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

   1 /*****************************************************************************
   2  * subsdec.c : text subtitles decoder
   3  *****************************************************************************
   4  * Copyright (C) 2000-2001 the VideoLAN team
   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 "vlc_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_shortname( _("Subtitles"));
  93     set_description( _("Text subtitles decoder") );
  94     set_capability( "decoder", 50 );
  95     set_callbacks( OpenDecoder, CloseDecoder );
  96     set_category( CAT_INPUT );
  97     set_subcategory( SUBCAT_INPUT_SCODEC );
  98
  99     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 100                  VLC_FALSE );
 101         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 102     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 103                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 104         change_string_list( ppsz_encodings, 0, 0 );
 105 vlc_module_end();
 106
 107 /*****************************************************************************
 108  * OpenDecoder: probe the decoder and return score
 109  *****************************************************************************
 110  * Tries to launch a decoder and return score so that the interface is able
 111  * to chose.
 112  *****************************************************************************/
 113 static int OpenDecoder( vlc_object_t *p_this )
 114 {
 115     decoder_t     *p_dec = (decoder_t*)p_this;
 116     decoder_sys_t *p_sys;
 117     vlc_value_t val;
 118
 119     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 120         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 121     {
 122         return VLC_EGENERIC;
 123     }
 124
 125     p_dec->pf_decode_sub = DecodeBlock;
 126
 127     /* Allocate the memory needed to store the decoder's structure */
 128     if( ( p_dec->p_sys = p_sys =
 129           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 130     {
 131         msg_Err( p_dec, "out of memory" );
 132         return VLC_EGENERIC;
 133     }
 134
 135     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 136     var_Get( p_dec, "subsdec-align", &val );
 137     p_sys->i_align = val.i_int;
 138
 139     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 140     {
 141         msg_Dbg( p_dec, "using character encoding: %s",
 142                  p_dec->fmt_in.subs.psz_encoding );
 143         p_sys->iconv_handle =
 144             vlc_iconv_open( "UTF-8", p_dec->fmt_in.subs.psz_encoding );
 145     }
 146     else
 147     {
 148         var_Create( p_dec, "subsdec-encoding",
 149                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 150         var_Get( p_dec, "subsdec-encoding", &val );
 151         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 152         {
 153             char *psz_charset =(char*)malloc( 100 );
 154             vlc_current_charset( &psz_charset );
 155             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", psz_charset );
 156             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 157             free( psz_charset );
 158         }
 159         else if( val.psz_string )
 160         {
 161             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 162             p_sys->iconv_handle = vlc_iconv_open( "UTF-8", val.psz_string );
 163         }
 164
 165         if( p_sys->iconv_handle == (vlc_iconv_t)-1 )
 166         {
 167             msg_Warn( p_dec, "unable to do requested conversion" );
 168         }
 169
 170         if( val.psz_string ) free( val.psz_string );
 171     }
 172
 173     return VLC_SUCCESS;
 174 }
 175
 176 /****************************************************************************
 177  * DecodeBlock: the whole thing
 178  ****************************************************************************
 179  * This function must be fed with complete subtitles units.
 180  ****************************************************************************/
 181 static subpicture_t *DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 182 {
 183     subpicture_t *p_spu;
 184
 185     if( !pp_block || *pp_block == NULL ) return NULL;
 186
 187     p_spu = ParseText( p_dec, *pp_block );
 188
 189     block_Release( *pp_block );
 190     *pp_block = NULL;
 191
 192     return p_spu;
 193 }
 194
 195 /*****************************************************************************
 196  * CloseDecoder: clean up the decoder
 197  *****************************************************************************/
 198 static void CloseDecoder( vlc_object_t *p_this )
 199 {
 200     decoder_t *p_dec = (decoder_t *)p_this;
 201     decoder_sys_t *p_sys = p_dec->p_sys;
 202
 203     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 204     {
 205         vlc_iconv_close( p_sys->iconv_handle );
 206     }
 207
 208     free( p_sys );
 209 }
 210
 211 /*****************************************************************************
 212  * ParseText: parse an text subtitle packet and send it to the video output
 213  *****************************************************************************/
 214 static subpicture_t *ParseText( decoder_t *p_dec, block_t *p_block )
 215 {
 216     decoder_sys_t *p_sys = p_dec->p_sys;
 217     subpicture_t *p_spu = 0;
 218     char *psz_subtitle;
 219     int i_align_h, i_align_v;
 220     video_format_t fmt;
 221
 222     /* We cannot display a subpicture with no date */
 223     if( p_block->i_pts == 0 )
 224     {
 225         msg_Warn( p_dec, "subtitle without a date" );
 226         return NULL;
 227     }
 228
 229     /* Check validity of packet data */
 230     if( p_block->i_buffer <= 1 || p_block->p_buffer[0] == '\0' )
 231     {
 232         msg_Warn( p_dec, "empty subtitle" );
 233         return NULL;
 234     }
 235
 236     /* Should be resiliant against bad subtitles */
 237     psz_subtitle = strndup( (const char *)p_block->p_buffer,
 238                             p_block->i_buffer );
 239
 240     i_align_h = p_sys->i_align ? 20 : 0;
 241     i_align_v = 10;
 242
 243     if( p_sys->iconv_handle != (vlc_iconv_t)-1 )
 244     {
 245         char *psz_new_subtitle;
 246         char *psz_convert_buffer_out;
 247         char *psz_convert_buffer_in;
 248         size_t ret, inbytes_left, outbytes_left;
 249
 250         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 251         psz_convert_buffer_out = psz_new_subtitle;
 252         psz_convert_buffer_in = psz_subtitle;
 253         inbytes_left = strlen( psz_subtitle );
 254         outbytes_left = 6 * inbytes_left;
 255         ret = vlc_iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 256                          &inbytes_left, &psz_convert_buffer_out,
 257                          &outbytes_left );
 258         *psz_convert_buffer_out = '\0';
 259
 260         if( inbytes_left )
 261         {
 262             msg_Warn( p_dec, "Failed to convert subtitle encoding, "
 263                       "dropping subtitle.\nTry setting a different "
 264                       "character-encoding for the subtitle." );
 265             free( psz_subtitle );
 266             return NULL;
 267         }
 268         else
 269         {
 270             free( psz_subtitle );
 271             psz_subtitle = psz_new_subtitle;
 272         }
 273     }
 274
 275     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 276     {
 277         /* Decode SSA strings */
 278         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR,
 279          * MarginV, Effect, Text */
 280         char *psz_new_subtitle;
 281         char *psz_buffer_sub;
 282         int         i_comma;
 283         int         i_text;
 284
 285         psz_buffer_sub = psz_subtitle;
 286         for( ;; )
 287         {
 288             i_comma = 0;
 289             while( i_comma < 8 &&
 290                 *psz_buffer_sub != '\0' )
 291             {
 292                 if( *psz_buffer_sub == ',' )
 293                 {
 294                     i_comma++;
 295                 }
 296                 psz_buffer_sub++;
 297             }
 298             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 299             i_text = 0;
 300             while( psz_buffer_sub[0] != '\0' )
 301             {
 302                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] == 'n' ||
 303                     psz_buffer_sub[1] == 'N' ) )
 304                 {
 305                     psz_new_subtitle[i_text] = '\n';
 306                     i_text++;
 307                     psz_buffer_sub += 2;
 308                 }
 309                 else if( psz_buffer_sub[0] == '{' &&
 310                          psz_buffer_sub[1] == '\\' )
 311                 {
 312                     /* SSA control code */
 313                     while( psz_buffer_sub[0] != '\0' &&
 314                            psz_buffer_sub[0] != '}' )
 315                     {
 316                         psz_buffer_sub++;
 317                     }
 318                     psz_buffer_sub++;
 319                 }
 320                 else
 321                 {
 322                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 323                     i_text++;
 324                     psz_buffer_sub++;
 325                 }
 326             }
 327             psz_new_subtitle[i_text] = '\0';
 328             free( psz_subtitle );
 329             psz_subtitle = psz_new_subtitle;
 330             break;
 331         }
 332     }
 333
 334     StripTags( psz_subtitle );
 335
 336     p_spu = p_dec->pf_spu_buffer_new( p_dec );
 337     if( !p_spu )
 338     {
 339         msg_Warn( p_dec, "can't get spu buffer" );
 340         free( psz_subtitle );
 341         return 0;
 342     }
 343
 344     /* Create a new subpicture region */
 345     memset( &fmt, 0, sizeof(video_format_t) );
 346     fmt.i_chroma = VLC_FOURCC('T','E','X','T');
 347     fmt.i_aspect = 0;
 348     fmt.i_width = fmt.i_height = 0;
 349     fmt.i_x_offset = fmt.i_y_offset = 0;
 350     p_spu->p_region = p_spu->pf_create_region( VLC_OBJECT(p_dec), &fmt );
 351     if( !p_spu->p_region )
 352     {
 353         msg_Err( p_dec, "cannot allocate SPU region" );
 354         free( psz_subtitle );
 355         p_dec->pf_spu_buffer_del( p_dec, p_spu );
 356         return 0;
 357     }
 358
 359     p_spu->p_region->psz_text = psz_subtitle;
 360     p_spu->i_start = p_block->i_pts;
 361     p_spu->i_stop = p_block->i_pts + p_block->i_length;
 362     p_spu->b_ephemer = (p_block->i_length == 0);
 363     p_spu->b_absolute = VLC_FALSE;
 364
 365     p_spu->i_flags = OSD_ALIGN_BOTTOM | p_sys->i_align;
 366     p_spu->i_x = i_align_h;
 367     p_spu->i_y = i_align_v;
 368
 369     return p_spu;
 370 }
 371
 372 static void StripTags( char *psz_text )
 373 {
 374     int i_left_moves = 0;
 375     vlc_bool_t b_inside_tag = VLC_FALSE;
 376     int i = 0;
 377     int i_tag_start = -1;
 378     while( psz_text[ i ] )
 379     {
 380         if( !b_inside_tag )
 381         {
 382             if( psz_text[ i ] == '<' )
 383             {
 384                 b_inside_tag = VLC_TRUE;
 385                 i_tag_start = i;
 386             }
 387             psz_text[ i - i_left_moves ] = psz_text[ i ];
 388         }
 389         else
 390         {
 391             if( ( psz_text[ i ] == ' ' ) ||
 392                 ( psz_text[ i ] == '\t' ) ||
 393                 ( psz_text[ i ] == '\n' ) ||
 394                 ( psz_text[ i ] == '\r' ) )
 395             {
 396                 b_inside_tag = VLC_FALSE;
 397                 i_tag_start = -1;
 398             }
 399             else if( psz_text[ i ] == '>' )
 400             {
 401                 i_left_moves += i - i_tag_start + 1;
 402                 i_tag_start = -1;
 403                 b_inside_tag = VLC_FALSE;
 404             }
 405             else
 406             {
 407                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 408             }
 409         }
 410         i++;
 411     }
 412     psz_text[ i - i_left_moves ] = '\0';
 413 }