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@videolan.org>
   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 #include "charset.h"
  36
  37 /*****************************************************************************
  38  * decoder_sys_t : decoder descriptor
  39  *****************************************************************************/
  40 struct decoder_sys_t
  41 {
  42     int                 i_align;          /* Subtitles alignment on the vout */
  43     vlc_iconv_t         iconv_handle;            /* handle to iconv instance */
  44 };
  45
  46 /*****************************************************************************
  47  * Local prototypes
  48  *****************************************************************************/
  49 static int  OpenDecoder   ( vlc_object_t * );
  50 static void CloseDecoder  ( vlc_object_t * );
  51
  52 static subpicture_t *DecodeBlock   ( decoder_t *, block_t ** );
  53 static subpicture_t *ParseText     ( decoder_t *, block_t * );
  54 static void         StripTags      ( char * );
  55
  56 #define DEFAULT_NAME "System Default"
  57
  58 /*****************************************************************************
  59  * Module descriptor.
  60  *****************************************************************************/
  61 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  62     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  63     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  64     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  65     "ISO-8859-6", "CP1256", "MacArabic", "",
  66     "ISO-8859-7", "CP1253", "MacGreek", "",
  67     "ISO-8859-8", "CP1255", "MacHebrew", "",
  68     "ISO-8859-9", "CP1254", "MacTurkish", "",
  69     "ISO-8859-13", "CP1257", "",
  70     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  71     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  72     "ISO-2022-KR", "EUC-KR", "",
  73     "MacThai", "KOI8-T", "",
  74     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  75     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  76     "Macintosh", "",
  77     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  78     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  79     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  80     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  81     "HPROMAN8", "NEXTSTEP" };
  82
  83 static int  pi_justification[] = { 0, 1, 2 };
  84 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  85
  86 #define ENCODING_TEXT N_("Subtitles text encoding")
  87 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
  88 #define ALIGN_TEXT N_("Subtitles justification")
  89 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
  90
  91 vlc_module_begin();
  92     set_description( _("text subtitles decoder") );
  93     set_capability( "decoder", 50 );
  94     set_callbacks( OpenDecoder, CloseDecoder );
  95     set_category( CAT_INPUT );
  96     set_subcategory( SUBCAT_INPUT_SCODEC );
  97
  98     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
  99                  VLC_TRUE );
 100         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 101     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 102                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 103         change_string_list( ppsz_encodings, 0, 0 );
 104 vlc_module_end();
 105
 106 /*****************************************************************************
 107  * OpenDecoder: probe the decoder and return score
 108  *****************************************************************************
 109  * Tries to launch a decoder and return score so that the interface is able
 110  * to chose.
 111  *****************************************************************************/
 112 static int OpenDecoder( vlc_object_t *p_this )
 113 {
 114     decoder_t     *p_dec = (decoder_t*)p_this;
 115     decoder_sys_t *p_sys;
 116     vlc_value_t val;
 117
 118     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 119         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 120     {
 121         return VLC_EGENERIC;
 122     }
 123
 124     p_dec->pf_decode_sub = DecodeBlock;
 125
 126     /* Allocate the memory needed to store the decoder's structure */
 127     if( ( p_dec->p_sys = p_sys =
 128           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 129     {
 130         msg_Err( p_dec, "out of memory" );
 131         return VLC_EGENERIC;
 132     }
 133
 134     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 135     var_Get( p_dec, "subsdec-align", &val );
 136     p_sys->i_align = val.i_int;
 137
 138     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 139     {
 140         msg_Dbg( p_dec, "using character encoding: %s",
 141                  p_dec->fmt_in.subs.psz_encoding );
 142         p_sys->iconv_handle =
 143             vlc_iconv_open( "UTF-8", p_dec->fmt_in.subs.psz_encoding );
 144     }
 145     else
 146     {
 147         var_Create( p_dec, "subsdec-encoding",
 148                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 149         var_Get( p_dec, "subsdec-encoding", &val );
 150         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 151         {
 152             char *psz_charset =(char*)malloc( 100 );
 153             vlc_current_charset( &psz_charset );
 154             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", psz_charset );
 155             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 156             free( psz_charset );
 157         }
 158         else if( val.psz_string )
 159         {
 160             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 161             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", val.psz_string );
 162         }
 163
 164         if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 165         {
 166             msg_Warn( p_dec, "unable to do requested conversion" );
 167         }
 168
 169         if( val.psz_string ) free( val.psz_string );
 170     }
 171
 172     return VLC_SUCCESS;
 173 }
 174
 175 /****************************************************************************
 176  * DecodeBlock: the whole thing
 177  ****************************************************************************
 178  * This function must be fed with complete subtitles units.
 179  ****************************************************************************/
 180 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 181 {
 182     subpicture_t *p_spu;
 183
 184     if( !pp_block || *pp_block == NULL ) return NULL;
 185
 186     p_spu = ParseText( p_dec, *pp_block );
 187
 188     block_Release( *pp_block );
 189     *pp_block = NULL;
 190
 191     return p_spu;
 192 }
 193
 194 /*****************************************************************************
 195  * CloseDecoder: clean up the decoder
 196  *****************************************************************************/
 197 static void CloseDecoder( vlc_object_t *p_this )
 198 {
 199     decoder_t *p_dec = (decoder_t *)p_this;
 200     decoder_sys_t *p_sys = p_dec->p_sys;
 201
 202     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 203     {
 204         vlc_iconv_close( p_sys->iconv_handle );
 205     }
 206
 207     free( p_sys );
 208 }
 209
 210 /*****************************************************************************
 211  * ParseText: parse an text subtitle packet and send it to the video output
 212  *****************************************************************************/
 213 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 214 {
 215     decoder_sys_t *p_sys = p_dec->p_sys;
 216     subpicture_t *p_spu = 0;
 217     char *psz_subtitle;
 218     int i_align_h, i_align_v;
 219     video_format_t fmt;
 220
 221     /* We cannot display a subpicture with no date */
 222     if( p_block->i_pts == 0 )
 223     {
 224         msg_Warn( p_dec, "subtitle without a date" );
 225         return NULL;
 226     }
 227
 228     /* Check validity of packet data */
 229     if( p_block->i_buffer <= 1 || p_block->p_buffer[0] == '\0' )
 230     {
 231         msg_Warn( p_dec, "empty subtitle" );
 232         return NULL;
 233     }
 234
 235     /* Should be resiliant against bad subtitles */
 236     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 237
 238     i_align_h = p_sys->i_align ? 20 : 0;
 239     i_align_v = 10;
 240
 241     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 242     {
 243         char *psz_new_subtitle;
 244         char *psz_convert_buffer_out;
 245         char *psz_convert_buffer_in;
 246         size_t ret, inbytes_left, outbytes_left;
 247
 248         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 249         psz_convert_buffer_out = psz_new_subtitle;
 250         psz_convert_buffer_in = psz_subtitle;
 251         inbytes_left = strlen( psz_subtitle );
 252         outbytes_left = 6 * inbytes_left;
 253         ret = vlc_iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 254                          &inbytes_left, &psz_convert_buffer_out,
 255                          &outbytes_left );
 256         *psz_convert_buffer_out = '\0';
 257
 258         if( inbytes_left )
 259         {
 260             msg_Warn( p_dec, "Failed to convert subtitle encoding, "
 261                       "dropping subtitle.\nTry setting a different "
 262                       "character-encoding for the subtitle." );
 263             free( psz_subtitle );
 264             return NULL;
 265         }
 266         else
 267         {
 268             free( psz_subtitle );
 269             psz_subtitle = psz_new_subtitle;
 270         }
 271     }
 272
 273     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 274     {
 275         /* Decode SSA strings */
 276         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR,
 277          * MarginV, Effect, Text */
 278         char *psz_new_subtitle;
 279         char *psz_buffer_sub;
 280         int         i_comma;
 281         int         i_text;
 282
 283         psz_buffer_sub = psz_subtitle;
 284         for( ;; )
 285         {
 286             i_comma = 0;
 287             while( i_comma < 8 &&
 288                 *psz_buffer_sub != '\0' )
 289             {
 290                 if( *psz_buffer_sub == ',' )
 291                 {
 292                     i_comma++;
 293                 }
 294                 psz_buffer_sub++;
 295             }
 296             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 297             i_text = 0;
 298             while( psz_buffer_sub[0] != '\0' )
 299             {
 300                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] == 'n' ||
 301                     psz_buffer_sub[1] == 'N' ) )
 302                 {
 303                     psz_new_subtitle[i_text] = '\n';
 304                     i_text++;
 305                     psz_buffer_sub += 2;
 306                 }
 307                 else if( psz_buffer_sub[0] == '{' &&
 308                          psz_buffer_sub[1] == '\\' )
 309                 {
 310                     /* SSA control code */
 311                     while( psz_buffer_sub[0] != '\0' &&
 312                            psz_buffer_sub[0] != '}' )
 313                     {
 314                         psz_buffer_sub++;
 315                     }
 316                     psz_buffer_sub++;
 317                 }
 318                 else
 319                 {
 320                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 321                     i_text++;
 322                     psz_buffer_sub++;
 323                 }
 324             }
 325             psz_new_subtitle[i_text] = '\0';
 326             free( psz_subtitle );
 327             psz_subtitle = psz_new_subtitle;
 328             break;
 329         }
 330     }
 331
 332     StripTags( psz_subtitle );
 333
 334     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 335     if( !p_spu )
 336     {
 337         msg_Warn( p_dec, "can't get spu buffer" );
 338         free( psz_subtitle );
 339         return 0;
 340     }
 341
 342     /* Create a new subpicture region */
 343     memset( &fmt, 0, sizeof(video_format_t) );
 344     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 345     fmt.i_aspect = 0;
 346     fmt.i_width = fmt.i_height = 0;
 347     fmt.i_x_offset = fmt.i_y_offset = 0;
 348     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 349     if( !p_spu->p_region )
 350     {
 351         msg_Err( p_dec, "cannot allocate SPU region" );
 352         free( psz_subtitle );
 353         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 354         return 0;
 355     }
 356
 357     p_spu->p_region->psz_text = psz_subtitle;
 358     p_spu->i_start = p_block->i_pts;
 359     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 360     p_spu->b_ephemer = (p_block->i_length == 0);
 361     p_spu->b_absolute = VLC_FALSE;
 362
 363     p_spu->i_flags = OSD_ALIGN_BOTTOM | p_sys->i_align;
 364     p_spu->i_x = i_align_h;
 365     p_spu->i_y = i_align_v;
 366
 367     return p_spu;
 368 }
 369
 370 static void StripTags( char *psz_text )
 371 {
 372     int i_left_moves = 0;
 373     vlc_bool_t b_inside_tag = VLC_FALSE;
 374     int i = 0;
 375     int i_tag_start = -1;
 376     while( psz_text[ i ] )
 377     {
 378         if( !b_inside_tag )
 379         {
 380             if( psz_text[ i ] == '<' )
 381             {
 382                 b_inside_tag = VLC_TRUE;
 383                 i_tag_start = i;
 384             }
 385             psz_text[ i - i_left_moves ] = psz_text[ i ];
 386         }
 387         else
 388         {
 389             if( ( psz_text[ i ] == ' ' ) ||
 390                 ( psz_text[ i ] == '\t' ) ||
 391                 ( psz_text[ i ] == '\n' ) ||
 392                 ( psz_text[ i ] == '\r' ) )
 393             {
 394                 b_inside_tag = VLC_FALSE;
 395                 i_tag_start = -1;
 396             }
 397             else if( psz_text[ i ] == '>' )
 398             {
 399                 i_left_moves += i - i_tag_start + 1;
 400                 i_tag_start = -1;
 401                 b_inside_tag = VLC_FALSE;
 402             }
 403             else
 404             {
 405                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 406             }
 407         }
 408         i++;
 409     }
 410     psz_text[ i - i_left_moves ] = '\0';
 411 }