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 #include <errno.h>
  36
  37 #include <vlc_common.h>
  38 #include <vlc_plugin.h>
  39 #include <vlc_codec.h>
  40 #include <vlc_charset.h>
  41
  42 #include "substext.h"
  43
  44 /*****************************************************************************
  45  * Module descriptor.
  46  *****************************************************************************/
  47 static const char *const ppsz_encodings[] = {
  48     "",
  49     "system",
  50     "UTF-8",
  51     "UTF-16",
  52     "UTF-16BE",
  53     "UTF-16LE",
  54     "GB18030",
  55     "ISO-8859-15",
  56     "Windows-1252",
  57     "IBM850",
  58     "ISO-8859-2",
  59     "Windows-1250",
  60     "ISO-8859-3",
  61     "ISO-8859-10",
  62     "Windows-1251",
  63     "KOI8-R",
  64     "KOI8-U",
  65     "ISO-8859-6",
  66     "Windows-1256",
  67     "ISO-8859-7",
  68     "Windows-1253",
  69     "ISO-8859-8",
  70     "Windows-1255",
  71     "ISO-8859-9",
  72     "Windows-1254",
  73     "ISO-8859-11",
  74     "Windows-874",
  75     "ISO-8859-13",
  76     "Windows-1257",
  77     "ISO-8859-14",
  78     "ISO-8859-16",
  79     "ISO-2022-CN-EXT",
  80     "EUC-CN",
  81     "ISO-2022-JP-2",
  82     "EUC-JP",
  83     "Shift_JIS",
  84     "CP949",
  85     "ISO-2022-KR",
  86     "Big5",
  87     "ISO-2022-TW",
  88     "Big5-HKSCS",
  89     "VISCII",
  90     "Windows-1258",
  91 };
  92
  93 static const char *const ppsz_encoding_names[] = {
  94     /* xgettext:
  95       The character encoding name in parenthesis corresponds to that used for
  96       the GetACP translation. "Windows-1252" applies to Western European
  97       languages using the Latin alphabet. */
  98     N_("Default (Windows-1252)"),
  99     N_("System codeset"),
 100     N_("Universal (UTF-8)"),
 101     N_("Universal (UTF-16)"),
 102     N_("Universal (big endian UTF-16)"),
 103     N_("Universal (little endian UTF-16)"),
 104     N_("Universal, Chinese (GB18030)"),
 105
 106   /* ISO 8859 and the likes */
 107     /* 1 */
 108     N_("Western European (Latin-9)"), /* mostly superset of Latin-1 */
 109     N_("Western European (Windows-1252)"),
 110     N_("Western European (IBM 00850)"),
 111     /* 2 */
 112     N_("Eastern European (Latin-2)"),
 113     N_("Eastern European (Windows-1250)"),
 114     /* 3 */
 115     N_("Esperanto (Latin-3)"),
 116     /* 4 */
 117     N_("Nordic (Latin-6)"), /* Latin 6 supersedes Latin 4 */
 118     /* 5 */
 119     N_("Cyrillic (Windows-1251)"), /* ISO 8859-5 is not practically used */
 120     N_("Russian (KOI8-R)"),
 121     N_("Ukrainian (KOI8-U)"),
 122     /* 6 */
 123     N_("Arabic (ISO 8859-6)"),
 124     N_("Arabic (Windows-1256)"),
 125     /* 7 */
 126     N_("Greek (ISO 8859-7)"),
 127     N_("Greek (Windows-1253)"),
 128     /* 8 */
 129     N_("Hebrew (ISO 8859-8)"),
 130     N_("Hebrew (Windows-1255)"),
 131     /* 9 */
 132     N_("Turkish (ISO 8859-9)"),
 133     N_("Turkish (Windows-1254)"),
 134     /* 10 -> 4 */
 135     /* 11 */
 136     N_("Thai (TIS 620-2533/ISO 8859-11)"),
 137     N_("Thai (Windows-874)"),
 138     /* 13 */
 139     N_("Baltic (Latin-7)"),
 140     N_("Baltic (Windows-1257)"),
 141     /* 12 -> /dev/null */
 142     /* 14 */
 143     N_("Celtic (Latin-8)"),
 144     /* 15 -> 1 */
 145     /* 16 */
 146     N_("South-Eastern European (Latin-10)"),
 147   /* CJK families */
 148     N_("Simplified Chinese (ISO-2022-CN-EXT)"),
 149     N_("Simplified Chinese Unix (EUC-CN)"),
 150     N_("Japanese (7-bits JIS/ISO-2022-JP-2)"),
 151     N_("Japanese Unix (EUC-JP)"),
 152     N_("Japanese (Shift JIS)"),
 153     N_("Korean (EUC-KR/CP949)"),
 154     N_("Korean (ISO-2022-KR)"),
 155     N_("Traditional Chinese (Big5)"),
 156     N_("Traditional Chinese Unix (EUC-TW)"),
 157     N_("Hong-Kong Supplementary (HKSCS)"),
 158   /* Other */
 159     N_("Vietnamese (VISCII)"),
 160     N_("Vietnamese (Windows-1258)"),
 161 };
 162
 163 static const int  pi_justification[] = { 0, 1, 2 };
 164 static const char *const ppsz_justification_text[] = {
 165     N_("Center"),N_("Left"),N_("Right")};
 166
 167 #define ENCODING_TEXT N_("Subtitle text encoding")
 168 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 169 #define ALIGN_TEXT N_("Subtitle justification")
 170 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 171 #define AUTODETECT_UTF8_TEXT N_("UTF-8 subtitle autodetection")
 172 #define AUTODETECT_UTF8_LONGTEXT N_("This enables automatic detection of " \
 173             "UTF-8 encoding within subtitle files.")
 174 #define FORMAT_TEXT N_("Formatted Subtitles")
 175 #define FORMAT_LONGTEXT N_("Some subtitle formats allow for text formatting. " \
 176  "VLC partly implements this, but you can choose to disable all formatting.")
 177
 178 static int  OpenDecoder   ( vlc_object_t * );
 179 static void CloseDecoder  ( vlc_object_t * );
 180
 181 vlc_module_begin ()
 182     set_shortname( N_("Subtitles"))
 183     set_description( N_("Text subtitle decoder") )
 184     set_capability( "decoder", 50 )
 185     set_callbacks( OpenDecoder, CloseDecoder )
 186     set_category( CAT_INPUT )
 187     set_subcategory( SUBCAT_INPUT_SCODEC )
 188
 189     add_integer( "subsdec-align", 0, ALIGN_TEXT, ALIGN_LONGTEXT,
 190                  false )
 191         change_integer_list( pi_justification, ppsz_justification_text )
 192     add_string( "subsdec-encoding", "",
 193                 ENCODING_TEXT, ENCODING_LONGTEXT, false )
 194         change_string_list( ppsz_encodings, ppsz_encoding_names )
 195     add_bool( "subsdec-autodetect-utf8", true,
 196               AUTODETECT_UTF8_TEXT, AUTODETECT_UTF8_LONGTEXT, false )
 197     add_bool( "subsdec-formatted", true, FORMAT_TEXT, FORMAT_LONGTEXT,
 198                  false )
 199 vlc_module_end ()
 200
 201 /*****************************************************************************
 202  * Local prototypes
 203  *****************************************************************************/
 204 #define NO_BREAKING_SPACE  "&#160;"
 205
 206 struct decoder_sys_t
 207 {
 208     int                 i_align;          /* Subtitles alignment on the vout */
 209
 210     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
 211     bool                b_autodetect_utf8;
 212 };
 213
 214
 215 static subpicture_t   *DecodeBlock   ( decoder_t *, block_t ** );
 216 static subpicture_t   *ParseText     ( decoder_t *, block_t * );
 217 static char           *StripTags      ( char * );
 218 static char           *CreateHtmlSubtitle( int *pi_align, char * );
 219
 220 /*****************************************************************************
 221  * OpenDecoder: probe the decoder and return score
 222  *****************************************************************************
 223  * Tries to launch a decoder and return score so that the interface is able
 224  * to chose.
 225  *****************************************************************************/
 226 static int OpenDecoder( vlc_object_t *p_this )
 227 {
 228     decoder_t     *p_dec = (decoder_t*)p_this;
 229     decoder_sys_t *p_sys;
 230
 231     switch( p_dec->fmt_in.i_codec )
 232     {
 233         case VLC_CODEC_SUBT:
 234         case VLC_CODEC_ITU_T140:
 235             break;
 236         default:
 237             return VLC_EGENERIC;
 238     }
 239
 240     p_dec->pf_decode_sub = DecodeBlock;
 241     p_dec->fmt_out.i_cat = SPU_ES;
 242     p_dec->fmt_out.i_codec = 0;
 243
 244     /* Allocate the memory needed to store the decoder's structure */
 245     p_dec->p_sys = p_sys = calloc( 1, sizeof( *p_sys ) );
 246     if( p_sys == NULL )
 247         return VLC_ENOMEM;
 248
 249     /* init of p_sys */
 250     p_sys->i_align = 0;
 251     p_sys->iconv_handle = (vlc_iconv_t)-1;
 252     p_sys->b_autodetect_utf8 = false;
 253
 254     const char *encoding;
 255     char *var = NULL;
 256
 257     /* First try demux-specified encoding */
 258     if( p_dec->fmt_in.i_codec == VLC_CODEC_ITU_T140 )
 259         encoding = "UTF-8"; /* IUT T.140 is always using UTF-8 */
 260     else
 261     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 262     {
 263         encoding = p_dec->fmt_in.subs.psz_encoding;
 264         msg_Dbg (p_dec, "trying demuxer-specified character encoding: %s",
 265                  encoding);
 266     }
 267     else
 268     {
 269         /* Second, try configured encoding */
 270         if ((var = var_InheritString (p_dec, "subsdec-encoding")) != NULL)
 271         {
 272             msg_Dbg (p_dec, "trying configured character encoding: %s", var);
 273             if (!strcmp (var, "system"))
 274             {
 275                 free (var);
 276                 var = NULL;
 277                 encoding = "";
 278                 /* ^ iconv() treats "" as nl_langinfo(CODESET) */
 279             }
 280             else
 281                 encoding = var;
 282         }
 283         else
 284         /* Third, try "local" encoding */
 285         {
 286         /* xgettext:
 287            The Windows ANSI code page most commonly used for this language.
 288            VLC uses this as a guess of the subtitle files character set
 289            (if UTF-8 and UTF-16 autodetection fails).
 290            Western European languages normally use "CP1252", which is a
 291            Microsoft-variant of ISO 8859-1. That suits the Latin alphabet.
 292            Other scripts use other code pages.
 293
 294            This MUST be a valid iconv character set. If unsure, please refer
 295            the VideoLAN translators mailing list. */
 296             encoding = vlc_pgettext("GetACP", "CP1252");
 297             msg_Dbg (p_dec, "trying default character encoding: %s", encoding);
 298         }
 299
 300         /* Check UTF-8 autodetection */
 301         if (var_InheritBool (p_dec, "subsdec-autodetect-utf8"))
 302         {
 303             msg_Dbg (p_dec, "using automatic UTF-8 detection");
 304             p_sys->b_autodetect_utf8 = true;
 305         }
 306     }
 307
 308     if (strcasecmp (encoding, "UTF-8") && strcasecmp (encoding, "utf8"))
 309     {
 310         p_sys->iconv_handle = vlc_iconv_open ("UTF-8", encoding);
 311         if (p_sys->iconv_handle == (vlc_iconv_t)(-1))
 312             msg_Err (p_dec, "cannot convert from %s: %s", encoding,
 313                      vlc_strerror_c(errno));
 314     }
 315     free (var);
 316
 317     p_sys->i_align = var_InheritInteger( p_dec, "subsdec-align" );
 318
 319     return VLC_SUCCESS;
 320 }
 321
 322 /****************************************************************************
 323  * DecodeBlock: the whole thing
 324  ****************************************************************************
 325  * This function must be fed with complete subtitles units.
 326  ****************************************************************************/
 327 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 328 {
 329     subpicture_t *p_spu;
 330     block_t *p_block;
 331
 332     if( !pp_block || *pp_block == NULL )
 333         return NULL;
 334
 335     p_block = *pp_block;
 336     if( p_block->i_flags & (BLOCK_FLAG_DISCONTINUITY|BLOCK_FLAG_CORRUPTED) )
 337     {
 338         block_Release( p_block );
 339         return NULL;
 340     }
 341
 342     p_spu = ParseText( p_dec, p_block );
 343
 344     block_Release( p_block );
 345     *pp_block = NULL;
 346
 347     return p_spu;
 348 }
 349
 350 /*****************************************************************************
 351  * CloseDecoder: clean up the decoder
 352  *****************************************************************************/
 353 static void CloseDecoder( vlc_object_t *p_this )
 354 {
 355     decoder_t *p_dec = (decoder_t *)p_this;
 356     decoder_sys_t *p_sys = p_dec->p_sys;
 357
 358     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 359         vlc_iconv_close( p_sys->iconv_handle );
 360
 361     free( p_sys );
 362 }
 363
 364 /*****************************************************************************
 365  * ParseText: parse an text subtitle packet and send it to the video output
 366  *****************************************************************************/
 367 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 368 {
 369     decoder_sys_t *p_sys = p_dec->p_sys;
 370     subpicture_t *p_spu = NULL;
 371     char *psz_subtitle = NULL;
 372
 373     /* We cannot display a subpicture with no date */
 374     if( p_block->i_pts <= VLC_TS_INVALID )
 375     {
 376         msg_Warn( p_dec, "subtitle without a date" );
 377         return NULL;
 378     }
 379
 380     /* Check validity of packet data */
 381     /* An "empty" line containing only \0 can be used to force
 382        and ephemer picture from the screen */
 383     if( p_block->i_buffer < 1 )
 384     {
 385         msg_Warn( p_dec, "no subtitle data" );
 386         return NULL;
 387     }
 388
 389     /* Should be resiliant against bad subtitles */
 390     psz_subtitle = malloc( p_block->i_buffer + 1 );
 391     if( psz_subtitle == NULL )
 392         return NULL;
 393     memcpy( psz_subtitle, p_block->p_buffer, p_block->i_buffer );
 394     psz_subtitle[p_block->i_buffer] = '\0';
 395
 396     if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 397     {
 398         if (EnsureUTF8( psz_subtitle ) == NULL)
 399         {
 400             msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 401                      "Try manually setting a character-encoding "
 402                      "before you open the file." );
 403         }
 404     }
 405     else
 406     {
 407
 408         if( p_sys->b_autodetect_utf8 )
 409         {
 410             if( IsUTF8( psz_subtitle ) == NULL )
 411             {
 412                 msg_Dbg( p_dec, "invalid UTF-8 sequence: "
 413                          "disabling UTF-8 subtitles autodetection" );
 414                 p_sys->b_autodetect_utf8 = false;
 415             }
 416         }
 417
 418         if( !p_sys->b_autodetect_utf8 )
 419         {
 420             size_t inbytes_left = strlen( psz_subtitle );
 421             size_t outbytes_left = 6 * inbytes_left;
 422             char *psz_new_subtitle = xmalloc( outbytes_left + 1 );
 423             char *psz_convert_buffer_out = psz_new_subtitle;
 424             const char *psz_convert_buffer_in = psz_subtitle;
 425
 426             size_t ret = vlc_iconv( p_sys->iconv_handle,
 427                                     &psz_convert_buffer_in, &inbytes_left,
 428                                     &psz_convert_buffer_out, &outbytes_left );
 429
 430             *psz_convert_buffer_out++ = '\0';
 431             free( psz_subtitle );
 432
 433             if( ( ret == (size_t)(-1) ) || inbytes_left )
 434             {
 435                 free( psz_new_subtitle );
 436                 msg_Err( p_dec, "failed to convert subtitle encoding.\n"
 437                         "Try manually setting a character-encoding "
 438                                 "before you open the file." );
 439                 return NULL;
 440             }
 441
 442             psz_subtitle = realloc( psz_new_subtitle,
 443                                     psz_convert_buffer_out - psz_new_subtitle );
 444             if( !psz_subtitle )
 445                 psz_subtitle = psz_new_subtitle;
 446         }
 447     }
 448
 449     /* Create the subpicture unit */
 450     p_spu = decoder_NewSubpictureText( p_dec );
 451     if( !p_spu )
 452     {
 453         free( psz_subtitle );
 454         return NULL;
 455     }
 456     p_spu->i_start    = p_block->i_pts;
 457     p_spu->i_stop     = p_block->i_pts + p_block->i_length;
 458     p_spu->b_ephemer  = (p_block->i_length == 0);
 459     p_spu->b_absolute = false;
 460
 461     subpicture_updater_sys_t *p_spu_sys = p_spu->updater.p_sys;
 462
 463     p_spu_sys->align = SUBPICTURE_ALIGN_BOTTOM | p_sys->i_align;
 464     p_spu_sys->text  = StripTags( psz_subtitle );
 465     if( var_InheritBool( p_dec, "subsdec-formatted" ) )
 466         p_spu_sys->html = CreateHtmlSubtitle( &p_spu_sys->align, psz_subtitle );
 467
 468     free( psz_subtitle );
 469
 470     return p_spu;
 471 }
 472
 473 /* Function now handles tags with attribute values, and tries
 474  * to deal with &' commands too. It no longer modifies the string
 475  * in place, so that the original text can be reused
 476  */
 477 static char *StripTags( char *psz_subtitle )
 478 {
 479     char *psz_text_start;
 480     char *psz_text;
 481
 482     psz_text = psz_text_start = malloc( strlen( psz_subtitle ) + 1 );
 483     if( !psz_text_start )
 484         return NULL;
 485
 486     while( *psz_subtitle )
 487     {
 488         if( *psz_subtitle == '<' )
 489         {
 490             if( strncasecmp( psz_subtitle, "<br/>", 5 ) == 0 )
 491                 *psz_text++ = '\n';
 492
 493             psz_subtitle += strcspn( psz_subtitle, ">" );
 494         }
 495         else if( *psz_subtitle == '&' )
 496         {
 497             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 498             {
 499                 *psz_text++ = '<';
 500                 psz_subtitle += strcspn( psz_subtitle, ";" );
 501             }
 502             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 503             {
 504                 *psz_text++ = '>';
 505                 psz_subtitle += strcspn( psz_subtitle, ";" );
 506             }
 507             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 508             {
 509                 *psz_text++ = '&';
 510                 psz_subtitle += strcspn( psz_subtitle, ";" );
 511             }
 512             else if( !strncasecmp( psz_subtitle, "&quot;", 6 ))
 513             {
 514                 *psz_text++ = '\"';
 515                 psz_subtitle += strcspn( psz_subtitle, ";" );
 516             }
 517             else
 518             {
 519                 /* Assume it is just a normal ampersand */
 520                 *psz_text++ = '&';
 521             }
 522         }
 523         else
 524         {
 525             *psz_text++ = *psz_subtitle;
 526         }
 527
 528         /* Security fix: Account for the case where input ends early */
 529         if( *psz_subtitle == '\0' ) break;
 530
 531         psz_subtitle++;
 532     }
 533     *psz_text = '\0';
 534     char *psz = realloc( psz_text_start, strlen( psz_text_start ) + 1 );
 535     if( psz ) psz_text_start = psz;
 536
 537     return psz_text_start;
 538 }
 539
 540 /* Try to respect any style tags present in the subtitle string. The main
 541  * problem here is a lack of adequate specs for the subtitle formats.
 542  * SSA/ASS and USF are both detail spec'ed -- but they are handled elsewhere.
 543  * SAMI has a detailed spec, but extensive rework is needed in the demux
 544  * code to prevent all this style information being excised, as it presently
 545  * does.
 546  * That leaves the others - none of which were (I guess) originally intended
 547  * to be carrying style information. Over time people have used them that way.
 548  * In the absence of specifications from which to work, the tags supported
 549  * have been restricted to the simple set permitted by the USF DTD, ie. :
 550  *  Basic: <br>, <i>, <b>, <u>, <s>
 551  *  Extended: <font>
 552  *    Attributes: face
 553  *                family
 554  *                size
 555  *                color
 556  *                outline-color
 557  *                shadow-color
 558  *                outline-level
 559  *                shadow-level
 560  *                back-color
 561  *                alpha
 562  * There is also the further restriction that the subtitle be well-formed
 563  * as an XML entity, ie. the HTML sentence:
 564  *        <b><i>Bold and Italics</b></i>
 565  * doesn't qualify because the tags aren't nested one inside the other.
 566  * <text> tags are automatically added to the output to ensure
 567  * well-formedness.
 568  * If the text doesn't qualify for any reason, a NULL string is
 569  * returned, and the rendering engine will fall back to the
 570  * plain text version of the subtitle.
 571  */
 572 /* TODO: highly suboptimal, offset should be cached */
 573 static void HtmlNPut( char **ppsz_html, const char *psz_text, int i_max )
 574 {
 575     char *psz_html = *ppsz_html;
 576     if( psz_html == NULL )
 577         return;
 578
 579     const size_t i_offset = strlen(psz_html);
 580     const size_t i_len = strnlen(psz_text, i_max);
 581
 582     psz_html = realloc( psz_html, i_offset + i_len + 1 );
 583     if( psz_html != NULL )
 584     {
 585         memcpy( psz_html + i_offset, psz_text, i_len );
 586         psz_html[i_offset + i_len] = '\0';
 587     }
 588     else
 589         free( *ppsz_html );
 590     *ppsz_html = psz_html;
 591 }
 592
 593 static void HtmlPut( char **ppsz_html, const char *psz_text )
 594 {
 595     HtmlNPut( ppsz_html, psz_text, INT_MAX );
 596 }
 597
 598 static void HtmlCopy( char **ppsz_html, char **ppsz_subtitle, const char *psz_text )
 599 {
 600     HtmlPut( ppsz_html, psz_text );
 601     *ppsz_subtitle += strlen(psz_text);
 602 }
 603
 604 static char *CreateHtmlSubtitle( int *pi_align, char *psz_subtitle )
 605 {
 606     char *psz_tag = malloc( 1 );
 607     if( psz_tag == NULL )
 608         return NULL;
 609
 610     char *psz_html = malloc( 1 );
 611     if( psz_html == NULL )
 612     {
 613         free( psz_tag );
 614         return NULL;
 615     }
 616     psz_tag[0] = '\0';
 617     psz_html[0] = '\0';
 618
 619     bool b_has_align = false;
 620
 621     HtmlPut( &psz_html, "<text>" );
 622
 623     /* */
 624     while( *psz_subtitle )
 625     {
 626         if( *psz_subtitle == '\n' )
 627         {
 628             HtmlPut( &psz_html, "<br/>" );
 629             psz_subtitle++;
 630         }
 631         else if( *psz_subtitle == '<' )
 632         {
 633             if( !strncasecmp( psz_subtitle, "<br/>", 5 ))
 634             {
 635                 HtmlCopy( &psz_html, &psz_subtitle, "<br/>" );
 636             }
 637             else if( !strncasecmp( psz_subtitle, "<b>", 3 ) )
 638             {
 639                 HtmlCopy( &psz_html, &psz_subtitle, "<b>" );
 640                 HtmlPut( &psz_tag, "b" );
 641             }
 642             else if( !strncasecmp( psz_subtitle, "<i>", 3 ) )
 643             {
 644                 HtmlCopy( &psz_html, &psz_subtitle, "<i>" );
 645                 HtmlPut( &psz_tag, "i" );
 646             }
 647             else if( !strncasecmp( psz_subtitle, "<u>", 3 ) )
 648             {
 649                 HtmlCopy( &psz_html, &psz_subtitle, "<u>" );
 650                 HtmlPut( &psz_tag, "u" );
 651             }
 652             else if( !strncasecmp( psz_subtitle, "<s>", 3 ) )
 653             {
 654                 HtmlCopy( &psz_html, &psz_subtitle, "<s>" );
 655                 HtmlPut( &psz_tag, "s" );
 656             }
 657             else if( !strncasecmp( psz_subtitle, "<font ", 6 ))
 658             {
 659                 const char *psz_attribs[] = { "face=", "family=", "size=",
 660                         "color=", "outline-color=", "shadow-color=",
 661                         "outline-level=", "shadow-level=", "back-color=",
 662                         "alpha=", NULL };
 663
 664                 HtmlCopy( &psz_html, &psz_subtitle, "<font " );
 665                 HtmlPut( &psz_tag, "f" );
 666
 667                 while( *psz_subtitle != '>' )
 668                 {
 669                     int  k;
 670
 671                     /* <font       color= */
 672                     while (*psz_subtitle == ' ')
 673                         psz_subtitle++;
 674
 675                     for( k=0; psz_attribs[ k ]; k++ )
 676                     {
 677                         int i_len = strlen( psz_attribs[ k ] );
 678
 679                         if( !strncasecmp( psz_subtitle, psz_attribs[k], i_len ) )
 680                         {
 681                             /* */
 682                             HtmlPut( &psz_html, psz_attribs[k] );
 683                             psz_subtitle += i_len;
 684
 685                             /* <font       color=   red */
 686                             while (*psz_subtitle == ' ')
 687                                 psz_subtitle++;
 688
 689                             /* */
 690                             if( *psz_subtitle == '"' )
 691                             {
 692                                 psz_subtitle++;
 693                                 i_len = strcspn( psz_subtitle, "\"" );
 694                             }
 695                             else if( *psz_subtitle == '\'' )
 696                             {
 697                                 psz_subtitle++;
 698                                 i_len = strcspn( psz_subtitle, "'" );
 699                             }
 700                             else
 701                             {
 702                                 i_len = strcspn( psz_subtitle, " \t>" );
 703                             }
 704                             HtmlPut( &psz_html, "\"" );
 705                             HtmlNPut( &psz_html, psz_subtitle, i_len );
 706                             HtmlPut( &psz_html, "\"" );
 707
 708                             psz_subtitle += i_len;
 709                             if( *psz_subtitle == '\"' || *psz_subtitle == '\'' )
 710                                 psz_subtitle++;
 711                             break;
 712                         }
 713                     }
 714                     if( psz_attribs[ k ] == NULL )
 715                     {
 716                         /* Jump over unrecognised tag */
 717                         int i_len = strcspn( psz_subtitle, "\"" );
 718                         if( psz_subtitle[i_len] == '\"' )
 719                         {
 720                             i_len += 1 + strcspn( &psz_subtitle[i_len + 1], "\"" );
 721                             if( psz_subtitle[i_len] == '\"' )
 722                                 i_len++;
 723                         }
 724                         /* Not a tag, something else we do not understand */
 725                         if( i_len == 0 )
 726                             psz_subtitle++;
 727
 728                         psz_subtitle += i_len;
 729                     }
 730
 731                     HtmlNPut( &psz_html, psz_subtitle, strspn(psz_subtitle, " ") );
 732                 }
 733                 HtmlPut( &psz_html, ">" );
 734                 psz_subtitle++;
 735             }
 736             else if( !strncmp( psz_subtitle, "</", 2 ))
 737             {
 738                 bool   b_match     = false;
 739                 bool   b_ignore    = false;
 740                 int    i_len       = (psz_tag ? strlen(psz_tag) : 0) - 1;
 741                 char  *psz_lastTag = NULL;
 742
 743                 if( i_len >= 0 )
 744                 {
 745                     psz_lastTag = psz_tag + i_len;
 746                     i_len = 0;
 747
 748                     switch( *psz_lastTag )
 749                     {
 750                     case 'b':
 751                         b_match = !strncasecmp( psz_subtitle, "</b>", 4 );
 752                         i_len   = 4;
 753                         break;
 754                     case 'i':
 755                         b_match = !strncasecmp( psz_subtitle, "</i>", 4 );
 756                         i_len   = 4;
 757                         break;
 758                     case 'u':
 759                         b_match = !strncasecmp( psz_subtitle, "</u>", 4 );
 760                         i_len   = 4;
 761                         break;
 762                     case 's':
 763                         b_match = !strncasecmp( psz_subtitle, "</s>", 4 );
 764                         i_len   = 4;
 765                         break;
 766                     case 'f':
 767                         b_match = !strncasecmp( psz_subtitle, "</font>", 7 );
 768                         i_len   = 7;
 769                         break;
 770                     case 'I':
 771                         i_len = strcspn( psz_subtitle, ">" );
 772                         b_match = psz_subtitle[i_len] == '>';
 773                         b_ignore = true;
 774                         if( b_match )
 775                             i_len++;
 776                         break;
 777                     }
 778                 }
 779                 if( !b_match )
 780                 {
 781                     /* Not well formed -- kill everything */
 782                     free( psz_html );
 783                     psz_html = NULL;
 784                     break;
 785                 }
 786                 *psz_lastTag = '\0';
 787                 if( !b_ignore )
 788                     HtmlNPut( &psz_html, psz_subtitle, i_len );
 789
 790                 psz_subtitle += i_len;
 791             }
 792             else if( ( psz_subtitle[1] < 'a' || psz_subtitle[1] > 'z' ) &&
 793                      ( psz_subtitle[1] < 'A' || psz_subtitle[1] > 'Z' ) )
 794             {
 795                 /* We have a single < */
 796                 HtmlPut( &psz_html, "&lt;" );
 797                 psz_subtitle++;
 798             }
 799             else
 800             {
 801                 /* We have an unknown tag or a single < */
 802
 803                 /* Search for the next tag or end of tag or end of string */
 804                 char *psz_stop = psz_subtitle + 1 + strcspn( &psz_subtitle[1], "<>" );
 805                 char *psz_closing = strstr( psz_subtitle, "/>" );
 806
 807                 if( psz_closing && psz_closing < psz_stop )
 808                 {
 809                     /* We have a self closed tag, remove it */
 810                     psz_subtitle = &psz_closing[2];
 811                 }
 812                 else if( *psz_stop == '>' )
 813                 {
 814                     char psz_match[256];
 815
 816                     snprintf( psz_match, sizeof(psz_match), "</%s", &psz_subtitle[1] );
 817                     psz_match[strcspn( psz_match, " \t>" )] = '\0';
 818
 819                     if( strstr( psz_subtitle, psz_match ) )
 820                     {
 821                         /* We have the closing tag, ignore it TODO */
 822                         psz_subtitle = &psz_stop[1];
 823                         HtmlPut( &psz_tag, "I" );
 824                     }
 825                     else
 826                     {
 827                         int i_len = psz_stop + 1 - psz_subtitle;
 828
 829                         /* Copy the whole data */
 830                         for( ; i_len > 0; i_len--, psz_subtitle++ )
 831                         {
 832                             if( *psz_subtitle == '<' )
 833                                 HtmlPut( &psz_html, "&lt;" );
 834                             else if( *psz_subtitle == '>' )
 835                                 HtmlPut( &psz_html, "&gt;" );
 836                             else
 837                                 HtmlNPut( &psz_html, psz_subtitle, 1 );
 838                         }
 839                     }
 840                 }
 841                 else
 842                 {
 843                     /* We have a single < */
 844                     HtmlPut( &psz_html, "&lt;" );
 845                     psz_subtitle++;
 846                 }
 847             }
 848         }
 849         else if( *psz_subtitle == '&' )
 850         {
 851             if( !strncasecmp( psz_subtitle, "&lt;", 4 ))
 852             {
 853                 HtmlCopy( &psz_html, &psz_subtitle, "&lt;" );
 854             }
 855             else if( !strncasecmp( psz_subtitle, "&gt;", 4 ))
 856             {
 857                 HtmlCopy( &psz_html, &psz_subtitle, "&gt;" );
 858             }
 859             else if( !strncasecmp( psz_subtitle, "&amp;", 5 ))
 860             {
 861                 HtmlCopy( &psz_html, &psz_subtitle, "&amp;" );
 862             }
 863             else
 864             {
 865                 HtmlPut( &psz_html, "&amp;" );
 866                 psz_subtitle++;
 867             }
 868         }
 869         else if( *psz_subtitle == '>' )
 870         {
 871             HtmlPut( &psz_html, "&gt;" );
 872             psz_subtitle++;
 873         }
 874         else if( psz_subtitle[0] == '{' && psz_subtitle[1] == '\\' &&
 875                  strchr( psz_subtitle, '}' ) )
 876         {
 877             /* Check for forced alignment */
 878             if( !b_has_align &&
 879                 !strncmp( psz_subtitle, "{\\an", 4 ) && psz_subtitle[4] >= '1' && psz_subtitle[4] <= '9' && psz_subtitle[5] == '}' )
 880             {
 881                 static const int pi_vertical[3] = { SUBPICTURE_ALIGN_BOTTOM, 0, SUBPICTURE_ALIGN_TOP };
 882                 static const int pi_horizontal[3] = { SUBPICTURE_ALIGN_LEFT, 0, SUBPICTURE_ALIGN_RIGHT };
 883                 const int i_id = psz_subtitle[4] - '1';
 884
 885                 b_has_align = true;
 886                 *pi_align = pi_vertical[i_id/3] | pi_horizontal[i_id%3];
 887             }
 888             /* TODO fr -> rotation */
 889
 890             /* Hide {\stupidity} */
 891             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 892         }
 893         else if( psz_subtitle[0] == '{' &&
 894                 ( psz_subtitle[1] == 'Y' || psz_subtitle[1] == 'y' )
 895                 && psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 896         {
 897             // FIXME: We don't do difference between Y and y, and we should.
 898             if( psz_subtitle[3] == 'i' )
 899             {
 900                 HtmlPut( &psz_html, "<i>" );
 901                 HtmlPut( &psz_tag, "i" );
 902             }
 903             if( psz_subtitle[3] == 'b' )
 904             {
 905                 HtmlPut( &psz_html, "<b>" );
 906                 HtmlPut( &psz_tag, "b" );
 907             }
 908             if( psz_subtitle[3] == 'u' )
 909             {
 910                 HtmlPut( &psz_html, "<u>" );
 911                 HtmlPut( &psz_tag, "u" );
 912             }
 913             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 914         }
 915         else if( psz_subtitle[0] == '{' &&  psz_subtitle[2] == ':' && strchr( psz_subtitle, '}' ) )
 916         {
 917             // Hide other {x:y} atrocities, like {c:$bbggrr} or {P:x}
 918             psz_subtitle = strchr( psz_subtitle, '}' ) + 1;
 919         }
 920         else if( psz_subtitle[0] == '\\' && psz_subtitle[1] )
 921         {
 922             if( psz_subtitle[1] == 'N' || psz_subtitle[1] == 'n' )
 923             {
 924                 HtmlPut( &psz_html, "<br/>" );
 925                 psz_subtitle += 2;
 926             }
 927             else if( psz_subtitle[1] == 'h' )
 928             {
 929                 /* Non breakable space */
 930                 HtmlPut( &psz_html, NO_BREAKING_SPACE );
 931                 psz_subtitle += 2;
 932             }
 933             else
 934             {
 935                 HtmlPut( &psz_html, "\\" );
 936                 psz_subtitle++;
 937             }
 938         }
 939         else
 940         {
 941             HtmlNPut( &psz_html, psz_subtitle, 1 );
 942 #if 0
 943             if( *psz_html )
 944             {
 945                 /* Check for double whitespace */
 946 # error This test does not make sense.
 947                 if( ( *psz_html == ' '  || *psz_html == '\t' ) &&
 948                     ( *(psz_html-1) == ' ' || *(psz_html-1) == '\t' ) )
 949                 {
 950                     HtmlPut( &psz_html, NO_BREAKING_SPACE );
 951                     psz_html--;
 952                 }
 953             }
 954 #endif
 955             psz_subtitle++;
 956         }
 957     }
 958
 959     while( psz_tag && *psz_tag )
 960     {
 961         /* */
 962         char *psz_last = &psz_tag[strlen(psz_tag)-1];
 963         switch( *psz_last )
 964         {
 965             case 'b':
 966                 HtmlPut( &psz_html, "</b>" );
 967                 break;
 968             case 'i':
 969                 HtmlPut( &psz_html, "</i>" );
 970                 break;
 971             case 'u':
 972                 HtmlPut( &psz_html, "</u>" );
 973                 break;
 974             case 's':
 975                 HtmlPut( &psz_html, "</s>" );
 976                 break;
 977             case 'f':
 978                 HtmlPut( &psz_html, "</font>" );
 979                 break;
 980             case 'I':
 981                 break;
 982         }
 983         *psz_last = '\0';
 984     }
 985     /* Close not well formed subtitle */
 986     HtmlPut( &psz_html, "</text>" );
 987
 988     free( psz_tag );
 989
 990     return psz_html;
 991 }