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, "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", "VISCII", "TCVN",
  73     "HPROMAN8",
  74 };
  75 /*
  76 SSA supports charset selection.
  77 The following known charsets are used:
  78
  79 0 = Ansi - Western European
  80 1 = default
  81 2 = symbol
  82 3 = invalid
  83 77 = Mac
  84 128 = Japanese (Shift JIS)
  85 129 = Hangul
  86 130 = Johab
  87 134 = GB2312 Simplified Chinese
  88 136 = Big5 Traditional Chinese
  89 161 = Greek
  90 162 = Turkish
  91 163 = Vietnamese
  92 177 = Hebrew
  93 178 = Arabic
  94 186 = Baltic
  95 204 = Russian (Cyrillic)
  96 222 = Thai
  97 238 = Eastern European
  98 254 = PC 437
  99 */
 100
 101 static const int  pi_justification[] = { 0, 1, 2 };
 102 static const char *const ppsz_justification_text[] = {
 103     N_("Center"),N_("Left"),N_("Right")};
 104
 105 #define ENCODING_TEXT N_("Subtitles text encoding")
 106 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 107 #define ALIGN_TEXT N_("Subtitles justification")
 108 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 109 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 110 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 111             "UTF-8 encoding within subtitles files.")
 112 #define FORMAT_TEXT N_("Formatted Subtitles")
 113 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 114  "VLC partly implements this, but you can choose to disable all formatting.")
 115
 116
 117 vlc_module_begin ()
 118     set_shortname( N_("Subtitles"))
 119     set_description( N_("Text subtitles decoder") )
 120     set_capability( "decoder", 50 )
 121     set_callbacks( OpenDecoder, CloseDecoder )
 122     set_category( CAT_INPUT )
 123     set_subcategory( SUBCAT_INPUT_SCODEC )
 124
 125     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 126                  false )
 127         change_integer_list( pi_justification, ppsz_justification_text, NULL )
 128     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 129                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 130         change_string_list( ppsz_encodings, 0, 0 )
 131     add_bool( "subsdec-autodetect-utf8", true, NULL,
 132               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 133     add_bool( "subsdec-formatted", true, NULL, FORMAT_TEXT, FORMAT_LONGTEXT,
 134                  false )
 135 vlc_module_end ()
 136
 137 /*****************************************************************************
 138  * OpenDecoder: probe the decoder and return score
 139  *****************************************************************************
 140  * Tries to launch a decoder and return score so that the interface is able
 141  * to chose.
 142  *****************************************************************************/
 143 static int OpenDecoder( vlc_object_t *p_this )
 144 {
 145     decoder_t     *p_dec = (decoder_t*)p_this;
 146     decoder_sys_t *p_sys;
 147     vlc_value_t    val;
 148
 149     switch( p_dec->fmt_in.i_codec )
 150     {
 151         case VLC_FOURCC('s','u','b','t'):
 152         case VLC_FOURCC('s','s','a',' '):
 153         case VLC_FOURCC('t','1','4','0'):
 154             break;
 155         default:
 156             return VLC_EGENERIC;
 157     }
 158
 159     p_dec->pf_decode_sub = DecodeBlock;
 160     p_dec->fmt_out.i_cat = SPU_ES;
 161     p_dec->fmt_out.i_codec = 0;
 162
 163     /* Allocate the memory needed to store the decoder's structure */
 164     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 165     if( p_sys == NULL )
 166         return VLC_ENOMEM;
 167
 168     /* init of p_sys */
 169     p_sys->i_align = 0;
 170     p_sys->iconv_handle = (vlc_iconv_t)-1;
 171     p_sys->b_autodetect_utf8 = false;
 172     p_sys->b_ass = false;
 173     p_sys->i_original_height = -1;
 174     p_sys->i_original_width = -1;
 175     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 176     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 177
 178     char *psz_charset = NULL;
 179
 180     /* First try demux-specified encoding */
 181     if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
 182         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 183     else
 184     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 185     {
 186         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 187         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 188                  p_dec->fmt_in.subs.psz_encoding ?: "not specified");
 189     }
 190
 191     /* Second, try configured encoding */
 192     if (psz_charset == NULL)
 193     {
 194         psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
 195         if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
 196         {
 197             free (psz_charset);
 198             psz_charset = NULL;
 199         }
 200
 201         msg_Dbg (p_dec, "trying configured character encoding: %s",
 202                  psz_charset ?: "not specified");
 203     }
 204
 205     /* Third, try "local" encoding with optional UTF-8 autodetection */
 206     if (psz_charset == NULL)
 207     {
 208         psz_charset = strdup (GetFallbackEncoding ());
 209         msg_Dbg (p_dec, "trying default character encoding: %s",
 210                  psz_charset ?: "not specified");
 211
 212         if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
 213         {
 214             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 215             p_sys->b_autodetect_utf8 = true;
 216         }
 217     }
 218
 219     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 220     if (psz_charset == NULL)
 221     {
 222         psz_charset = strdup ("UTF-8");
 223         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 224     }
 225
 226     if ((psz_charset != NULL)
 227      && strcasecmp (psz_charset, "UTF-8")
 228      && strcasecmp (psz_charset, "utf8"))
 229     {
 230         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 231         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 232             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 233     }
 234     free (psz_charset);
 235
 236     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 237     var_Get( p_dec, "subsdec-align", &val );
 238     p_sys->i_align = val.i_int;
 239
 240     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 241      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 242     {
 243         if( p_dec->fmt_in.i_extra > 0 )
 244             ParseSSAHeader( p_dec );
 245     }
 246
 247     return VLC_SUCCESS;
 248 }
 249
 250 /****************************************************************************
 251  * DecodeBlock: the whole thing
 252  ****************************************************************************
 253  * This function must be fed with complete subtitles units.
 254  ****************************************************************************/
 255 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 256 {
 257     subpicture_t *p_spu;
 258     block_t *p_block;
 259
 260     if( !pp_block || *pp_block == NULL )
 261         return NULL;
 262
 263     p_block = *pp_block;
 264     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 265     {
 266         block_Release( p_block );
 267         return NULL;
 268     }
 269
 270     p_spu = ParseText( p_dec, p_block );
 271
 272     block_Release( p_block );
 273     *pp_block = NULL;
 274
 275     return p_spu;
 276 }
 277
 278 /*****************************************************************************
 279  * CloseDecoder: clean up the decoder
 280  *****************************************************************************/
 281 static void CloseDecoder( vlc_object_t *p_this )
 282 {
 283     decoder_t *p_dec = (decoder_t *)p_this;
 284     decoder_sys_t *p_sys = p_dec->p_sys;
 285
 286     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 287         vlc_iconv_close( p_sys->iconv_handle );
 288
 289     if( p_sys->pp_ssa_styles )
 290     {
 291         int i;
 292         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 293         {
 294             if( !p_sys->pp_ssa_styles[i] )
 295                 continue;
 296
 297             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 298             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 299             free( p_sys->pp_ssa_styles[i] );
 300         }
 301         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 302     }
 303     if( p_sys->pp_images )
 304     {
 305         int i;
 306         for( i = 0; i < p_sys->i_images; i++ )
 307         {
 308             if( !p_sys->pp_images[i] )
 309                 continue;
 310
 311             if( p_sys->pp_images[i]->p_pic )
 312                 picture_Release( p_sys->pp_images[i]->p_pic );
 313             free( p_sys->pp_images[i]->psz_filename );
 314
 315             free( p_sys->pp_images[i] );
 316         }
 317         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 318     }
 319
 320     free( p_sys );
 321 }
 322
 323 /*****************************************************************************
 324  * ParseText: parse an text subtitle packet and send it to the video output
 325  *****************************************************************************/
 326 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 327 {
 328     decoder_sys_t *p_sys = p_dec->p_sys;
 329     subpicture_t *p_spu = NULL;
 330     char *psz_subtitle = NULL;
 331     video_format_t fmt;
 332
 333     /* We cannot display a subpicture with no date */
 334     if( p_block->i_pts == 0 )
 335     {
 336         msg_Warn( p_dec, "subtitle without a date" );
 337         return NULL;
 338     }
 339
 340     /* Check validity of packet data */
 341     /* An "empty" line containing only \0 can be used to force
 342        and ephemer picture from the screen */
 343     if( p_block->i_buffer < 1 )
 344     {
 345         msg_Warn( p_dec, "no subtitle data" );
 346         return NULL;
 347     }
 348
 349     /* Should be resiliant against bad subtitles */
 350     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 351                             p_block->i_buffer );
 352     if( psz_subtitle == NULL )
 353         return NULL;
 354
 355     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 356     {
 357         if (EnsureUTF8( psz_subtitle ) == NULL)
 358         {
 359             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 360                      "Try manually setting a character-encoding "
 361                      "before you open the file." );
 362         }
 363     }
 364     else
 365     {
 366
 367         if( p_sys->b_autodetect_utf8 )
 368         {
 369             if( IsUTF8( psz_subtitle ) == NULL )
 370             {
 371                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 372                          "disabling UTF-8 subtitles autodetection" );
 373                 p_sys->b_autodetect_utf8 = false;
 374             }
 375         }
 376
 377         if( !p_sys->b_autodetect_utf8 )
 378         {
 379             size_t inbytes_left = strlen( psz_subtitle );
 380             size_t outbytes_left = 6 * inbytes_left;
 381             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 382             char *psz_convert_buffer_out = psz_new_subtitle;
 383             const char *psz_convert_buffer_in = psz_subtitle;
 384
 385             size_t ret = vlc_iconv( p_sys->iconv_handle,
 386                                     &psz_convert_buffer_in, &inbytes_left,
 387                                     &psz_convert_buffer_out, &outbytes_left );
 388
 389             *psz_convert_buffer_out++ = '\0';
 390             free( psz_subtitle );
 391
 392             if( ( ret == (size_t)(-1) ) || inbytes_left )
 393             {
 394                 free( psz_new_subtitle );
 395                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 396                         "Try manually setting a character-encoding "
 397                                 "before you open the file." );
 398                 return NULL;
 399             }
 400
 401             psz_subtitle = realloc( psz_new_subtitle,
 402                                     psz_convert_buffer_out - psz_new_subtitle );
 403         }
 404     }
 405
 406     /* Create the subpicture unit */
 407     p_spu = decoder_NewSubpicture( p_dec );
 408     if( !p_spu )
 409     {
 410         msg_Warn( p_dec, "can't get spu buffer" );
 411         free( psz_subtitle );
 412         return NULL;
 413     }
 414
 415     /* Create a new subpicture region */
 416     memset( &fmt, 0, sizeof(video_format_t) );
 417     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 418     fmt.i_aspect = 0;
 419     fmt.i_width = fmt.i_height = 0;
 420     fmt.i_x_offset = fmt.i_y_offset = 0;
 421     p_spu->p_region = subpicture_region_New( &fmt );
 422     if( !p_spu->p_region )
 423     {
 424         msg_Err( p_dec, "cannot allocate SPU region" );
 425         free( psz_subtitle );
 426         decoder_DeleteSubpicture( p_dec, p_spu );
 427         return NULL;
 428     }
 429
 430     /* Decode and format the subpicture unit */
 431     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 432     {
 433         /* Normal text subs, easy markup */
 434         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 435         p_spu->p_region->i_x = p_sys->i_align ? 20 : 0;
 436         p_spu->p_region->i_y = 10;
 437
 438         /* Remove formatting from string */
 439
 440         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 441         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 442         {
 443             p_spu->p_region->psz_html = CreateHtmlSubtitle( &p_spu->p_region->i_align, psz_subtitle );
 444         }
 445
 446         p_spu->i_start = p_block->i_pts;
 447         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 448         p_spu->b_ephemer = (p_block->i_length == 0);
 449         p_spu->b_absolute = false;
 450     }
 451     else
 452     {
 453         /* Decode SSA/USF strings */
 454         ParseSSAString( p_dec, psz_subtitle, p_spu );
 455
 456         p_spu->i_start = p_block->i_pts;
 457         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 458         p_spu->b_ephemer = (p_block->i_length == 0);
 459         p_spu->b_absolute = false;
 460         p_spu->i_original_picture_width = p_sys->i_original_width;
 461         p_spu->i_original_picture_height = p_sys->i_original_height;
 462     }
 463     free( psz_subtitle );
 464
 465     return p_spu;
 466 }
 467
 468 char* GotoNextLine( char *psz_text )
 469 {
 470     char *p_newline = psz_text;
 471
 472     while( p_newline[0] != '\0' )
 473     {
 474         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 475         {
 476             p_newline++;
 477             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 478                 p_newline++;
 479             break;
 480         }
 481         else p_newline++;
 482     }
 483     return p_newline;
 484 }
 485
 486 /* Function now handles tags with attribute values, and tries
 487  * to deal with &' commands too. It no longer modifies the string
 488  * in place, so that the original text can be reused
 489  */
 490 static char *StripTags( char *psz_subtitle )
 491 {
 492     char *psz_text_start;
 493     char *psz_text;
 494
 495     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 496     if( !psz_text_start )
 497         return NULL;
 498
 499     while( *psz_subtitle )
 500     {
 501         if( *psz_subtitle == '<' )
 502         {
 503             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 504                 *psz_text++ = '\n';
 505
 506             psz_subtitle += strcspn( psz_subtitle, ">" );
 507         }
 508         else if( *psz_subtitle == '&' )
 509         {
 510             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 511             {
 512                 *psz_text++ = '<';
 513                 psz_subtitle += strcspn( psz_subtitle, ";" );
 514             }
 515             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 516             {
 517                 *psz_text++ = '>';
 518                 psz_subtitle += strcspn( psz_subtitle, ";" );
 519             }
 520             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 521             {
 522                 *psz_text++ = '&';
 523                 psz_subtitle += strcspn( psz_subtitle, ";" );
 524             }
 525             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 526             {
 527                 *psz_text++ = '\"';
 528                 psz_subtitle += strcspn( psz_subtitle, ";" );
 529             }
 530             else
 531             {
 532                 /* Assume it is just a normal ampersand */
 533                 *psz_text++ = '&';
 534             }
 535         }
 536         else
 537         {
 538             *psz_text++ = *psz_subtitle;
 539         }
 540
 541         psz_subtitle++;
 542     }
 543     *psz_text = '\0';
 544     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 545
 546     return psz_text_start;
 547 }
 548
 549 /* Try to respect any style tags present in the subtitle string. The main
 550  * problem here is a lack of adequate specs for the subtitle formats.
 551  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 552  * SAMI has a detailed spec, but extensive rework is needed in the demux
 553  * code to prevent all this style information being excised, as it presently
 554  * does.
 555  * That leaves the others - none of which were (I guess) originally intended
 556  * to be carrying style information. Over time people have used them that way.
 557  * In the absence of specifications from which to work, the tags supported
 558  * have been restricted to the simple set permitted by the USF DTD, ie. :
 559  *  Basic: <br>, <i>, <b>, <u>
 560  *  Extended: <font>
 561  *    Attributes: face
 562  *                family
 563  *                size
 564  *                color
 565  *                outline-color
 566  *                shadow-color
 567  *                outline-level
 568  *                shadow-level
 569  *                back-color
 570  *                alpha
 571  * There is also the further restriction that the subtitle be well-formed
 572  * as an XML entity, ie. the HTML sentence:
 573  *        <b><i>Bold and Italics</b></i>
 574  * doesn't qualify because the tags aren't nested one inside the other.
 575  * <text> tags are automatically added to the output to ensure
 576  * well-formedness.
 577  * If the text doesn't qualify for any reason, a NULL string is
 578  * returned, and the rendering engine will fall back to the
 579  * plain text version of the subtitle.
 580  */
 581 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 582 {
 583     const int i_len = strlen(psz_text);
 584
 585     strncpy( *ppsz_html, psz_text, i_max );
 586     *ppsz_html += __MIN(i_max,i_len);
 587 }
 588
 589 static void HtmlPut( char **ppsz_html, const char *psz_text )
 590 {
 591     strcpy( *ppsz_html, psz_text );
 592     *ppsz_html += strlen(psz_text);
 593 }
 594 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 595 {
 596     HtmlPut( ppsz_html, psz_text );
 597     *ppsz_subtitle += strlen(psz_text);
 598 }
 599
 600 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 601 {
 602     /* */
 603     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 604     if( !psz_tag )
 605         return NULL;
 606     psz_tag[ 0 ] = '\0';
 607
 608     /* */
 609     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 610     char   *psz_html_start = malloc( i_buf_size );
 611     char   *psz_html = psz_html_start;
 612     if( psz_html_start == NULL )
 613     {
 614         free( psz_tag );
 615         return NULL;
 616     }
 617     psz_html[0] = '\0';
 618
 619     bool b_has_align = false;
 620
 621     HtmlPut( &psz_html, "<text>" );
 622
 623     /* */
 624     while( *psz_subtitle )
 625     {
 626         if( *psz_subtitle == '\n' )
 627         {
 628             HtmlPut( &psz_html, "<br/>" );
 629             psz_subtitle++;
 630         }
 631         else if( *psz_subtitle == '<' )
 632         {
 633             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 634             {
 635                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 636             }
 637             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 638             {
 639                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 640                 strcat( psz_tag, "b" );
 641             }
 642             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 643             {
 644                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 645                 strcat( psz_tag, "i" );
 646             }
 647             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 648             {
 649                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 650                 strcat( psz_tag, "u" );
 651             }
 652             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 653             {
 654                 const char *psz_attribs[] = { "face=", "family=", "size=",
 655                         "color=", "outline-color=", "shadow-color=",
 656                         "outline-level=", "shadow-level=", "back-color=",
 657                         "alpha=", NULL };
 658
 659                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 660                 strcat( psz_tag, "f" );
 661
 662                 while( *psz_subtitle != '>' )
 663                 {
 664                     int  k;
 665
 666                     for( k=0; psz_attribs[ k ]; k++ )
 667                     {
 668                         int i_len = strlen( psz_attribs[ k ] );
 669
 670                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 671                         {
 672                             /* */
 673                             HtmlPut( &psz_html, psz_attribs[k] );
 674                             psz_subtitle += i_len;
 675
 676                             /* */
 677                             if( *psz_subtitle == '"' )
 678                             {
 679                                 psz_subtitle++;
 680                                 i_len = strcspn( psz_subtitle, "\"" );
 681                             }
 682                             else
 683                             {
 684                                 i_len = strcspn( psz_subtitle, " \t>" );
 685                             }
 686                             HtmlPut( &psz_html, "\"" );
 687                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 688                                 HtmlPut( &psz_html, "#" );
 689                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 690                             HtmlPut( &psz_html, "\"" );
 691
 692                             psz_subtitle += i_len;
 693                             if( *psz_subtitle == '\"' )
 694                                 psz_subtitle++;
 695                             break;
 696                         }
 697                     }
 698                     if( psz_attribs[ k ] == NULL )
 699                     {
 700                         /* Jump over unrecognised tag */
 701                         int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 702
 703                         i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 704                         psz_subtitle += i_len;
 705                     }
 706                     while (*psz_subtitle == ' ')
 707                         *psz_html++ = *psz_subtitle++;
 708                 }
 709                 *psz_html++ = *psz_subtitle++;
 710             }
 711             else if( !strncmp( psz_subtitle, "</", 2 ))
 712             {
 713                 bool   b_match     = false;
 714                 bool   b_ignore    = false;
 715                 int    i_len       = strlen( psz_tag ) - 1;
 716                 char  *psz_lastTag = NULL;
 717
 718                 if( i_len >= 0 )
 719                 {
 720                     psz_lastTag = psz_tag + i_len;
 721                     i_len = 0;
 722
 723                     switch( *psz_lastTag )
 724                     {
 725                     case 'b':
 726                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 727                         i_len   = 4;
 728                         break;
 729                     case 'i':
 730                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 731                         i_len   = 4;
 732                         break;
 733                     case 'u':
 734                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 735                         i_len   = 4;
 736                         break;
 737                     case 'f':
 738                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 739                         i_len   = 7;
 740                         break;
 741                     case 'I':
 742                         i_len = strcspn( psz_subtitle, ">" );
 743                         b_match = psz_subtitle[i_len] == '>';
 744                         b_ignore = true;
 745                         if( b_match )
 746                             i_len++;
 747                         break;
 748                     }
 749                 }
 750                 if( !b_match )
 751                 {
 752                     /* Not well formed -- kill everything */
 753                     free( psz_html_start );
 754                     psz_html_start = NULL;
 755                     break;
 756                 }
 757                 *psz_lastTag = '\0';
 758                 if( !b_ignore )
 759                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 760
 761                 psz_subtitle += i_len;
 762             }
 763             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 764                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 765             {
 766                 /* We have a single < */
 767                 HtmlPut( &psz_html, "&lt;" );
 768                 psz_subtitle++;
 769             }
 770             else
 771             {
 772                 /* We have an unknown tag or a single < */
 773
 774                 /* Search for the next tag or end of tag or end of string */
 775                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 776                 char *psz_closing = strstr( psz_subtitle, "/>" );
 777
 778                 if( psz_closing && psz_closing < psz_stop )
 779                 {
 780                     /* We have a self closed tag, remove it */
 781                     psz_subtitle = &psz_closing[2];
 782                 }
 783                 else if( *psz_stop == '>' )
 784                 {
 785                     char psz_match[256];
 786
 787                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 788                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 789
 790                     if( strstr( psz_subtitle, psz_match ) )
 791                     {
 792                         /* We have the closing tag, ignore it TODO */
 793                         psz_subtitle = &psz_stop[1];
 794                         strcat( psz_tag, "I" );
 795                     }
 796                     else
 797                     {
 798                         int i_len = psz_stop + 1 - psz_subtitle;
 799
 800                         /* Copy the whole data */
 801                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 802                         {
 803                             if( *psz_subtitle == '<' )
 804                                 HtmlPut( &psz_html, "&lt;" );
 805                             else if( *psz_subtitle == '>' )
 806                                 HtmlPut( &psz_html, "&gt;" );
 807                             else
 808                                 *psz_html++ = *psz_subtitle;
 809                         }
 810                     }
 811                 }
 812                 else
 813                 {
 814                     /* We have a single < */
 815                     HtmlPut( &psz_html, "&lt;" );
 816                     psz_subtitle++;
 817                 }
 818             }
 819         }
 820         else if( *psz_subtitle == '&' )
 821         {
 822             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 823             {
 824                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 825             }
 826             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 827             {
 828                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 829             }
 830             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 831             {
 832                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 833             }
 834             else
 835             {
 836                 HtmlPut( &psz_html, "&amp;" );
 837                 psz_subtitle++;
 838             }
 839         }
 840         else if( *psz_subtitle == '>' )
 841         {
 842             HtmlPut( &psz_html, "&gt;" );
 843             psz_subtitle++;
 844         }
 845         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 846                  strchr( psz_subtitle, '}' ) )
 847         {
 848             /* Check for forced alignment */
 849             if( !b_has_align &&
 850                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 851             {
 852                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 853                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 854                 const int i_id = psz_subtitle[4] - '1';
 855
 856                 b_has_align = true;
 857                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 858             }
 859             /* TODO fr -> rotation */
 860
 861             /* Hide {\stupidity} */
 862             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 863         }
 864         else
 865         {
 866             *psz_html = *psz_subtitle;
 867             if( psz_html > psz_html_start )
 868             {
 869                 /* Check for double whitespace */
 870                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 871                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 872                 {
 873                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 874                     psz_html--;
 875                 }
 876             }
 877             psz_html++;
 878             psz_subtitle++;
 879         }
 880
 881         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 882         {
 883             const int i_len = psz_html - psz_html_start;
 884
 885             i_buf_size += 200;
 886             char *psz_new = realloc( psz_html_start, i_buf_size );
 887             if( !psz_new )
 888                 break;
 889             psz_html_start = psz_new;
 890             psz_html = &psz_new[i_len];
 891         }
 892     }
 893     if( psz_html_start )
 894     {
 895         static const char *psz_text_close = "</text>";
 896         static const char *psz_tag_long = "/font>";
 897
 898         /* Realloc for closing tags and shrink memory */
 899         const size_t i_length = (size_t)( psz_html - psz_html_start );
 900
 901         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 902         char *psz_new = realloc( psz_html_start, i_size );
 903         if( psz_new )
 904         {
 905             psz_html_start = psz_new;
 906             psz_html = &psz_new[i_length];
 907
 908             /* Close not well formed subtitle */
 909             while( *psz_tag )
 910             {
 911                 /* */
 912                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 913                 switch( *psz_last )
 914                 {
 915                 case 'b':
 916                     HtmlPut( &psz_html, "</b>" );
 917                     break;
 918                 case 'i':
 919                     HtmlPut( &psz_html, "</i>" );
 920                     break;
 921                 case 'u':
 922                     HtmlPut( &psz_html, "</u>" );
 923                     break;
 924                 case 'f':
 925                     HtmlPut( &psz_html, "/font>" );
 926                     break;
 927                 case 'I':
 928                     break;
 929                 }
 930
 931                 *psz_last = '\0';
 932             }
 933             HtmlPut( &psz_html, psz_text_close );
 934         }
 935     }
 936     free( psz_tag );
 937
 938     return psz_html_start;
 939 }
 940