modules/codec/subsdec.c

   1 /*****************************************************************************
   2  * subsdec.c : text subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2006 the VideoLAN team
   5  * $Id$
   6  *
   7  * Authors: Gildas Bazin <gbazin@videolan.org>
   8  *          Samuel Hocevar <sam@zoy.org>
   9  *          Derk-Jan Hartman <hartman at videolan dot org>
  10  *          Bernie Purcell <bitmap@videolan.org>
  11  *
  12  * This program is free software; you can redistribute it and/or modify
  13  * it under the terms of the GNU General Public License as published by
  14  * the Free Software Foundation; either version 2 of the License, or
  15  * (at your option) any later version.
  16  *
  17  * This program is distributed in the hope that it will be useful,
  18  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  19  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  20  * GNU General Public License for more details.
  21  *
  22  * You should have received a copy of the GNU General Public License
  23  * along with this program; if not, write to the Free Software
  24  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  25  *****************************************************************************/
  26
  27 /*****************************************************************************
  28  * Preamble
  29  *****************************************************************************/
  30 #ifdef HAVE_CONFIG_H
  31 # include "config.h"
  32 #endif
  33
  34 #include <vlc_common.h>
  35 #include <vlc_plugin.h>
  36 #include <vlc_codec.h>
  37 #include <vlc_charset.h>
  38
  39 #include "substext.h"
  40
  41 /*****************************************************************************
  42  * Module descriptor.
  43  *****************************************************************************/
  44 static const char *const ppsz_encodings[] = {
  45     "",
  46     "system",
  47     "UTF-8",
  48     "UTF-16",
  49     "UTF-16BE",
  50     "UTF-16LE",
  51     "GB18030",
  52     "ISO-8859-15",
  53     "Windows-1252",
  54     "ISO-8859-2",
  55     "Windows-1250",
  56     "ISO-8859-3",
  57     "ISO-8859-10",
  58     "Windows-1251",
  59     "KOI8-R",
  60     "KOI8-U",
  61     "ISO-8859-6",
  62     "Windows-1256",
  63     "ISO-8859-7",
  64     "Windows-1253",
  65     "ISO-8859-8",
  66     "Windows-1255",
  67     "ISO-8859-9",
  68     "Windows-1254",
  69     "ISO-8859-11",
  70     "Windows-874",
  71     "ISO-8859-13",
  72     "Windows-1257",
  73     "ISO-8859-14",
  74     "ISO-8859-16",
  75     "ISO-2022-CN-EXT",
  76     "EUC-CN",
  77     "ISO-2022-JP-2",
  78     "EUC-JP",
  79     "Shift_JIS",
  80     "CP949",
  81     "ISO-2022-KR",
  82     "Big5",
  83     "ISO-2022-TW",
  84     "Big5-HKSCS",
  85     "VISCII",
  86     "Windows-1258",
  87 };
  88
  89 static const char *const ppsz_encoding_names[] = {
  90     /* xgettext:
  91       The character encoding name in parenthesis corresponds to that used for
  92       the GetACP translation. "Windows-1252" applies to Western European
  93       languages using the Latin alphabet. */
  94     N_("Default (Windows-1252)"),
  95     N_("System codeset"),
  96     N_("Universal (UTF-8)"),
  97     N_("Universal (UTF-16)"),
  98     N_("Universal (big endian UTF-16)"),
  99     N_("Universal (little endian UTF-16)"),
 100     N_("Universal, Chinese (GB18030)"),
 101
 102   /* ISO 8859 and the likes */
 103     /* 1 */
 104     N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
 105     N_("Western European (Windows-1252)"),
 106     /* 2 */
 107     N_("Eastern European (Latin-2)"),
 108     N_("Eastern European (Windows-1250)"),
 109     /* 3 */
 110     N_("Esperanto (Latin-3)"),
 111     /* 4 */
 112     N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
 113     /* 5 */
 114     N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
 115     N_("Russian (KOI8-R)"),
 116     N_("Ukrainian (KOI8-U)"),
 117     /* 6 */
 118     N_("Arabic (ISO 8859-6)"),
 119     N_("Arabic (Windows-1256)"),
 120     /* 7 */
 121     N_("Greek (ISO 8859-7)"),
 122     N_("Greek (Windows-1253)"),
 123     /* 8 */
 124     N_("Hebrew (ISO 8859-8)"),
 125     N_("Hebrew (Windows-1255)"),
 126     /* 9 */
 127     N_("Turkish (ISO 8859-9)"),
 128     N_("Turkish (Windows-1254)"),
 129     /* 10 -> 4 */
 130     /* 11 */
 131     N_("Thai (TIS 620-2533/ISO 8859-11)"),
 132     N_("Thai (Windows-874)"),
 133     /* 13 */
 134     N_("Baltic (Latin-7)"),
 135     N_("Baltic (Windows-1257)"),
 136     /* 12 -> /dev/null */
 137     /* 14 */
 138     N_("Celtic (Latin-8)"),
 139     /* 15 -> 1 */
 140     /* 16 */
 141     N_("South-Eastern European (Latin-10)"),
 142   /* CJK families */
 143     N_("Simplified Chinese (ISO-2022-CN-EXT)"),
 144     N_("Simplified Chinese Unix (EUC-CN)"),
 145     N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
 146     N_("Japanese Unix (EUC-JP)"),
 147     N_("Japanese (Shift JIS)"),
 148     N_("Korean (EUC-KR/CP949)"),
 149     N_("Korean (ISO-2022-KR)"),
 150     N_("Traditional Chinese (Big5)"),
 151     N_("Traditional Chinese Unix (EUC-TW)"),
 152     N_("Hong-Kong Supplementary (HKSCS)"),
 153   /* Other */
 154     N_("Vietnamese (VISCII)"),
 155     N_("Vietnamese (Windows-1258)"),
 156 };
 157
 158 static const int  pi_justification[] = { 0, 1, 2 };
 159 static const char *const ppsz_justification_text[] = {
 160     N_("Center"),N_("Left"),N_("Right")};
 161
 162 #define ENCODING_TEXT N_("Subtitles text encoding")
 163 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 164 #define ALIGN_TEXT N_("Subtitles justification")
 165 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 166 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitles autodetection")
 167 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 168             "UTF-8 encoding within subtitles files.")
 169 #define FORMAT_TEXT N_("Formatted Subtitles")
 170 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 171  "VLC partly implements this, but you can choose to disable all formatting.")
 172
 173 static int  OpenDecoder   ( vlc_object_t * );
 174 static void CloseDecoder  ( vlc_object_t * );
 175
 176 vlc_module_begin ()
 177     set_shortname( N_("Subtitles"))
 178     set_description( N_("Text subtitles decoder") )
 179     set_capability( "decoder", 50 )
 180     set_callbacks( OpenDecoder, CloseDecoder )
 181     set_category( CAT_INPUT )
 182     set_subcategory( SUBCAT_INPUT_SCODEC )
 183
 184     add_integer( "subsdec-align", 0, ALIGN_TEXT, ALIGN_LONGTEXT,
 185                  false )
 186         change_integer_list( pi_justification, ppsz_justification_text )
 187     add_string( "subsdec-encoding", "",
 188                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 189         change_string_list( ppsz_encodings, ppsz_encoding_names, 0 )
 190     add_bool( "subsdec-autodetect-utf8", true,
 191               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 192     add_bool( "subsdec-formatted", true, FORMAT_TEXT, FORMAT_LONGTEXT,
 193                  false )
 194 vlc_module_end ()
 195
 196 /*****************************************************************************
 197  * Local prototypes
 198  *****************************************************************************/
 199 #define NO_BREAKING_SPACE  "&#160;"
 200
 201 struct decoder_sys_t
 202 {
 203     int                 i_align;          /* Subtitles alignment on the vout */
 204
 205     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
 206     bool                b_autodetect_utf8;
 207 };
 208
 209
 210 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
 211 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
 212 static char           *StripTags      ( char * );
 213 static char           *CreateHtmlSubtitle( int *pi_align, char * );
 214
 215 /*****************************************************************************
 216  * OpenDecoder: probe the decoder and return score
 217  *****************************************************************************
 218  * Tries to launch a decoder and return score so that the interface is able
 219  * to chose.
 220  *****************************************************************************/
 221 static int OpenDecoder( vlc_object_t *p_this )
 222 {
 223     decoder_t     *p_dec = (decoder_t*)p_this;
 224     decoder_sys_t *p_sys;
 225
 226     switch( p_dec->fmt_in.i_codec )
 227     {
 228         case VLC_CODEC_SUBT:
 229         case VLC_CODEC_ITU_T140:
 230             break;
 231         default:
 232             return VLC_EGENERIC;
 233     }
 234
 235     p_dec->pf_decode_sub = DecodeBlock;
 236     p_dec->fmt_out.i_cat = SPU_ES;
 237     p_dec->fmt_out.i_codec = 0;
 238
 239     /* Allocate the memory needed to store the decoder's structure */
 240     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 241     if( p_sys == NULL )
 242         return VLC_ENOMEM;
 243
 244     /* init of p_sys */
 245     p_sys->i_align = 0;
 246     p_sys->iconv_handle = (vlc_iconv_t)-1;
 247     p_sys->b_autodetect_utf8 = false;
 248
 249     char *psz_charset = NULL;
 250
 251     /* First try demux-specified encoding */
 252     if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
 253         psz_charset = strdup( "UTF-8" ); /* IUT T.140 is always using UTF-8 */
 254     else
 255     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 256     {
 257         psz_charset = strdup (p_dec->fmt_in.subs.psz_encoding);
 258         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 259                  p_dec->fmt_in.subs.psz_encoding ?
 260                  p_dec->fmt_in.subs.psz_encoding : "not specified");
 261     }
 262
 263     /* Second, try configured encoding */
 264     if (psz_charset == NULL)
 265     {
 266         psz_charset = var_InheritString (p_dec, "subsdec-encoding");
 267         msg_Dbg (p_dec, "trying configured character encoding: %s",
 268                  psz_charset ? psz_charset : "not specified");
 269         if (psz_charset != NULL && !strcmp (psz_charset, "system"))
 270         {
 271             free (psz_charset);
 272             psz_charset = strdup ("");
 273             /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 274         }
 275     }
 276
 277     /* Third, try "local" encoding with optional UTF-8 autodetection */
 278     if (psz_charset == NULL)
 279     {
 280         /* xgettext:
 281            The Windows ANSI code page most commonly used for this language.
 282            VLC uses this as a guess of the subtitle files character set
 283            (if UTF-8 and UTF-16 autodetection fails).
 284            Western European languages normally use "CP1252", which is a
 285            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 286            Other scripts use other code pages.
 287
 288            This MUST be a valid iconv character set. If unsure, please refer
 289            the VideoLAN translators mailing list. */
 290         const char *acp = vlc_pgettext("GetACP", "CP1252");
 291
 292         psz_charset = strdup (acp);
 293         msg_Dbg (p_dec, "trying default character encoding: %s",
 294                  psz_charset ? psz_charset : "not specified");
 295
 296         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 297         {
 298             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 299             p_sys->b_autodetect_utf8 = true;
 300         }
 301     }
 302
 303     /* Forth, don't do character decoding, i.e. assume UTF-8 */
 304     if (psz_charset == NULL)
 305     {
 306         psz_charset = strdup ("UTF-8");
 307         msg_Dbg (p_dec, "using UTF-8 character encoding" );
 308     }
 309
 310     if ((psz_charset != NULL)
 311      && strcasecmp (psz_charset, "UTF-8")
 312      && strcasecmp (psz_charset, "utf8"))
 313     {
 314         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", psz_charset);
 315         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 316             msg_Err (p_dec, "cannot convert from %s: %m", psz_charset);
 317     }
 318     free (psz_charset);
 319
 320     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 321
 322     return VLC_SUCCESS;
 323 }
 324
 325 /****************************************************************************
 326  * DecodeBlock: the whole thing
 327  ****************************************************************************
 328  * This function must be fed with complete subtitles units.
 329  ****************************************************************************/
 330 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 331 {
 332     subpicture_t *p_spu;
 333     block_t *p_block;
 334
 335     if( !pp_block || *pp_block == NULL )
 336         return NULL;
 337
 338     p_block = *pp_block;
 339     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 340     {
 341         block_Release( p_block );
 342         return NULL;
 343     }
 344
 345     p_spu = ParseText( p_dec, p_block );
 346
 347     block_Release( p_block );
 348     *pp_block = NULL;
 349
 350     return p_spu;
 351 }
 352
 353 /*****************************************************************************
 354  * CloseDecoder: clean up the decoder
 355  *****************************************************************************/
 356 static void CloseDecoder( vlc_object_t *p_this )
 357 {
 358     decoder_t *p_dec = (decoder_t *)p_this;
 359     decoder_sys_t *p_sys = p_dec->p_sys;
 360
 361     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 362         vlc_iconv_close( p_sys->iconv_handle );
 363
 364     free( p_sys );
 365 }
 366
 367 /*****************************************************************************
 368  * ParseText: parse an text subtitle packet and send it to the video output
 369  *****************************************************************************/
 370 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 371 {
 372     decoder_sys_t *p_sys = p_dec->p_sys;
 373     subpicture_t *p_spu = NULL;
 374     char *psz_subtitle = NULL;
 375
 376     /* We cannot display a subpicture with no date */
 377     if( p_block->i_pts <= VLC_TS_INVALID )
 378     {
 379         msg_Warn( p_dec, "subtitle without a date" );
 380         return NULL;
 381     }
 382
 383     /* Check validity of packet data */
 384     /* An "empty" line containing only \0 can be used to force
 385        and ephemer picture from the screen */
 386     if( p_block->i_buffer < 1 )
 387     {
 388         msg_Warn( p_dec, "no subtitle data" );
 389         return NULL;
 390     }
 391
 392     /* Should be resiliant against bad subtitles */
 393     psz_subtitle = malloc( p_block->i_buffer + 1 );
 394     if( psz_subtitle == NULL )
 395         return NULL;
 396     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 397     psz_subtitle[p_block->i_buffer] = '\0';
 398
 399     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 400     {
 401         if (EnsureUTF8( psz_subtitle ) == NULL)
 402         {
 403             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 404                      "Try manually setting a character-encoding "
 405                      "before you open the file." );
 406         }
 407     }
 408     else
 409     {
 410
 411         if( p_sys->b_autodetect_utf8 )
 412         {
 413             if( IsUTF8( psz_subtitle ) == NULL )
 414             {
 415                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 416                          "disabling UTF-8 subtitles autodetection" );
 417                 p_sys->b_autodetect_utf8 = false;
 418             }
 419         }
 420
 421         if( !p_sys->b_autodetect_utf8 )
 422         {
 423             size_t inbytes_left = strlen( psz_subtitle );
 424             size_t outbytes_left = 6 * inbytes_left;
 425             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 426             char *psz_convert_buffer_out = psz_new_subtitle;
 427             const char *psz_convert_buffer_in = psz_subtitle;
 428
 429             size_t ret = vlc_iconv( p_sys->iconv_handle,
 430                                     &psz_convert_buffer_in, &inbytes_left,
 431                                     &psz_convert_buffer_out, &outbytes_left );
 432
 433             *psz_convert_buffer_out++ = '\0';
 434             free( psz_subtitle );
 435
 436             if( ( ret == (size_t)(-1) ) || inbytes_left )
 437             {
 438                 free( psz_new_subtitle );
 439                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 440                         "Try manually setting a character-encoding "
 441                                 "before you open the file." );
 442                 return NULL;
 443             }
 444
 445             psz_subtitle = realloc( psz_new_subtitle,
 446                                     psz_convert_buffer_out - psz_new_subtitle );
 447             if( !psz_subtitle )
 448                 psz_subtitle = psz_new_subtitle;
 449         }
 450     }
 451
 452     /* Create the subpicture unit */
 453     p_spu = decoder_NewSubpictureText( p_dec );
 454     if( !p_spu )
 455     {
 456         free( psz_subtitle );
 457         return NULL;
 458     }
 459     p_spu->i_start    = p_block->i_pts;
 460     p_spu->i_stop     = p_block->i_pts + p_block->i_length;
 461     p_spu->b_ephemer  = (p_block->i_length == 0);
 462     p_spu->b_absolute = false;
 463
 464     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 465
 466     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 467     p_spu_sys->text  = StripTags( psz_subtitle );
 468     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 469         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 470
 471     free( psz_subtitle );
 472
 473     return p_spu;
 474 }
 475
 476 /* Function now handles tags with attribute values, and tries
 477  * to deal with &' commands too. It no longer modifies the string
 478  * in place, so that the original text can be reused
 479  */
 480 static char *StripTags( char *psz_subtitle )
 481 {
 482     char *psz_text_start;
 483     char *psz_text;
 484
 485     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 486     if( !psz_text_start )
 487         return NULL;
 488
 489     while( *psz_subtitle )
 490     {
 491         if( *psz_subtitle == '<' )
 492         {
 493             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 494                 *psz_text++ = '\n';
 495
 496             psz_subtitle += strcspn( psz_subtitle, ">" );
 497         }
 498         else if( *psz_subtitle == '&' )
 499         {
 500             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 501             {
 502                 *psz_text++ = '<';
 503                 psz_subtitle += strcspn( psz_subtitle, ";" );
 504             }
 505             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 506             {
 507                 *psz_text++ = '>';
 508                 psz_subtitle += strcspn( psz_subtitle, ";" );
 509             }
 510             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 511             {
 512                 *psz_text++ = '&';
 513                 psz_subtitle += strcspn( psz_subtitle, ";" );
 514             }
 515             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 516             {
 517                 *psz_text++ = '\"';
 518                 psz_subtitle += strcspn( psz_subtitle, ";" );
 519             }
 520             else
 521             {
 522                 /* Assume it is just a normal ampersand */
 523                 *psz_text++ = '&';
 524             }
 525         }
 526         else
 527         {
 528             *psz_text++ = *psz_subtitle;
 529         }
 530
 531         /* Security fix: Account for the case where input ends early */
 532         if( *psz_subtitle == '\0' ) break;
 533
 534         psz_subtitle++;
 535     }
 536     *psz_text = '\0';
 537     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 538     if( psz ) psz_text_start = psz;
 539
 540     return psz_text_start;
 541 }
 542
 543 /* Try to respect any style tags present in the subtitle string. The main
 544  * problem here is a lack of adequate specs for the subtitle formats.
 545  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 546  * SAMI has a detailed spec, but extensive rework is needed in the demux
 547  * code to prevent all this style information being excised, as it presently
 548  * does.
 549  * That leaves the others - none of which were (I guess) originally intended
 550  * to be carrying style information. Over time people have used them that way.
 551  * In the absence of specifications from which to work, the tags supported
 552  * have been restricted to the simple set permitted by the USF DTD, ie. :
 553  *  Basic: <br>, <i>, <b>, <u>, <s>
 554  *  Extended: <font>
 555  *    Attributes: face
 556  *                family
 557  *                size
 558  *                color
 559  *                outline-color
 560  *                shadow-color
 561  *                outline-level
 562  *                shadow-level
 563  *                back-color
 564  *                alpha
 565  * There is also the further restriction that the subtitle be well-formed
 566  * as an XML entity, ie. the HTML sentence:
 567  *        <b><i>Bold and Italics</b></i>
 568  * doesn't qualify because the tags aren't nested one inside the other.
 569  * <text> tags are automatically added to the output to ensure
 570  * well-formedness.
 571  * If the text doesn't qualify for any reason, a NULL string is
 572  * returned, and the rendering engine will fall back to the
 573  * plain text version of the subtitle.
 574  */
 575 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 576 {
 577     const int i_len = strlen(psz_text);
 578
 579     strncpy( *ppsz_html, psz_text, i_max );
 580     *ppsz_html += __MIN(i_max,i_len);
 581 }
 582
 583 static void HtmlPut( char **ppsz_html, const char *psz_text )
 584 {
 585     strcpy( *ppsz_html, psz_text );
 586     *ppsz_html += strlen(psz_text);
 587 }
 588 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 589 {
 590     HtmlPut( ppsz_html, psz_text );
 591     *ppsz_subtitle += strlen(psz_text);
 592 }
 593
 594 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 595 {
 596     /* */
 597     char *psz_tag = malloc( ( strlen( psz_subtitle ) / 3 ) + 1 );
 598     if( !psz_tag )
 599         return NULL;
 600     psz_tag[ 0 ] = '\0';
 601
 602     /* */
 603     //Oo + 100 ???
 604     size_t i_buf_size = strlen( psz_subtitle ) + 100;
 605     char   *psz_html_start = malloc( i_buf_size );
 606     char   *psz_html = psz_html_start;
 607     if( psz_html_start == NULL )
 608     {
 609         free( psz_tag );
 610         return NULL;
 611     }
 612     psz_html[0] = '\0';
 613
 614     bool b_has_align = false;
 615
 616     HtmlPut( &psz_html, "<text>" );
 617
 618     /* */
 619     while( *psz_subtitle )
 620     {
 621         if( *psz_subtitle == '\n' )
 622         {
 623             HtmlPut( &psz_html, "<br/>" );
 624             psz_subtitle++;
 625         }
 626         else if( *psz_subtitle == '<' )
 627         {
 628             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 629             {
 630                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 631             }
 632             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 633             {
 634                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 635                 strcat( psz_tag, "b" );
 636             }
 637             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 638             {
 639                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 640                 strcat( psz_tag, "i" );
 641             }
 642             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 643             {
 644                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 645                 strcat( psz_tag, "u" );
 646             }
 647             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 648             {
 649                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 650                 strcat( psz_tag, "s" );
 651             }
 652             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 653             {
 654                 const char *psz_attribs[] = { "face=", "family=", "size=",
 655                         "color=", "outline-color=", "shadow-color=",
 656                         "outline-level=", "shadow-level=", "back-color=",
 657                         "alpha=", NULL };
 658
 659                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 660                 strcat( psz_tag, "f" );
 661
 662                 while( *psz_subtitle != '>' )
 663                 {
 664                     int  k;
 665
 666                     for( k=0; psz_attribs[ k ]; k++ )
 667                     {
 668                         int i_len = strlen( psz_attribs[ k ] );
 669
 670                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 671                         {
 672                             /* */
 673                             HtmlPut( &psz_html, psz_attribs[k] );
 674                             psz_subtitle += i_len;
 675
 676                             /* */
 677                             if( *psz_subtitle == '"' )
 678                             {
 679                                 psz_subtitle++;
 680                                 i_len = strcspn( psz_subtitle, "\"" );
 681                             }
 682                             else
 683                             {
 684                                 i_len = strcspn( psz_subtitle, " \t>" );
 685                             }
 686                             HtmlPut( &psz_html, "\"" );
 687                             if( !strcmp( psz_attribs[ k ], "color=" ) && *psz_subtitle >= '0' && *psz_subtitle <= '9' )
 688                                 HtmlPut( &psz_html, "#" );
 689                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 690                             HtmlPut( &psz_html, "\"" );
 691
 692                             psz_subtitle += i_len;
 693                             if( *psz_subtitle == '\"' )
 694                                 psz_subtitle++;
 695                             break;
 696                         }
 697                     }
 698                     if( psz_attribs[ k ] == NULL )
 699                     {
 700                         /* Jump over unrecognised tag */
 701                         int i_len = strcspn( psz_subtitle, "\"" );
 702                         if( psz_subtitle[i_len] == '\"' )
 703                         {
 704                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 705                             if( psz_subtitle[i_len] == '\"' )
 706                                 i_len++;
 707                         }
 708                         psz_subtitle += i_len;
 709                     }
 710                     while (*psz_subtitle == ' ')
 711                         *psz_html++ = *psz_subtitle++;
 712                 }
 713                 *psz_html++ = *psz_subtitle++;
 714             }
 715             else if( !strncmp( psz_subtitle, "</", 2 ))
 716             {
 717                 bool   b_match     = false;
 718                 bool   b_ignore    = false;
 719                 int    i_len       = strlen( psz_tag ) - 1;
 720                 char  *psz_lastTag = NULL;
 721
 722                 if( i_len >= 0 )
 723                 {
 724                     psz_lastTag = psz_tag + i_len;
 725                     i_len = 0;
 726
 727                     switch( *psz_lastTag )
 728                     {
 729                     case 'b':
 730                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 731                         i_len   = 4;
 732                         break;
 733                     case 'i':
 734                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 735                         i_len   = 4;
 736                         break;
 737                     case 'u':
 738                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 739                         i_len   = 4;
 740                         break;
 741                     case 's':
 742                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 743                         i_len   = 4;
 744                         break;
 745                     case 'f':
 746                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 747                         i_len   = 7;
 748                         break;
 749                     case 'I':
 750                         i_len = strcspn( psz_subtitle, ">" );
 751                         b_match = psz_subtitle[i_len] == '>';
 752                         b_ignore = true;
 753                         if( b_match )
 754                             i_len++;
 755                         break;
 756                     }
 757                 }
 758                 if( !b_match )
 759                 {
 760                     /* Not well formed -- kill everything */
 761                     free( psz_html_start );
 762                     psz_html_start = NULL;
 763                     break;
 764                 }
 765                 *psz_lastTag = '\0';
 766                 if( !b_ignore )
 767                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 768
 769                 psz_subtitle += i_len;
 770             }
 771             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 772                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 773             {
 774                 /* We have a single < */
 775                 HtmlPut( &psz_html, "&lt;" );
 776                 psz_subtitle++;
 777             }
 778             else
 779             {
 780                 /* We have an unknown tag or a single < */
 781
 782                 /* Search for the next tag or end of tag or end of string */
 783                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 784                 char *psz_closing = strstr( psz_subtitle, "/>" );
 785
 786                 if( psz_closing && psz_closing < psz_stop )
 787                 {
 788                     /* We have a self closed tag, remove it */
 789                     psz_subtitle = &psz_closing[2];
 790                 }
 791                 else if( *psz_stop == '>' )
 792                 {
 793                     char psz_match[256];
 794
 795                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 796                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 797
 798                     if( strstr( psz_subtitle, psz_match ) )
 799                     {
 800                         /* We have the closing tag, ignore it TODO */
 801                         psz_subtitle = &psz_stop[1];
 802                         strcat( psz_tag, "I" );
 803                     }
 804                     else
 805                     {
 806                         int i_len = psz_stop + 1 - psz_subtitle;
 807
 808                         /* Copy the whole data */
 809                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 810                         {
 811                             if( *psz_subtitle == '<' )
 812                                 HtmlPut( &psz_html, "&lt;" );
 813                             else if( *psz_subtitle == '>' )
 814                                 HtmlPut( &psz_html, "&gt;" );
 815                             else
 816                                 *psz_html++ = *psz_subtitle;
 817                         }
 818                     }
 819                 }
 820                 else
 821                 {
 822                     /* We have a single < */
 823                     HtmlPut( &psz_html, "&lt;" );
 824                     psz_subtitle++;
 825                 }
 826             }
 827         }
 828         else if( *psz_subtitle == '&' )
 829         {
 830             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 831             {
 832                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 833             }
 834             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 835             {
 836                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 837             }
 838             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 839             {
 840                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 841             }
 842             else
 843             {
 844                 HtmlPut( &psz_html, "&amp;" );
 845                 psz_subtitle++;
 846             }
 847         }
 848         else if( *psz_subtitle == '>' )
 849         {
 850             HtmlPut( &psz_html, "&gt;" );
 851             psz_subtitle++;
 852         }
 853         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 854                  strchr( psz_subtitle, '}' ) )
 855         {
 856             /* Check for forced alignment */
 857             if( !b_has_align &&
 858                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 859             {
 860                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 861                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 862                 const int i_id = psz_subtitle[4] - '1';
 863
 864                 b_has_align = true;
 865                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 866             }
 867             /* TODO fr -> rotation */
 868
 869             /* Hide {\stupidity} */
 870             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 871         }
 872         else if( psz_subtitle[0] == '{' &&
 873                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 874                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 875         {
 876             // FIXME: We don't do difference between Y and y, and we should.
 877             if( psz_subtitle[3] == 'i' )
 878             {
 879                 HtmlPut( &psz_html, "<i>" );
 880                 strcat( psz_tag, "i" );
 881             }
 882             if( psz_subtitle[3] == 'b' )
 883             {
 884                 HtmlPut( &psz_html, "<b>" );
 885                 strcat( psz_tag, "b" );
 886             }
 887             if( psz_subtitle[3] == 'u' )
 888             {
 889                 HtmlPut( &psz_html, "<u>" );
 890                 strcat( psz_tag, "u" );
 891             }
 892             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 893         }
 894         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 895         {
 896             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 897             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 898         }
 899         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 900         {
 901             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 902             {
 903                 HtmlPut( &psz_html, "<br/>" );
 904                 psz_subtitle += 2;
 905             }
 906             else if( psz_subtitle[1] == 'h' )
 907             {
 908                 /* Non breakable space */
 909                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 910                 psz_subtitle += 2;
 911             }
 912             else
 913             {
 914                 HtmlPut( &psz_html, "\\" );
 915                 psz_subtitle++;
 916             }
 917         }
 918         else
 919         {
 920             *psz_html = *psz_subtitle;
 921             if( psz_html > psz_html_start )
 922             {
 923                 /* Check for double whitespace */
 924                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 925                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 926                 {
 927                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 928                     psz_html--;
 929                 }
 930             }
 931             psz_html++;
 932             psz_subtitle++;
 933         }
 934
 935         if( ( size_t )( psz_html - psz_html_start ) > i_buf_size - 50 )
 936         {
 937             const int i_len = psz_html - psz_html_start;
 938
 939             i_buf_size += 200;
 940             char *psz_new = realloc( psz_html_start, i_buf_size );
 941             if( !psz_new )
 942                 break;
 943             psz_html_start = psz_new;
 944             psz_html = &psz_new[i_len];
 945         }
 946     }
 947     if( psz_html_start )
 948     {
 949         static const char *psz_text_close = "</text>";
 950         static const char *psz_tag_long = "/font>";
 951
 952         /* Realloc for closing tags and shrink memory */
 953         const size_t i_length = (size_t)( psz_html - psz_html_start );
 954
 955         const size_t i_size = i_length + strlen(psz_tag_long) * strlen(psz_tag) + strlen(psz_text_close) + 1;
 956         char *psz_new = realloc( psz_html_start, i_size );
 957         if( psz_new )
 958         {
 959             psz_html_start = psz_new;
 960             psz_html = &psz_new[i_length];
 961
 962             /* Close not well formed subtitle */
 963             while( *psz_tag )
 964             {
 965                 /* */
 966                 char *psz_last = &psz_tag[strlen(psz_tag)-1];
 967                 switch( *psz_last )
 968                 {
 969                 case 'b':
 970                     HtmlPut( &psz_html, "</b>" );
 971                     break;
 972                 case 'i':
 973                     HtmlPut( &psz_html, "</i>" );
 974                     break;
 975                 case 'u':
 976                     HtmlPut( &psz_html, "</u>" );
 977                     break;
 978                 case 's':
 979                     HtmlPut( &psz_html, "</s>" );
 980                     break;
 981                 case 'f':
 982                     HtmlPut( &psz_html, "/font>" );
 983                     break;
 984                 case 'I':
 985                     break;
 986                 }
 987
 988                 *psz_last = '\0';
 989             }
 990             HtmlPut( &psz_html, psz_text_close );
 991         }
 992     }
 993     free( psz_tag );
 994
 995     return psz_html_start;
 996 }
 997