git.sesse.net Git - vlc/blob - modules/codec/subsdec.c

   1 /*****************************************************************************
   2  * subsdec.c : text subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 VLC authors and VideoLAN
   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 it
  13  * under the terms of the GNU Lesser General Public License as published by
  14  * the Free Software Foundation; either version 2.1 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 Lesser General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU Lesser General Public License
  23  * along with this program; if not, write to the Free Software Foundation,
  24  * 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     "IBM850",
  55     "ISO-8859-2",
  56     "Windows-1250",
  57     "ISO-8859-3",
  58     "ISO-8859-10",
  59     "Windows-1251",
  60     "KOI8-R",
  61     "KOI8-U",
  62     "ISO-8859-6",
  63     "Windows-1256",
  64     "ISO-8859-7",
  65     "Windows-1253",
  66     "ISO-8859-8",
  67     "Windows-1255",
  68     "ISO-8859-9",
  69     "Windows-1254",
  70     "ISO-8859-11",
  71     "Windows-874",
  72     "ISO-8859-13",
  73     "Windows-1257",
  74     "ISO-8859-14",
  75     "ISO-8859-16",
  76     "ISO-2022-CN-EXT",
  77     "EUC-CN",
  78     "ISO-2022-JP-2",
  79     "EUC-JP",
  80     "Shift_JIS",
  81     "CP949",
  82     "ISO-2022-KR",
  83     "Big5",
  84     "ISO-2022-TW",
  85     "Big5-HKSCS",
  86     "VISCII",
  87     "Windows-1258",
  88 };
  89
  90 static const char *const ppsz_encoding_names[] = {
  91     /* xgettext:
  92       The character encoding name in parenthesis corresponds to that used for
  93       the GetACP translation. "Windows-1252" applies to Western European
  94       languages using the Latin alphabet. */
  95     N_("Default (Windows-1252)"),
  96     N_("System codeset"),
  97     N_("Universal (UTF-8)"),
  98     N_("Universal (UTF-16)"),
  99     N_("Universal (big endian UTF-16)"),
 100     N_("Universal (little endian UTF-16)"),
 101     N_("Universal, Chinese (GB18030)"),
 102
 103   /* ISO 8859 and the likes */
 104     /* 1 */
 105     N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
 106     N_("Western European (Windows-1252)"),
 107     N_("Western European (IBM 00850)"),
 108     /* 2 */
 109     N_("Eastern European (Latin-2)"),
 110     N_("Eastern European (Windows-1250)"),
 111     /* 3 */
 112     N_("Esperanto (Latin-3)"),
 113     /* 4 */
 114     N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
 115     /* 5 */
 116     N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
 117     N_("Russian (KOI8-R)"),
 118     N_("Ukrainian (KOI8-U)"),
 119     /* 6 */
 120     N_("Arabic (ISO 8859-6)"),
 121     N_("Arabic (Windows-1256)"),
 122     /* 7 */
 123     N_("Greek (ISO 8859-7)"),
 124     N_("Greek (Windows-1253)"),
 125     /* 8 */
 126     N_("Hebrew (ISO 8859-8)"),
 127     N_("Hebrew (Windows-1255)"),
 128     /* 9 */
 129     N_("Turkish (ISO 8859-9)"),
 130     N_("Turkish (Windows-1254)"),
 131     /* 10 -> 4 */
 132     /* 11 */
 133     N_("Thai (TIS 620-2533/ISO 8859-11)"),
 134     N_("Thai (Windows-874)"),
 135     /* 13 */
 136     N_("Baltic (Latin-7)"),
 137     N_("Baltic (Windows-1257)"),
 138     /* 12 -> /dev/null */
 139     /* 14 */
 140     N_("Celtic (Latin-8)"),
 141     /* 15 -> 1 */
 142     /* 16 */
 143     N_("South-Eastern European (Latin-10)"),
 144   /* CJK families */
 145     N_("Simplified Chinese (ISO-2022-CN-EXT)"),
 146     N_("Simplified Chinese Unix (EUC-CN)"),
 147     N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
 148     N_("Japanese Unix (EUC-JP)"),
 149     N_("Japanese (Shift JIS)"),
 150     N_("Korean (EUC-KR/CP949)"),
 151     N_("Korean (ISO-2022-KR)"),
 152     N_("Traditional Chinese (Big5)"),
 153     N_("Traditional Chinese Unix (EUC-TW)"),
 154     N_("Hong-Kong Supplementary (HKSCS)"),
 155   /* Other */
 156     N_("Vietnamese (VISCII)"),
 157     N_("Vietnamese (Windows-1258)"),
 158 };
 159
 160 static const int  pi_justification[] = { 0, 1, 2 };
 161 static const char *const ppsz_justification_text[] = {
 162     N_("Center"),N_("Left"),N_("Right")};
 163
 164 #define ENCODING_TEXT N_("Subtitles text encoding")
 165 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 166 #define ALIGN_TEXT N_("Subtitles justification")
 167 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 168 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 169 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 170             "UTF-8 encoding within subtitles files.")
 171 #define FORMAT_TEXT N_("Formatted Subtitles")
 172 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 173  "VLC partly implements this, but you can choose to disable all formatting.")
 174
 175 static int  OpenDecoder   ( vlc_object_t * );
 176 static void CloseDecoder  ( vlc_object_t * );
 177
 178 vlc_module_begin ()
 179     set_shortname( N_("Subtitles"))
 180     set_description( N_("Text subtitles decoder") )
 181     set_capability( "decoder", 50 )
 182     set_callbacks( OpenDecoder, CloseDecoder )
 183     set_category( CAT_INPUT )
 184     set_subcategory( SUBCAT_INPUT_SCODEC )
 185
 186     add_integer( "subsdec-align", 0, ALIGN_TEXT, ALIGN_LONGTEXT,
 187                  false )
 188         change_integer_list( pi_justification, ppsz_justification_text )
 189     add_string( "subsdec-encoding", "",
 190                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 191         change_string_list( ppsz_encodings, ppsz_encoding_names )
 192     add_bool( "subsdec-autodetect-utf8", true,
 193               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 194     add_bool( "subsdec-formatted", true, FORMAT_TEXT, FORMAT_LONGTEXT,
 195                  false )
 196 vlc_module_end ()
 197
 198 /*****************************************************************************
 199  * Local prototypes
 200  *****************************************************************************/
 201 #define NO_BREAKING_SPACE  "&#160;"
 202
 203 struct decoder_sys_t
 204 {
 205     int                 i_align;          /* Subtitles alignment on the vout */
 206
 207     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
 208     bool                b_autodetect_utf8;
 209 };
 210
 211
 212 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
 213 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
 214 static char           *StripTags      ( char * );
 215 static char           *CreateHtmlSubtitle( int *pi_align, char * );
 216
 217 /*****************************************************************************
 218  * OpenDecoder: probe the decoder and return score
 219  *****************************************************************************
 220  * Tries to launch a decoder and return score so that the interface is able
 221  * to chose.
 222  *****************************************************************************/
 223 static int OpenDecoder( vlc_object_t *p_this )
 224 {
 225     decoder_t     *p_dec = (decoder_t*)p_this;
 226     decoder_sys_t *p_sys;
 227
 228     switch( p_dec->fmt_in.i_codec )
 229     {
 230         case VLC_CODEC_SUBT:
 231         case VLC_CODEC_ITU_T140:
 232             break;
 233         default:
 234             return VLC_EGENERIC;
 235     }
 236
 237     p_dec->pf_decode_sub = DecodeBlock;
 238     p_dec->fmt_out.i_cat = SPU_ES;
 239     p_dec->fmt_out.i_codec = 0;
 240
 241     /* Allocate the memory needed to store the decoder's structure */
 242     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 243     if( p_sys == NULL )
 244         return VLC_ENOMEM;
 245
 246     /* init of p_sys */
 247     p_sys->i_align = 0;
 248     p_sys->iconv_handle = (vlc_iconv_t)-1;
 249     p_sys->b_autodetect_utf8 = false;
 250
 251     const char *encoding;
 252     char *var = NULL;
 253
 254     /* First try demux-specified encoding */
 255     if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
 256         encoding = "UTF-8"; /* IUT T.140 is always using UTF-8 */
 257     else
 258     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 259     {
 260         encoding = p_dec->fmt_in.subs.psz_encoding;
 261         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 262                  encoding);
 263     }
 264     else
 265     {
 266         /* Second, try configured encoding */
 267         if ((var = var_InheritString (p_dec, "subsdec-encoding")) != NULL)
 268         {
 269             msg_Dbg (p_dec, "trying configured character encoding: %s", var);
 270             if (!strcmp (var, "system"))
 271             {
 272                 free (var);
 273                 var = NULL;
 274                 encoding = "";
 275                 /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 276             }
 277             else
 278                 encoding = var;
 279         }
 280         else
 281         /* Third, try "local" encoding */
 282         {
 283         /* xgettext:
 284            The Windows ANSI code page most commonly used for this language.
 285            VLC uses this as a guess of the subtitle files character set
 286            (if UTF-8 and UTF-16 autodetection fails).
 287            Western European languages normally use "CP1252", which is a
 288            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 289            Other scripts use other code pages.
 290
 291            This MUST be a valid iconv character set. If unsure, please refer
 292            the VideoLAN translators mailing list. */
 293             encoding = vlc_pgettext("GetACP", "CP1252");
 294             msg_Dbg (p_dec, "trying default character encoding: %s", encoding);
 295         }
 296
 297         /* Check UTF-8 autodetection */
 298         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 299         {
 300             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 301             p_sys->b_autodetect_utf8 = true;
 302         }
 303     }
 304
 305     if (strcasecmp (encoding, "UTF-8") && strcasecmp (encoding, "utf8"))
 306     {
 307         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", encoding);
 308         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 309             msg_Err (p_dec, "cannot convert from %s: %m", encoding);
 310     }
 311     free (var);
 312
 313     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 314
 315     return VLC_SUCCESS;
 316 }
 317
 318 /****************************************************************************
 319  * DecodeBlock: the whole thing
 320  ****************************************************************************
 321  * This function must be fed with complete subtitles units.
 322  ****************************************************************************/
 323 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 324 {
 325     subpicture_t *p_spu;
 326     block_t *p_block;
 327
 328     if( !pp_block || *pp_block == NULL )
 329         return NULL;
 330
 331     p_block = *pp_block;
 332     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 333     {
 334         block_Release( p_block );
 335         return NULL;
 336     }
 337
 338     p_spu = ParseText( p_dec, p_block );
 339
 340     block_Release( p_block );
 341     *pp_block = NULL;
 342
 343     return p_spu;
 344 }
 345
 346 /*****************************************************************************
 347  * CloseDecoder: clean up the decoder
 348  *****************************************************************************/
 349 static void CloseDecoder( vlc_object_t *p_this )
 350 {
 351     decoder_t *p_dec = (decoder_t *)p_this;
 352     decoder_sys_t *p_sys = p_dec->p_sys;
 353
 354     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 355         vlc_iconv_close( p_sys->iconv_handle );
 356
 357     free( p_sys );
 358 }
 359
 360 /*****************************************************************************
 361  * ParseText: parse an text subtitle packet and send it to the video output
 362  *****************************************************************************/
 363 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 364 {
 365     decoder_sys_t *p_sys = p_dec->p_sys;
 366     subpicture_t *p_spu = NULL;
 367     char *psz_subtitle = NULL;
 368
 369     /* We cannot display a subpicture with no date */
 370     if( p_block->i_pts <= VLC_TS_INVALID )
 371     {
 372         msg_Warn( p_dec, "subtitle without a date" );
 373         return NULL;
 374     }
 375
 376     /* Check validity of packet data */
 377     /* An "empty" line containing only \0 can be used to force
 378        and ephemer picture from the screen */
 379     if( p_block->i_buffer < 1 )
 380     {
 381         msg_Warn( p_dec, "no subtitle data" );
 382         return NULL;
 383     }
 384
 385     /* Should be resiliant against bad subtitles */
 386     psz_subtitle = malloc( p_block->i_buffer + 1 );
 387     if( psz_subtitle == NULL )
 388         return NULL;
 389     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 390     psz_subtitle[p_block->i_buffer] = '\0';
 391
 392     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 393     {
 394         if (EnsureUTF8( psz_subtitle ) == NULL)
 395         {
 396             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 397                      "Try manually setting a character-encoding "
 398                      "before you open the file." );
 399         }
 400     }
 401     else
 402     {
 403
 404         if( p_sys->b_autodetect_utf8 )
 405         {
 406             if( IsUTF8( psz_subtitle ) == NULL )
 407             {
 408                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 409                          "disabling UTF-8 subtitles autodetection" );
 410                 p_sys->b_autodetect_utf8 = false;
 411             }
 412         }
 413
 414         if( !p_sys->b_autodetect_utf8 )
 415         {
 416             size_t inbytes_left = strlen( psz_subtitle );
 417             size_t outbytes_left = 6 * inbytes_left;
 418             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 419             char *psz_convert_buffer_out = psz_new_subtitle;
 420             const char *psz_convert_buffer_in = psz_subtitle;
 421
 422             size_t ret = vlc_iconv( p_sys->iconv_handle,
 423                                     &psz_convert_buffer_in, &inbytes_left,
 424                                     &psz_convert_buffer_out, &outbytes_left );
 425
 426             *psz_convert_buffer_out++ = '\0';
 427             free( psz_subtitle );
 428
 429             if( ( ret == (size_t)(-1) ) || inbytes_left )
 430             {
 431                 free( psz_new_subtitle );
 432                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 433                         "Try manually setting a character-encoding "
 434                                 "before you open the file." );
 435                 return NULL;
 436             }
 437
 438             psz_subtitle = realloc( psz_new_subtitle,
 439                                     psz_convert_buffer_out - psz_new_subtitle );
 440             if( !psz_subtitle )
 441                 psz_subtitle = psz_new_subtitle;
 442         }
 443     }
 444
 445     /* Create the subpicture unit */
 446     p_spu = decoder_NewSubpictureText( p_dec );
 447     if( !p_spu )
 448     {
 449         free( psz_subtitle );
 450         return NULL;
 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
 457     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 458
 459     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 460     p_spu_sys->text  = StripTags( psz_subtitle );
 461     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 462         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 463
 464     free( psz_subtitle );
 465
 466     return p_spu;
 467 }
 468
 469 /* Function now handles tags with attribute values, and tries
 470  * to deal with &' commands too. It no longer modifies the string
 471  * in place, so that the original text can be reused
 472  */
 473 static char *StripTags( char *psz_subtitle )
 474 {
 475     char *psz_text_start;
 476     char *psz_text;
 477
 478     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 479     if( !psz_text_start )
 480         return NULL;
 481
 482     while( *psz_subtitle )
 483     {
 484         if( *psz_subtitle == '<' )
 485         {
 486             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 487                 *psz_text++ = '\n';
 488
 489             psz_subtitle += strcspn( psz_subtitle, ">" );
 490         }
 491         else if( *psz_subtitle == '&' )
 492         {
 493             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 494             {
 495                 *psz_text++ = '<';
 496                 psz_subtitle += strcspn( psz_subtitle, ";" );
 497             }
 498             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 499             {
 500                 *psz_text++ = '>';
 501                 psz_subtitle += strcspn( psz_subtitle, ";" );
 502             }
 503             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 504             {
 505                 *psz_text++ = '&';
 506                 psz_subtitle += strcspn( psz_subtitle, ";" );
 507             }
 508             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 509             {
 510                 *psz_text++ = '\"';
 511                 psz_subtitle += strcspn( psz_subtitle, ";" );
 512             }
 513             else
 514             {
 515                 /* Assume it is just a normal ampersand */
 516                 *psz_text++ = '&';
 517             }
 518         }
 519         else
 520         {
 521             *psz_text++ = *psz_subtitle;
 522         }
 523
 524         /* Security fix: Account for the case where input ends early */
 525         if( *psz_subtitle == '\0' ) break;
 526
 527         psz_subtitle++;
 528     }
 529     *psz_text = '\0';
 530     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 531     if( psz ) psz_text_start = psz;
 532
 533     return psz_text_start;
 534 }
 535
 536 /* Try to respect any style tags present in the subtitle string. The main
 537  * problem here is a lack of adequate specs for the subtitle formats.
 538  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 539  * SAMI has a detailed spec, but extensive rework is needed in the demux
 540  * code to prevent all this style information being excised, as it presently
 541  * does.
 542  * That leaves the others - none of which were (I guess) originally intended
 543  * to be carrying style information. Over time people have used them that way.
 544  * In the absence of specifications from which to work, the tags supported
 545  * have been restricted to the simple set permitted by the USF DTD, ie. :
 546  *  Basic: <br>, <i>, <b>, <u>, <s>
 547  *  Extended: <font>
 548  *    Attributes: face
 549  *                family
 550  *                size
 551  *                color
 552  *                outline-color
 553  *                shadow-color
 554  *                outline-level
 555  *                shadow-level
 556  *                back-color
 557  *                alpha
 558  * There is also the further restriction that the subtitle be well-formed
 559  * as an XML entity, ie. the HTML sentence:
 560  *        <b><i>Bold and Italics</b></i>
 561  * doesn't qualify because the tags aren't nested one inside the other.
 562  * <text> tags are automatically added to the output to ensure
 563  * well-formedness.
 564  * If the text doesn't qualify for any reason, a NULL string is
 565  * returned, and the rendering engine will fall back to the
 566  * plain text version of the subtitle.
 567  */
 568 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 569 {
 570     const int i_len = strlen(psz_text);
 571
 572     strncpy( *ppsz_html, psz_text, i_max );
 573     *ppsz_html += __MIN(i_max,i_len);
 574 }
 575
 576 static void HtmlPut( char **ppsz_html, const char *psz_text )
 577 {
 578     strcpy( *ppsz_html, psz_text );
 579     *ppsz_html += strlen(psz_text);
 580 }
 581 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 582 {
 583     HtmlPut( ppsz_html, psz_text );
 584     *ppsz_subtitle += strlen(psz_text);
 585 }
 586
 587 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 588 {
 589     /* */
 590     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 591     if( !psz_tag )
 592         return NULL;
 593     psz_tag[ 0 ] = '\0';
 594
 595     /* */
 596     //Oo + 100 ???
 597     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 598     char   *psz_html_start = malloc( i_buf_size );
 599     char   *psz_html = psz_html_start;
 600     if( psz_html_start == NULL )
 601     {
 602         free( psz_tag );
 603         return NULL;
 604     }
 605     psz_html[0] = '\0';
 606
 607     bool b_has_align = false;
 608
 609     HtmlPut( &psz_html, "<text>" );
 610
 611     /* */
 612     while( *psz_subtitle )
 613     {
 614         if( *psz_subtitle == '\n' )
 615         {
 616             HtmlPut( &psz_html, "<br/>" );
 617             psz_subtitle++;
 618         }
 619         else if( *psz_subtitle == '<' )
 620         {
 621             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 622             {
 623                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 624             }
 625             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 626             {
 627                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 628                 strcat( psz_tag, "b" );
 629             }
 630             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 631             {
 632                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 633                 strcat( psz_tag, "i" );
 634             }
 635             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 636             {
 637                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 638                 strcat( psz_tag, "u" );
 639             }
 640             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 641             {
 642                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 643                 strcat( psz_tag, "s" );
 644             }
 645             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 646             {
 647                 const char *psz_attribs[] = { "face=", "family=", "size=",
 648                         "color=", "outline-color=", "shadow-color=",
 649                         "outline-level=", "shadow-level=", "back-color=",
 650                         "alpha=", NULL };
 651
 652                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 653                 strcat( psz_tag, "f" );
 654
 655                 /* <font       color= */
 656                 while (*psz_subtitle == ' ')
 657                     psz_subtitle++;
 658
 659                 while( *psz_subtitle != '>' )
 660                 {
 661                     int  k;
 662
 663                     for( k=0; psz_attribs[ k ]; k++ )
 664                     {
 665                         int i_len = strlen( psz_attribs[ k ] );
 666
 667                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 668                         {
 669                             /* */
 670                             HtmlPut( &psz_html, psz_attribs[k] );
 671                             psz_subtitle += i_len;
 672
 673                             /* <font       color=   red */
 674                             while (*psz_subtitle == ' ')
 675                                 psz_subtitle++;
 676
 677                             /* */
 678                             if( *psz_subtitle == '"' )
 679                             {
 680                                 psz_subtitle++;
 681                                 i_len = strcspn( psz_subtitle, "\"" );
 682                             }
 683                             else if( *psz_subtitle == '\'' )
 684                             {
 685                                 psz_subtitle++;
 686                                 i_len = strcspn( psz_subtitle, "'" );
 687                             }
 688                             else
 689                             {
 690                                 i_len = strcspn( psz_subtitle, " \t>" );
 691                             }
 692                             HtmlPut( &psz_html, "\"" );
 693                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 694                             HtmlPut( &psz_html, "\"" );
 695
 696                             psz_subtitle += i_len;
 697                             if( *psz_subtitle == '\"' || *psz_subtitle == '\'' )
 698                                 psz_subtitle++;
 699                             break;
 700                         }
 701                     }
 702                     if( psz_attribs[ k ] == NULL )
 703                     {
 704                         /* Jump over unrecognised tag */
 705                         int i_len = strcspn( psz_subtitle, "\"" );
 706                         if( psz_subtitle[i_len] == '\"' )
 707                         {
 708                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 709                             if( psz_subtitle[i_len] == '\"' )
 710                                 i_len++;
 711                         }
 712                         /* Not a tag, something else we do not understand */
 713                         if( i_len == 0 )
 714                             psz_subtitle++;
 715
 716                         psz_subtitle += i_len;
 717                     }
 718
 719                     while (*psz_subtitle == ' ')
 720                         *psz_html++ = *psz_subtitle++;
 721                 }
 722                 *psz_html++ = '>';
 723                 psz_subtitle++;
 724             }
 725             else if( !strncmp( psz_subtitle, "</", 2 ))
 726             {
 727                 bool   b_match     = false;
 728                 bool   b_ignore    = false;
 729                 int    i_len       = strlen( psz_tag ) - 1;
 730                 char  *psz_lastTag = NULL;
 731
 732                 if( i_len >= 0 )
 733                 {
 734                     psz_lastTag = psz_tag + i_len;
 735                     i_len = 0;
 736
 737                     switch( *psz_lastTag )
 738                     {
 739                     case 'b':
 740                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 741                         i_len   = 4;
 742                         break;
 743                     case 'i':
 744                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 745                         i_len   = 4;
 746                         break;
 747                     case 'u':
 748                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 749                         i_len   = 4;
 750                         break;
 751                     case 's':
 752                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 753                         i_len   = 4;
 754                         break;
 755                     case 'f':
 756                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 757                         i_len   = 7;
 758                         break;
 759                     case 'I':
 760                         i_len = strcspn( psz_subtitle, ">" );
 761                         b_match = psz_subtitle[i_len] == '>';
 762                         b_ignore = true;
 763                         if( b_match )
 764                             i_len++;
 765                         break;
 766                     }
 767                 }
 768                 if( !b_match )
 769                 {
 770                     /* Not well formed -- kill everything */
 771                     free( psz_html_start );
 772                     psz_html_start = NULL;
 773                     break;
 774                 }
 775                 *psz_lastTag = '\0';
 776                 if( !b_ignore )
 777                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 778
 779                 psz_subtitle += i_len;
 780             }
 781             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 782                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 783             {
 784                 /* We have a single < */
 785                 HtmlPut( &psz_html, "&lt;" );
 786                 psz_subtitle++;
 787             }
 788             else
 789             {
 790                 /* We have an unknown tag or a single < */
 791
 792                 /* Search for the next tag or end of tag or end of string */
 793                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 794                 char *psz_closing = strstr( psz_subtitle, "/>" );
 795
 796                 if( psz_closing && psz_closing < psz_stop )
 797                 {
 798                     /* We have a self closed tag, remove it */
 799                     psz_subtitle = &psz_closing[2];
 800                 }
 801                 else if( *psz_stop == '>' )
 802                 {
 803                     char psz_match[256];
 804
 805                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 806                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 807
 808                     if( strstr( psz_subtitle, psz_match ) )
 809                     {
 810                         /* We have the closing tag, ignore it TODO */
 811                         psz_subtitle = &psz_stop[1];
 812                         strcat( psz_tag, "I" );
 813                     }
 814                     else
 815                     {
 816                         int i_len = psz_stop + 1 - psz_subtitle;
 817
 818                         /* Copy the whole data */
 819                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 820                         {
 821                             if( *psz_subtitle == '<' )
 822                                 HtmlPut( &psz_html, "&lt;" );
 823                             else if( *psz_subtitle == '>' )
 824                                 HtmlPut( &psz_html, "&gt;" );
 825                             else
 826                                 *psz_html++ = *psz_subtitle;
 827                         }
 828                     }
 829                 }
 830                 else
 831                 {
 832                     /* We have a single < */
 833                     HtmlPut( &psz_html, "&lt;" );
 834                     psz_subtitle++;
 835                 }
 836             }
 837         }
 838         else if( *psz_subtitle == '&' )
 839         {
 840             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 841             {
 842                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 843             }
 844             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 845             {
 846                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 847             }
 848             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 849             {
 850                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 851             }
 852             else
 853             {
 854                 HtmlPut( &psz_html, "&amp;" );
 855                 psz_subtitle++;
 856             }
 857         }
 858         else if( *psz_subtitle == '>' )
 859         {
 860             HtmlPut( &psz_html, "&gt;" );
 861             psz_subtitle++;
 862         }
 863         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 864                  strchr( psz_subtitle, '}' ) )
 865         {
 866             /* Check for forced alignment */
 867             if( !b_has_align &&
 868                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 869             {
 870                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 871                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 872                 const int i_id = psz_subtitle[4] - '1';
 873
 874                 b_has_align = true;
 875                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 876             }
 877             /* TODO fr -> rotation */
 878
 879             /* Hide {\stupidity} */
 880             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 881         }
 882         else if( psz_subtitle[0] == '{' &&
 883                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 884                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 885         {
 886             // FIXME: We don't do difference between Y and y, and we should.
 887             if( psz_subtitle[3] == 'i' )
 888             {
 889                 HtmlPut( &psz_html, "<i>" );
 890                 strcat( psz_tag, "i" );
 891             }
 892             if( psz_subtitle[3] == 'b' )
 893             {
 894                 HtmlPut( &psz_html, "<b>" );
 895                 strcat( psz_tag, "b" );
 896             }
 897             if( psz_subtitle[3] == 'u' )
 898             {
 899                 HtmlPut( &psz_html, "<u>" );
 900                 strcat( psz_tag, "u" );
 901             }
 902             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 903         }
 904         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 905         {
 906             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 907             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 908         }
 909         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 910         {
 911             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 912             {
 913                 HtmlPut( &psz_html, "<br/>" );
 914                 psz_subtitle += 2;
 915             }
 916             else if( psz_subtitle[1] == 'h' )
 917             {
 918                 /* Non breakable space */
 919                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 920                 psz_subtitle += 2;
 921             }
 922             else
 923             {
 924                 HtmlPut( &psz_html, "\\" );
 925                 psz_subtitle++;
 926             }
 927         }
 928         else
 929         {
 930             *psz_html = *psz_subtitle;
 931             if( psz_html > psz_html_start )
 932             {
 933                 /* Check for double whitespace */
 934                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 935                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 936                 {
 937                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 938                     psz_html--;
 939                 }
 940             }
 941             psz_html++;
 942             psz_subtitle++;
 943         }
 944
 945         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 946         {
 947             const int i_len = psz_html - psz_html_start;
 948
 949             i_buf_size += 200;
 950             char *psz_new = realloc( psz_html_start, i_buf_size );
 951             if( !psz_new )
 952                 break;
 953             psz_html_start = psz_new;
 954             psz_html = &psz_new[i_len];
 955         }
 956     }
 957     if( psz_html_start )
 958     {
 959         static const char *psz_text_close = "</text>";
 960         static const char *psz_tag_long = "/font>";
 961
 962         /* Realloc for closing tags and shrink memory */
 963         const size_t i_length = (size_t)( psz_html - psz_html_start );
 964
 965         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 966         char *psz_new = realloc( psz_html_start, i_size );
 967         if( psz_new )
 968         {
 969             psz_html_start = psz_new;
 970             psz_html = &psz_new[i_length];
 971
 972             /* Close not well formed subtitle */
 973             while( *psz_tag )
 974             {
 975                 /* */
 976                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 977                 switch( *psz_last )
 978                 {
 979                 case 'b':
 980                     HtmlPut( &psz_html, "</b>" );
 981                     break;
 982                 case 'i':
 983                     HtmlPut( &psz_html, "</i>" );
 984                     break;
 985                 case 'u':
 986                     HtmlPut( &psz_html, "</u>" );
 987                     break;
 988                 case 's':
 989                     HtmlPut( &psz_html, "</s>" );
 990                     break;
 991                 case 'f':
 992                     HtmlPut( &psz_html, "</font>" );
 993                     break;
 994                 case 'I':
 995                     break;
 996                 }
 997
 998                 *psz_last = '\0';
 999             }
1000             HtmlPut( &psz_html, psz_text_close );
1001         }
1002     }
1003     free( psz_tag );
1004
1005     return psz_html_start;
1006 }
1007