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, "trying hard-coded character encoding: %s",
 219                  psz_charset ? psz_charset : "error");
 220     }
 221
 222     /* Fifth, fail */
 223     if (psz_charset == NULL)
 224     {
 225         free (p_sys);
 226         return VLC_ENOMEM;
 227     }
 228
 229     if (strcasecmp (psz_charset, "UTF-8") && strcasecmp (psz_charset, "utf8"))
 230     {
 231         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 232         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 233             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 234     }
 235     free (psz_charset);
 236
 237     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 238     var_Get( p_dec, "subsdec-align", &val );
 239     p_sys->i_align = val.i_int;
 240
 241     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 242      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 243     {
 244         if( p_dec->fmt_in.i_extra > 0 )
 245             ParseSSAHeader( p_dec );
 246     }
 247
 248     return VLC_SUCCESS;
 249 }
 250
 251 /****************************************************************************
 252  * DecodeBlock: the whole thing
 253  ****************************************************************************
 254  * This function must be fed with complete subtitles units.
 255  ****************************************************************************/
 256 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 257 {
 258     subpicture_t *p_spu = NULL;
 259
 260     if( !pp_block || *pp_block == NULL ) return NULL;
 261
 262     p_spu = ParseText( p_dec, *pp_block );
 263
 264     block_Release( *pp_block );
 265     *pp_block = NULL;
 266
 267     return p_spu;
 268 }
 269
 270 /*****************************************************************************
 271  * CloseDecoder: clean up the decoder
 272  *****************************************************************************/
 273 static void CloseDecoder( vlc_object_t *p_this )
 274 {
 275     decoder_t *p_dec = (decoder_t *)p_this;
 276     decoder_sys_t *p_sys = p_dec->p_sys;
 277
 278     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 279         vlc_iconv_close( p_sys->iconv_handle );
 280
 281     if( p_sys->pp_ssa_styles )
 282     {
 283         int i;
 284         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 285         {
 286             if( !p_sys->pp_ssa_styles[i] )
 287                 continue;
 288
 289             if( p_sys->pp_ssa_styles[i]->psz_stylename )
 290                 free( p_sys->pp_ssa_styles[i]->psz_stylename );
 291             if( p_sys->pp_ssa_styles[i]->font_style.psz_fontname )
 292                 free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 293             if( p_sys->pp_ssa_styles[i] )
 294                 free( p_sys->pp_ssa_styles[i] );
 295         }
 296         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 297     }
 298     if( p_sys->pp_images )
 299     {
 300         int i;
 301         for( i = 0; i < p_sys->i_images; i++ )
 302         {
 303             if( !p_sys->pp_images[i] )
 304                 continue;
 305
 306             if( p_sys->pp_images[i]->p_pic )
 307                 p_sys->pp_images[i]->p_pic->pf_release( p_sys->pp_images[i]->p_pic );
 308             if( p_sys->pp_images[i]->psz_filename )
 309                 free( p_sys->pp_images[i]->psz_filename );
 310
 311             free( p_sys->pp_images[i] );
 312         }
 313         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 314     }
 315
 316     free( p_sys );
 317 }
 318
 319 /*****************************************************************************
 320  * ParseText: parse an text subtitle packet and send it to the video output
 321  *****************************************************************************/
 322 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 323 {
 324     decoder_sys_t *p_sys = p_dec->p_sys;
 325     subpicture_t *p_spu = NULL;
 326     char *psz_subtitle = NULL;
 327     video_format_t fmt;
 328
 329     /* We cannot display a subpicture with no date */
 330     if( p_block->i_pts == 0 )
 331     {
 332         msg_Warn( p_dec, "subtitle without a date" );
 333         return NULL;
 334     }
 335
 336     /* Check validity of packet data */
 337     /* An "empty" line containing only \0 can be used to force
 338        and ephemer picture from the screen */
 339     if( p_block->i_buffer < 1 )
 340     {
 341         msg_Warn( p_dec, "no subtitle data" );
 342         return NULL;
 343     }
 344
 345     /* Should be resiliant against bad subtitles */
 346     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 347                             p_block->i_buffer );
 348     if( psz_subtitle == NULL )
 349         return NULL;
 350
 351     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 352     {
 353         if (EnsureUTF8( psz_subtitle ) == NULL)
 354         {
 355             msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 356                      "Try manually setting a character-encoding "
 357                      "before you open the file.") );
 358         }
 359     }
 360     else
 361     {
 362
 363         if( p_sys->b_autodetect_utf8 )
 364         {
 365             if( IsUTF8( psz_subtitle ) == NULL )
 366             {
 367                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 368                          "disabling UTF-8 subtitles autodetection" );
 369                 p_sys->b_autodetect_utf8 = VLC_FALSE;
 370             }
 371         }
 372
 373         if( !p_sys->b_autodetect_utf8 )
 374         {
 375             size_t inbytes_left = strlen( psz_subtitle );
 376             size_t outbytes_left = 6 * inbytes_left;
 377             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 378             char *psz_convert_buffer_out = psz_new_subtitle;
 379             const char *psz_convert_buffer_in = psz_subtitle;
 380
 381             size_t ret = vlc_iconv( p_sys->iconv_handle,
 382                                     &psz_convert_buffer_in, &inbytes_left,
 383                                     &psz_convert_buffer_out, &outbytes_left );
 384
 385             *psz_convert_buffer_out++ = '\0';
 386             free( psz_subtitle );
 387
 388             if( ( ret == (size_t)(-1) ) || inbytes_left )
 389             {
 390                 free( psz_new_subtitle );
 391                 msg_Err( p_dec, _("failed to convert subtitle encoding.\n"
 392                         "Try manually setting a character-encoding "
 393                                 "before you open the file.") );
 394                 return NULL;
 395             }
 396
 397             psz_subtitle = realloc( psz_new_subtitle,
 398                                     psz_convert_buffer_out - psz_new_subtitle );
 399         }
 400     }
 401
 402     /* Create the subpicture unit */
 403     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 404     if( !p_spu )
 405     {
 406         msg_Warn( p_dec, "can't get spu buffer" );
 407         if( psz_subtitle ) free( psz_subtitle );
 408         return NULL;
 409     }
 410
 411     p_spu->b_pausable = VLC_TRUE;
 412
 413     /* Create a new subpicture region */
 414     memset( &fmt, 0, sizeof(video_format_t) );
 415     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 416     fmt.i_aspect = 0;
 417     fmt.i_width = fmt.i_height = 0;
 418     fmt.i_x_offset = fmt.i_y_offset = 0;
 419     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 420     if( !p_spu->p_region )
 421     {
 422         msg_Err( p_dec, "cannot allocate SPU region" );
 423         if( psz_subtitle ) free( psz_subtitle );
 424         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 425         return NULL;
 426     }
 427
 428     /* Decode and format the subpicture unit */
 429     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 430     {
 431         /* Normal text subs, easy markup */
 432         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 433         p_spu->i_x = p_sys->i_align ? 20 : 0;
 434         p_spu->i_y = 10;
 435
 436         /* Remove formatting from string */
 437
 438         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 439         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 440         {
 441             p_spu->p_region->psz_html = CreateHtmlSubtitle( psz_subtitle );
 442         }
 443
 444         p_spu->i_start = p_block->i_pts;
 445         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 446         p_spu->b_ephemer = (p_block->i_length == 0);
 447         p_spu->b_absolute = VLC_FALSE;
 448     }
 449     else
 450     {
 451         /* Decode SSA/USF strings */
 452         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 453             ParseSSAString( p_dec, psz_subtitle, p_spu );
 454
 455         p_spu->i_start = p_block->i_pts;
 456         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 457         p_spu->b_ephemer = (p_block->i_length == 0);
 458         p_spu->b_absolute = VLC_FALSE;
 459         p_spu->i_original_picture_width = p_sys->i_original_width;
 460         p_spu->i_original_picture_height = p_sys->i_original_height;
 461     }
 462     if( psz_subtitle ) free( psz_subtitle );
 463
 464     return p_spu;
 465 }
 466
 467 char* GotoNextLine( char *psz_text )
 468 {
 469     char *p_newline = psz_text;
 470
 471     while( p_newline[0] != '\0' )
 472     {
 473         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 474         {
 475             p_newline++;
 476             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 477                 p_newline++;
 478             break;
 479         }
 480         else p_newline++;
 481     }
 482     return p_newline;
 483 }
 484
 485 /* Function now handles tags with attribute values, and tries
 486  * to deal with &' commands too. It no longer modifies the string
 487  * in place, so that the original text can be reused
 488  */
 489 static char *StripTags( char *psz_subtitle )
 490 {
 491     char *psz_text_start;
 492     char *psz_text;
 493
 494     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 495     if( !psz_text_start )
 496         return NULL;
 497
 498     while( *psz_subtitle )
 499     {
 500         if( *psz_subtitle == '<' )
 501         {
 502             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 503                 *psz_text++ = '\n';
 504
 505             psz_subtitle += strcspn( psz_subtitle, ">" );
 506         }
 507         else if( *psz_subtitle == '&' )
 508         {
 509             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 510             {
 511                 *psz_text++ = '<';
 512                 psz_subtitle += strcspn( psz_subtitle, ";" );
 513             }
 514             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 515             {
 516                 *psz_text++ = '>';
 517                 psz_subtitle += strcspn( psz_subtitle, ";" );
 518             }
 519             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 520             {
 521                 *psz_text++ = '&';
 522                 psz_subtitle += strcspn( psz_subtitle, ";" );
 523             }
 524             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 525             {
 526                 *psz_text++ = '\"';
 527                 psz_subtitle += strcspn( psz_subtitle, ";" );
 528             }
 529             else
 530             {
 531                 /* Assume it is just a normal ampersand */
 532                 *psz_text++ = '&';
 533             }
 534         }
 535         else
 536         {
 537             *psz_text++ = *psz_subtitle;
 538         }
 539
 540         psz_subtitle++;
 541     }
 542     *psz_text = '\0';
 543     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 544
 545     return psz_text_start;
 546 }
 547
 548 /* Try to respect any style tags present in the subtitle string. The main
 549  * problem here is a lack of adequate specs for the subtitle formats.
 550  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 551  * SAMI has a detailed spec, but extensive rework is needed in the demux
 552  * code to prevent all this style information being excised, as it presently
 553  * does.
 554  * That leaves the others - none of which were (I guess) originally intended
 555  * to be carrying style information. Over time people have used them that way.
 556  * In the absence of specifications from which to work, the tags supported
 557  * have been restricted to the simple set permitted by the USF DTD, ie. :
 558  *  Basic: <br>, <i>, <b>, <u>
 559  *  Extended: <font>
 560  *    Attributes: face
 561  *                family
 562  *                size
 563  *                color
 564  *                outline-color
 565  *                shadow-color
 566  *                outline-level
 567  *                shadow-level
 568  *                back-color
 569  *                alpha
 570  * There is also the further restriction that the subtitle be well-formed
 571  * as an XML entity, ie. the HTML sentence:
 572  *        <b><i>Bold and Italics</b></i>
 573  * doesn't qualify because the tags aren't nested one inside the other.
 574  * <text> tags are automatically added to the output to ensure
 575  * well-formedness.
 576  * If the text doesn't qualify for any reason, a NULL string is
 577  * returned, and the rendering engine will fall back to the
 578  * plain text version of the subtitle.
 579  */
 580 static char *CreateHtmlSubtitle( char *psz_subtitle )
 581 {
 582     char    psz_tagStack[ 100 ];
 583     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 584     char   *psz_html_start = malloc( i_buf_size );
 585
 586     psz_tagStack[ 0 ] = '\0';
 587
 588     if( psz_html_start != NULL )
 589     {
 590         char *psz_html = psz_html_start;
 591
 592         strcpy( psz_html, "<text>" );
 593         psz_html += 6;
 594
 595         while( *psz_subtitle )
 596         {
 597             if( *psz_subtitle == '\n' )
 598             {
 599                 strcpy( psz_html, "<br/>" );
 600                 psz_html += 5;
 601                 psz_subtitle++;
 602             }
 603             else if( *psz_subtitle == '<' )
 604             {
 605                 if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 606                 {
 607                     strcpy( psz_html, "<br/>" );
 608                     psz_html += 5;
 609                     psz_subtitle += 5;
 610                 }
 611                 else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 612                 {
 613                     strcpy( psz_html, "<b>" );
 614                     strcat( psz_tagStack, "b" );
 615                     psz_html += 3;
 616                     psz_subtitle += 3;
 617                 }
 618                 else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 619                 {
 620                     strcpy( psz_html, "<i>" );
 621                     strcat( psz_tagStack, "i" );
 622                     psz_html += 3;
 623                     psz_subtitle += 3;
 624                 }
 625                 else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 626                 {
 627                     strcpy( psz_html, "<u>" );
 628                     strcat( psz_tagStack, "u" );
 629                     psz_html += 3;
 630                     psz_subtitle += 3;
 631                 }
 632                 else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 633                 {
 634                     const char *psz_attribs[] = { "face=\"", "family=\"", "size=\"",
 635                             "color=\"", "outline-color=\"", "shadow-color=\"",
 636                             "outline-level=\"", "shadow-level=\"", "back-color=\"",
 637                             "alpha=\"", NULL };
 638
 639                     strcpy( psz_html, "<font " );
 640                     strcat( psz_tagStack, "f" );
 641                     psz_html += 6;
 642                     psz_subtitle += 6;
 643
 644                     while( *psz_subtitle != '>' )
 645                     {
 646                         int  k;
 647
 648                         for( k=0; psz_attribs[ k ]; k++ )
 649                         {
 650                             int i_len = strlen( psz_attribs[ k ] );
 651
 652                             if( !strncasecmp( psz_subtitle, psz_attribs[ k ], i_len ))
 653                             {
 654                                 i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 655
 656                                 strncpy( psz_html, psz_subtitle, i_len );
 657                                 psz_html += i_len;
 658                                 psz_subtitle += i_len;
 659                                 break;
 660                             }
 661                         }
 662                         if( psz_attribs[ k ] == NULL )
 663                         {
 664                             /* Jump over unrecognised tag */
 665                             int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 666
 667                             i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 668                             psz_subtitle += i_len;
 669                         }
 670                         while (*psz_subtitle == ' ')
 671                             *psz_html++ = *psz_subtitle++;
 672                     }
 673                     *psz_html++ = *psz_subtitle++;
 674                 }
 675                 else if( !strncmp( psz_subtitle, "</", 2 ))
 676                 {
 677                     vlc_bool_t  b_match     = VLC_FALSE;
 678                     int         i_len       = strlen( psz_tagStack ) - 1;
 679                     char       *psz_lastTag = NULL;
 680
 681                     if( i_len >= 0 )
 682                     {
 683                         psz_lastTag = psz_tagStack + i_len;
 684                         i_len = 0;
 685
 686                         switch( *psz_lastTag )
 687                         {
 688                             case 'b':
 689                                 b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 690                                 i_len   = 4;
 691                                 break;
 692                             case 'i':
 693                                 b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 694                                 i_len   = 4;
 695                                 break;
 696                             case 'u':
 697                                 b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 698                                 i_len   = 4;
 699                                 break;
 700                             case 'f':
 701                                 b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 702                                 i_len   = 7;
 703                                 break;
 704                         }
 705                     }
 706                     if( ! b_match )
 707                     {
 708                         /* Not well formed -- kill everything */
 709                         free( psz_html_start );
 710                         psz_html_start = NULL;
 711                         break;
 712                     }
 713                     *psz_lastTag = '\0';
 714                     strncpy( psz_html, psz_subtitle, i_len );
 715                     psz_html += i_len;
 716                     psz_subtitle += i_len;
 717                 }
 718                 else
 719                 {
 720                     psz_subtitle += strcspn( psz_subtitle, ">" );
 721                 }
 722             }
 723             else if( *psz_subtitle == '&' )
 724             {
 725                 if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 726                 {
 727                     strcpy( psz_html, "&lt;" );
 728                     psz_html += 4;
 729                     psz_subtitle += 4;
 730                 }
 731                 else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 732                 {
 733                     strcpy( psz_html, "&gt;" );
 734                     psz_html += 4;
 735                     psz_subtitle += 4;
 736                 }
 737                 else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 738                 {
 739                     strcpy( psz_html, "&amp;" );
 740                     psz_html += 5;
 741                     psz_subtitle += 5;
 742                 }
 743                 else
 744                 {
 745                     strcpy( psz_html, "&amp;" );
 746                     psz_html += 5;
 747                     psz_subtitle++;
 748                 }
 749             }
 750             else
 751             {
 752                 *psz_html = *psz_subtitle;
 753                 if( psz_html > psz_html_start )
 754                 {
 755                     /* Check for double whitespace */
 756                     if((( *psz_html == ' ' ) ||
 757                         ( *psz_html == '\t' )) &&
 758                        (( *(psz_html-1) == ' ' ) ||
 759                         ( *(psz_html-1) == '\t' )))
 760                     {
 761                         strcpy( psz_html, NO_BREAKING_SPACE );
 762                         psz_html += strlen( NO_BREAKING_SPACE ) - 1;
 763                     }
 764                 }
 765                 psz_html++;
 766                 psz_subtitle++;
 767             }
 768
 769             if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 10 )
 770             {
 771                 int i_len = psz_html - psz_html_start;
 772
 773                 i_buf_size += 100;
 774                 psz_html_start = realloc( psz_html_start, i_buf_size );
 775                 psz_html = psz_html_start + i_len;
 776                 *psz_html = '\0';
 777             }
 778         }
 779         strcpy( psz_html, "</text>" );
 780         psz_html += 7;
 781
 782         if( psz_tagStack[ 0 ] != '\0' )
 783         {
 784             /* Not well formed -- kill everything */
 785             free( psz_html_start );
 786             psz_html_start = NULL;
 787         }
 788         else if( psz_html_start )
 789         {
 790             /* Shrink the memory requirements */
 791             psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 792         }
 793     }
 794     return psz_html_start;
 795 }