modules/codec/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 <vlc_common.h>
  35 #include <vlc_plugin.h>
  36 #include <vlc_codec.h>
  37 #include <vlc_charset.h>
  38
  39 #include "substext.h"
  40
  41 /*****************************************************************************
  42  * Module descriptor.
  43  *****************************************************************************/
  44 static const char *const ppsz_encodings[] = {
  45     "",
  46     "system",
  47     "UTF-8",
  48     "UTF-16",
  49     "UTF-16BE",
  50     "UTF-16LE",
  51     "GB18030",
  52     "ISO-8859-15",
  53     "Windows-1252",
  54     "ISO-8859-2",
  55     "Windows-1250",
  56     "ISO-8859-3",
  57     "ISO-8859-10",
  58     "Windows-1251",
  59     "KOI8-R",
  60     "KOI8-U",
  61     "ISO-8859-6",
  62     "Windows-1256",
  63     "ISO-8859-7",
  64     "Windows-1253",
  65     "ISO-8859-8",
  66     "Windows-1255",
  67     "ISO-8859-9",
  68     "Windows-1254",
  69     "ISO-8859-11",
  70     "Windows-874",
  71     "ISO-8859-13",
  72     "Windows-1257",
  73     "ISO-8859-14",
  74     "ISO-8859-16",
  75     "ISO-2022-CN-EXT",
  76     "EUC-CN",
  77     "ISO-2022-JP-2",
  78     "EUC-JP",
  79     "Shift_JIS",
  80     "CP949",
  81     "ISO-2022-KR",
  82     "Big5",
  83     "ISO-2022-TW",
  84     "Big5-HKSCS",
  85     "VISCII",
  86     "Windows-1258",
  87 };
  88
  89 static const char *const ppsz_encoding_names[] = {
  90     /* xgettext:
  91       The character encoding name in parenthesis corresponds to that used for
  92       the GetACP translation. "Windows-1252" applies to Western European
  93       languages using the Latin alphabet. */
  94     N_("Default (Windows-1252)"),
  95     N_("System codeset"),
  96     N_("Universal (UTF-8)"),
  97     N_("Universal (UTF-16)"),
  98     N_("Universal (big endian UTF-16)"),
  99     N_("Universal (little endian UTF-16)"),
 100     N_("Universal, Chinese (GB18030)"),
 101
 102   /* ISO 8859 and the likes */
 103     /* 1 */
 104     N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
 105     N_("Western European (Windows-1252)"),
 106     /* 2 */
 107     N_("Eastern European (Latin-2)"),
 108     N_("Eastern European (Windows-1250)"),
 109     /* 3 */
 110     N_("Esperanto (Latin-3)"),
 111     /* 4 */
 112     N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
 113     /* 5 */
 114     N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
 115     N_("Russian (KOI8-R)"),
 116     N_("Ukrainian (KOI8-U)"),
 117     /* 6 */
 118     N_("Arabic (ISO 8859-6)"),
 119     N_("Arabic (Windows-1256)"),
 120     /* 7 */
 121     N_("Greek (ISO 8859-7)"),
 122     N_("Greek (Windows-1253)"),
 123     /* 8 */
 124     N_("Hebrew (ISO 8859-8)"),
 125     N_("Hebrew (Windows-1255)"),
 126     /* 9 */
 127     N_("Turkish (ISO 8859-9)"),
 128     N_("Turkish (Windows-1254)"),
 129     /* 10 -> 4 */
 130     /* 11 */
 131     N_("Thai (TIS 620-2533/ISO 8859-11)"),
 132     N_("Thai (Windows-874)"),
 133     /* 13 */
 134     N_("Baltic (Latin-7)"),
 135     N_("Baltic (Windows-1257)"),
 136     /* 12 -> /dev/null */
 137     /* 14 */
 138     N_("Celtic (Latin-8)"),
 139     /* 15 -> 1 */
 140     /* 16 */
 141     N_("South-Eastern European (Latin-10)"),
 142   /* CJK families */
 143     N_("Simplified Chinese (ISO-2022-CN-EXT)"),
 144     N_("Simplified Chinese Unix (EUC-CN)"),
 145     N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
 146     N_("Japanese Unix (EUC-JP)"),
 147     N_("Japanese (Shift JIS)"),
 148     N_("Korean (EUC-KR/CP949)"),
 149     N_("Korean (ISO-2022-KR)"),
 150     N_("Traditional Chinese (Big5)"),
 151     N_("Traditional Chinese Unix (EUC-TW)"),
 152     N_("Hong-Kong Supplementary (HKSCS)"),
 153   /* Other */
 154     N_("Vietnamese (VISCII)"),
 155     N_("Vietnamese (Windows-1258)"),
 156 };
 157
 158 static const int  pi_justification[] = { 0, 1, 2 };
 159 static const char *const ppsz_justification_text[] = {
 160     N_("Center"),N_("Left"),N_("Right")};
 161
 162 #define ENCODING_TEXT N_("Subtitles text encoding")
 163 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 164 #define ALIGN_TEXT N_("Subtitles justification")
 165 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 166 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 167 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 168             "UTF-8 encoding within subtitles files.")
 169 #define FORMAT_TEXT N_("Formatted Subtitles")
 170 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 171  "VLC partly implements this, but you can choose to disable all formatting.")
 172
 173 static int  OpenDecoder   ( vlc_object_t * );
 174 static void CloseDecoder  ( vlc_object_t * );
 175
 176 vlc_module_begin ()
 177     set_shortname( N_("Subtitles"))
 178     set_description( N_("Text subtitles decoder") )
 179     set_capability( "decoder", 50 )
 180     set_callbacks( OpenDecoder, CloseDecoder )
 181     set_category( CAT_INPUT )
 182     set_subcategory( SUBCAT_INPUT_SCODEC )
 183
 184     add_integer( "subsdec-align", 0, ALIGN_TEXT, ALIGN_LONGTEXT,
 185                  false )
 186         change_integer_list( pi_justification, ppsz_justification_text )
 187     add_string( "subsdec-encoding", "",
 188                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 189         change_string_list( ppsz_encodings, ppsz_encoding_names, 0 )
 190     add_bool( "subsdec-autodetect-utf8", true,
 191               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 192     add_bool( "subsdec-formatted", true, FORMAT_TEXT, FORMAT_LONGTEXT,
 193                  false )
 194 vlc_module_end ()
 195
 196 /*****************************************************************************
 197  * Local prototypes
 198  *****************************************************************************/
 199 #define NO_BREAKING_SPACE  "&#160;"
 200
 201 struct decoder_sys_t
 202 {
 203     int                 i_align;          /* Subtitles alignment on the vout */
 204
 205     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
 206     bool                b_autodetect_utf8;
 207 };
 208
 209
 210 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
 211 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
 212 static char           *StripTags      ( char * );
 213 static char           *CreateHtmlSubtitle( int *pi_align, char * );
 214
 215 /*****************************************************************************
 216  * OpenDecoder: probe the decoder and return score
 217  *****************************************************************************
 218  * Tries to launch a decoder and return score so that the interface is able
 219  * to chose.
 220  *****************************************************************************/
 221 static int OpenDecoder( vlc_object_t *p_this )
 222 {
 223     decoder_t     *p_dec = (decoder_t*)p_this;
 224     decoder_sys_t *p_sys;
 225
 226     switch( p_dec->fmt_in.i_codec )
 227     {
 228         case VLC_CODEC_SUBT:
 229         case VLC_CODEC_ITU_T140:
 230             break;
 231         default:
 232             return VLC_EGENERIC;
 233     }
 234
 235     p_dec->pf_decode_sub = DecodeBlock;
 236     p_dec->fmt_out.i_cat = SPU_ES;
 237     p_dec->fmt_out.i_codec = 0;
 238
 239     /* Allocate the memory needed to store the decoder's structure */
 240     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 241     if( p_sys == NULL )
 242         return VLC_ENOMEM;
 243
 244     /* init of p_sys */
 245     p_sys->i_align = 0;
 246     p_sys->iconv_handle = (vlc_iconv_t)-1;
 247     p_sys->b_autodetect_utf8 = false;
 248
 249     const char *encoding;
 250     char *var = NULL;
 251
 252     /* First try demux-specified encoding */
 253     if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
 254         encoding = "UTF-8"; /* IUT T.140 is always using UTF-8 */
 255     else
 256     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 257     {
 258         encoding = p_dec->fmt_in.subs.psz_encoding;
 259         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 260                  encoding);
 261     }
 262     else
 263     /* Second, try configured encoding */
 264     if ((var = var_InheritString (p_dec, "subsdec-encoding")) != NULL)
 265     {
 266         msg_Dbg (p_dec, "trying configured character encoding: %s", var);
 267         if (!strcmp (var, "system"))
 268         {
 269             free (var);
 270             var = NULL;
 271             encoding = "";
 272             /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 273         }
 274         else
 275             encoding = var;
 276     }
 277     else
 278     /* Third, try "local" encoding with optional UTF-8 autodetection */
 279     {
 280         /* xgettext:
 281            The Windows ANSI code page most commonly used for this language.
 282            VLC uses this as a guess of the subtitle files character set
 283            (if UTF-8 and UTF-16 autodetection fails).
 284            Western European languages normally use "CP1252", which is a
 285            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 286            Other scripts use other code pages.
 287
 288            This MUST be a valid iconv character set. If unsure, please refer
 289            the VideoLAN translators mailing list. */
 290         encoding = vlc_pgettext("GetACP", "CP1252");
 291         msg_Dbg (p_dec, "trying default character encoding: %s", encoding);
 292         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 293         {
 294             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 295             p_sys->b_autodetect_utf8 = true;
 296         }
 297     }
 298
 299     if (strcasecmp (encoding, "UTF-8") && strcasecmp (encoding, "utf8"))
 300     {
 301         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", encoding);
 302         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 303             msg_Err (p_dec, "cannot convert from %s: %m", encoding);
 304     }
 305     free (var);
 306
 307     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 308
 309     return VLC_SUCCESS;
 310 }
 311
 312 /****************************************************************************
 313  * DecodeBlock: the whole thing
 314  ****************************************************************************
 315  * This function must be fed with complete subtitles units.
 316  ****************************************************************************/
 317 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 318 {
 319     subpicture_t *p_spu;
 320     block_t *p_block;
 321
 322     if( !pp_block || *pp_block == NULL )
 323         return NULL;
 324
 325     p_block = *pp_block;
 326     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 327     {
 328         block_Release( p_block );
 329         return NULL;
 330     }
 331
 332     p_spu = ParseText( p_dec, p_block );
 333
 334     block_Release( p_block );
 335     *pp_block = NULL;
 336
 337     return p_spu;
 338 }
 339
 340 /*****************************************************************************
 341  * CloseDecoder: clean up the decoder
 342  *****************************************************************************/
 343 static void CloseDecoder( vlc_object_t *p_this )
 344 {
 345     decoder_t *p_dec = (decoder_t *)p_this;
 346     decoder_sys_t *p_sys = p_dec->p_sys;
 347
 348     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 349         vlc_iconv_close( p_sys->iconv_handle );
 350
 351     free( p_sys );
 352 }
 353
 354 /*****************************************************************************
 355  * ParseText: parse an text subtitle packet and send it to the video output
 356  *****************************************************************************/
 357 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 358 {
 359     decoder_sys_t *p_sys = p_dec->p_sys;
 360     subpicture_t *p_spu = NULL;
 361     char *psz_subtitle = NULL;
 362
 363     /* We cannot display a subpicture with no date */
 364     if( p_block->i_pts <= VLC_TS_INVALID )
 365     {
 366         msg_Warn( p_dec, "subtitle without a date" );
 367         return NULL;
 368     }
 369
 370     /* Check validity of packet data */
 371     /* An "empty" line containing only \0 can be used to force
 372        and ephemer picture from the screen */
 373     if( p_block->i_buffer < 1 )
 374     {
 375         msg_Warn( p_dec, "no subtitle data" );
 376         return NULL;
 377     }
 378
 379     /* Should be resiliant against bad subtitles */
 380     psz_subtitle = malloc( p_block->i_buffer + 1 );
 381     if( psz_subtitle == NULL )
 382         return NULL;
 383     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 384     psz_subtitle[p_block->i_buffer] = '\0';
 385
 386     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 387     {
 388         if (EnsureUTF8( psz_subtitle ) == NULL)
 389         {
 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         }
 394     }
 395     else
 396     {
 397
 398         if( p_sys->b_autodetect_utf8 )
 399         {
 400             if( IsUTF8( psz_subtitle ) == NULL )
 401             {
 402                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 403                          "disabling UTF-8 subtitles autodetection" );
 404                 p_sys->b_autodetect_utf8 = false;
 405             }
 406         }
 407
 408         if( !p_sys->b_autodetect_utf8 )
 409         {
 410             size_t inbytes_left = strlen( psz_subtitle );
 411             size_t outbytes_left = 6 * inbytes_left;
 412             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 413             char *psz_convert_buffer_out = psz_new_subtitle;
 414             const char *psz_convert_buffer_in = psz_subtitle;
 415
 416             size_t ret = vlc_iconv( p_sys->iconv_handle,
 417                                     &psz_convert_buffer_in, &inbytes_left,
 418                                     &psz_convert_buffer_out, &outbytes_left );
 419
 420             *psz_convert_buffer_out++ = '\0';
 421             free( psz_subtitle );
 422
 423             if( ( ret == (size_t)(-1) ) || inbytes_left )
 424             {
 425                 free( psz_new_subtitle );
 426                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 427                         "Try manually setting a character-encoding "
 428                                 "before you open the file." );
 429                 return NULL;
 430             }
 431
 432             psz_subtitle = realloc( psz_new_subtitle,
 433                                     psz_convert_buffer_out - psz_new_subtitle );
 434             if( !psz_subtitle )
 435                 psz_subtitle = psz_new_subtitle;
 436         }
 437     }
 438
 439     /* Create the subpicture unit */
 440     p_spu = decoder_NewSubpictureText( p_dec );
 441     if( !p_spu )
 442     {
 443         free( psz_subtitle );
 444         return NULL;
 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     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 452
 453     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 454     p_spu_sys->text  = StripTags( psz_subtitle );
 455     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 456         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 457
 458     free( psz_subtitle );
 459
 460     return p_spu;
 461 }
 462
 463 /* Function now handles tags with attribute values, and tries
 464  * to deal with &' commands too. It no longer modifies the string
 465  * in place, so that the original text can be reused
 466  */
 467 static char *StripTags( char *psz_subtitle )
 468 {
 469     char *psz_text_start;
 470     char *psz_text;
 471
 472     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 473     if( !psz_text_start )
 474         return NULL;
 475
 476     while( *psz_subtitle )
 477     {
 478         if( *psz_subtitle == '<' )
 479         {
 480             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 481                 *psz_text++ = '\n';
 482
 483             psz_subtitle += strcspn( psz_subtitle, ">" );
 484         }
 485         else if( *psz_subtitle == '&' )
 486         {
 487             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 488             {
 489                 *psz_text++ = '<';
 490                 psz_subtitle += strcspn( psz_subtitle, ";" );
 491             }
 492             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 493             {
 494                 *psz_text++ = '>';
 495                 psz_subtitle += strcspn( psz_subtitle, ";" );
 496             }
 497             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 498             {
 499                 *psz_text++ = '&';
 500                 psz_subtitle += strcspn( psz_subtitle, ";" );
 501             }
 502             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 503             {
 504                 *psz_text++ = '\"';
 505                 psz_subtitle += strcspn( psz_subtitle, ";" );
 506             }
 507             else
 508             {
 509                 /* Assume it is just a normal ampersand */
 510                 *psz_text++ = '&';
 511             }
 512         }
 513         else
 514         {
 515             *psz_text++ = *psz_subtitle;
 516         }
 517
 518         /* Security fix: Account for the case where input ends early */
 519         if( *psz_subtitle == '\0' ) break;
 520
 521         psz_subtitle++;
 522     }
 523     *psz_text = '\0';
 524     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 525     if( psz ) psz_text_start = psz;
 526
 527     return psz_text_start;
 528 }
 529
 530 /* Try to respect any style tags present in the subtitle string. The main
 531  * problem here is a lack of adequate specs for the subtitle formats.
 532  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 533  * SAMI has a detailed spec, but extensive rework is needed in the demux
 534  * code to prevent all this style information being excised, as it presently
 535  * does.
 536  * That leaves the others - none of which were (I guess) originally intended
 537  * to be carrying style information. Over time people have used them that way.
 538  * In the absence of specifications from which to work, the tags supported
 539  * have been restricted to the simple set permitted by the USF DTD, ie. :
 540  *  Basic: <br>, <i>, <b>, <u>, <s>
 541  *  Extended: <font>
 542  *    Attributes: face
 543  *                family
 544  *                size
 545  *                color
 546  *                outline-color
 547  *                shadow-color
 548  *                outline-level
 549  *                shadow-level
 550  *                back-color
 551  *                alpha
 552  * There is also the further restriction that the subtitle be well-formed
 553  * as an XML entity, ie. the HTML sentence:
 554  *        <b><i>Bold and Italics</b></i>
 555  * doesn't qualify because the tags aren't nested one inside the other.
 556  * <text> tags are automatically added to the output to ensure
 557  * well-formedness.
 558  * If the text doesn't qualify for any reason, a NULL string is
 559  * returned, and the rendering engine will fall back to the
 560  * plain text version of the subtitle.
 561  */
 562 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 563 {
 564     const int i_len = strlen(psz_text);
 565
 566     strncpy( *ppsz_html, psz_text, i_max );
 567     *ppsz_html += __MIN(i_max,i_len);
 568 }
 569
 570 static void HtmlPut( char **ppsz_html, const char *psz_text )
 571 {
 572     strcpy( *ppsz_html, psz_text );
 573     *ppsz_html += strlen(psz_text);
 574 }
 575 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 576 {
 577     HtmlPut( ppsz_html, psz_text );
 578     *ppsz_subtitle += strlen(psz_text);
 579 }
 580
 581 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 582 {
 583     /* */
 584     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 585     if( !psz_tag )
 586         return NULL;
 587     psz_tag[ 0 ] = '\0';
 588
 589     /* */
 590     //Oo + 100 ???
 591     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 592     char   *psz_html_start = malloc( i_buf_size );
 593     char   *psz_html = psz_html_start;
 594     if( psz_html_start == NULL )
 595     {
 596         free( psz_tag );
 597         return NULL;
 598     }
 599     psz_html[0] = '\0';
 600
 601     bool b_has_align = false;
 602
 603     HtmlPut( &psz_html, "<text>" );
 604
 605     /* */
 606     while( *psz_subtitle )
 607     {
 608         if( *psz_subtitle == '\n' )
 609         {
 610             HtmlPut( &psz_html, "<br/>" );
 611             psz_subtitle++;
 612         }
 613         else if( *psz_subtitle == '<' )
 614         {
 615             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 616             {
 617                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 618             }
 619             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 620             {
 621                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 622                 strcat( psz_tag, "b" );
 623             }
 624             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 625             {
 626                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 627                 strcat( psz_tag, "i" );
 628             }
 629             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 630             {
 631                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 632                 strcat( psz_tag, "u" );
 633             }
 634             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 635             {
 636                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 637                 strcat( psz_tag, "s" );
 638             }
 639             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 640             {
 641                 const char *psz_attribs[] = { "face=", "family=", "size=",
 642                         "color=", "outline-color=", "shadow-color=",
 643                         "outline-level=", "shadow-level=", "back-color=",
 644                         "alpha=", NULL };
 645
 646                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 647                 strcat( psz_tag, "f" );
 648
 649                 while( *psz_subtitle != '>' )
 650                 {
 651                     int  k;
 652
 653                     for( k=0; psz_attribs[ k ]; k++ )
 654                     {
 655                         int i_len = strlen( psz_attribs[ k ] );
 656
 657                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 658                         {
 659                             /* */
 660                             HtmlPut( &psz_html, psz_attribs[k] );
 661                             psz_subtitle += i_len;
 662
 663                             /* */
 664                             if( *psz_subtitle == '"' )
 665                             {
 666                                 psz_subtitle++;
 667                                 i_len = strcspn( psz_subtitle, "\"" );
 668                             }
 669                             else
 670                             {
 671                                 i_len = strcspn( psz_subtitle, " \t>" );
 672                             }
 673                             HtmlPut( &psz_html, "\"" );
 674                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 675                             HtmlPut( &psz_html, "\"" );
 676
 677                             psz_subtitle += i_len;
 678                             if( *psz_subtitle == '\"' )
 679                                 psz_subtitle++;
 680                             break;
 681                         }
 682                     }
 683                     if( psz_attribs[ k ] == NULL )
 684                     {
 685                         /* Jump over unrecognised tag */
 686                         int i_len = strcspn( psz_subtitle, "\"" );
 687                         if( psz_subtitle[i_len] == '\"' )
 688                         {
 689                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 690                             if( psz_subtitle[i_len] == '\"' )
 691                                 i_len++;
 692                         }
 693                         psz_subtitle += i_len;
 694                     }
 695                     while (*psz_subtitle == ' ')
 696                         *psz_html++ = *psz_subtitle++;
 697                 }
 698                 *psz_html++ = *psz_subtitle++;
 699             }
 700             else if( !strncmp( psz_subtitle, "</", 2 ))
 701             {
 702                 bool   b_match     = false;
 703                 bool   b_ignore    = false;
 704                 int    i_len       = strlen( psz_tag ) - 1;
 705                 char  *psz_lastTag = NULL;
 706
 707                 if( i_len >= 0 )
 708                 {
 709                     psz_lastTag = psz_tag + i_len;
 710                     i_len = 0;
 711
 712                     switch( *psz_lastTag )
 713                     {
 714                     case 'b':
 715                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 716                         i_len   = 4;
 717                         break;
 718                     case 'i':
 719                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 720                         i_len   = 4;
 721                         break;
 722                     case 'u':
 723                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 724                         i_len   = 4;
 725                         break;
 726                     case 's':
 727                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 728                         i_len   = 4;
 729                         break;
 730                     case 'f':
 731                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 732                         i_len   = 7;
 733                         break;
 734                     case 'I':
 735                         i_len = strcspn( psz_subtitle, ">" );
 736                         b_match = psz_subtitle[i_len] == '>';
 737                         b_ignore = true;
 738                         if( b_match )
 739                             i_len++;
 740                         break;
 741                     }
 742                 }
 743                 if( !b_match )
 744                 {
 745                     /* Not well formed -- kill everything */
 746                     free( psz_html_start );
 747                     psz_html_start = NULL;
 748                     break;
 749                 }
 750                 *psz_lastTag = '\0';
 751                 if( !b_ignore )
 752                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 753
 754                 psz_subtitle += i_len;
 755             }
 756             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 757                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 758             {
 759                 /* We have a single < */
 760                 HtmlPut( &psz_html, "&lt;" );
 761                 psz_subtitle++;
 762             }
 763             else
 764             {
 765                 /* We have an unknown tag or a single < */
 766
 767                 /* Search for the next tag or end of tag or end of string */
 768                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 769                 char *psz_closing = strstr( psz_subtitle, "/>" );
 770
 771                 if( psz_closing && psz_closing < psz_stop )
 772                 {
 773                     /* We have a self closed tag, remove it */
 774                     psz_subtitle = &psz_closing[2];
 775                 }
 776                 else if( *psz_stop == '>' )
 777                 {
 778                     char psz_match[256];
 779
 780                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 781                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 782
 783                     if( strstr( psz_subtitle, psz_match ) )
 784                     {
 785                         /* We have the closing tag, ignore it TODO */
 786                         psz_subtitle = &psz_stop[1];
 787                         strcat( psz_tag, "I" );
 788                     }
 789                     else
 790                     {
 791                         int i_len = psz_stop + 1 - psz_subtitle;
 792
 793                         /* Copy the whole data */
 794                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 795                         {
 796                             if( *psz_subtitle == '<' )
 797                                 HtmlPut( &psz_html, "&lt;" );
 798                             else if( *psz_subtitle == '>' )
 799                                 HtmlPut( &psz_html, "&gt;" );
 800                             else
 801                                 *psz_html++ = *psz_subtitle;
 802                         }
 803                     }
 804                 }
 805                 else
 806                 {
 807                     /* We have a single < */
 808                     HtmlPut( &psz_html, "&lt;" );
 809                     psz_subtitle++;
 810                 }
 811             }
 812         }
 813         else if( *psz_subtitle == '&' )
 814         {
 815             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 816             {
 817                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 818             }
 819             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 820             {
 821                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 822             }
 823             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 824             {
 825                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 826             }
 827             else
 828             {
 829                 HtmlPut( &psz_html, "&amp;" );
 830                 psz_subtitle++;
 831             }
 832         }
 833         else if( *psz_subtitle == '>' )
 834         {
 835             HtmlPut( &psz_html, "&gt;" );
 836             psz_subtitle++;
 837         }
 838         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 839                  strchr( psz_subtitle, '}' ) )
 840         {
 841             /* Check for forced alignment */
 842             if( !b_has_align &&
 843                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 844             {
 845                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 846                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 847                 const int i_id = psz_subtitle[4] - '1';
 848
 849                 b_has_align = true;
 850                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 851             }
 852             /* TODO fr -> rotation */
 853
 854             /* Hide {\stupidity} */
 855             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 856         }
 857         else if( psz_subtitle[0] == '{' &&
 858                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 859                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 860         {
 861             // FIXME: We don't do difference between Y and y, and we should.
 862             if( psz_subtitle[3] == 'i' )
 863             {
 864                 HtmlPut( &psz_html, "<i>" );
 865                 strcat( psz_tag, "i" );
 866             }
 867             if( psz_subtitle[3] == 'b' )
 868             {
 869                 HtmlPut( &psz_html, "<b>" );
 870                 strcat( psz_tag, "b" );
 871             }
 872             if( psz_subtitle[3] == 'u' )
 873             {
 874                 HtmlPut( &psz_html, "<u>" );
 875                 strcat( psz_tag, "u" );
 876             }
 877             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 878         }
 879         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 880         {
 881             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 882             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 883         }
 884         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 885         {
 886             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 887             {
 888                 HtmlPut( &psz_html, "<br/>" );
 889                 psz_subtitle += 2;
 890             }
 891             else if( psz_subtitle[1] == 'h' )
 892             {
 893                 /* Non breakable space */
 894                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 895                 psz_subtitle += 2;
 896             }
 897             else
 898             {
 899                 HtmlPut( &psz_html, "\\" );
 900                 psz_subtitle++;
 901             }
 902         }
 903         else
 904         {
 905             *psz_html = *psz_subtitle;
 906             if( psz_html > psz_html_start )
 907             {
 908                 /* Check for double whitespace */
 909                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 910                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 911                 {
 912                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 913                     psz_html--;
 914                 }
 915             }
 916             psz_html++;
 917             psz_subtitle++;
 918         }
 919
 920         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 921         {
 922             const int i_len = psz_html - psz_html_start;
 923
 924             i_buf_size += 200;
 925             char *psz_new = realloc( psz_html_start, i_buf_size );
 926             if( !psz_new )
 927                 break;
 928             psz_html_start = psz_new;
 929             psz_html = &psz_new[i_len];
 930         }
 931     }
 932     if( psz_html_start )
 933     {
 934         static const char *psz_text_close = "</text>";
 935         static const char *psz_tag_long = "/font>";
 936
 937         /* Realloc for closing tags and shrink memory */
 938         const size_t i_length = (size_t)( psz_html - psz_html_start );
 939
 940         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 941         char *psz_new = realloc( psz_html_start, i_size );
 942         if( psz_new )
 943         {
 944             psz_html_start = psz_new;
 945             psz_html = &psz_new[i_length];
 946
 947             /* Close not well formed subtitle */
 948             while( *psz_tag )
 949             {
 950                 /* */
 951                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 952                 switch( *psz_last )
 953                 {
 954                 case 'b':
 955                     HtmlPut( &psz_html, "</b>" );
 956                     break;
 957                 case 'i':
 958                     HtmlPut( &psz_html, "</i>" );
 959                     break;
 960                 case 'u':
 961                     HtmlPut( &psz_html, "</u>" );
 962                     break;
 963                 case 's':
 964                     HtmlPut( &psz_html, "</s>" );
 965                     break;
 966                 case 'f':
 967                     HtmlPut( &psz_html, "/font>" );
 968                     break;
 969                 case 'I':
 970                     break;
 971                 }
 972
 973                 *psz_last = '\0';
 974             }
 975             HtmlPut( &psz_html, psz_text_close );
 976         }
 977     }
 978     free( psz_tag );
 979
 980     return psz_html_start;
 981 }
 982