512             {
 513                 *psz_text++ = '>';
 514                 psz_subtitle += strcspn( psz_subtitle, ";" );
 515             }
 516             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 517             {
 518                 *psz_text++ = '&';
 519                 psz_subtitle += strcspn( psz_subtitle, ";" );
 520             }
 521             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 522             {
 523                 *psz_text++ = '\"';
 524                 psz_subtitle += strcspn( psz_subtitle, ";" );
 525             }
 526             else
 527             {
 528                 /* Assume it is just a normal ampersand */
 529                 *psz_text++ = '&';
 530             }
 531         }
 532         else
 533         {
 534             *psz_text++ = *psz_subtitle;
 535         }
 536
 537         psz_subtitle++;
 538     }
 539     *psz_text = '\0';
 540     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 541
 542     return psz_text_start;
 543 }
 544
 545 /* Try to respect any style tags present in the subtitle string. The main
 546  * problem here is a lack of adequate specs for the subtitle formats.
 547  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 548  * SAMI has a detailed spec, but extensive rework is needed in the demux
 549  * code to prevent all this style information being excised, as it presently
 550  * does.
 551  * That leaves the others - none of which were (I guess) originally intended
 552  * to be carrying style information. Over time people have used them that way.
 553  * In the absence of specifications from which to work, the tags supported
 554  * have been restricted to the simple set permitted by the USF DTD, ie. :
 555  *  Basic: <br>, <i>, <b>, <u>
 556  *  Extended: <font>
 557  *    Attributes: face
 558  *                family
 559  *                size
 560  *                color
 561  *                outline-color
 562  *                shadow-color
 563  *                outline-level
 564  *                shadow-level
 565  *                back-color
 566  *                alpha
 567  * There is also the further restriction that the subtitle be well-formed
 568  * as an XML entity, ie. the HTML sentence:
 569  *        <b><i>Bold and Italics</b></i>
 570  * doesn't qualify because the tags aren't nested one inside the other.
 571  * <text> tags are automatically added to the output to ensure
 572  * well-formedness.
 573  * If the text doesn't qualify for any reason, a NULL string is
 574  * returned, and the rendering engine will fall back to the
 575  * plain text version of the subtitle.
 576  */
 577 static char *CreateHtmlSubtitle( char *psz_subtitle )
 578 {
 579     char    psz_tagStack[ 100 ];
 580     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 581     char   *psz_html_start = malloc( i_buf_size );
 582
 583     psz_tagStack[ 0 ] = '\0';
 584
 585     if( psz_html_start != NULL )
 586     {
 587         char *psz_html = psz_html_start;
 588
 589         strcpy( psz_html, "<text>" );
 590         psz_html += 6;
 591
 592         while( *psz_subtitle )
 593         {
 594             if( *psz_subtitle == '\n' )
 595             {
 596                 strcpy( psz_html, "<br/>" );
 597                 psz_html += 5;
 598                 psz_subtitle++;
 599             }
 600             else if( *psz_subtitle == '<' )
 601             {
 602                 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 603                 {
 604                     strcpy( psz_html, "<br/>" );
 605                     psz_html += 5;
 606                     psz_subtitle += 5;
 607                 }
 608                 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 609                 {
 610                     strcpy( psz_html, "<b>" );
 611                     strcat( psz_tagStack, "b" );
 612                     psz_html += 3;
 613                     psz_subtitle += 3;
 614                 }
 615                 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 616                 {
 617                     strcpy( psz_html, "<i>" );
 618                     strcat( psz_tagStack, "i" );
 619                     psz_html += 3;
 620                     psz_subtitle += 3;
 621                 }
 622                 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 623                 {
 624                     strcpy( psz_html, "<u>" );
 625                     strcat( psz_tagStack, "u" );
 626                     psz_html += 3;
 627                     psz_subtitle += 3;
 628                 }
 629                 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 630                 {
 631                     const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
 632                             "color=\"", "outline-color=\"", "shadow-color=\"",
 633                             "outline-level=\"", "shadow-level=\"", "back-color=\"",
 634                             "alpha=\"", NULL };
 635
 636                     strcpy( psz_html, "<font " );
 637                     strcat( psz_tagStack, "f" );
 638                     psz_html += 6;
 639                     psz_subtitle += 6;
 640
 641                     while( *psz_subtitle != '>' )
 642                     {
 643                         int  k;
 644
 645                         for( k=0; psz_attribs[ k ]; k++ )
 646                         {
 647                             int i_len = strlen( psz_attribs[ k ] );
 648
 649                             if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
 650                             {
 651                                 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 652
 653                                 strncpy( psz_html, psz_subtitle, i_len );
 654                                 psz_html += i_len;
 655                                 psz_subtitle += i_len;
 656                                 break;
 657                             }
 658                         }
 659                         if( psz_attribs[ k ] == NULL )
 660                         {
 661                             /* Jump over unrecognised tag */
 662                             int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 663
 664                             i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 665                             psz_subtitle += i_len;
 666                         }
 667                         while (*psz_subtitle == ' ')
 668                             *psz_html++ = *psz_subtitle++;
 669                     }
 670                     *psz_html++ = *psz_subtitle++;
 671                 }
 672                 else if( !strncmp( psz_subtitle, "</", 2 ))
 673                 {
 674                     vlc_bool_t  b_match     = VLC_FALSE;
 675                     int         i_len       = strlen( psz_tagStack ) - 1;
 676                     char       *psz_lastTag = NULL;
 677
 678                     if( i_len >= 0 )
 679                     {
 680                         psz_lastTag = psz_tagStack + i_len;
 681                         i_len = 0;
 682
 683                         switch( *psz_lastTag )
 684                         {
 685                             case 'b':
 686                                 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 687                                 i_len   = 4;
 688                                 break;
 689                             case 'i':
 690                                 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 691                                 i_len   = 4;
 692                                 break;
 693                             case 'u':
 694                                 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 695                                 i_len   = 4;
 696                                 break;
 697                             case 'f':
 698                                 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 699                                 i_len   = 7;
 700                                 break;
 701                         }
 702                     }
 703                     if( ! b_match )
 704                     {
 705                         /* Not well formed -- kill everything */
 706                         free( psz_html_start );
 707                         psz_html_start = NULL;
 708                         break;
 709                     }
 710                     *psz_lastTag = '\0';
 711                     strncpy( psz_html, psz_subtitle, i_len );
 712                     psz_html += i_len;
 713                     psz_subtitle += i_len;
 714                 }
 715                 else
 716                 {
 717                     psz_subtitle += strcspn( psz_subtitle, ">" );
 718                 }
 719             }
 720             else if( *psz_subtitle == '&' )
 721             {
 722                 if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 723                 {
 724                     strcpy( psz_html, "&lt;" );
 725                     psz_html += 4;
 726                     psz_subtitle += 4;
 727                 }
 728                 else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 729                 {
 730                     strcpy( psz_html, "&gt;" );
 731                     psz_html += 4;
 732                     psz_subtitle += 4;
 733                 }
 734                 else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 735                 {
 736                     strcpy( psz_html, "&amp;" );
 737                     psz_html += 5;
 738                     psz_subtitle += 5;
 739                 }
 740                 else
 741                 {
 742                     strcpy( psz_html, "&amp;" );
 743                     psz_html += 5;
 744                     psz_subtitle++;
 745                 }
 746             }
 747             else
 748             {
 749                 *psz_html = *psz_subtitle;
 750                 if( psz_html > psz_html_start )
 751                 {
 752                     /* Check for double whitespace */
 753                     if((( *psz_html == ' ' ) ||
 754                         ( *psz_html == '\t' )) &&
 755                        (( *(psz_html-1) == ' ' ) ||
 756                         ( *(psz_html-1) == '\t' )))
 757                     {
 758                         strcpy( psz_html, NO_BREAKING_SPACE );
 759                         psz_html += strlen( NO_BREAKING_SPACE ) - 1;
 760                     }
 761                 }
 762                 psz_html++;
 763                 psz_subtitle++;
 764             }
 765
 766             if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
 767             {
 768                 int i_len = psz_html - psz_html_start;
 769
 770                 i_buf_size += 100;
 771                 psz_html_start = realloc( psz_html_start, i_buf_size );
 772                 psz_html = psz_html_start + i_len;
 773                 *psz_html = '\0';
 774             }
 775         }
 776         strcpy( psz_html, "</text>" );
 777         psz_html += 7;
 778
 779         if( psz_tagStack[ 0 ] != '\0' )
 780         {
 781             /* Not well formed -- kill everything */
 782             free( psz_html_start );
 783             psz_html_start = NULL;
 784         }
 785         else if( psz_html_start )
 786         {
 787             /* Shrink the memory requirements */
 788             psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 789         }
 790     }
 791     return psz_html_start;
 792 }