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_rate != 0 )
 263         p_block->i_length = p_block->i_length * p_block->i_rate / INPUT_RATE_DEFAULT;
 264
 265     p_spu = ParseText( p_dec, p_block );
 266
 267     block_Release( p_block );
 268     *pp_block = NULL;
 269
 270     return p_spu;
 271 }
 272
 273 /*****************************************************************************
 274  * CloseDecoder: clean up the decoder
 275  *****************************************************************************/
 276 static void CloseDecoder( vlc_object_t *p_this )
 277 {
 278     decoder_t *p_dec = (decoder_t *)p_this;
 279     decoder_sys_t *p_sys = p_dec->p_sys;
 280
 281     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 282         vlc_iconv_close( p_sys->iconv_handle );
 283
 284     if( p_sys->pp_ssa_styles )
 285     {
 286         int i;
 287         for( i = 0; i < p_sys->i_ssa_styles; i++ )
 288         {
 289             if( !p_sys->pp_ssa_styles[i] )
 290                 continue;
 291
 292             free( p_sys->pp_ssa_styles[i]->psz_stylename );
 293             free( p_sys->pp_ssa_styles[i]->font_style.psz_fontname );
 294             free( p_sys->pp_ssa_styles[i] );
 295         }
 296         TAB_CLEAN( p_sys->i_ssa_styles, p_sys->pp_ssa_styles );
 297     }
 298     if( p_sys->pp_images )
 299     {
 300         int i;
 301         for( i = 0; i < p_sys->i_images; i++ )
 302         {
 303             if( !p_sys->pp_images[i] )
 304                 continue;
 305
 306             if( p_sys->pp_images[i]->p_pic )
 307                 picture_Release( p_sys->pp_images[i]->p_pic );
 308             free( p_sys->pp_images[i]->psz_filename );
 309
 310             free( p_sys->pp_images[i] );
 311         }
 312         TAB_CLEAN( p_sys->i_images, p_sys->pp_images );
 313     }
 314
 315     free( p_sys );
 316 }
 317
 318 /*****************************************************************************
 319  * ParseText: parse an text subtitle packet and send it to the video output
 320  *****************************************************************************/
 321 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 322 {
 323     decoder_sys_t *p_sys = p_dec->p_sys;
 324     subpicture_t *p_spu = NULL;
 325     char *psz_subtitle = NULL;
 326     video_format_t fmt;
 327
 328     /* We cannot display a subpicture with no date */
 329     if( p_block->i_pts == 0 )
 330     {
 331         msg_Warn( p_dec, "subtitle without a date" );
 332         return NULL;
 333     }
 334
 335     /* Check validity of packet data */
 336     /* An "empty" line containing only \0 can be used to force
 337        and ephemer picture from the screen */
 338     if( p_block->i_buffer < 1 )
 339     {
 340         msg_Warn( p_dec, "no subtitle data" );
 341         return NULL;
 342     }
 343
 344     /* Should be resiliant against bad subtitles */
 345     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 346                             p_block->i_buffer );
 347     if( psz_subtitle == NULL )
 348         return NULL;
 349
 350     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 351     {
 352         if (EnsureUTF8( psz_subtitle ) == NULL)
 353         {
 354             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 355                      "Try manually setting a character-encoding "
 356                      "before you open the file." );
 357         }
 358     }
 359     else
 360     {
 361
 362         if( p_sys->b_autodetect_utf8 )
 363         {
 364             if( IsUTF8( psz_subtitle ) == NULL )
 365             {
 366                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 367                          "disabling UTF-8 subtitles autodetection" );
 368                 p_sys->b_autodetect_utf8 = false;
 369             }
 370         }
 371
 372         if( !p_sys->b_autodetect_utf8 )
 373         {
 374             size_t inbytes_left = strlen( psz_subtitle );
 375             size_t outbytes_left = 6 * inbytes_left;
 376             char *psz_new_subtitle = malloc( outbytes_left + 1 );
 377             char *psz_convert_buffer_out = psz_new_subtitle;
 378             const char *psz_convert_buffer_in = psz_subtitle;
 379
 380             size_t ret = vlc_iconv( p_sys->iconv_handle,
 381                                     &psz_convert_buffer_in, &inbytes_left,
 382                                     &psz_convert_buffer_out, &outbytes_left );
 383
 384             *psz_convert_buffer_out++ = '\0';
 385             free( psz_subtitle );
 386
 387             if( ( ret == (size_t)(-1) ) || inbytes_left )
 388             {
 389                 free( psz_new_subtitle );
 390                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 391                         "Try manually setting a character-encoding "
 392                                 "before you open the file." );
 393                 return NULL;
 394             }
 395
 396             psz_subtitle = realloc( psz_new_subtitle,
 397                                     psz_convert_buffer_out - psz_new_subtitle );
 398         }
 399     }
 400
 401     /* Create the subpicture unit */
 402     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 403     if( !p_spu )
 404     {
 405         msg_Warn( p_dec, "can't get spu buffer" );
 406         free( psz_subtitle );
 407         return NULL;
 408     }
 409
 410     /* Create a new subpicture region */
 411     memset( &fmt, 0, sizeof(video_format_t) );
 412     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 413     fmt.i_aspect = 0;
 414     fmt.i_width = fmt.i_height = 0;
 415     fmt.i_x_offset = fmt.i_y_offset = 0;
 416     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 417     if( !p_spu->p_region )
 418     {
 419         msg_Err( p_dec, "cannot allocate SPU region" );
 420         free( psz_subtitle );
 421         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 422         return NULL;
 423     }
 424
 425     /* Decode and format the subpicture unit */
 426     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 427     {
 428         /* Normal text subs, easy markup */
 429         p_spu->p_region->i_align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 430         p_spu->p_region->i_x = p_sys->i_align ? 20 : 0;
 431         p_spu->p_region->i_y = 10;
 432
 433         /* Remove formatting from string */
 434
 435         p_spu->p_region->psz_text = StripTags( psz_subtitle );
 436         if( var_CreateGetBool( p_dec, "subsdec-formatted" ) )
 437         {
 438             p_spu->p_region->psz_html = CreateHtmlSubtitle( &p_spu->p_region->i_align, psz_subtitle );
 439         }
 440
 441         p_spu->i_start = p_block->i_pts;
 442         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 443         p_spu->b_ephemer = (p_block->i_length == 0);
 444         p_spu->b_absolute = false;
 445     }
 446     else
 447     {
 448         /* Decode SSA/USF strings */
 449         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 450             ParseSSAString( p_dec, psz_subtitle, p_spu );
 451
 452         p_spu->i_start = p_block->i_pts;
 453         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 454         p_spu->b_ephemer = (p_block->i_length == 0);
 455         p_spu->b_absolute = false;
 456         p_spu->i_original_picture_width = p_sys->i_original_width;
 457         p_spu->i_original_picture_height = p_sys->i_original_height;
 458     }
 459     free( psz_subtitle );
 460
 461     return p_spu;
 462 }
 463
 464 char* GotoNextLine( char *psz_text )
 465 {
 466     char *p_newline = psz_text;
 467
 468     while( p_newline[0] != '\0' )
 469     {
 470         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 471         {
 472             p_newline++;
 473             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 474                 p_newline++;
 475             break;
 476         }
 477         else p_newline++;
 478     }
 479     return p_newline;
 480 }
 481
 482 /* Function now handles tags with attribute values, and tries
 483  * to deal with &' commands too. It no longer modifies the string
 484  * in place, so that the original text can be reused
 485  */
 486 static char *StripTags( char *psz_subtitle )
 487 {
 488     char *psz_text_start;
 489     char *psz_text;
 490
 491     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 492     if( !psz_text_start )
 493         return NULL;
 494
 495     while( *psz_subtitle )
 496     {
 497         if( *psz_subtitle == '<' )
 498         {
 499             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 500                 *psz_text++ = '\n';
 501
 502             psz_subtitle += strcspn( psz_subtitle, ">" );
 503         }
 504         else if( *psz_subtitle == '&' )
 505         {
 506             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 507             {
 508                 *psz_text++ = '<';
 509                 psz_subtitle += strcspn( psz_subtitle, ";" );
 510             }
 511             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 512             {
 513                 *psz_text++ = '>';
 514                 psz_subtitle += strcspn( psz_subtitle, ";" );
 515             }
 516             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 517             {
 518                 *psz_text++ = '&';
 519                 psz_subtitle += strcspn( psz_subtitle, ";" );
 520             }
 521             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 522             {
 523                 *psz_text++ = '\"';
 524                 psz_subtitle += strcspn( psz_subtitle, ";" );
 525             }
 526             else
 527             {
 528                 /* Assume it is just a normal ampersand */
 529                 *psz_text++ = '&';
 530             }
 531         }
 532         else
 533         {
 534             *psz_text++ = *psz_subtitle;
 535         }
 536
 537         psz_subtitle++;
 538     }
 539     *psz_text = '\0';
 540     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 541
 542     return psz_text_start;
 543 }
 544
 545 /* Try to respect any style tags present in the subtitle string. The main
 546  * problem here is a lack of adequate specs for the subtitle formats.
 547  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 548  * SAMI has a detailed spec, but extensive rework is needed in the demux
 549  * code to prevent all this style information being excised, as it presently
 550  * does.
 551  * That leaves the others - none of which were (I guess) originally intended
 552  * to be carrying style information. Over time people have used them that way.
 553  * In the absence of specifications from which to work, the tags supported
 554  * have been restricted to the simple set permitted by the USF DTD, ie. :
 555  *  Basic: <br>, <i>, <b>, <u>
 556  *  Extended: <font>
 557  *    Attributes: face
 558  *                family
 559  *                size
 560  *                color
 561  *                outline-color
 562  *                shadow-color
 563  *                outline-level
 564  *                shadow-level
 565  *                back-color
 566  *                alpha
 567  * There is also the further restriction that the subtitle be well-formed
 568  * as an XML entity, ie. the HTML sentence:
 569  *        <b><i>Bold and Italics</b></i>
 570  * doesn't qualify because the tags aren't nested one inside the other.
 571  * <text> tags are automatically added to the output to ensure
 572  * well-formedness.
 573  * If the text doesn't qualify for any reason, a NULL string is
 574  * returned, and the rendering engine will fall back to the
 575  * plain text version of the subtitle.
 576  */
 577 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 578 {
 579     const int i_len = strlen(psz_text);
 580
 581     strncpy( *ppsz_html, psz_text, i_max );
 582     *ppsz_html += __MIN(i_max,i_len);
 583 }
 584
 585 static void HtmlPut( char **ppsz_html, const char *psz_text )
 586 {
 587     strcpy( *ppsz_html, psz_text );
 588     *ppsz_html += strlen(psz_text);
 589 }
 590 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 591 {
 592     HtmlPut( ppsz_html, psz_text );
 593     *ppsz_subtitle += strlen(psz_text);
 594 }
 595
 596 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 597 {
 598     char   *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 599     if( !psz_tag ) return NULL;
 600     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 601     char   *psz_html_start = malloc( i_buf_size );
 602
 603     psz_tag[ 0 ] = '\0';
 604
 605     if( psz_html_start == NULL )
 606     {
 607         free( psz_tag );
 608         return NULL;
 609     }
 610
 611     char *psz_html = psz_html_start;
 612
 613     strcpy( psz_html, "<text>" );
 614     psz_html += 6;
 615
 616     /* Check for forced alignment */
 617     if( !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 618     {
 619         static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 620         static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 621         const int i_id = psz_subtitle[4] - '1';
 622
 623         *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 624
 625         psz_subtitle += 6;
 626     }
 627
 628     /* */
 629     while( *psz_subtitle )
 630     {
 631         if( *psz_subtitle == '\n' )
 632         {
 633             HtmlPut( &psz_html, "<br/>" );
 634             psz_subtitle++;
 635         }
 636         else if( *psz_subtitle == '<' )
 637         {
 638             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 639             {
 640                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 641             }
 642             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 643             {
 644                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 645                 strcat( psz_tag, "b" );
 646             }
 647             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 648             {
 649                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 650                 strcat( psz_tag, "i" );
 651             }
 652             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 653             {
 654                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 655                 strcat( psz_tag, "u" );
 656             }
 657             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 658             {
 659                 const char *psz_attribs[] = { "face=", "family=", "size=",
 660                         "color=", "outline-color=", "shadow-color=",
 661                         "outline-level=", "shadow-level=", "back-color=",
 662                         "alpha=", NULL };
 663
 664                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 665                 strcat( psz_tag, "f" );
 666
 667                 while( *psz_subtitle != '>' )
 668                 {
 669                     int  k;
 670
 671                     for( k=0; psz_attribs[ k ]; k++ )
 672                     {
 673                         int i_len = strlen( psz_attribs[ k ] );
 674
 675                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 676                         {
 677                             /* */
 678                             HtmlPut( &psz_html, psz_attribs[k] );
 679                             psz_subtitle += i_len;
 680
 681                             /* */
 682                             if( *psz_subtitle == '"' )
 683                             {
 684                                 psz_subtitle++;
 685                                 i_len = strcspn( psz_subtitle, "\"" );
 686                             }
 687                             else
 688                             {
 689                                 i_len = strcspn( psz_subtitle, " \t>" );
 690                             }
 691                             HtmlPut( &psz_html, "\"" );
 692                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 693                                 HtmlPut( &psz_html, "#" );
 694                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 695                             HtmlPut( &psz_html, "\"" );
 696
 697                             psz_subtitle += i_len;
 698                             if( *psz_subtitle == '\"' )
 699                                 psz_subtitle++;
 700                             break;
 701                         }
 702                     }
 703                     if( psz_attribs[ k ] == NULL )
 704                     {
 705                         /* Jump over unrecognised tag */
 706                         int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 707
 708                         i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 709                         psz_subtitle += i_len;
 710                     }
 711                     while (*psz_subtitle == ' ')
 712                         *psz_html++ = *psz_subtitle++;
 713                 }
 714                 *psz_html++ = *psz_subtitle++;
 715             }
 716             else if( !strncmp( psz_subtitle, "</", 2 ))
 717             {
 718                 bool   b_match     = false;
 719                 int    i_len       = strlen( psz_tag ) - 1;
 720                 char  *psz_lastTag = NULL;
 721
 722                 if( i_len >= 0 )
 723                 {
 724                     psz_lastTag = psz_tag + i_len;
 725                     i_len = 0;
 726
 727                     switch( *psz_lastTag )
 728                     {
 729                     case 'b':
 730                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 731                         i_len   = 4;
 732                         break;
 733                     case 'i':
 734                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 735                         i_len   = 4;
 736                         break;
 737                     case 'u':
 738                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 739                         i_len   = 4;
 740                         break;
 741                     case 'f':
 742                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 743                         i_len   = 7;
 744                         break;
 745                     }
 746                 }
 747                 if( ! b_match )
 748                 {
 749                     /* Not well formed -- kill everything */
 750                     free( psz_html_start );
 751                     psz_html_start = NULL;
 752                     break;
 753                 }
 754                 *psz_lastTag = '\0';
 755                 strncpy( psz_html, psz_subtitle, i_len );
 756                 psz_html += i_len;
 757                 psz_subtitle += i_len;
 758             }
 759             else
 760             {
 761                 psz_subtitle += strcspn( psz_subtitle, ">" );
 762             }
 763         }
 764         else if( *psz_subtitle == '&' )
 765         {
 766             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 767             {
 768                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 769             }
 770             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 771             {
 772                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 773             }
 774             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 775             {
 776                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 777             }
 778             else
 779             {
 780                 HtmlPut( &psz_html, "&amp;" );
 781                 psz_subtitle++;
 782             }
 783         }
 784         else
 785         {
 786             *psz_html = *psz_subtitle;
 787             if( psz_html > psz_html_start )
 788             {
 789                 /* Check for double whitespace */
 790                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 791                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 792                 {
 793                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 794                     psz_html--;
 795                 }
 796             }
 797             psz_html++;
 798             psz_subtitle++;
 799         }
 800
 801         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 802         {
 803             int i_len = psz_html - psz_html_start;
 804
 805             i_buf_size += 200;
 806             psz_html_start = realloc( psz_html_start, i_buf_size );
 807             psz_html = psz_html_start + i_len;
 808             *psz_html = '\0';
 809         }
 810     }
 811     strcpy( psz_html, "</text>" );
 812     psz_html += 7;
 813
 814     if( psz_tag[ 0 ] != '\0' )
 815     {
 816         /* Not well formed -- kill everything */
 817         free( psz_html_start );
 818         psz_html_start = NULL;
 819     }
 820     else if( psz_html_start )
 821     {
 822         /* Shrink the memory requirements */
 823         psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 824     }
 825     free( psz_tag );
 826
 827     return psz_html_start;
 828 }
 829