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     if( p_block->i_rate != 0 )
 268         p_block->i_length = p_block->i_length * p_block->i_rate / INPUT_RATE_DEFAULT;
 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 = p_dec->pf_spu_buffer_new( 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 = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 422     if( !p_spu->p_region )
 423     {
 424         msg_Err( p_dec, "cannot allocate SPU region" );
 425         free( psz_subtitle );
 426         p_dec->pf_spu_buffer_del( 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         if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 455             ParseSSAString( p_dec, psz_subtitle, p_spu );
 456
 457         p_spu->i_start = p_block->i_pts;
 458         p_spu->i_stop = p_block->i_pts + p_block->i_length;
 459         p_spu->b_ephemer = (p_block->i_length == 0);
 460         p_spu->b_absolute = false;
 461         p_spu->i_original_picture_width = p_sys->i_original_width;
 462         p_spu->i_original_picture_height = p_sys->i_original_height;
 463     }
 464     free( psz_subtitle );
 465
 466     return p_spu;
 467 }
 468
 469 char* GotoNextLine( char *psz_text )
 470 {
 471     char *p_newline = psz_text;
 472
 473     while( p_newline[0] != '\0' )
 474     {
 475         if( p_newline[0] == '\n' || p_newline[0] == '\r' )
 476         {
 477             p_newline++;
 478             while( p_newline[0] == '\n' || p_newline[0] == '\r' )
 479                 p_newline++;
 480             break;
 481         }
 482         else p_newline++;
 483     }
 484     return p_newline;
 485 }
 486
 487 /* Function now handles tags with attribute values, and tries
 488  * to deal with &' commands too. It no longer modifies the string
 489  * in place, so that the original text can be reused
 490  */
 491 static char *StripTags( char *psz_subtitle )
 492 {
 493     char *psz_text_start;
 494     char *psz_text;
 495
 496     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 497     if( !psz_text_start )
 498         return NULL;
 499
 500     while( *psz_subtitle )
 501     {
 502         if( *psz_subtitle == '<' )
 503         {
 504             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 505                 *psz_text++ = '\n';
 506
 507             psz_subtitle += strcspn( psz_subtitle, ">" );
 508         }
 509         else if( *psz_subtitle == '&' )
 510         {
 511             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 512             {
 513                 *psz_text++ = '<';
 514                 psz_subtitle += strcspn( psz_subtitle, ";" );
 515             }
 516             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 517             {
 518                 *psz_text++ = '>';
 519                 psz_subtitle += strcspn( psz_subtitle, ";" );
 520             }
 521             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 522             {
 523                 *psz_text++ = '&';
 524                 psz_subtitle += strcspn( psz_subtitle, ";" );
 525             }
 526             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 527             {
 528                 *psz_text++ = '\"';
 529                 psz_subtitle += strcspn( psz_subtitle, ";" );
 530             }
 531             else
 532             {
 533                 /* Assume it is just a normal ampersand */
 534                 *psz_text++ = '&';
 535             }
 536         }
 537         else
 538         {
 539             *psz_text++ = *psz_subtitle;
 540         }
 541
 542         psz_subtitle++;
 543     }
 544     *psz_text = '\0';
 545     psz_text_start = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 546
 547     return psz_text_start;
 548 }
 549
 550 /* Try to respect any style tags present in the subtitle string. The main
 551  * problem here is a lack of adequate specs for the subtitle formats.
 552  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 553  * SAMI has a detailed spec, but extensive rework is needed in the demux
 554  * code to prevent all this style information being excised, as it presently
 555  * does.
 556  * That leaves the others - none of which were (I guess) originally intended
 557  * to be carrying style information. Over time people have used them that way.
 558  * In the absence of specifications from which to work, the tags supported
 559  * have been restricted to the simple set permitted by the USF DTD, ie. :
 560  *  Basic: <br>, <i>, <b>, <u>
 561  *  Extended: <font>
 562  *    Attributes: face
 563  *                family
 564  *                size
 565  *                color
 566  *                outline-color
 567  *                shadow-color
 568  *                outline-level
 569  *                shadow-level
 570  *                back-color
 571  *                alpha
 572  * There is also the further restriction that the subtitle be well-formed
 573  * as an XML entity, ie. the HTML sentence:
 574  *        <b><i>Bold and Italics</b></i>
 575  * doesn't qualify because the tags aren't nested one inside the other.
 576  * <text> tags are automatically added to the output to ensure
 577  * well-formedness.
 578  * If the text doesn't qualify for any reason, a NULL string is
 579  * returned, and the rendering engine will fall back to the
 580  * plain text version of the subtitle.
 581  */
 582 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 583 {
 584     const int i_len = strlen(psz_text);
 585
 586     strncpy( *ppsz_html, psz_text, i_max );
 587     *ppsz_html += __MIN(i_max,i_len);
 588 }
 589
 590 static void HtmlPut( char **ppsz_html, const char *psz_text )
 591 {
 592     strcpy( *ppsz_html, psz_text );
 593     *ppsz_html += strlen(psz_text);
 594 }
 595 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 596 {
 597     HtmlPut( ppsz_html, psz_text );
 598     *ppsz_subtitle += strlen(psz_text);
 599 }
 600
 601 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 602 {
 603     char   *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 604     if( !psz_tag ) return NULL;
 605     size_t  i_buf_size     = strlen( psz_subtitle ) + 100;
 606     char   *psz_html_start = malloc( i_buf_size );
 607
 608     psz_tag[ 0 ] = '\0';
 609
 610     if( psz_html_start == NULL )
 611     {
 612         free( psz_tag );
 613         return NULL;
 614     }
 615
 616     char *psz_html = psz_html_start;
 617
 618     strcpy( psz_html, "<text>" );
 619     psz_html += 6;
 620
 621     /* Check for forced alignment */
 622     if( !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 623     {
 624         static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 625         static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 626         const int i_id = psz_subtitle[4] - '1';
 627
 628         *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 629
 630         psz_subtitle += 6;
 631     }
 632
 633     /* */
 634     while( *psz_subtitle )
 635     {
 636         if( *psz_subtitle == '\n' )
 637         {
 638             HtmlPut( &psz_html, "<br/>" );
 639             psz_subtitle++;
 640         }
 641         else if( *psz_subtitle == '<' )
 642         {
 643             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 644             {
 645                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 646             }
 647             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 648             {
 649                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 650                 strcat( psz_tag, "b" );
 651             }
 652             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 653             {
 654                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 655                 strcat( psz_tag, "i" );
 656             }
 657             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 658             {
 659                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 660                 strcat( psz_tag, "u" );
 661             }
 662             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 663             {
 664                 const char *psz_attribs[] = { "face=", "family=", "size=",
 665                         "color=", "outline-color=", "shadow-color=",
 666                         "outline-level=", "shadow-level=", "back-color=",
 667                         "alpha=", NULL };
 668
 669                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 670                 strcat( psz_tag, "f" );
 671
 672                 while( *psz_subtitle != '>' )
 673                 {
 674                     int  k;
 675
 676                     for( k=0; psz_attribs[ k ]; k++ )
 677                     {
 678                         int i_len = strlen( psz_attribs[ k ] );
 679
 680                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 681                         {
 682                             /* */
 683                             HtmlPut( &psz_html, psz_attribs[k] );
 684                             psz_subtitle += i_len;
 685
 686                             /* */
 687                             if( *psz_subtitle == '"' )
 688                             {
 689                                 psz_subtitle++;
 690                                 i_len = strcspn( psz_subtitle, "\"" );
 691                             }
 692                             else
 693                             {
 694                                 i_len = strcspn( psz_subtitle, " \t>" );
 695                             }
 696                             HtmlPut( &psz_html, "\"" );
 697                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 698                                 HtmlPut( &psz_html, "#" );
 699                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 700                             HtmlPut( &psz_html, "\"" );
 701
 702                             psz_subtitle += i_len;
 703                             if( *psz_subtitle == '\"' )
 704                                 psz_subtitle++;
 705                             break;
 706                         }
 707                     }
 708                     if( psz_attribs[ k ] == NULL )
 709                     {
 710                         /* Jump over unrecognised tag */
 711                         int i_len = strcspn( psz_subtitle, "\"" ) + 1;
 712
 713                         i_len += strcspn( psz_subtitle + i_len, "\"" ) + 1;
 714                         psz_subtitle += i_len;
 715                     }
 716                     while (*psz_subtitle == ' ')
 717                         *psz_html++ = *psz_subtitle++;
 718                 }
 719                 *psz_html++ = *psz_subtitle++;
 720             }
 721             else if( !strncmp( psz_subtitle, "</", 2 ))
 722             {
 723                 bool   b_match     = false;
 724                 int    i_len       = strlen( psz_tag ) - 1;
 725                 char  *psz_lastTag = NULL;
 726
 727                 if( i_len >= 0 )
 728                 {
 729                     psz_lastTag = psz_tag + i_len;
 730                     i_len = 0;
 731
 732                     switch( *psz_lastTag )
 733                     {
 734                     case 'b':
 735                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 736                         i_len   = 4;
 737                         break;
 738                     case 'i':
 739                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 740                         i_len   = 4;
 741                         break;
 742                     case 'u':
 743                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 744                         i_len   = 4;
 745                         break;
 746                     case 'f':
 747                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 748                         i_len   = 7;
 749                         break;
 750                     }
 751                 }
 752                 if( ! b_match )
 753                 {
 754                     /* Not well formed -- kill everything */
 755                     free( psz_html_start );
 756                     psz_html_start = NULL;
 757                     break;
 758                 }
 759                 *psz_lastTag = '\0';
 760                 strncpy( psz_html, psz_subtitle, i_len );
 761                 psz_html += i_len;
 762                 psz_subtitle += i_len;
 763             }
 764             else
 765             {
 766                 psz_subtitle += strcspn( psz_subtitle, ">" );
 767             }
 768         }
 769         else if( *psz_subtitle == '&' )
 770         {
 771             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 772             {
 773                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 774             }
 775             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 776             {
 777                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 778             }
 779             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 780             {
 781                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 782             }
 783             else
 784             {
 785                 HtmlPut( &psz_html, "&amp;" );
 786                 psz_subtitle++;
 787             }
 788         }
 789         else
 790         {
 791             *psz_html = *psz_subtitle;
 792             if( psz_html > psz_html_start )
 793             {
 794                 /* Check for double whitespace */
 795                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 796                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 797                 {
 798                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 799                     psz_html--;
 800                 }
 801             }
 802             psz_html++;
 803             psz_subtitle++;
 804         }
 805
 806         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 807         {
 808             int i_len = psz_html - psz_html_start;
 809
 810             i_buf_size += 200;
 811             psz_html_start = realloc( psz_html_start, i_buf_size );
 812             psz_html = psz_html_start + i_len;
 813             *psz_html = '\0';
 814         }
 815     }
 816     strcpy( psz_html, "</text>" );
 817     psz_html += 7;
 818
 819     if( psz_tag[ 0 ] != '\0' )
 820     {
 821         /* Not well formed -- kill everything */
 822         free( psz_html_start );
 823         psz_html_start = NULL;
 824     }
 825     else if( psz_html_start )
 826     {
 827         /* Shrink the memory requirements */
 828         psz_html_start = realloc( psz_html_start,  psz_html - psz_html_start + 1 );
 829     }
 830     free( psz_tag );
 831
 832     return psz_html_start;
 833 }
 834