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     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  71     "HZ", "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
 160     /* Allocate the memory needed to store the decoder's structure */
 161     p_dec->p_sys = p_sys = malloc( sizeof( *p_sys ) );
 162     if( p_sys == NULL )
 163         return VLC_ENOMEM;
 164
 165     /* init of p_sys */
 166     memset( p_sys, 0, sizeof( *p_sys ) );
 167     p_sys->i_align = 0;
 168     p_sys->iconv_handle = (vlc_iconv_t)-1;
 169     p_sys->b_autodetect_utf8 = false;
 170     p_sys->b_ass = false;
 171     p_sys->i_original_height = -1;
 172     p_sys->i_original_width = -1;
 173     TAB_INIT( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 174     TAB_INIT( p_sys->i_images, p_sys->pp_images );
 175
 176     char *psz_charset = NULL;
 177
 178     /* First try demux-specified encoding */
 179     if( p_dec->fmt_in.i_codec == VLC_FOURCC('t','1','4','0') )
 180         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 181     else
 182     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 183     {
 184         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 185         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 186                  p_dec->fmt_in.subs.psz_encoding ?: "not specified");
 187     }
 188
 189     /* Second, try configured encoding */
 190     if (psz_charset == NULL)
 191     {
 192         psz_charset = var_CreateGetNonEmptyString (p_dec, "subsdec-encoding");
 193         if ((psz_charset != NULL) && !strcasecmp (psz_charset, DEFAULT_NAME))
 194         {
 195             free (psz_charset);
 196             psz_charset = NULL;
 197         }
 198
 199         msg_Dbg (p_dec, "trying configured character encoding: %s",
 200                  psz_charset ?: "not specified");
 201     }
 202
 203     /* Third, try "local" encoding with optional UTF-8 autodetection */
 204     if (psz_charset == NULL)
 205     {
 206         psz_charset = strdup (GetFallbackEncoding ());
 207         msg_Dbg (p_dec, "trying default character encoding: %s",
 208                  psz_charset ?: "not specified");
 209
 210         if (var_CreateGetBool (p_dec, "subsdec-autodetect-utf8"))
 211         {
 212             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 213             p_sys->b_autodetect_utf8 = true;
 214         }
 215     }
 216
 217     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 218     if (psz_charset == NULL)
 219     {
 220         psz_charset = strdup ("UTF-8");
 221         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 222     }
 223
 224     if ((psz_charset != NULL)
 225      && strcasecmp (psz_charset, "UTF-8")
 226      && strcasecmp (psz_charset, "utf8"))
 227     {
 228         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 229         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 230             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 231     }
 232     free (psz_charset);
 233
 234     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 235     var_Get( p_dec, "subsdec-align", &val );
 236     p_sys->i_align = val.i_int;
 237
 238     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ')
 239      && var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 240     {
 241         if( p_dec->fmt_in.i_extra > 0 )
 242             ParseSSAHeader( p_dec );
 243     }
 244
 245     return VLC_SUCCESS;
 246 }
 247
 248 /****************************************************************************
 249  * DecodeBlock: the whole thing
 250  ****************************************************************************
 251  * This function must be fed with complete subtitles units.
 252  ****************************************************************************/
 253 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 254 {
 255     subpicture_t *p_spu;
 256     block_t *p_block;
 257
 258     if( !pp_block || *pp_block == NULL )
 259         return NULL;
 260
 261     p_block = *pp_block;
 262     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 263     {
 264         block_Release( p_block );
 265         return NULL;
 266     }
 267
 268     p_spu = ParseText( p_dec, p_block );
 269
 270     block_Release( p_block );
 271     *pp_block = NULL;
 272
 273     return p_spu;
 274 }
 275
 276 /*****************************************************************************
 277  * CloseDecoder: clean up the decoder
 278  *****************************************************************************/
 279 static void CloseDecoder( vlc_object_t *p_this )
 280 {
 281     decoder_t *p_dec = (decoder_t *)p_this;
 282     decoder_sys_t *p_sys = p_dec->p_sys;
 283
 284     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 285         vlc_iconv_close( p_sys->iconv_handle );
 286
 287     if( p_sys->pp_ssa_styles )
 288     {
 289         int i;
 290         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 291         {
 292             if( !p_sys->pp_ssa_styles[i] )
 293                 continue;
 294
 295             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 296             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 297             free( p_sys->pp_ssa_styles[i] );
 298         }
 299         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 300     }
 301     if( p_sys->pp_images )
 302     {
 303         int i;
 304         for( i = 0; i < p_sys->i_images; i++ )
 305         {
 306             if( !p_sys->pp_images[i] )
 307                 continue;
 308
 309             if( p_sys->pp_images[i]->p_pic )
 310                 picture_Release( p_sys->pp_images[i]->p_pic );
 311             free( p_sys->pp_images[i]->psz_filename );
 312
 313             free( p_sys->pp_images[i] );
 314         }
 315         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 316     }
 317
 318     free( p_sys );
 319 }
 320
 321 /*****************************************************************************
 322  * ParseText: parse an text subtitle packet and send it to the video output
 323  *****************************************************************************/
 324 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 325 {
 326     decoder_sys_t *p_sys = p_dec->p_sys;
 327     subpicture_t *p_spu = NULL;
 328     char *psz_subtitle = NULL;
 329     video_format_t fmt;
 330
 331     /* We cannot display a subpicture with no date */
 332     if( p_block->i_pts == 0 )
 333     {
 334         msg_Warn( p_dec, "subtitle without a date" );
 335         return NULL;
 336     }
 337
 338     /* Check validity of packet data */
 339     /* An "empty" line containing only \0 can be used to force
 340        and ephemer picture from the screen */
 341     if( p_block->i_buffer < 1 )
 342     {
 343         msg_Warn( p_dec, "no subtitle data" );
 344         return NULL;
 345     }
 346
 347     /* Should be resiliant against bad subtitles */
 348     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 349                             p_block->i_buffer );
 350     if( psz_subtitle == NULL )
 351         return NULL;
 352
 353     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 354     {
 355         if (EnsureUTF8( psz_subtitle ) == NULL)
 356         {
 357             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 358                      "Try manually setting a character-encoding "
 359                      "before you open the file." );
 360         }
 361     }
 362     else
 363     {
 364
 365         if( p_sys->b_autodetect_utf8 )
 366         {
 367             if( IsUTF8( psz_subtitle ) == NULL )
 368             {
 369                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 370                          "disabling UTF-8 subtitles autodetection" );
 371                 p_sys->b_autodetect_utf8 = false;
 372             }
 373         }
 374
 375         if( !p_sys->b_autodetect_utf8 )
 376         {
 377             size_t inbytes_left = strlen( psz_subtitle );
 378             size_t outbytes_left = 6 * inbytes_left;
 379             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 380             char *psz_convert_buffer_out = psz_new_subtitle;
 381             const char *psz_convert_buffer_in = psz_subtitle;
 382
 383             size_t ret = vlc_iconv( p_sys->iconv_handle,
 384                                     &psz_convert_buffer_in, &inbytes_left,
 385                                     &psz_convert_buffer_out, &outbytes_left );
 386
 387             *psz_convert_buffer_out++ = '\0';
 388             free( psz_subtitle );
 389
 390             if( ( ret == (size_t)(-1) ) || inbytes_left )
 391             {
 392                 free( psz_new_subtitle );
 393                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 394                         "Try manually setting a character-encoding "
 395                                 "before you open the file." );
 396                 return NULL;
 397             }
 398
 399             psz_subtitle = realloc( psz_new_subtitle,
 400                                     psz_convert_buffer_out - psz_new_subtitle );
 401         }
 402     }
 403
 404     /* Create the subpicture unit */
 405     p_spu = decoder_NewSubpicture( p_dec );
 406     if( !p_spu )
 407     {
 408         msg_Warn( p_dec, "can't get spu buffer" );
 409         free( psz_subtitle );
 410         return NULL;
 411     }
 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 = subpicture_region_New( &fmt );
 420     if( !p_spu->p_region )
 421     {
 422         msg_Err( p_dec, "cannot allocate SPU region" );
 423         free( psz_subtitle );
 424         decoder_DeleteSubpicture( 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->p_region->i_x = p_sys->i_align ? 20 : 0;
 434         p_spu->p_region->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( &p_spu->p_region->i_align, 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 = false;
 448     }
 449     else
 450     {
 451         /* Decode SSA/USF strings */
 452         ParseSSAString( p_dec, psz_subtitle, p_spu );
 453
 454         p_spu->i_start = p_block->i_pts;
 455         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 456         p_spu->b_ephemer = (p_block->i_length == 0);
 457         p_spu->b_absolute = false;
 458         p_spu->i_original_picture_width = p_sys->i_original_width;
 459         p_spu->i_original_picture_height = p_sys->i_original_height;
 460     }
 461     free( psz_subtitle );
 462
 463     return p_spu;
 464 }
 465
 466 char* GotoNextLine( char *psz_text )
 467 {
 468     char *p_newline = psz_text;
 469
 470     while( p_newline[0] != '\0' )
 471     {
 472         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 473         {
 474             p_newline++;
 475             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 476                 p_newline++;
 477             break;
 478         }
 479         else p_newline++;
 480     }
 481     return p_newline;
 482 }
 483
 484 /* Function now handles tags with attribute values, and tries
 485  * to deal with &' commands too. It no longer modifies the string
 486  * in place, so that the original text can be reused
 487  */
 488 static char *StripTags( char *psz_subtitle )
 489 {
 490     char *psz_text_start;
 491     char *psz_text;
 492
 493     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 494     if( !psz_text_start )
 495         return NULL;
 496
 497     while( *psz_subtitle )
 498     {
 499         if( *psz_subtitle == '<' )
 500         {
 501             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 502                 *psz_text++ = '\n';
 503
 504             psz_subtitle += strcspn( psz_subtitle, ">" );
 505         }
 506         else if( *psz_subtitle == '&' )
 507         {
 508             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 509             {
 510                 *psz_text++ = '<';
 511                 psz_subtitle += strcspn( psz_subtitle, ";" );
 512             }
 513             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 514             {
 515                 *psz_text++ = '>';
 516                 psz_subtitle += strcspn( psz_subtitle, ";" );
 517             }
 518             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 519             {
 520                 *psz_text++ = '&';
 521                 psz_subtitle += strcspn( psz_subtitle, ";" );
 522             }
 523             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 524             {
 525                 *psz_text++ = '\"';
 526                 psz_subtitle += strcspn( psz_subtitle, ";" );
 527             }
 528             else
 529             {
 530                 /* Assume it is just a normal ampersand */
 531                 *psz_text++ = '&';
 532             }
 533         }
 534         else
 535         {
 536             *psz_text++ = *psz_subtitle;
 537         }
 538
 539         psz_subtitle++;
 540     }
 541     *psz_text = '\0';
 542     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 543
 544     return psz_text_start;
 545 }
 546
 547 /* Try to respect any style tags present in the subtitle string. The main
 548  * problem here is a lack of adequate specs for the subtitle formats.
 549  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 550  * SAMI has a detailed spec, but extensive rework is needed in the demux
 551  * code to prevent all this style information being excised, as it presently
 552  * does.
 553  * That leaves the others - none of which were (I guess) originally intended
 554  * to be carrying style information. Over time people have used them that way.
 555  * In the absence of specifications from which to work, the tags supported
 556  * have been restricted to the simple set permitted by the USF DTD, ie. :
 557  *  Basic: <br>, <i>, <b>, <u>
 558  *  Extended: <font>
 559  *    Attributes: face
 560  *                family
 561  *                size
 562  *                color
 563  *                outline-color
 564  *                shadow-color
 565  *                outline-level
 566  *                shadow-level
 567  *                back-color
 568  *                alpha
 569  * There is also the further restriction that the subtitle be well-formed
 570  * as an XML entity, ie. the HTML sentence:
 571  *        <b><i>Bold and Italics</b></i>
 572  * doesn't qualify because the tags aren't nested one inside the other.
 573  * <text> tags are automatically added to the output to ensure
 574  * well-formedness.
 575  * If the text doesn't qualify for any reason, a NULL string is
 576  * returned, and the rendering engine will fall back to the
 577  * plain text version of the subtitle.
 578  */
 579 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 580 {
 581     const int i_len = strlen(psz_text);
 582
 583     strncpy( *ppsz_html, psz_text, i_max );
 584     *ppsz_html += __MIN(i_max,i_len);
 585 }
 586
 587 static void HtmlPut( char **ppsz_html, const char *psz_text )
 588 {
 589     strcpy( *ppsz_html, psz_text );
 590     *ppsz_html += strlen(psz_text);
 591 }
 592 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 593 {
 594     HtmlPut( ppsz_html, psz_text );
 595     *ppsz_subtitle += strlen(psz_text);
 596 }
 597
 598 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 599 {
 600     char   *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 601     if( !psz_tag ) return NULL;
 602     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 603     char   *psz_html_start = malloc( i_buf_size );
 604     bool b_has_align = false;
 605
 606     psz_tag[ 0 ] = '\0';
 607
 608     if( psz_html_start == NULL )
 609     {
 610         free( psz_tag );
 611         return NULL;
 612     }
 613
 614     char *psz_html = psz_html_start;
 615
 616     strcpy( psz_html, "<text>" );
 617     psz_html += 6;
 618
 619     /* */
 620     while( *psz_subtitle )
 621     {
 622         if( *psz_subtitle == '\n' )
 623         {
 624             HtmlPut( &psz_html, "<br/>" );
 625             psz_subtitle++;
 626         }
 627         else if( *psz_subtitle == '<' )
 628         {
 629             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 630             {
 631                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 632             }
 633             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 634             {
 635                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 636                 strcat( psz_tag, "b" );
 637             }
 638             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 639             {
 640                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 641                 strcat( psz_tag, "i" );
 642             }
 643             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 644             {
 645                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 646                 strcat( psz_tag, "u" );
 647             }
 648             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 649             {
 650                 const char *psz_attribs[] = { "face=", "family=", "size=",
 651                         "color=", "outline-color=", "shadow-color=",
 652                         "outline-level=", "shadow-level=", "back-color=",
 653                         "alpha=", NULL };
 654
 655                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 656                 strcat( psz_tag, "f" );
 657
 658                 while( *psz_subtitle != '>' )
 659                 {
 660                     int  k;
 661
 662                     for( k=0; psz_attribs[ k ]; k++ )
 663                     {
 664                         int i_len = strlen( psz_attribs[ k ] );
 665
 666                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 667                         {
 668                             /* */
 669                             HtmlPut( &psz_html, psz_attribs[k] );
 670                             psz_subtitle += i_len;
 671
 672                             /* */
 673                             if( *psz_subtitle == '"' )
 674                             {
 675                                 psz_subtitle++;
 676                                 i_len = strcspn( psz_subtitle, "\"" );
 677                             }
 678                             else
 679                             {
 680                                 i_len = strcspn( psz_subtitle, " \t>" );
 681                             }
 682                             HtmlPut( &psz_html, "\"" );
 683                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 684                                 HtmlPut( &psz_html, "#" );
 685                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 686                             HtmlPut( &psz_html, "\"" );
 687
 688                             psz_subtitle += i_len;
 689                             if( *psz_subtitle == '\"' )
 690                                 psz_subtitle++;
 691                             break;
 692                         }
 693                     }
 694                     if( psz_attribs[ k ] == NULL )
 695                     {
 696                         /* Jump over unrecognised tag */
 697                         int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 698
 699                         i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 700                         psz_subtitle += i_len;
 701                     }
 702                     while (*psz_subtitle == ' ')
 703                         *psz_html++ = *psz_subtitle++;
 704                 }
 705                 *psz_html++ = *psz_subtitle++;
 706             }
 707             else if( !strncmp( psz_subtitle, "</", 2 ))
 708             {
 709                 bool   b_match     = false;
 710                 bool   b_ignore    = false;
 711                 int    i_len       = strlen( psz_tag ) - 1;
 712                 char  *psz_lastTag = NULL;
 713
 714                 if( i_len >= 0 )
 715                 {
 716                     psz_lastTag = psz_tag + i_len;
 717                     i_len = 0;
 718
 719                     switch( *psz_lastTag )
 720                     {
 721                     case 'b':
 722                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 723                         i_len   = 4;
 724                         break;
 725                     case 'i':
 726                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 727                         i_len   = 4;
 728                         break;
 729                     case 'u':
 730                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 731                         i_len   = 4;
 732                         break;
 733                     case 'f':
 734                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 735                         i_len   = 7;
 736                         break;
 737                     case 'I':
 738                         i_len = strcspn( psz_subtitle, ">" );
 739                         b_match = psz_subtitle[i_len] == '>';
 740                         b_ignore = true;
 741                         if( b_match )
 742                             i_len++;
 743                         break;
 744                     }
 745                 }
 746                 if( !b_match )
 747                 {
 748                     /* Not well formed -- kill everything */
 749                     free( psz_html_start );
 750                     psz_html_start = NULL;
 751                     break;
 752                 }
 753                 *psz_lastTag = '\0';
 754                 if( !b_ignore )
 755                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 756
 757                 psz_subtitle += i_len;
 758             }
 759             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 760                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 761             {
 762                 /* We have a single < */
 763                 HtmlPut( &psz_html, "&lt;" );
 764                 psz_subtitle++;
 765             }
 766             else
 767             {
 768                 /* We have an unknown tag or a single < */
 769
 770                 /* Search for the next tag or end of tag or end of string */
 771                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 772                 char *psz_closing = strstr( psz_subtitle, "/>" );
 773
 774                 if( psz_closing && psz_closing < psz_stop )
 775                 {
 776                     /* We have a self closed tag, remove it */
 777                     psz_subtitle = &psz_closing[2];
 778                 }
 779                 else if( *psz_stop == '>' )
 780                 {
 781                     char psz_match[256];
 782
 783                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 784                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 785
 786                     if( strstr( psz_subtitle, psz_match ) )
 787                     {
 788                         /* We have the closing tag, ignore it TODO */
 789                         psz_subtitle = &psz_stop[1];
 790                         strcat( psz_tag, "I" );
 791                     }
 792                     else
 793                     {
 794                         int i_len = psz_stop + 1 - psz_subtitle;
 795
 796                         /* Copy the whole data */
 797                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 798                         {
 799                             if( *psz_subtitle == '<' )
 800                                 HtmlPut( &psz_html, "&lt;" );
 801                             else if( *psz_subtitle == '>' )
 802                                 HtmlPut( &psz_html, "&gt;" );
 803                             else
 804                                 *psz_html++ = *psz_subtitle;
 805                         }
 806                     }
 807                 }
 808                 else
 809                 {
 810                     /* We have a single < */
 811                     HtmlPut( &psz_html, "&lt;" );
 812                     psz_subtitle++;
 813                 }
 814             }
 815         }
 816         else if( *psz_subtitle == '&' )
 817         {
 818             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 819             {
 820                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 821             }
 822             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 823             {
 824                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 825             }
 826             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 827             {
 828                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 829             }
 830             else
 831             {
 832                 HtmlPut( &psz_html, "&amp;" );
 833                 psz_subtitle++;
 834             }
 835         }
 836         else if( *psz_subtitle == '>' )
 837         {
 838             HtmlPut( &psz_html, "&gt;" );
 839             psz_subtitle++;
 840         }
 841         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 842                  strchr( psz_subtitle, '}' ) )
 843         {
 844             /* Check for forced alignment */
 845             if( !b_has_align &&
 846                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 847             {
 848                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 849                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 850                 const int i_id = psz_subtitle[4] - '1';
 851
 852                 b_has_align = true;
 853                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 854             }
 855             /* TODO fr -> rotation */
 856
 857             /* Hide {\stupidity} */
 858             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 859         }
 860         else
 861         {
 862             *psz_html = *psz_subtitle;
 863             if( psz_html > psz_html_start )
 864             {
 865                 /* Check for double whitespace */
 866                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 867                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 868                 {
 869                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 870                     psz_html--;
 871                 }
 872             }
 873             psz_html++;
 874             psz_subtitle++;
 875         }
 876
 877         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 878         {
 879             int i_len = psz_html - psz_html_start;
 880
 881             i_buf_size += 200;
 882             psz_html_start = realloc( psz_html_start, i_buf_size );
 883             psz_html = psz_html_start + i_len;
 884             *psz_html = '\0';
 885         }
 886     }
 887     strcpy( psz_html, "</text>" );
 888     psz_html += 7;
 889
 890     if( psz_tag[ 0 ] != '\0' )
 891     {
 892         /* Not well formed -- kill everything */
 893         free( psz_html_start );
 894         psz_html_start = NULL;
 895     }
 896     else if( psz_html_start )
 897     {
 898         /* Shrink the memory requirements */
 899         psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 900     }
 901     free( psz_tag );
 902
 903     return psz_html_start;
 904 }
 905