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

   1 /*****************************************************************************
   2  * subsdec.c : text subtitle 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 <limits.h>
  35
  36 #include <vlc_common.h>
  37 #include <vlc_plugin.h>
  38 #include <vlc_codec.h>
  39 #include <vlc_charset.h>
  40
  41 #include "substext.h"
  42
  43 /*****************************************************************************
  44  * Module descriptor.
  45  *****************************************************************************/
  46 static const char *const ppsz_encodings[] = {
  47     "",
  48     "system",
  49     "UTF-8",
  50     "UTF-16",
  51     "UTF-16BE",
  52     "UTF-16LE",
  53     "GB18030",
  54     "ISO-8859-15",
  55     "Windows-1252",
  56     "IBM850",
  57     "ISO-8859-2",
  58     "Windows-1250",
  59     "ISO-8859-3",
  60     "ISO-8859-10",
  61     "Windows-1251",
  62     "KOI8-R",
  63     "KOI8-U",
  64     "ISO-8859-6",
  65     "Windows-1256",
  66     "ISO-8859-7",
  67     "Windows-1253",
  68     "ISO-8859-8",
  69     "Windows-1255",
  70     "ISO-8859-9",
  71     "Windows-1254",
  72     "ISO-8859-11",
  73     "Windows-874",
  74     "ISO-8859-13",
  75     "Windows-1257",
  76     "ISO-8859-14",
  77     "ISO-8859-16",
  78     "ISO-2022-CN-EXT",
  79     "EUC-CN",
  80     "ISO-2022-JP-2",
  81     "EUC-JP",
  82     "Shift_JIS",
  83     "CP949",
  84     "ISO-2022-KR",
  85     "Big5",
  86     "ISO-2022-TW",
  87     "Big5-HKSCS",
  88     "VISCII",
  89     "Windows-1258",
  90 };
  91
  92 static const char *const ppsz_encoding_names[] = {
  93     /* xgettext:
  94       The character encoding name in parenthesis corresponds to that used for
  95       the GetACP translation. "Windows-1252" applies to Western European
  96       languages using the Latin alphabet. */
  97     N_("Default (Windows-1252)"),
  98     N_("System codeset"),
  99     N_("Universal (UTF-8)"),
 100     N_("Universal (UTF-16)"),
 101     N_("Universal (big endian UTF-16)"),
 102     N_("Universal (little endian UTF-16)"),
 103     N_("Universal, Chinese (GB18030)"),
 104
 105   /* ISO 8859 and the likes */
 106     /* 1 */
 107     N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
 108     N_("Western European (Windows-1252)"),
 109     N_("Western European (IBM 00850)"),
 110     /* 2 */
 111     N_("Eastern European (Latin-2)"),
 112     N_("Eastern European (Windows-1250)"),
 113     /* 3 */
 114     N_("Esperanto (Latin-3)"),
 115     /* 4 */
 116     N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
 117     /* 5 */
 118     N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
 119     N_("Russian (KOI8-R)"),
 120     N_("Ukrainian (KOI8-U)"),
 121     /* 6 */
 122     N_("Arabic (ISO 8859-6)"),
 123     N_("Arabic (Windows-1256)"),
 124     /* 7 */
 125     N_("Greek (ISO 8859-7)"),
 126     N_("Greek (Windows-1253)"),
 127     /* 8 */
 128     N_("Hebrew (ISO 8859-8)"),
 129     N_("Hebrew (Windows-1255)"),
 130     /* 9 */
 131     N_("Turkish (ISO 8859-9)"),
 132     N_("Turkish (Windows-1254)"),
 133     /* 10 -> 4 */
 134     /* 11 */
 135     N_("Thai (TIS 620-2533/ISO 8859-11)"),
 136     N_("Thai (Windows-874)"),
 137     /* 13 */
 138     N_("Baltic (Latin-7)"),
 139     N_("Baltic (Windows-1257)"),
 140     /* 12 -> /dev/null */
 141     /* 14 */
 142     N_("Celtic (Latin-8)"),
 143     /* 15 -> 1 */
 144     /* 16 */
 145     N_("South-Eastern European (Latin-10)"),
 146   /* CJK families */
 147     N_("Simplified Chinese (ISO-2022-CN-EXT)"),
 148     N_("Simplified Chinese Unix (EUC-CN)"),
 149     N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
 150     N_("Japanese Unix (EUC-JP)"),
 151     N_("Japanese (Shift JIS)"),
 152     N_("Korean (EUC-KR/CP949)"),
 153     N_("Korean (ISO-2022-KR)"),
 154     N_("Traditional Chinese (Big5)"),
 155     N_("Traditional Chinese Unix (EUC-TW)"),
 156     N_("Hong-Kong Supplementary (HKSCS)"),
 157   /* Other */
 158     N_("Vietnamese (VISCII)"),
 159     N_("Vietnamese (Windows-1258)"),
 160 };
 161
 162 static const int  pi_justification[] = { 0, 1, 2 };
 163 static const char *const ppsz_justification_text[] = {
 164     N_("Center"),N_("Left"),N_("Right")};
 165
 166 #define ENCODING_TEXT N_("Subtitle text encoding")
 167 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 168 #define ALIGN_TEXT N_("Subtitle justification")
 169 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 170 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitle autodetection")
 171 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 172             "UTF-8 encoding within subtitle files.")
 173 #define FORMAT_TEXT N_("Formatted Subtitles")
 174 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 175  "VLC partly implements this, but you can choose to disable all formatting.")
 176
 177 static int  OpenDecoder   ( vlc_object_t * );
 178 static void CloseDecoder  ( vlc_object_t * );
 179
 180 vlc_module_begin ()
 181     set_shortname( N_("Subtitles"))
 182     set_description( N_("Text subtitle decoder") )
 183     set_capability( "decoder", 50 )
 184     set_callbacks( OpenDecoder, CloseDecoder )
 185     set_category( CAT_INPUT )
 186     set_subcategory( SUBCAT_INPUT_SCODEC )
 187
 188     add_integer( "subsdec-align", 0, ALIGN_TEXT, ALIGN_LONGTEXT,
 189                  false )
 190         change_integer_list( pi_justification, ppsz_justification_text )
 191     add_string( "subsdec-encoding", "",
 192                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 193         change_string_list( ppsz_encodings, ppsz_encoding_names )
 194     add_bool( "subsdec-autodetect-utf8", true,
 195               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 196     add_bool( "subsdec-formatted", true, FORMAT_TEXT, FORMAT_LONGTEXT,
 197                  false )
 198 vlc_module_end ()
 199
 200 /*****************************************************************************
 201  * Local prototypes
 202  *****************************************************************************/
 203 #define NO_BREAKING_SPACE  "&#160;"
 204
 205 struct decoder_sys_t
 206 {
 207     int                 i_align;          /* Subtitles alignment on the vout */
 208
 209     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
 210     bool                b_autodetect_utf8;
 211 };
 212
 213
 214 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
 215 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
 216 static char           *StripTags      ( char * );
 217 static char           *CreateHtmlSubtitle( int *pi_align, char * );
 218
 219 /*****************************************************************************
 220  * OpenDecoder: probe the decoder and return score
 221  *****************************************************************************
 222  * Tries to launch a decoder and return score so that the interface is able
 223  * to chose.
 224  *****************************************************************************/
 225 static int OpenDecoder( vlc_object_t *p_this )
 226 {
 227     decoder_t     *p_dec = (decoder_t*)p_this;
 228     decoder_sys_t *p_sys;
 229
 230     switch( p_dec->fmt_in.i_codec )
 231     {
 232         case VLC_CODEC_SUBT:
 233         case VLC_CODEC_ITU_T140:
 234             break;
 235         default:
 236             return VLC_EGENERIC;
 237     }
 238
 239     p_dec->pf_decode_sub = DecodeBlock;
 240     p_dec->fmt_out.i_cat = SPU_ES;
 241     p_dec->fmt_out.i_codec = 0;
 242
 243     /* Allocate the memory needed to store the decoder's structure */
 244     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 245     if( p_sys == NULL )
 246         return VLC_ENOMEM;
 247
 248     /* init of p_sys */
 249     p_sys->i_align = 0;
 250     p_sys->iconv_handle = (vlc_iconv_t)-1;
 251     p_sys->b_autodetect_utf8 = false;
 252
 253     const char *encoding;
 254     char *var = NULL;
 255
 256     /* First try demux-specified encoding */
 257     if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
 258         encoding = "UTF-8"; /* IUT T.140 is always using UTF-8 */
 259     else
 260     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 261     {
 262         encoding = p_dec->fmt_in.subs.psz_encoding;
 263         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 264                  encoding);
 265     }
 266     else
 267     {
 268         /* Second, try configured encoding */
 269         if ((var = var_InheritString (p_dec, "subsdec-encoding")) != NULL)
 270         {
 271             msg_Dbg (p_dec, "trying configured character encoding: %s", var);
 272             if (!strcmp (var, "system"))
 273             {
 274                 free (var);
 275                 var = NULL;
 276                 encoding = "";
 277                 /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 278             }
 279             else
 280                 encoding = var;
 281         }
 282         else
 283         /* Third, try "local" encoding */
 284         {
 285         /* xgettext:
 286            The Windows ANSI code page most commonly used for this language.
 287            VLC uses this as a guess of the subtitle files character set
 288            (if UTF-8 and UTF-16 autodetection fails).
 289            Western European languages normally use "CP1252", which is a
 290            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 291            Other scripts use other code pages.
 292
 293            This MUST be a valid iconv character set. If unsure, please refer
 294            the VideoLAN translators mailing list. */
 295             encoding = vlc_pgettext("GetACP", "CP1252");
 296             msg_Dbg (p_dec, "trying default character encoding: %s", encoding);
 297         }
 298
 299         /* Check UTF-8 autodetection */
 300         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 301         {
 302             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 303             p_sys->b_autodetect_utf8 = true;
 304         }
 305     }
 306
 307     if (strcasecmp (encoding, "UTF-8") && strcasecmp (encoding, "utf8"))
 308     {
 309         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", encoding);
 310         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 311             msg_Err (p_dec, "cannot convert from %s: %m", encoding);
 312     }
 313     free (var);
 314
 315     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 316
 317     return VLC_SUCCESS;
 318 }
 319
 320 /****************************************************************************
 321  * DecodeBlock: the whole thing
 322  ****************************************************************************
 323  * This function must be fed with complete subtitles units.
 324  ****************************************************************************/
 325 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 326 {
 327     subpicture_t *p_spu;
 328     block_t *p_block;
 329
 330     if( !pp_block || *pp_block == NULL )
 331         return NULL;
 332
 333     p_block = *pp_block;
 334     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 335     {
 336         block_Release( p_block );
 337         return NULL;
 338     }
 339
 340     p_spu = ParseText( p_dec, p_block );
 341
 342     block_Release( p_block );
 343     *pp_block = NULL;
 344
 345     return p_spu;
 346 }
 347
 348 /*****************************************************************************
 349  * CloseDecoder: clean up the decoder
 350  *****************************************************************************/
 351 static void CloseDecoder( vlc_object_t *p_this )
 352 {
 353     decoder_t *p_dec = (decoder_t *)p_this;
 354     decoder_sys_t *p_sys = p_dec->p_sys;
 355
 356     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 357         vlc_iconv_close( p_sys->iconv_handle );
 358
 359     free( p_sys );
 360 }
 361
 362 /*****************************************************************************
 363  * ParseText: parse an text subtitle packet and send it to the video output
 364  *****************************************************************************/
 365 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 366 {
 367     decoder_sys_t *p_sys = p_dec->p_sys;
 368     subpicture_t *p_spu = NULL;
 369     char *psz_subtitle = NULL;
 370
 371     /* We cannot display a subpicture with no date */
 372     if( p_block->i_pts <= VLC_TS_INVALID )
 373     {
 374         msg_Warn( p_dec, "subtitle without a date" );
 375         return NULL;
 376     }
 377
 378     /* Check validity of packet data */
 379     /* An "empty" line containing only \0 can be used to force
 380        and ephemer picture from the screen */
 381     if( p_block->i_buffer < 1 )
 382     {
 383         msg_Warn( p_dec, "no subtitle data" );
 384         return NULL;
 385     }
 386
 387     /* Should be resiliant against bad subtitles */
 388     psz_subtitle = malloc( p_block->i_buffer + 1 );
 389     if( psz_subtitle == NULL )
 390         return NULL;
 391     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 392     psz_subtitle[p_block->i_buffer] = '\0';
 393
 394     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 395     {
 396         if (EnsureUTF8( psz_subtitle ) == NULL)
 397         {
 398             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 399                      "Try manually setting a character-encoding "
 400                      "before you open the file." );
 401         }
 402     }
 403     else
 404     {
 405
 406         if( p_sys->b_autodetect_utf8 )
 407         {
 408             if( IsUTF8( psz_subtitle ) == NULL )
 409             {
 410                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 411                          "disabling UTF-8 subtitles autodetection" );
 412                 p_sys->b_autodetect_utf8 = false;
 413             }
 414         }
 415
 416         if( !p_sys->b_autodetect_utf8 )
 417         {
 418             size_t inbytes_left = strlen( psz_subtitle );
 419             size_t outbytes_left = 6 * inbytes_left;
 420             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 421             char *psz_convert_buffer_out = psz_new_subtitle;
 422             const char *psz_convert_buffer_in = psz_subtitle;
 423
 424             size_t ret = vlc_iconv( p_sys->iconv_handle,
 425                                     &psz_convert_buffer_in, &inbytes_left,
 426                                     &psz_convert_buffer_out, &outbytes_left );
 427
 428             *psz_convert_buffer_out++ = '\0';
 429             free( psz_subtitle );
 430
 431             if( ( ret == (size_t)(-1) ) || inbytes_left )
 432             {
 433                 free( psz_new_subtitle );
 434                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 435                         "Try manually setting a character-encoding "
 436                                 "before you open the file." );
 437                 return NULL;
 438             }
 439
 440             psz_subtitle = realloc( psz_new_subtitle,
 441                                     psz_convert_buffer_out - psz_new_subtitle );
 442             if( !psz_subtitle )
 443                 psz_subtitle = psz_new_subtitle;
 444         }
 445     }
 446
 447     /* Create the subpicture unit */
 448     p_spu = decoder_NewSubpictureText( p_dec );
 449     if( !p_spu )
 450     {
 451         free( psz_subtitle );
 452         return NULL;
 453     }
 454     p_spu->i_start    = p_block->i_pts;
 455     p_spu->i_stop     = p_block->i_pts + p_block->i_length;
 456     p_spu->b_ephemer  = (p_block->i_length == 0);
 457     p_spu->b_absolute = false;
 458
 459     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 460
 461     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 462     p_spu_sys->text  = StripTags( psz_subtitle );
 463     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 464         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 465
 466     free( psz_subtitle );
 467
 468     return p_spu;
 469 }
 470
 471 /* Function now handles tags with attribute values, and tries
 472  * to deal with &' commands too. It no longer modifies the string
 473  * in place, so that the original text can be reused
 474  */
 475 static char *StripTags( char *psz_subtitle )
 476 {
 477     char *psz_text_start;
 478     char *psz_text;
 479
 480     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 481     if( !psz_text_start )
 482         return NULL;
 483
 484     while( *psz_subtitle )
 485     {
 486         if( *psz_subtitle == '<' )
 487         {
 488             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 489                 *psz_text++ = '\n';
 490
 491             psz_subtitle += strcspn( psz_subtitle, ">" );
 492         }
 493         else if( *psz_subtitle == '&' )
 494         {
 495             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 496             {
 497                 *psz_text++ = '<';
 498                 psz_subtitle += strcspn( psz_subtitle, ";" );
 499             }
 500             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 501             {
 502                 *psz_text++ = '>';
 503                 psz_subtitle += strcspn( psz_subtitle, ";" );
 504             }
 505             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 506             {
 507                 *psz_text++ = '&';
 508                 psz_subtitle += strcspn( psz_subtitle, ";" );
 509             }
 510             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 511             {
 512                 *psz_text++ = '\"';
 513                 psz_subtitle += strcspn( psz_subtitle, ";" );
 514             }
 515             else
 516             {
 517                 /* Assume it is just a normal ampersand */
 518                 *psz_text++ = '&';
 519             }
 520         }
 521         else
 522         {
 523             *psz_text++ = *psz_subtitle;
 524         }
 525
 526         /* Security fix: Account for the case where input ends early */
 527         if( *psz_subtitle == '\0' ) break;
 528
 529         psz_subtitle++;
 530     }
 531     *psz_text = '\0';
 532     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 533     if( psz ) psz_text_start = psz;
 534
 535     return psz_text_start;
 536 }
 537
 538 /* Try to respect any style tags present in the subtitle string. The main
 539  * problem here is a lack of adequate specs for the subtitle formats.
 540  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 541  * SAMI has a detailed spec, but extensive rework is needed in the demux
 542  * code to prevent all this style information being excised, as it presently
 543  * does.
 544  * That leaves the others - none of which were (I guess) originally intended
 545  * to be carrying style information. Over time people have used them that way.
 546  * In the absence of specifications from which to work, the tags supported
 547  * have been restricted to the simple set permitted by the USF DTD, ie. :
 548  *  Basic: <br>, <i>, <b>, <u>, <s>
 549  *  Extended: <font>
 550  *    Attributes: face
 551  *                family
 552  *                size
 553  *                color
 554  *                outline-color
 555  *                shadow-color
 556  *                outline-level
 557  *                shadow-level
 558  *                back-color
 559  *                alpha
 560  * There is also the further restriction that the subtitle be well-formed
 561  * as an XML entity, ie. the HTML sentence:
 562  *        <b><i>Bold and Italics</b></i>
 563  * doesn't qualify because the tags aren't nested one inside the other.
 564  * <text> tags are automatically added to the output to ensure
 565  * well-formedness.
 566  * If the text doesn't qualify for any reason, a NULL string is
 567  * returned, and the rendering engine will fall back to the
 568  * plain text version of the subtitle.
 569  */
 570 /* TODO: highly suboptimal, offset should be cached */
 571 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 572 {
 573     char *psz_html = *ppsz_html;
 574     if( psz_html == NULL )
 575         return;
 576
 577     const size_t i_offset = strlen(psz_html);
 578     const size_t i_len = strnlen(psz_text, i_max);
 579
 580     psz_html = realloc( psz_html, i_offset + i_len + 1 );
 581     if( psz_html != NULL )
 582     {
 583         memcpy( psz_html + i_offset, psz_text, i_len );
 584         psz_html[i_offset + i_len] = '\0';
 585     }
 586     else
 587         free( *ppsz_html );
 588     *ppsz_html = psz_html;
 589 }
 590
 591 static void HtmlPut( char **ppsz_html, const char *psz_text )
 592 {
 593     HtmlNPut( ppsz_html, psz_text, INT_MAX );
 594 }
 595
 596 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 597 {
 598     HtmlPut( ppsz_html, psz_text );
 599     *ppsz_subtitle += strlen(psz_text);
 600 }
 601
 602 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 603 {
 604     char *psz_tag = malloc( 1 );
 605     if( psz_tag == NULL )
 606         return NULL;
 607
 608     char *psz_html = malloc( 1 );
 609     if( psz_html == NULL )
 610     {
 611         free( psz_tag );
 612         return NULL;
 613     }
 614     psz_tag[0] = '\0';
 615     psz_html[0] = '\0';
 616
 617     bool b_has_align = false;
 618
 619     HtmlPut( &psz_html, "<text>" );
 620
 621     /* */
 622     while( *psz_subtitle )
 623     {
 624         if( *psz_subtitle == '\n' )
 625         {
 626             HtmlPut( &psz_html, "<br/>" );
 627             psz_subtitle++;
 628         }
 629         else if( *psz_subtitle == '<' )
 630         {
 631             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 632             {
 633                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 634             }
 635             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 636             {
 637                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 638                 HtmlPut( &psz_tag, "b" );
 639             }
 640             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 641             {
 642                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 643                 HtmlPut( &psz_tag, "i" );
 644             }
 645             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 646             {
 647                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 648                 HtmlPut( &psz_tag, "u" );
 649             }
 650             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 651             {
 652                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 653                 HtmlPut( &psz_tag, "s" );
 654             }
 655             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 656             {
 657                 const char *psz_attribs[] = { "face=", "family=", "size=",
 658                         "color=", "outline-color=", "shadow-color=",
 659                         "outline-level=", "shadow-level=", "back-color=",
 660                         "alpha=", NULL };
 661
 662                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 663                 HtmlPut( &psz_tag, "f" );
 664
 665                 while( *psz_subtitle != '>' )
 666                 {
 667                     int  k;
 668
 669                     /* <font       color= */
 670                     while (*psz_subtitle == ' ')
 671                         psz_subtitle++;
 672
 673                     for( k=0; psz_attribs[ k ]; k++ )
 674                     {
 675                         int i_len = strlen( psz_attribs[ k ] );
 676
 677                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 678                         {
 679                             /* */
 680                             HtmlPut( &psz_html, psz_attribs[k] );
 681                             psz_subtitle += i_len;
 682
 683                             /* <font       color=   red */
 684                             while (*psz_subtitle == ' ')
 685                                 psz_subtitle++;
 686
 687                             /* */
 688                             if( *psz_subtitle == '"' )
 689                             {
 690                                 psz_subtitle++;
 691                                 i_len = strcspn( psz_subtitle, "\"" );
 692                             }
 693                             else if( *psz_subtitle == '\'' )
 694                             {
 695                                 psz_subtitle++;
 696                                 i_len = strcspn( psz_subtitle, "'" );
 697                             }
 698                             else
 699                             {
 700                                 i_len = strcspn( psz_subtitle, " \t>" );
 701                             }
 702                             HtmlPut( &psz_html, "\"" );
 703                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 704                             HtmlPut( &psz_html, "\"" );
 705
 706                             psz_subtitle += i_len;
 707                             if( *psz_subtitle == '\"' || *psz_subtitle == '\'' )
 708                                 psz_subtitle++;
 709                             break;
 710                         }
 711                     }
 712                     if( psz_attribs[ k ] == NULL )
 713                     {
 714                         /* Jump over unrecognised tag */
 715                         int i_len = strcspn( psz_subtitle, "\"" );
 716                         if( psz_subtitle[i_len] == '\"' )
 717                         {
 718                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 719                             if( psz_subtitle[i_len] == '\"' )
 720                                 i_len++;
 721                         }
 722                         /* Not a tag, something else we do not understand */
 723                         if( i_len == 0 )
 724                             psz_subtitle++;
 725
 726                         psz_subtitle += i_len;
 727                     }
 728
 729                     HtmlNPut( &psz_html, psz_subtitle, strspn(psz_subtitle, " ") );
 730                 }
 731                 HtmlPut( &psz_html, ">" );
 732                 psz_subtitle++;
 733             }
 734             else if( !strncmp( psz_subtitle, "</", 2 ))
 735             {
 736                 bool   b_match     = false;
 737                 bool   b_ignore    = false;
 738                 int    i_len       = (psz_tag ? strlen(psz_tag) : 0) - 1;
 739                 char  *psz_lastTag = NULL;
 740
 741                 if( i_len >= 0 )
 742                 {
 743                     psz_lastTag = psz_tag + i_len;
 744                     i_len = 0;
 745
 746                     switch( *psz_lastTag )
 747                     {
 748                     case 'b':
 749                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 750                         i_len   = 4;
 751                         break;
 752                     case 'i':
 753                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 754                         i_len   = 4;
 755                         break;
 756                     case 'u':
 757                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 758                         i_len   = 4;
 759                         break;
 760                     case 's':
 761                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 762                         i_len   = 4;
 763                         break;
 764                     case 'f':
 765                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 766                         i_len   = 7;
 767                         break;
 768                     case 'I':
 769                         i_len = strcspn( psz_subtitle, ">" );
 770                         b_match = psz_subtitle[i_len] == '>';
 771                         b_ignore = true;
 772                         if( b_match )
 773                             i_len++;
 774                         break;
 775                     }
 776                 }
 777                 if( !b_match )
 778                 {
 779                     /* Not well formed -- kill everything */
 780                     free( psz_html );
 781                     psz_html = NULL;
 782                     break;
 783                 }
 784                 *psz_lastTag = '\0';
 785                 if( !b_ignore )
 786                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 787
 788                 psz_subtitle += i_len;
 789             }
 790             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 791                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 792             {
 793                 /* We have a single < */
 794                 HtmlPut( &psz_html, "&lt;" );
 795                 psz_subtitle++;
 796             }
 797             else
 798             {
 799                 /* We have an unknown tag or a single < */
 800
 801                 /* Search for the next tag or end of tag or end of string */
 802                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 803                 char *psz_closing = strstr( psz_subtitle, "/>" );
 804
 805                 if( psz_closing && psz_closing < psz_stop )
 806                 {
 807                     /* We have a self closed tag, remove it */
 808                     psz_subtitle = &psz_closing[2];
 809                 }
 810                 else if( *psz_stop == '>' )
 811                 {
 812                     char psz_match[256];
 813
 814                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 815                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 816
 817                     if( strstr( psz_subtitle, psz_match ) )
 818                     {
 819                         /* We have the closing tag, ignore it TODO */
 820                         psz_subtitle = &psz_stop[1];
 821                         HtmlPut( &psz_tag, "I" );
 822                     }
 823                     else
 824                     {
 825                         int i_len = psz_stop + 1 - psz_subtitle;
 826
 827                         /* Copy the whole data */
 828                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 829                         {
 830                             if( *psz_subtitle == '<' )
 831                                 HtmlPut( &psz_html, "&lt;" );
 832                             else if( *psz_subtitle == '>' )
 833                                 HtmlPut( &psz_html, "&gt;" );
 834                             else
 835                                 HtmlNPut( &psz_html, psz_subtitle, 1 );
 836                         }
 837                     }
 838                 }
 839                 else
 840                 {
 841                     /* We have a single < */
 842                     HtmlPut( &psz_html, "&lt;" );
 843                     psz_subtitle++;
 844                 }
 845             }
 846         }
 847         else if( *psz_subtitle == '&' )
 848         {
 849             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 850             {
 851                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 852             }
 853             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 854             {
 855                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 856             }
 857             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 858             {
 859                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 860             }
 861             else
 862             {
 863                 HtmlPut( &psz_html, "&amp;" );
 864                 psz_subtitle++;
 865             }
 866         }
 867         else if( *psz_subtitle == '>' )
 868         {
 869             HtmlPut( &psz_html, "&gt;" );
 870             psz_subtitle++;
 871         }
 872         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 873                  strchr( psz_subtitle, '}' ) )
 874         {
 875             /* Check for forced alignment */
 876             if( !b_has_align &&
 877                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 878             {
 879                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 880                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 881                 const int i_id = psz_subtitle[4] - '1';
 882
 883                 b_has_align = true;
 884                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 885             }
 886             /* TODO fr -> rotation */
 887
 888             /* Hide {\stupidity} */
 889             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 890         }
 891         else if( psz_subtitle[0] == '{' &&
 892                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 893                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 894         {
 895             // FIXME: We don't do difference between Y and y, and we should.
 896             if( psz_subtitle[3] == 'i' )
 897             {
 898                 HtmlPut( &psz_html, "<i>" );
 899                 HtmlPut( &psz_tag, "i" );
 900             }
 901             if( psz_subtitle[3] == 'b' )
 902             {
 903                 HtmlPut( &psz_html, "<b>" );
 904                 HtmlPut( &psz_tag, "b" );
 905             }
 906             if( psz_subtitle[3] == 'u' )
 907             {
 908                 HtmlPut( &psz_html, "<u>" );
 909                 HtmlPut( &psz_tag, "u" );
 910             }
 911             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 912         }
 913         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 914         {
 915             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 916             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 917         }
 918         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 919         {
 920             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 921             {
 922                 HtmlPut( &psz_html, "<br/>" );
 923                 psz_subtitle += 2;
 924             }
 925             else if( psz_subtitle[1] == 'h' )
 926             {
 927                 /* Non breakable space */
 928                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 929                 psz_subtitle += 2;
 930             }
 931             else
 932             {
 933                 HtmlPut( &psz_html, "\\" );
 934                 psz_subtitle++;
 935             }
 936         }
 937         else
 938         {
 939             HtmlNPut( &psz_html, psz_subtitle, 1 );
 940 #if 0
 941             if( *psz_html )
 942             {
 943                 /* Check for double whitespace */
 944 # error This test does not make sense.
 945                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 946                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 947                 {
 948                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 949                     psz_html--;
 950                 }
 951             }
 952 #endif
 953             psz_subtitle++;
 954         }
 955     }
 956
 957     while( psz_tag && *psz_tag )
 958     {
 959         /* */
 960         char *psz_last = &psz_tag[strlen(psz_tag)-1];
 961         switch( *psz_last )
 962         {
 963             case 'b':
 964                 HtmlPut( &psz_html, "</b>" );
 965                 break;
 966             case 'i':
 967                 HtmlPut( &psz_html, "</i>" );
 968                 break;
 969             case 'u':
 970                 HtmlPut( &psz_html, "</u>" );
 971                 break;
 972             case 's':
 973                 HtmlPut( &psz_html, "</s>" );
 974                 break;
 975             case 'f':
 976                 HtmlPut( &psz_html, "</font>" );
 977                 break;
 978             case 'I':
 979                 break;
 980         }
 981         *psz_last = '\0';
 982     }
 983     /* Close not well formed subtitle */
 984     HtmlPut( &psz_html, "</text>" );
 985
 986     free( psz_tag );
 987
 988     return psz_html;
 989 }