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

   1 /*****************************************************************************
   2  * subsdec.c : text subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2001 VideoLAN
   5  * $Id$
   6  *
   7  * Authors: Gildas Bazin <gbazin@netcourrier.com>
   8  *          Samuel Hocevar <sam@zoy.org>
   9  *
  10  * This program is free software; you can redistribute it and/or modify
  11  * it under the terms of the GNU General Public License as published by
  12  * the Free Software Foundation; either version 2 of the License, or
  13  * (at your option) any later version.
  14  *
  15  * This program is distributed in the hope that it will be useful,
  16  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  17  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18  * GNU General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU General Public License
  21  * along with this program; if not, write to the Free Software
  22  * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
  23  *****************************************************************************/
  24
  25 /*****************************************************************************
  26  * Preamble
  27  *****************************************************************************/
  28 #include <vlc/vlc.h>
  29 #include <vlc/vout.h>
  30 #include <vlc/decoder.h>
  31
  32 #include "osd.h"
  33 #include "vlc_filter.h"
  34
  35 #if defined(HAVE_ICONV)
  36 #include <iconv.h>
  37 #endif
  38
  39 #include "charset.h"
  40
  41 /*****************************************************************************
  42  * decoder_sys_t : decoder descriptor
  43  *****************************************************************************/
  44 struct decoder_sys_t
  45 {
  46     int                 i_align;          /* Subtitles alignment on the vout */
  47
  48 #if defined(HAVE_ICONV)
  49     iconv_t             iconv_handle;            /* handle to iconv instance */
  50 #endif
  51
  52     filter_t *p_render;                              /* text renderer filter */
  53 };
  54
  55 /*****************************************************************************
  56  * Local prototypes
  57  *****************************************************************************/
  58 static int  OpenDecoder   ( vlc_object_t * );
  59 static void CloseDecoder  ( vlc_object_t * );
  60
  61 static subpicture_t *DecodeBlock   ( decoder_t *, block_t ** );
  62 static subpicture_t *ParseText     ( decoder_t *, block_t * );
  63 static void         StripTags      ( char * );
  64
  65 static subpicture_t *spu_new_buffer( filter_t * );
  66 static void spu_del_buffer( filter_t *, subpicture_t * );
  67
  68 #define DEFAULT_NAME "System Default"
  69
  70 /*****************************************************************************
  71  * Module descriptor.
  72  *****************************************************************************/
  73 #if defined(HAVE_ICONV)
  74 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  75     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  76     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  77     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  78     "ISO-8859-6", "CP1256", "MacArabic", "",
  79     "ISO-8859-7", "CP1253", "MacGreek", "",
  80     "ISO-8859-8", "CP1255", "MacHebrew", "",
  81     "ISO-8859-9", "CP1254", "MacTurkish", "",
  82     "ISO-8859-13", "CP1257", "",
  83     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  84     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  85     "ISO-2022-KR", "EUC-KR", "",
  86     "MacThai", "KOI8-T", "",
  87     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  88     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  89     "Macintosh", "",
  90     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  91     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  92     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  93     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  94     "HPROMAN8", "NEXTSTEP" };
  95 #endif
  96
  97 static int  pi_justification[] = { 0, 1, 2 };
  98 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  99
 100 #define ENCODING_TEXT N_("Subtitles text encoding")
 101 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 102 #define ALIGN_TEXT N_("Subtitles justification")
 103 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 104
 105 vlc_module_begin();
 106     set_description( _("text subtitles decoder") );
 107     set_capability( "decoder", 50 );
 108     set_callbacks( OpenDecoder, CloseDecoder );
 109
 110     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 111                  VLC_TRUE );
 112         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 113 #if defined(HAVE_ICONV)
 114     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 115                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 116         change_string_list( ppsz_encodings, 0, 0 );
 117 #endif
 118 vlc_module_end();
 119
 120 /*****************************************************************************
 121  * OpenDecoder: probe the decoder and return score
 122  *****************************************************************************
 123  * Tries to launch a decoder and return score so that the interface is able
 124  * to chose.
 125  *****************************************************************************/
 126 static int OpenDecoder( vlc_object_t *p_this )
 127 {
 128     decoder_t     *p_dec = (decoder_t*)p_this;
 129     decoder_sys_t *p_sys;
 130     vlc_value_t val;
 131
 132     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 133         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 134     {
 135         return VLC_EGENERIC;
 136     }
 137
 138     p_dec->pf_decode_sub = DecodeBlock;
 139
 140     /* Allocate the memory needed to store the decoder's structure */
 141     if( ( p_dec->p_sys = p_sys =
 142           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 143     {
 144         msg_Err( p_dec, "out of memory" );
 145         return VLC_EGENERIC;
 146     }
 147
 148     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 149     var_Get( p_dec, "subsdec-align", &val );
 150     p_sys->i_align = val.i_int;
 151
 152 #if defined(HAVE_ICONV)
 153     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 154     {
 155         msg_Dbg( p_dec, "using character encoding: %s",
 156                  p_dec->fmt_in.subs.psz_encoding );
 157         p_sys->iconv_handle =
 158             iconv_open( "UTF-8", p_dec->fmt_in.subs.psz_encoding );
 159     }
 160     else
 161     {
 162         var_Create( p_dec, "subsdec-encoding",
 163                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 164         var_Get( p_dec, "subsdec-encoding", &val );
 165         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 166         {
 167             char *psz_charset =(char*)malloc( 100 );
 168             vlc_current_charset( &psz_charset );
 169             p_sys->iconv_handle = iconv_open( "UTF-8", psz_charset );
 170             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 171             free( psz_charset );
 172         }
 173         else if( val.psz_string )
 174         {
 175             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 176             p_sys->iconv_handle = iconv_open( "UTF-8", val.psz_string );
 177         }
 178
 179         if( p_sys->iconv_handle == (iconv_t)-1 )
 180         {
 181             msg_Warn( p_dec, "unable to do requested conversion" );
 182         }
 183
 184         if( val.psz_string ) free( val.psz_string );
 185     }
 186 #else
 187
 188     msg_Dbg( p_dec, "no iconv support available" );
 189 #endif
 190
 191     /* Load the text rendering module */
 192     p_sys->p_render = vlc_object_create( p_dec, sizeof(filter_t) );
 193     p_sys->p_render->pf_spu_buffer_new = spu_new_buffer;
 194     p_sys->p_render->pf_spu_buffer_del = spu_del_buffer;
 195     p_sys->p_render->p_owner = (filter_owner_sys_t *)p_dec;
 196     vlc_object_attach( p_sys->p_render, p_dec );
 197     p_sys->p_render->p_module =
 198         module_Need( p_sys->p_render, "text renderer", 0, 0 );
 199     if( p_sys->p_render->p_module == NULL )
 200     {
 201         msg_Warn( p_dec, "no suitable text renderer module" );
 202         vlc_object_detach( p_sys->p_render );
 203         vlc_object_destroy( p_sys->p_render );
 204         p_sys->p_render = NULL;
 205     }
 206
 207     return VLC_SUCCESS;
 208 }
 209
 210 /****************************************************************************
 211  * DecodeBlock: the whole thing
 212  ****************************************************************************
 213  * This function must be fed with complete subtitles units.
 214  ****************************************************************************/
 215 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 216 {
 217     subpicture_t *p_spu;
 218
 219     if( !pp_block || *pp_block == NULL ) return NULL;
 220
 221     p_spu = ParseText( p_dec, *pp_block );
 222
 223     block_Release( *pp_block );
 224     *pp_block = NULL;
 225
 226     return p_spu;
 227 }
 228
 229 /*****************************************************************************
 230  * CloseDecoder: clean up the decoder
 231  *****************************************************************************/
 232 static void CloseDecoder( vlc_object_t *p_this )
 233 {
 234     decoder_t *p_dec = (decoder_t *)p_this;
 235     decoder_sys_t *p_sys = p_dec->p_sys;
 236
 237 #if defined(HAVE_ICONV)
 238     if( p_sys->iconv_handle != (iconv_t)-1 )
 239     {
 240         iconv_close( p_sys->iconv_handle );
 241     }
 242 #endif
 243
 244     if( p_sys->p_render )
 245     {
 246         if( p_sys->p_render->p_module )
 247             module_Unneed( p_sys->p_render, p_sys->p_render->p_module );
 248
 249         vlc_object_detach( p_sys->p_render );
 250         vlc_object_destroy( p_sys->p_render );
 251     }
 252
 253     free( p_sys );
 254 }
 255
 256 /*****************************************************************************
 257  * ParseText: parse an text subtitle packet and send it to the video output
 258  *****************************************************************************/
 259 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 260 {
 261     decoder_sys_t *p_sys = p_dec->p_sys;
 262     subpicture_t *p_spu = 0;
 263     char *psz_subtitle;
 264     int i_align_h, i_align_v;
 265
 266     /* We cannot display a subpicture with no date */
 267     if( p_block->i_pts == 0 )
 268     {
 269         msg_Warn( p_dec, "subtitle without a date" );
 270         return NULL;
 271     }
 272
 273     /* Check validity of packet data */
 274     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 275     {
 276         msg_Warn( p_dec, "empty subtitle" );
 277         return NULL;
 278     }
 279
 280     /* Should be resiliant against bad subtitles */
 281     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 282
 283     i_align_h = p_sys->i_align ? 20 : 0;
 284     i_align_v = 10;
 285
 286 #if defined(HAVE_ICONV)
 287     if( p_sys->iconv_handle != (iconv_t)-1 )
 288     {
 289         char *psz_new_subtitle;
 290         char *psz_convert_buffer_out;
 291         char *psz_convert_buffer_in;
 292         size_t ret, inbytes_left, outbytes_left;
 293
 294         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 295         psz_convert_buffer_out = psz_new_subtitle;
 296         psz_convert_buffer_in = psz_subtitle;
 297         inbytes_left = strlen( psz_subtitle );
 298         outbytes_left = 6 * inbytes_left;
 299         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 300                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 301         *psz_convert_buffer_out = '\0';
 302
 303         if( inbytes_left )
 304         {
 305             msg_Warn( p_dec, "Failed to convert subtitle encoding, "
 306                       "dropping subtitle.\nTry setting a different "
 307                       "character-encoding for the subtitle." );
 308             free( psz_subtitle );
 309             return NULL;
 310         }
 311         else
 312         {
 313             free( psz_subtitle );
 314             psz_subtitle = psz_new_subtitle;
 315         }
 316     }
 317 #endif
 318
 319     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 320     {
 321         /* Decode SSA strings */
 322         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR,
 323          * MarginV, Effect, Text */
 324         char *psz_new_subtitle;
 325         char *psz_buffer_sub;
 326         int         i_comma;
 327         int         i_text;
 328
 329         psz_buffer_sub = psz_subtitle;
 330         for( ;; )
 331         {
 332             i_comma = 0;
 333             while( i_comma < 8 &&
 334                 *psz_buffer_sub != '\0' )
 335             {
 336                 if( *psz_buffer_sub == ',' )
 337                 {
 338                     i_comma++;
 339                 }
 340                 psz_buffer_sub++;
 341             }
 342             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 343             i_text = 0;
 344             while( psz_buffer_sub[0] != '\0' )
 345             {
 346                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] == 'n' ||
 347                     psz_buffer_sub[1] == 'N' ) )
 348                 {
 349                     psz_new_subtitle[i_text] = '\n';
 350                     i_text++;
 351                     psz_buffer_sub += 2;
 352                 }
 353                 else if( psz_buffer_sub[0] == '{' &&
 354                          psz_buffer_sub[1] == '\\' )
 355                 {
 356                     /* SSA control code */
 357                     while( psz_buffer_sub[0] != '\0' &&
 358                            psz_buffer_sub[0] != '}' )
 359                     {
 360                         psz_buffer_sub++;
 361                     }
 362                     psz_buffer_sub++;
 363                 }
 364                 else
 365                 {
 366                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 367                     i_text++;
 368                     psz_buffer_sub++;
 369                 }
 370             }
 371             psz_new_subtitle[i_text] = '\0';
 372             free( psz_subtitle );
 373             psz_subtitle = psz_new_subtitle;
 374             break;
 375         }
 376     }
 377
 378     StripTags( psz_subtitle );
 379
 380     if( p_sys->p_render && p_sys->p_render->p_module &&
 381         p_sys->p_render->pf_render_string )
 382     {
 383         block_t *p_new_block = block_New( p_dec, strlen(psz_subtitle) + 1 );
 384         if( p_new_block )
 385         {
 386             memcpy( p_new_block->p_buffer, psz_subtitle,
 387                     p_new_block->i_buffer );
 388             p_new_block->i_pts = p_new_block->i_dts = p_block->i_pts;
 389             p_new_block->i_length = p_block->i_length;
 390             p_spu = p_sys->p_render->pf_render_string( p_sys->p_render,
 391                                                        p_new_block );
 392         }
 393     }
 394
 395     if( p_spu )
 396     {
 397         p_spu->i_flags = OSD_ALIGN_BOTTOM | p_sys->i_align;
 398         p_spu->i_x = i_align_h;
 399         p_spu->i_y = i_align_v;
 400     }
 401
 402     free( psz_subtitle );
 403     return p_spu;
 404 }
 405
 406 static void StripTags( char *psz_text )
 407 {
 408     int i_left_moves = 0;
 409     vlc_bool_t b_inside_tag = VLC_FALSE;
 410     int i = 0;
 411     int i_tag_start = -1;
 412     while( psz_text[ i ] )
 413     {
 414         if( !b_inside_tag )
 415         {
 416             if( psz_text[ i ] == '<' )
 417             {
 418                 b_inside_tag = VLC_TRUE;
 419                 i_tag_start = i;
 420             }
 421             psz_text[ i - i_left_moves ] = psz_text[ i ];
 422         }
 423         else
 424         {
 425             if( ( psz_text[ i ] == ' ' ) ||
 426                 ( psz_text[ i ] == '\t' ) ||
 427                 ( psz_text[ i ] == '\n' ) ||
 428                 ( psz_text[ i ] == '\r' ) )
 429             {
 430                 b_inside_tag = VLC_FALSE;
 431                 i_tag_start = -1;
 432             }
 433             else if( psz_text[ i ] == '>' )
 434             {
 435                 i_left_moves += i - i_tag_start + 1;
 436                 i_tag_start = -1;
 437                 b_inside_tag = VLC_FALSE;
 438             }
 439             else
 440             {
 441                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 442             }
 443         }
 444         i++;
 445     }
 446     psz_text[ i - i_left_moves ] = '\0';
 447 }
 448
 449 /*****************************************************************************
 450  * Buffers allocation callbacks for the filters
 451  *****************************************************************************/
 452 static subpicture_t *spu_new_buffer( filter_t *p_filter )
 453 {
 454     decoder_t *p_dec = (decoder_t *)p_filter->p_owner;
 455     return p_dec->pf_spu_buffer_new( p_dec );
 456 }
 457
 458 static void spu_del_buffer( filter_t *p_filter, subpicture_t *p_spu )
 459 {
 460     decoder_t *p_dec = (decoder_t *)p_filter->p_owner;
 461     p_dec->pf_spu_buffer_del( p_dec, p_spu );
 462 }