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 #include <vlc_plugin.h>
  36
  37 /*****************************************************************************
  38  * Local prototypes
  39  *****************************************************************************/
  40 static int  OpenDecoder   ( vlc_object_t * );
  41 static void CloseDecoder  ( vlc_object_t * );
  42
  43 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
  44 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
  45 static char           *StripTags      ( char * );
  46 static char           *CreateHtmlSubtitle( int *pi_align, char * );
  47
  48
  49 /*****************************************************************************
  50  * Module descriptor.
  51  *****************************************************************************/
  52 static const char *const ppsz_encodings[] = {
  53     DEFAULT_NAME, "ASCII", "UTF-8", "",
  54     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  55     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  56     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  57     "ISO-8859-6", "CP1256", "MacArabic", "",
  58     "ISO-8859-7", "CP1253", "MacGreek", "",
  59     "ISO-8859-8", "CP1255", "MacHebrew", "",
  60     "ISO-8859-9", "CP1254", "MacTurkish", "",
  61     "ISO-8859-13", "CP1257", "",
  62     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  63     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  64     "ISO-2022-KR", "EUC-KR", "",
  65     "MacThai", "KOI8-T", "",
  66     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  67     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  68     "Macintosh", "",
  69     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  70     "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  71     "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  72     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  73     "HPROMAN8", "NEXTSTEP" };
  74 /*
  75 SSA supports charset selection.
  76 The following known charsets are used:
  77
  78 0 = Ansi - Western European
  79 1 = default
  80 2 = symbol
  81 3 = invalid
  82 77 = Mac
  83 128 = Japanese (Shift JIS)
  84 129 = Hangul
  85 130 = Johab
  86 134 = GB2312 Simplified Chinese
  87 136 = Big5 Traditional Chinese
  88 161 = Greek
  89 162 = Turkish
  90 163 = Vietnamese
  91 177 = Hebrew
  92 178 = Arabic
  93 186 = Baltic
  94 204 = Russian (Cyrillic)
  95 222 = Thai
  96 238 = Eastern European
  97 254 = PC 437
  98 */
  99
 100 static const int  pi_justification[] = { 0, 1, 2 };
 101 static const char *const ppsz_justification_text[] = {
 102     N_("Center"),N_("Left"),N_("Right")};
 103
 104 #define ENCODING_TEXT N_("Subtitles text encoding")
 105 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 106 #define ALIGN_TEXT N_("Subtitles justification")
 107 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 108 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 109 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 110             "UTF-8 encoding within subtitles files.")
 111 #define FORMAT_TEXT N_("Formatted Subtitles")
 112 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 113  "VLC partly implements this, but you can choose to disable all formatting.")
 114
 115
 116 vlc_module_begin ()
 117     set_shortname( N_("Subtitles"))
 118     set_description( N_("Text subtitles decoder") )
 119     set_capability( "decoder", 50 )
 120     set_callbacks( OpenDecoder, CloseDecoder )
 121     set_category( CAT_INPUT )
 122     set_subcategory( SUBCAT_INPUT_SCODEC )
 123
 124     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 125                  false )
 126         change_integer_list( pi_justification, ppsz_justification_text, NULL )
 127     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 128                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 129         change_string_list( ppsz_encodings, 0, 0 )
 130     add_bool( "subsdec-autodetect-utf8", true, NULL,
 131               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 132     add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
 133                  false )
 134 vlc_module_end ()
 135
 136 /*****************************************************************************
 137  * OpenDecoder: probe the decoder and return score
 138  *****************************************************************************
 139  * Tries to launch a decoder and return score so that the interface is able
 140  * to chose.
 141  *****************************************************************************/
 142 static int OpenDecoder( vlc_object_t *p_this )
 143 {
 144     decoder_t     *p_dec = (decoder_t*)p_this;
 145     decoder_sys_t *p_sys;
 146     vlc_value_t    val;
 147
 148     switch( p_dec->fmt_in.i_codec )
 149     {
 150         case VLC_FOURCC('s','u','b','t'):
 151         case VLC_FOURCC('s','s','a',' '):
 152         case VLC_FOURCC('t','1','4','0'):
 153             break;
 154         default:
 155             return VLC_EGENERIC;
 156     }
 157
 158     p_dec->pf_decode_sub = DecodeBlock;
 159     p_dec->fmt_out.i_cat = SPU_ES;
 160     p_dec->fmt_out.i_codec = 0;
 161
 162     /* Allocate the memory needed to store the decoder's structure */
 163     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 164     if( p_sys == NULL )
 165         return VLC_ENOMEM;
 166
 167     /* init of p_sys */
 168     p_sys->i_align = 0;
 169     p_sys->iconv_handle = (vlc_iconv_t)-1;
 170     p_sys->b_autodetect_utf8 = false;
 171     p_sys->b_ass = false;
 172     p_sys->i_original_height = -1;
 173     p_sys->i_original_width = -1;
 174     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 175     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 176
 177     char *psz_charset = NULL;
 178
 179     /* First try demux-specified encoding */
 180     if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
 181         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 182     else
 183     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 184     {
 185         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 186         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 187                  p_dec->fmt_in.subs.psz_encoding ?: "not specified");
 188     }
 189
 190     /* Second, try configured encoding */
 191     if (psz_charset == NULL)
 192     {
 193         psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
 194         if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
 195         {
 196             free (psz_charset);
 197             psz_charset = NULL;
 198         }
 199
 200         msg_Dbg (p_dec, "trying configured character encoding: %s",
 201                  psz_charset ?: "not specified");
 202     }
 203
 204     /* Third, try "local" encoding with optional UTF-8 autodetection */
 205     if (psz_charset == NULL)
 206     {
 207         psz_charset = strdup (GetFallbackEncoding ());
 208         msg_Dbg (p_dec, "trying default character encoding: %s",
 209                  psz_charset ?: "not specified");
 210
 211         if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
 212         {
 213             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 214             p_sys->b_autodetect_utf8 = true;
 215         }
 216     }
 217
 218     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 219     if (psz_charset == NULL)
 220     {
 221         psz_charset = strdup ("UTF-8");
 222         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 223     }
 224
 225     if ((psz_charset != NULL)
 226      && strcasecmp (psz_charset, "UTF-8")
 227      && strcasecmp (psz_charset, "utf8"))
 228     {
 229         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 230         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 231             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 232     }
 233     free (psz_charset);
 234
 235     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 236     var_Get( p_dec, "subsdec-align", &val );
 237     p_sys->i_align = val.i_int;
 238
 239     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 240      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 241     {
 242         if( p_dec->fmt_in.i_extra > 0 )
 243             ParseSSAHeader( p_dec );
 244     }
 245
 246     return VLC_SUCCESS;
 247 }
 248
 249 /****************************************************************************
 250  * DecodeBlock: the whole thing
 251  ****************************************************************************
 252  * This function must be fed with complete subtitles units.
 253  ****************************************************************************/
 254 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 255 {
 256     subpicture_t *p_spu;
 257     block_t *p_block;
 258
 259     if( !pp_block || *pp_block == NULL )
 260         return NULL;
 261
 262     p_block = *pp_block;
 263     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 264     {
 265         block_Release( p_block );
 266         return NULL;
 267     }
 268
 269     p_spu = ParseText( p_dec, p_block );
 270
 271     block_Release( p_block );
 272     *pp_block = NULL;
 273
 274     return p_spu;
 275 }
 276
 277 /*****************************************************************************
 278  * CloseDecoder: clean up the decoder
 279  *****************************************************************************/
 280 static void CloseDecoder( vlc_object_t *p_this )
 281 {
 282     decoder_t *p_dec = (decoder_t *)p_this;
 283     decoder_sys_t *p_sys = p_dec->p_sys;
 284
 285     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 286         vlc_iconv_close( p_sys->iconv_handle );
 287
 288     if( p_sys->pp_ssa_styles )
 289     {
 290         int i;
 291         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 292         {
 293             if( !p_sys->pp_ssa_styles[i] )
 294                 continue;
 295
 296             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 297             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 298             free( p_sys->pp_ssa_styles[i] );
 299         }
 300         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 301     }
 302     if( p_sys->pp_images )
 303     {
 304         int i;
 305         for( i = 0; i < p_sys->i_images; i++ )
 306         {
 307             if( !p_sys->pp_images[i] )
 308                 continue;
 309
 310             if( p_sys->pp_images[i]->p_pic )
 311                 picture_Release( p_sys->pp_images[i]->p_pic );
 312             free( p_sys->pp_images[i]->psz_filename );
 313
 314             free( p_sys->pp_images[i] );
 315         }
 316         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 317     }
 318
 319     free( p_sys );
 320 }
 321
 322 /*****************************************************************************
 323  * ParseText: parse an text subtitle packet and send it to the video output
 324  *****************************************************************************/
 325 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 326 {
 327     decoder_sys_t *p_sys = p_dec->p_sys;
 328     subpicture_t *p_spu = NULL;
 329     char *psz_subtitle = NULL;
 330     video_format_t fmt;
 331
 332     /* We cannot display a subpicture with no date */
 333     if( p_block->i_pts == 0 )
 334     {
 335         msg_Warn( p_dec, "subtitle without a date" );
 336         return NULL;
 337     }
 338
 339     /* Check validity of packet data */
 340     /* An "empty" line containing only \0 can be used to force
 341        and ephemer picture from the screen */
 342     if( p_block->i_buffer < 1 )
 343     {
 344         msg_Warn( p_dec, "no subtitle data" );
 345         return NULL;
 346     }
 347
 348     /* Should be resiliant against bad subtitles */
 349     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 350                             p_block->i_buffer );
 351     if( psz_subtitle == NULL )
 352         return NULL;
 353
 354     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 355     {
 356         if (EnsureUTF8( psz_subtitle ) == NULL)
 357         {
 358             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 359                      "Try manually setting a character-encoding "
 360                      "before you open the file." );
 361         }
 362     }
 363     else
 364     {
 365
 366         if( p_sys->b_autodetect_utf8 )
 367         {
 368             if( IsUTF8( psz_subtitle ) == NULL )
 369             {
 370                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 371                          "disabling UTF-8 subtitles autodetection" );
 372                 p_sys->b_autodetect_utf8 = false;
 373             }
 374         }
 375
 376         if( !p_sys->b_autodetect_utf8 )
 377         {
 378             size_t inbytes_left = strlen( psz_subtitle );
 379             size_t outbytes_left = 6 * inbytes_left;
 380             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 381             char *psz_convert_buffer_out = psz_new_subtitle;
 382             const char *psz_convert_buffer_in = psz_subtitle;
 383
 384             size_t ret = vlc_iconv( p_sys->iconv_handle,
 385                                     &psz_convert_buffer_in, &inbytes_left,
 386                                     &psz_convert_buffer_out, &outbytes_left );
 387
 388             *psz_convert_buffer_out++ = '\0';
 389             free( psz_subtitle );
 390
 391             if( ( ret == (size_t)(-1) ) || inbytes_left )
 392             {
 393                 free( psz_new_subtitle );
 394                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 395                         "Try manually setting a character-encoding "
 396                                 "before you open the file." );
 397                 return NULL;
 398             }
 399
 400             psz_subtitle = realloc( psz_new_subtitle,
 401                                     psz_convert_buffer_out - psz_new_subtitle );
 402         }
 403     }
 404
 405     /* Create the subpicture unit */
 406     p_spu = decoder_NewSubpicture( p_dec );
 407     if( !p_spu )
 408     {
 409         msg_Warn( p_dec, "can't get spu buffer" );
 410         free( psz_subtitle );
 411         return NULL;
 412     }
 413
 414     /* Create a new subpicture region */
 415     memset( &fmt, 0, sizeof(video_format_t) );
 416     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 417     fmt.i_aspect = 0;
 418     fmt.i_width = fmt.i_height = 0;
 419     fmt.i_x_offset = fmt.i_y_offset = 0;
 420     p_spu->p_region = subpicture_region_New( &fmt );
 421     if( !p_spu->p_region )
 422     {
 423         msg_Err( p_dec, "cannot allocate SPU region" );
 424         free( psz_subtitle );
 425         decoder_DeleteSubpicture( p_dec, p_spu );
 426         return NULL;
 427     }
 428
 429     /* Decode and format the subpicture unit */
 430     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 431     {
 432         /* Normal text subs, easy markup */
 433         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 434         p_spu->p_region->i_x = p_sys->i_align ? 20 : 0;
 435         p_spu->p_region->i_y = 10;
 436
 437         /* Remove formatting from string */
 438
 439         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 440         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 441         {
 442             p_spu->p_region->psz_html = CreateHtmlSubtitle( &p_spu->p_region->i_align, psz_subtitle );
 443         }
 444
 445         p_spu->i_start = p_block->i_pts;
 446         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 447         p_spu->b_ephemer = (p_block->i_length == 0);
 448         p_spu->b_absolute = false;
 449     }
 450     else
 451     {
 452         /* Decode SSA/USF strings */
 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 = 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     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 void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 581 {
 582     const int i_len = strlen(psz_text);
 583
 584     strncpy( *ppsz_html, psz_text, i_max );
 585     *ppsz_html += __MIN(i_max,i_len);
 586 }
 587
 588 static void HtmlPut( char **ppsz_html, const char *psz_text )
 589 {
 590     strcpy( *ppsz_html, psz_text );
 591     *ppsz_html += strlen(psz_text);
 592 }
 593 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 594 {
 595     HtmlPut( ppsz_html, psz_text );
 596     *ppsz_subtitle += strlen(psz_text);
 597 }
 598
 599 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 600 {
 601     /* */
 602     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 603     if( !psz_tag )
 604         return NULL;
 605     psz_tag[ 0 ] = '\0';
 606
 607     /* */
 608     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 609     char   *psz_html_start = malloc( i_buf_size );
 610     char   *psz_html = psz_html_start;
 611     if( psz_html_start == NULL )
 612     {
 613         free( psz_tag );
 614         return NULL;
 615     }
 616     psz_html[0] = '\0';
 617
 618     bool b_has_align = false;
 619
 620     HtmlPut( &psz_html, "<text>" );
 621
 622     /* */
 623     while( *psz_subtitle )
 624     {
 625         if( *psz_subtitle == '\n' )
 626         {
 627             HtmlPut( &psz_html, "<br/>" );
 628             psz_subtitle++;
 629         }
 630         else if( *psz_subtitle == '<' )
 631         {
 632             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 633             {
 634                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 635             }
 636             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 637             {
 638                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 639                 strcat( psz_tag, "b" );
 640             }
 641             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 642             {
 643                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 644                 strcat( psz_tag, "i" );
 645             }
 646             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 647             {
 648                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 649                 strcat( psz_tag, "u" );
 650             }
 651             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 652             {
 653                 const char *psz_attribs[] = { "face=", "family=", "size=",
 654                         "color=", "outline-color=", "shadow-color=",
 655                         "outline-level=", "shadow-level=", "back-color=",
 656                         "alpha=", NULL };
 657
 658                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 659                 strcat( psz_tag, "f" );
 660
 661                 while( *psz_subtitle != '>' )
 662                 {
 663                     int  k;
 664
 665                     for( k=0; psz_attribs[ k ]; k++ )
 666                     {
 667                         int i_len = strlen( psz_attribs[ k ] );
 668
 669                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 670                         {
 671                             /* */
 672                             HtmlPut( &psz_html, psz_attribs[k] );
 673                             psz_subtitle += i_len;
 674
 675                             /* */
 676                             if( *psz_subtitle == '"' )
 677                             {
 678                                 psz_subtitle++;
 679                                 i_len = strcspn( psz_subtitle, "\"" );
 680                             }
 681                             else
 682                             {
 683                                 i_len = strcspn( psz_subtitle, " \t>" );
 684                             }
 685                             HtmlPut( &psz_html, "\"" );
 686                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 687                                 HtmlPut( &psz_html, "#" );
 688                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 689                             HtmlPut( &psz_html, "\"" );
 690
 691                             psz_subtitle += i_len;
 692                             if( *psz_subtitle == '\"' )
 693                                 psz_subtitle++;
 694                             break;
 695                         }
 696                     }
 697                     if( psz_attribs[ k ] == NULL )
 698                     {
 699                         /* Jump over unrecognised tag */
 700                         int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 701
 702                         i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 703                         psz_subtitle += i_len;
 704                     }
 705                     while (*psz_subtitle == ' ')
 706                         *psz_html++ = *psz_subtitle++;
 707                 }
 708                 *psz_html++ = *psz_subtitle++;
 709             }
 710             else if( !strncmp( psz_subtitle, "</", 2 ))
 711             {
 712                 bool   b_match     = false;
 713                 bool   b_ignore    = false;
 714                 int    i_len       = strlen( psz_tag ) - 1;
 715                 char  *psz_lastTag = NULL;
 716
 717                 if( i_len >= 0 )
 718                 {
 719                     psz_lastTag = psz_tag + i_len;
 720                     i_len = 0;
 721
 722                     switch( *psz_lastTag )
 723                     {
 724                     case 'b':
 725                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 726                         i_len   = 4;
 727                         break;
 728                     case 'i':
 729                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 730                         i_len   = 4;
 731                         break;
 732                     case 'u':
 733                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 734                         i_len   = 4;
 735                         break;
 736                     case 'f':
 737                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 738                         i_len   = 7;
 739                         break;
 740                     case 'I':
 741                         i_len = strcspn( psz_subtitle, ">" );
 742                         b_match = psz_subtitle[i_len] == '>';
 743                         b_ignore = true;
 744                         if( b_match )
 745                             i_len++;
 746                         break;
 747                     }
 748                 }
 749                 if( !b_match )
 750                 {
 751                     /* Not well formed -- kill everything */
 752                     free( psz_html_start );
 753                     psz_html_start = NULL;
 754                     break;
 755                 }
 756                 *psz_lastTag = '\0';
 757                 if( !b_ignore )
 758                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 759
 760                 psz_subtitle += i_len;
 761             }
 762             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 763                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 764             {
 765                 /* We have a single < */
 766                 HtmlPut( &psz_html, "&lt;" );
 767                 psz_subtitle++;
 768             }
 769             else
 770             {
 771                 /* We have an unknown tag or a single < */
 772
 773                 /* Search for the next tag or end of tag or end of string */
 774                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 775                 char *psz_closing = strstr( psz_subtitle, "/>" );
 776
 777                 if( psz_closing && psz_closing < psz_stop )
 778                 {
 779                     /* We have a self closed tag, remove it */
 780                     psz_subtitle = &psz_closing[2];
 781                 }
 782                 else if( *psz_stop == '>' )
 783                 {
 784                     char psz_match[256];
 785
 786                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 787                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 788
 789                     if( strstr( psz_subtitle, psz_match ) )
 790                     {
 791                         /* We have the closing tag, ignore it TODO */
 792                         psz_subtitle = &psz_stop[1];
 793                         strcat( psz_tag, "I" );
 794                     }
 795                     else
 796                     {
 797                         int i_len = psz_stop + 1 - psz_subtitle;
 798
 799                         /* Copy the whole data */
 800                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 801                         {
 802                             if( *psz_subtitle == '<' )
 803                                 HtmlPut( &psz_html, "&lt;" );
 804                             else if( *psz_subtitle == '>' )
 805                                 HtmlPut( &psz_html, "&gt;" );
 806                             else
 807                                 *psz_html++ = *psz_subtitle;
 808                         }
 809                     }
 810                 }
 811                 else
 812                 {
 813                     /* We have a single < */
 814                     HtmlPut( &psz_html, "&lt;" );
 815                     psz_subtitle++;
 816                 }
 817             }
 818         }
 819         else if( *psz_subtitle == '&' )
 820         {
 821             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 822             {
 823                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 824             }
 825             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 826             {
 827                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 828             }
 829             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 830             {
 831                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 832             }
 833             else
 834             {
 835                 HtmlPut( &psz_html, "&amp;" );
 836                 psz_subtitle++;
 837             }
 838         }
 839         else if( *psz_subtitle == '>' )
 840         {
 841             HtmlPut( &psz_html, "&gt;" );
 842             psz_subtitle++;
 843         }
 844         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 845                  strchr( psz_subtitle, '}' ) )
 846         {
 847             /* Check for forced alignment */
 848             if( !b_has_align &&
 849                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 850             {
 851                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 852                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 853                 const int i_id = psz_subtitle[4] - '1';
 854
 855                 b_has_align = true;
 856                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 857             }
 858             /* TODO fr -> rotation */
 859
 860             /* Hide {\stupidity} */
 861             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 862         }
 863         else
 864         {
 865             *psz_html = *psz_subtitle;
 866             if( psz_html > psz_html_start )
 867             {
 868                 /* Check for double whitespace */
 869                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 870                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 871                 {
 872                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 873                     psz_html--;
 874                 }
 875             }
 876             psz_html++;
 877             psz_subtitle++;
 878         }
 879
 880         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 881         {
 882             const int i_len = psz_html - psz_html_start;
 883
 884             i_buf_size += 200;
 885             char *psz_new = realloc( psz_html_start, i_buf_size );
 886             if( !psz_new )
 887                 break;
 888             psz_html_start = psz_new;
 889             psz_html = &psz_new[i_len];
 890         }
 891     }
 892     if( psz_html_start )
 893     {
 894         static const char *psz_text_close = "</text>";
 895         static const char *psz_tag_long = "/font>";
 896
 897         /* Realloc for closing tags and shrink memory */
 898         const size_t i_length = (size_t)( psz_html - psz_html_start );
 899
 900         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 901         char *psz_new = realloc( psz_html_start, i_size );
 902         if( psz_new )
 903         {
 904             psz_html_start = psz_new;
 905             psz_html = &psz_new[i_length];
 906
 907             /* Close not well formed subtitle */
 908             while( *psz_tag )
 909             {
 910                 /* */
 911                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 912                 switch( *psz_last )
 913                 {
 914                 case 'b':
 915                     HtmlPut( &psz_html, "</b>" );
 916                     break;
 917                 case 'i':
 918                     HtmlPut( &psz_html, "</i>" );
 919                     break;
 920                 case 'u':
 921                     HtmlPut( &psz_html, "</u>" );
 922                     break;
 923                 case 'f':
 924                     HtmlPut( &psz_html, "/font>" );
 925                     break;
 926                 case 'I':
 927                     break;
 928                 }
 929
 930                 *psz_last = '\0';
 931             }
 932             HtmlPut( &psz_html, psz_text_close );
 933         }
 934     }
 935     free( psz_tag );
 936
 937     return psz_html_start;
 938 }
 939