git.sesse.net Git - vlc/blob - 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     "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, 0 )
 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     /* Second, try configured encoding */
 266     if ((var = var_InheritString (p_dec, "subsdec-encoding")) != NULL)
 267     {
 268         msg_Dbg (p_dec, "trying configured character encoding: %s", var);
 269         if (!strcmp (var, "system"))
 270         {
 271             free (var);
 272             var = NULL;
 273             encoding = "";
 274             /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 275         }
 276         else
 277             encoding = var;
 278     }
 279     else
 280     /* Third, try "local" encoding with optional UTF-8 autodetection */
 281     {
 282         /* xgettext:
 283            The Windows ANSI code page most commonly used for this language.
 284            VLC uses this as a guess of the subtitle files character set
 285            (if UTF-8 and UTF-16 autodetection fails).
 286            Western European languages normally use "CP1252", which is a
 287            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 288            Other scripts use other code pages.
 289
 290            This MUST be a valid iconv character set. If unsure, please refer
 291            the VideoLAN translators mailing list. */
 292         encoding = vlc_pgettext("GetACP", "CP1252");
 293         msg_Dbg (p_dec, "trying default character encoding: %s", encoding);
 294         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 295         {
 296             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 297             p_sys->b_autodetect_utf8 = true;
 298         }
 299     }
 300
 301     if (strcasecmp (encoding, "UTF-8") && strcasecmp (encoding, "utf8"))
 302     {
 303         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", encoding);
 304         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 305             msg_Err (p_dec, "cannot convert from %s: %m", encoding);
 306     }
 307     free (var);
 308
 309     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 310
 311     return VLC_SUCCESS;
 312 }
 313
 314 /****************************************************************************
 315  * DecodeBlock: the whole thing
 316  ****************************************************************************
 317  * This function must be fed with complete subtitles units.
 318  ****************************************************************************/
 319 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 320 {
 321     subpicture_t *p_spu;
 322     block_t *p_block;
 323
 324     if( !pp_block || *pp_block == NULL )
 325         return NULL;
 326
 327     p_block = *pp_block;
 328     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 329     {
 330         block_Release( p_block );
 331         return NULL;
 332     }
 333
 334     p_spu = ParseText( p_dec, p_block );
 335
 336     block_Release( p_block );
 337     *pp_block = NULL;
 338
 339     return p_spu;
 340 }
 341
 342 /*****************************************************************************
 343  * CloseDecoder: clean up the decoder
 344  *****************************************************************************/
 345 static void CloseDecoder( vlc_object_t *p_this )
 346 {
 347     decoder_t *p_dec = (decoder_t *)p_this;
 348     decoder_sys_t *p_sys = p_dec->p_sys;
 349
 350     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 351         vlc_iconv_close( p_sys->iconv_handle );
 352
 353     free( p_sys );
 354 }
 355
 356 /*****************************************************************************
 357  * ParseText: parse an text subtitle packet and send it to the video output
 358  *****************************************************************************/
 359 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 360 {
 361     decoder_sys_t *p_sys = p_dec->p_sys;
 362     subpicture_t *p_spu = NULL;
 363     char *psz_subtitle = NULL;
 364
 365     /* We cannot display a subpicture with no date */
 366     if( p_block->i_pts <= VLC_TS_INVALID )
 367     {
 368         msg_Warn( p_dec, "subtitle without a date" );
 369         return NULL;
 370     }
 371
 372     /* Check validity of packet data */
 373     /* An "empty" line containing only \0 can be used to force
 374        and ephemer picture from the screen */
 375     if( p_block->i_buffer < 1 )
 376     {
 377         msg_Warn( p_dec, "no subtitle data" );
 378         return NULL;
 379     }
 380
 381     /* Should be resiliant against bad subtitles */
 382     psz_subtitle = malloc( p_block->i_buffer + 1 );
 383     if( psz_subtitle == NULL )
 384         return NULL;
 385     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 386     psz_subtitle[p_block->i_buffer] = '\0';
 387
 388     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 389     {
 390         if (EnsureUTF8( psz_subtitle ) == NULL)
 391         {
 392             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 393                      "Try manually setting a character-encoding "
 394                      "before you open the file." );
 395         }
 396     }
 397     else
 398     {
 399
 400         if( p_sys->b_autodetect_utf8 )
 401         {
 402             if( IsUTF8( psz_subtitle ) == NULL )
 403             {
 404                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 405                          "disabling UTF-8 subtitles autodetection" );
 406                 p_sys->b_autodetect_utf8 = false;
 407             }
 408         }
 409
 410         if( !p_sys->b_autodetect_utf8 )
 411         {
 412             size_t inbytes_left = strlen( psz_subtitle );
 413             size_t outbytes_left = 6 * inbytes_left;
 414             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 415             char *psz_convert_buffer_out = psz_new_subtitle;
 416             const char *psz_convert_buffer_in = psz_subtitle;
 417
 418             size_t ret = vlc_iconv( p_sys->iconv_handle,
 419                                     &psz_convert_buffer_in, &inbytes_left,
 420                                     &psz_convert_buffer_out, &outbytes_left );
 421
 422             *psz_convert_buffer_out++ = '\0';
 423             free( psz_subtitle );
 424
 425             if( ( ret == (size_t)(-1) ) || inbytes_left )
 426             {
 427                 free( psz_new_subtitle );
 428                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 429                         "Try manually setting a character-encoding "
 430                                 "before you open the file." );
 431                 return NULL;
 432             }
 433
 434             psz_subtitle = realloc( psz_new_subtitle,
 435                                     psz_convert_buffer_out - psz_new_subtitle );
 436             if( !psz_subtitle )
 437                 psz_subtitle = psz_new_subtitle;
 438         }
 439     }
 440
 441     /* Create the subpicture unit */
 442     p_spu = decoder_NewSubpictureText( p_dec );
 443     if( !p_spu )
 444     {
 445         free( psz_subtitle );
 446         return NULL;
 447     }
 448     p_spu->i_start    = p_block->i_pts;
 449     p_spu->i_stop     = p_block->i_pts + p_block->i_length;
 450     p_spu->b_ephemer  = (p_block->i_length == 0);
 451     p_spu->b_absolute = false;
 452
 453     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 454
 455     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 456     p_spu_sys->text  = StripTags( psz_subtitle );
 457     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 458         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 459
 460     free( psz_subtitle );
 461
 462     return p_spu;
 463 }
 464
 465 /* Function now handles tags with attribute values, and tries
 466  * to deal with &' commands too. It no longer modifies the string
 467  * in place, so that the original text can be reused
 468  */
 469 static char *StripTags( char *psz_subtitle )
 470 {
 471     char *psz_text_start;
 472     char *psz_text;
 473
 474     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 475     if( !psz_text_start )
 476         return NULL;
 477
 478     while( *psz_subtitle )
 479     {
 480         if( *psz_subtitle == '<' )
 481         {
 482             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 483                 *psz_text++ = '\n';
 484
 485             psz_subtitle += strcspn( psz_subtitle, ">" );
 486         }
 487         else if( *psz_subtitle == '&' )
 488         {
 489             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 490             {
 491                 *psz_text++ = '<';
 492                 psz_subtitle += strcspn( psz_subtitle, ";" );
 493             }
 494             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 495             {
 496                 *psz_text++ = '>';
 497                 psz_subtitle += strcspn( psz_subtitle, ";" );
 498             }
 499             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 500             {
 501                 *psz_text++ = '&';
 502                 psz_subtitle += strcspn( psz_subtitle, ";" );
 503             }
 504             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 505             {
 506                 *psz_text++ = '\"';
 507                 psz_subtitle += strcspn( psz_subtitle, ";" );
 508             }
 509             else
 510             {
 511                 /* Assume it is just a normal ampersand */
 512                 *psz_text++ = '&';
 513             }
 514         }
 515         else
 516         {
 517             *psz_text++ = *psz_subtitle;
 518         }
 519
 520         /* Security fix: Account for the case where input ends early */
 521         if( *psz_subtitle == '\0' ) break;
 522
 523         psz_subtitle++;
 524     }
 525     *psz_text = '\0';
 526     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 527     if( psz ) psz_text_start = psz;
 528
 529     return psz_text_start;
 530 }
 531
 532 /* Try to respect any style tags present in the subtitle string. The main
 533  * problem here is a lack of adequate specs for the subtitle formats.
 534  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 535  * SAMI has a detailed spec, but extensive rework is needed in the demux
 536  * code to prevent all this style information being excised, as it presently
 537  * does.
 538  * That leaves the others - none of which were (I guess) originally intended
 539  * to be carrying style information. Over time people have used them that way.
 540  * In the absence of specifications from which to work, the tags supported
 541  * have been restricted to the simple set permitted by the USF DTD, ie. :
 542  *  Basic: <br>, <i>, <b>, <u>, <s>
 543  *  Extended: <font>
 544  *    Attributes: face
 545  *                family
 546  *                size
 547  *                color
 548  *                outline-color
 549  *                shadow-color
 550  *                outline-level
 551  *                shadow-level
 552  *                back-color
 553  *                alpha
 554  * There is also the further restriction that the subtitle be well-formed
 555  * as an XML entity, ie. the HTML sentence:
 556  *        <b><i>Bold and Italics</b></i>
 557  * doesn't qualify because the tags aren't nested one inside the other.
 558  * <text> tags are automatically added to the output to ensure
 559  * well-formedness.
 560  * If the text doesn't qualify for any reason, a NULL string is
 561  * returned, and the rendering engine will fall back to the
 562  * plain text version of the subtitle.
 563  */
 564 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 565 {
 566     const int i_len = strlen(psz_text);
 567
 568     strncpy( *ppsz_html, psz_text, i_max );
 569     *ppsz_html += __MIN(i_max,i_len);
 570 }
 571
 572 static void HtmlPut( char **ppsz_html, const char *psz_text )
 573 {
 574     strcpy( *ppsz_html, psz_text );
 575     *ppsz_html += strlen(psz_text);
 576 }
 577 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 578 {
 579     HtmlPut( ppsz_html, psz_text );
 580     *ppsz_subtitle += strlen(psz_text);
 581 }
 582
 583 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 584 {
 585     /* */
 586     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 587     if( !psz_tag )
 588         return NULL;
 589     psz_tag[ 0 ] = '\0';
 590
 591     /* */
 592     //Oo + 100 ???
 593     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 594     char   *psz_html_start = malloc( i_buf_size );
 595     char   *psz_html = psz_html_start;
 596     if( psz_html_start == NULL )
 597     {
 598         free( psz_tag );
 599         return NULL;
 600     }
 601     psz_html[0] = '\0';
 602
 603     bool b_has_align = false;
 604
 605     HtmlPut( &psz_html, "<text>" );
 606
 607     /* */
 608     while( *psz_subtitle )
 609     {
 610         if( *psz_subtitle == '\n' )
 611         {
 612             HtmlPut( &psz_html, "<br/>" );
 613             psz_subtitle++;
 614         }
 615         else if( *psz_subtitle == '<' )
 616         {
 617             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 618             {
 619                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 620             }
 621             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 622             {
 623                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 624                 strcat( psz_tag, "b" );
 625             }
 626             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 627             {
 628                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 629                 strcat( psz_tag, "i" );
 630             }
 631             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 632             {
 633                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 634                 strcat( psz_tag, "u" );
 635             }
 636             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 637             {
 638                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 639                 strcat( psz_tag, "s" );
 640             }
 641             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 642             {
 643                 const char *psz_attribs[] = { "face=", "family=", "size=",
 644                         "color=", "outline-color=", "shadow-color=",
 645                         "outline-level=", "shadow-level=", "back-color=",
 646                         "alpha=", NULL };
 647
 648                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 649                 strcat( psz_tag, "f" );
 650
 651                 while( *psz_subtitle != '>' )
 652                 {
 653                     int  k;
 654
 655                     for( k=0; psz_attribs[ k ]; k++ )
 656                     {
 657                         int i_len = strlen( psz_attribs[ k ] );
 658
 659                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 660                         {
 661                             /* */
 662                             HtmlPut( &psz_html, psz_attribs[k] );
 663                             psz_subtitle += i_len;
 664
 665                             /* */
 666                             if( *psz_subtitle == '"' )
 667                             {
 668                                 psz_subtitle++;
 669                                 i_len = strcspn( psz_subtitle, "\"" );
 670                             }
 671                             else if( *psz_subtitle == '\'' )
 672                             {
 673                                 psz_subtitle++;
 674                                 i_len = strcspn( psz_subtitle, "'" );
 675                             }
 676                             else
 677                             {
 678                                 i_len = strcspn( psz_subtitle, " \t>" );
 679                             }
 680                             HtmlPut( &psz_html, "\"" );
 681                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 682                             HtmlPut( &psz_html, "\"" );
 683
 684                             psz_subtitle += i_len;
 685                             if( *psz_subtitle == '\"' || *psz_subtitle == '\'' )
 686                                 psz_subtitle++;
 687                             break;
 688                         }
 689                     }
 690                     if( psz_attribs[ k ] == NULL )
 691                     {
 692                         /* Jump over unrecognised tag */
 693                         int i_len = strcspn( psz_subtitle, "\"" );
 694                         if( psz_subtitle[i_len] == '\"' )
 695                         {
 696                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 697                             if( psz_subtitle[i_len] == '\"' )
 698                                 i_len++;
 699                         }
 700                         /* Not a tag, something else we do not understand */
 701                         if( i_len == 0 )
 702                             *psz_subtitle++;
 703
 704                         psz_subtitle += i_len;
 705                     }
 706                     while (*psz_subtitle == ' ')
 707                         *psz_html++ = *psz_subtitle++;
 708                 }
 709                 *psz_html++ = '>';
 710                 *psz_subtitle++;
 711             }
 712             else if( !strncmp( psz_subtitle, "</", 2 ))
 713             {
 714                 bool   b_match     = false;
 715                 bool   b_ignore    = false;
 716                 int    i_len       = strlen( psz_tag ) - 1;
 717                 char  *psz_lastTag = NULL;
 718
 719                 if( i_len >= 0 )
 720                 {
 721                     psz_lastTag = psz_tag + i_len;
 722                     i_len = 0;
 723
 724                     switch( *psz_lastTag )
 725                     {
 726                     case 'b':
 727                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 728                         i_len   = 4;
 729                         break;
 730                     case 'i':
 731                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 732                         i_len   = 4;
 733                         break;
 734                     case 'u':
 735                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 736                         i_len   = 4;
 737                         break;
 738                     case 's':
 739                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 740                         i_len   = 4;
 741                         break;
 742                     case 'f':
 743                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 744                         i_len   = 7;
 745                         break;
 746                     case 'I':
 747                         i_len = strcspn( psz_subtitle, ">" );
 748                         b_match = psz_subtitle[i_len] == '>';
 749                         b_ignore = true;
 750                         if( b_match )
 751                             i_len++;
 752                         break;
 753                     }
 754                 }
 755                 if( !b_match )
 756                 {
 757                     /* Not well formed -- kill everything */
 758                     free( psz_html_start );
 759                     psz_html_start = NULL;
 760                     break;
 761                 }
 762                 *psz_lastTag = '\0';
 763                 if( !b_ignore )
 764                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 765
 766                 psz_subtitle += i_len;
 767             }
 768             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 769                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 770             {
 771                 /* We have a single < */
 772                 HtmlPut( &psz_html, "&lt;" );
 773                 psz_subtitle++;
 774             }
 775             else
 776             {
 777                 /* We have an unknown tag or a single < */
 778
 779                 /* Search for the next tag or end of tag or end of string */
 780                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 781                 char *psz_closing = strstr( psz_subtitle, "/>" );
 782
 783                 if( psz_closing && psz_closing < psz_stop )
 784                 {
 785                     /* We have a self closed tag, remove it */
 786                     psz_subtitle = &psz_closing[2];
 787                 }
 788                 else if( *psz_stop == '>' )
 789                 {
 790                     char psz_match[256];
 791
 792                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 793                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 794
 795                     if( strstr( psz_subtitle, psz_match ) )
 796                     {
 797                         /* We have the closing tag, ignore it TODO */
 798                         psz_subtitle = &psz_stop[1];
 799                         strcat( psz_tag, "I" );
 800                     }
 801                     else
 802                     {
 803                         int i_len = psz_stop + 1 - psz_subtitle;
 804
 805                         /* Copy the whole data */
 806                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 807                         {
 808                             if( *psz_subtitle == '<' )
 809                                 HtmlPut( &psz_html, "&lt;" );
 810                             else if( *psz_subtitle == '>' )
 811                                 HtmlPut( &psz_html, "&gt;" );
 812                             else
 813                                 *psz_html++ = *psz_subtitle;
 814                         }
 815                     }
 816                 }
 817                 else
 818                 {
 819                     /* We have a single < */
 820                     HtmlPut( &psz_html, "&lt;" );
 821                     psz_subtitle++;
 822                 }
 823             }
 824         }
 825         else if( *psz_subtitle == '&' )
 826         {
 827             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 828             {
 829                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 830             }
 831             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 832             {
 833                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 834             }
 835             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 836             {
 837                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 838             }
 839             else
 840             {
 841                 HtmlPut( &psz_html, "&amp;" );
 842                 psz_subtitle++;
 843             }
 844         }
 845         else if( *psz_subtitle == '>' )
 846         {
 847             HtmlPut( &psz_html, "&gt;" );
 848             psz_subtitle++;
 849         }
 850         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 851                  strchr( psz_subtitle, '}' ) )
 852         {
 853             /* Check for forced alignment */
 854             if( !b_has_align &&
 855                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 856             {
 857                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 858                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 859                 const int i_id = psz_subtitle[4] - '1';
 860
 861                 b_has_align = true;
 862                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 863             }
 864             /* TODO fr -> rotation */
 865
 866             /* Hide {\stupidity} */
 867             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 868         }
 869         else if( psz_subtitle[0] == '{' &&
 870                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 871                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 872         {
 873             // FIXME: We don't do difference between Y and y, and we should.
 874             if( psz_subtitle[3] == 'i' )
 875             {
 876                 HtmlPut( &psz_html, "<i>" );
 877                 strcat( psz_tag, "i" );
 878             }
 879             if( psz_subtitle[3] == 'b' )
 880             {
 881                 HtmlPut( &psz_html, "<b>" );
 882                 strcat( psz_tag, "b" );
 883             }
 884             if( psz_subtitle[3] == 'u' )
 885             {
 886                 HtmlPut( &psz_html, "<u>" );
 887                 strcat( psz_tag, "u" );
 888             }
 889             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 890         }
 891         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 892         {
 893             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 894             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 895         }
 896         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 897         {
 898             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 899             {
 900                 HtmlPut( &psz_html, "<br/>" );
 901                 psz_subtitle += 2;
 902             }
 903             else if( psz_subtitle[1] == 'h' )
 904             {
 905                 /* Non breakable space */
 906                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 907                 psz_subtitle += 2;
 908             }
 909             else
 910             {
 911                 HtmlPut( &psz_html, "\\" );
 912                 psz_subtitle++;
 913             }
 914         }
 915         else
 916         {
 917             *psz_html = *psz_subtitle;
 918             if( psz_html > psz_html_start )
 919             {
 920                 /* Check for double whitespace */
 921                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 922                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 923                 {
 924                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 925                     psz_html--;
 926                 }
 927             }
 928             psz_html++;
 929             psz_subtitle++;
 930         }
 931
 932         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 933         {
 934             const int i_len = psz_html - psz_html_start;
 935
 936             i_buf_size += 200;
 937             char *psz_new = realloc( psz_html_start, i_buf_size );
 938             if( !psz_new )
 939                 break;
 940             psz_html_start = psz_new;
 941             psz_html = &psz_new[i_len];
 942         }
 943     }
 944     if( psz_html_start )
 945     {
 946         static const char *psz_text_close = "</text>";
 947         static const char *psz_tag_long = "/font>";
 948
 949         /* Realloc for closing tags and shrink memory */
 950         const size_t i_length = (size_t)( psz_html - psz_html_start );
 951
 952         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 953         char *psz_new = realloc( psz_html_start, i_size );
 954         if( psz_new )
 955         {
 956             psz_html_start = psz_new;
 957             psz_html = &psz_new[i_length];
 958
 959             /* Close not well formed subtitle */
 960             while( *psz_tag )
 961             {
 962                 /* */
 963                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 964                 switch( *psz_last )
 965                 {
 966                 case 'b':
 967                     HtmlPut( &psz_html, "</b>" );
 968                     break;
 969                 case 'i':
 970                     HtmlPut( &psz_html, "</i>" );
 971                     break;
 972                 case 'u':
 973                     HtmlPut( &psz_html, "</u>" );
 974                     break;
 975                 case 's':
 976                     HtmlPut( &psz_html, "</s>" );
 977                     break;
 978                 case 'f':
 979                     HtmlPut( &psz_html, "/font>" );
 980                     break;
 981                 case 'I':
 982                     break;
 983                 }
 984
 985                 *psz_last = '\0';
 986             }
 987             HtmlPut( &psz_html, psz_text_close );
 988         }
 989     }
 990     free( psz_tag );
 991
 992     return psz_html_start;
 993 }
 994