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
  96     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
  97                  VLC_TRUE );
  98         change_integer_list( pi_justification, ppsz_justification_text, 0 );
  99     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 100                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 101         change_string_list( ppsz_encodings, 0, 0 );
 102 vlc_module_end();
 103
 104 /*****************************************************************************
 105  * OpenDecoder: probe the decoder and return score
 106  *****************************************************************************
 107  * Tries to launch a decoder and return score so that the interface is able
 108  * to chose.
 109  *****************************************************************************/
 110 static int OpenDecoder( vlc_object_t *p_this )
 111 {
 112     decoder_t     *p_dec = (decoder_t*)p_this;
 113     decoder_sys_t *p_sys;
 114     vlc_value_t val;
 115
 116     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 117         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 118     {
 119         return VLC_EGENERIC;
 120     }
 121
 122     p_dec->pf_decode_sub = DecodeBlock;
 123
 124     /* Allocate the memory needed to store the decoder's structure */
 125     if( ( p_dec->p_sys = p_sys =
 126           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 127     {
 128         msg_Err( p_dec, "out of memory" );
 129         return VLC_EGENERIC;
 130     }
 131
 132     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 133     var_Get( p_dec, "subsdec-align", &val );
 134     p_sys->i_align = val.i_int;
 135
 136     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 137     {
 138         msg_Dbg( p_dec, "using character encoding: %s",
 139                  p_dec->fmt_in.subs.psz_encoding );
 140         p_sys->iconv_handle =
 141             vlc_iconv_open( "UTF-8", p_dec->fmt_in.subs.psz_encoding );
 142     }
 143     else
 144     {
 145         var_Create( p_dec, "subsdec-encoding",
 146                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 147         var_Get( p_dec, "subsdec-encoding", &val );
 148         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 149         {
 150             char *psz_charset =(char*)malloc( 100 );
 151             vlc_current_charset( &psz_charset );
 152             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", psz_charset );
 153             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 154             free( psz_charset );
 155         }
 156         else if( val.psz_string )
 157         {
 158             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 159             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", val.psz_string );
 160         }
 161
 162         if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 163         {
 164             msg_Warn( p_dec, "unable to do requested conversion" );
 165         }
 166
 167         if( val.psz_string ) free( val.psz_string );
 168     }
 169
 170     return VLC_SUCCESS;
 171 }
 172
 173 /****************************************************************************
 174  * DecodeBlock: the whole thing
 175  ****************************************************************************
 176  * This function must be fed with complete subtitles units.
 177  ****************************************************************************/
 178 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 179 {
 180     subpicture_t *p_spu;
 181
 182     if( !pp_block || *pp_block == NULL ) return NULL;
 183
 184     p_spu = ParseText( p_dec, *pp_block );
 185
 186     block_Release( *pp_block );
 187     *pp_block = NULL;
 188
 189     return p_spu;
 190 }
 191
 192 /*****************************************************************************
 193  * CloseDecoder: clean up the decoder
 194  *****************************************************************************/
 195 static void CloseDecoder( vlc_object_t *p_this )
 196 {
 197     decoder_t *p_dec = (decoder_t *)p_this;
 198     decoder_sys_t *p_sys = p_dec->p_sys;
 199
 200     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 201     {
 202         vlc_iconv_close( p_sys->iconv_handle );
 203     }
 204
 205     free( p_sys );
 206 }
 207
 208 /*****************************************************************************
 209  * ParseText: parse an text subtitle packet and send it to the video output
 210  *****************************************************************************/
 211 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 212 {
 213     decoder_sys_t *p_sys = p_dec->p_sys;
 214     subpicture_t *p_spu = 0;
 215     char *psz_subtitle;
 216     int i_align_h, i_align_v;
 217     video_format_t fmt;
 218
 219     /* We cannot display a subpicture with no date */
 220     if( p_block->i_pts == 0 )
 221     {
 222         msg_Warn( p_dec, "subtitle without a date" );
 223         return NULL;
 224     }
 225
 226     /* Check validity of packet data */
 227     if( p_block->i_buffer <= 1 || p_block->p_buffer[0] == '\0' )
 228     {
 229         msg_Warn( p_dec, "empty subtitle" );
 230         return NULL;
 231     }
 232
 233     /* Should be resiliant against bad subtitles */
 234     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 235
 236     i_align_h = p_sys->i_align ? 20 : 0;
 237     i_align_v = 10;
 238
 239     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 240     {
 241         char *psz_new_subtitle;
 242         char *psz_convert_buffer_out;
 243         char *psz_convert_buffer_in;
 244         size_t ret, inbytes_left, outbytes_left;
 245
 246         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 247         psz_convert_buffer_out = psz_new_subtitle;
 248         psz_convert_buffer_in = psz_subtitle;
 249         inbytes_left = strlen( psz_subtitle );
 250         outbytes_left = 6 * inbytes_left;
 251         ret = vlc_iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 252                          &inbytes_left, &psz_convert_buffer_out,
 253                          &outbytes_left );
 254         *psz_convert_buffer_out = '\0';
 255
 256         if( inbytes_left )
 257         {
 258             msg_Warn( p_dec, "Failed to convert subtitle encoding, "
 259                       "dropping subtitle.\nTry setting a different "
 260                       "character-encoding for the subtitle." );
 261             free( psz_subtitle );
 262             return NULL;
 263         }
 264         else
 265         {
 266             free( psz_subtitle );
 267             psz_subtitle = psz_new_subtitle;
 268         }
 269     }
 270
 271     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 272     {
 273         /* Decode SSA strings */
 274         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR,
 275          * MarginV, Effect, Text */
 276         char *psz_new_subtitle;
 277         char *psz_buffer_sub;
 278         int         i_comma;
 279         int         i_text;
 280
 281         psz_buffer_sub = psz_subtitle;
 282         for( ;; )
 283         {
 284             i_comma = 0;
 285             while( i_comma < 8 &&
 286                 *psz_buffer_sub != '\0' )
 287             {
 288                 if( *psz_buffer_sub == ',' )
 289                 {
 290                     i_comma++;
 291                 }
 292                 psz_buffer_sub++;
 293             }
 294             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 295             i_text = 0;
 296             while( psz_buffer_sub[0] != '\0' )
 297             {
 298                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] == 'n' ||
 299                     psz_buffer_sub[1] == 'N' ) )
 300                 {
 301                     psz_new_subtitle[i_text] = '\n';
 302                     i_text++;
 303                     psz_buffer_sub += 2;
 304                 }
 305                 else if( psz_buffer_sub[0] == '{' &&
 306                          psz_buffer_sub[1] == '\\' )
 307                 {
 308                     /* SSA control code */
 309                     while( psz_buffer_sub[0] != '\0' &&
 310                            psz_buffer_sub[0] != '}' )
 311                     {
 312                         psz_buffer_sub++;
 313                     }
 314                     psz_buffer_sub++;
 315                 }
 316                 else
 317                 {
 318                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 319                     i_text++;
 320                     psz_buffer_sub++;
 321                 }
 322             }
 323             psz_new_subtitle[i_text] = '\0';
 324             free( psz_subtitle );
 325             psz_subtitle = psz_new_subtitle;
 326             break;
 327         }
 328     }
 329
 330     StripTags( psz_subtitle );
 331
 332     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 333     if( !p_spu )
 334     {
 335         msg_Warn( p_dec, "can't get spu buffer" );
 336         free( psz_subtitle );
 337         return 0;
 338     }
 339
 340     /* Create a new subpicture region */
 341     memset( &fmt, 0, sizeof(video_format_t) );
 342     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 343     fmt.i_aspect = 0;
 344     fmt.i_width = fmt.i_height = 0;
 345     fmt.i_x_offset = fmt.i_y_offset = 0;
 346     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 347     if( !p_spu->p_region )
 348     {
 349         msg_Err( p_dec, "cannot allocate SPU region" );
 350         free( psz_subtitle );
 351         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 352         return 0;
 353     }
 354
 355     p_spu->p_region->psz_text = psz_subtitle;
 356     p_spu->i_start = p_block->i_pts;
 357     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 358     p_spu->b_ephemer = (p_block->i_length == 0);
 359     p_spu->b_absolute = VLC_FALSE;
 360
 361     p_spu->i_flags = OSD_ALIGN_BOTTOM | p_sys->i_align;
 362     p_spu->i_x = i_align_h;
 363     p_spu->i_y = i_align_v;
 364
 365     return p_spu;
 366 }
 367
 368 static void StripTags( char *psz_text )
 369 {
 370     int i_left_moves = 0;
 371     vlc_bool_t b_inside_tag = VLC_FALSE;
 372     int i = 0;
 373     int i_tag_start = -1;
 374     while( psz_text[ i ] )
 375     {
 376         if( !b_inside_tag )
 377         {
 378             if( psz_text[ i ] == '<' )
 379             {
 380                 b_inside_tag = VLC_TRUE;
 381                 i_tag_start = i;
 382             }
 383             psz_text[ i - i_left_moves ] = psz_text[ i ];
 384         }
 385         else
 386         {
 387             if( ( psz_text[ i ] == ' ' ) ||
 388                 ( psz_text[ i ] == '\t' ) ||
 389                 ( psz_text[ i ] == '\n' ) ||
 390                 ( psz_text[ i ] == '\r' ) )
 391             {
 392                 b_inside_tag = VLC_FALSE;
 393                 i_tag_start = -1;
 394             }
 395             else if( psz_text[ i ] == '>' )
 396             {
 397                 i_left_moves += i - i_tag_start + 1;
 398                 i_tag_start = -1;
 399                 b_inside_tag = VLC_FALSE;
 400             }
 401             else
 402             {
 403                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 404             }
 405         }
 406         i++;
 407     }
 408     psz_text[ i - i_left_moves ] = '\0';
 409 }