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
  34 #if defined(HAVE_ICONV)
  35 #include <iconv.h>
  36 #endif
  37
  38 #include "charset.h"
  39
  40 /*****************************************************************************
  41  * decoder_sys_t : decoder descriptor
  42  *****************************************************************************/
  43 struct decoder_sys_t
  44 {
  45     int                 i_align;          /* Subtitles alignment on the vout */
  46
  47 #if defined(HAVE_ICONV)
  48     iconv_t             iconv_handle;            /* handle to iconv instance */
  49 #endif
  50 };
  51
  52 /*****************************************************************************
  53  * Local prototypes
  54  *****************************************************************************/
  55 static int  OpenDecoder   ( vlc_object_t * );
  56 static void CloseDecoder  ( vlc_object_t * );
  57
  58 static void DecodeBlock   ( decoder_t *, block_t ** );
  59
  60 static void ParseText     ( decoder_t *, block_t *, vout_thread_t * );
  61 static void StripTags     ( char * );
  62
  63 #define DEFAULT_NAME "System Default"
  64
  65 /*****************************************************************************
  66  * Module descriptor.
  67  *****************************************************************************/
  68 #if defined(HAVE_ICONV)
  69 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  70     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  71     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  72     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  73     "ISO-8859-6", "CP1256", "MacArabic", "",
  74     "ISO-8859-7", "CP1253", "MacGreek", "",
  75     "ISO-8859-8", "CP1255", "MacHebrew", "",
  76     "ISO-8859-9", "CP1254", "MacTurkish", "",
  77     "ISO-8859-13", "CP1257", "",
  78     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  79     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  80     "ISO-2022-KR", "EUC-KR", "",
  81     "MacThai", "KOI8-T", "",
  82     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  83     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  84     "Macintosh", "",
  85     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  86     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  87     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  88     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  89     "HPROMAN8", "NEXTSTEP" };
  90 #endif
  91
  92 static int  pi_justification[] = { 0, 1, 2 };
  93 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  94
  95 #define ENCODING_TEXT N_("Subtitles text encoding")
  96 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
  97 #define ALIGN_TEXT N_("Subtitles justification")
  98 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
  99
 100 vlc_module_begin();
 101     set_description( _("text subtitles decoder") );
 102     set_capability( "decoder", 50 );
 103     set_callbacks( OpenDecoder, CloseDecoder );
 104
 105     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 106                  VLC_TRUE );
 107         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 108 #if defined(HAVE_ICONV)
 109     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 110                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 111         change_string_list( ppsz_encodings, 0, 0 );
 112 #endif
 113 vlc_module_end();
 114
 115 /*****************************************************************************
 116  * OpenDecoder: probe the decoder and return score
 117  *****************************************************************************
 118  * Tries to launch a decoder and return score so that the interface is able
 119  * to chose.
 120  *****************************************************************************/
 121 static int OpenDecoder( vlc_object_t *p_this )
 122 {
 123     decoder_t     *p_dec = (decoder_t*)p_this;
 124     decoder_sys_t *p_sys;
 125     vlc_value_t val;
 126
 127     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 128         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 129     {
 130         return VLC_EGENERIC;
 131     }
 132
 133     p_dec->pf_decode_sub = DecodeBlock;
 134
 135     /* Allocate the memory needed to store the decoder's structure */
 136     if( ( p_dec->p_sys = p_sys =
 137           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 138     {
 139         msg_Err( p_dec, "out of memory" );
 140         return VLC_EGENERIC;
 141     }
 142
 143     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 144     var_Get( p_dec, "subsdec-align", &val );
 145     p_sys->i_align = val.i_int;
 146
 147 #if defined(HAVE_ICONV)
 148     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 149     {
 150         msg_Dbg( p_dec, "using character encoding: %s",
 151                  p_dec->fmt_in.subs.psz_encoding );
 152         p_sys->iconv_handle = iconv_open( "UTF-8",
 153                                           p_dec->fmt_in.subs.psz_encoding );
 154     }
 155     else
 156     {
 157         var_Create( p_dec, "subsdec-encoding",
 158                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 159         var_Get( p_dec, "subsdec-encoding", &val );
 160         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 161         {
 162             char *psz_charset =(char*)malloc( 100 );
 163             vlc_current_charset( &psz_charset );
 164             p_sys->iconv_handle = iconv_open( "UTF-8", psz_charset );
 165             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 166             free( psz_charset );
 167         }
 168         else if( val.psz_string )
 169         {
 170             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 171             p_sys->iconv_handle = iconv_open( "UTF-8", val.psz_string );
 172         }
 173
 174         if( p_sys->iconv_handle == (iconv_t)-1 )
 175         {
 176             msg_Warn( p_dec, "unable to do requested conversion" );
 177         }
 178
 179         if( val.psz_string ) free( val.psz_string );
 180     }
 181 #else
 182
 183     msg_Dbg( p_dec, "no iconv support available" );
 184 #endif
 185
 186     return VLC_SUCCESS;
 187 }
 188
 189 /****************************************************************************
 190  * DecodeBlock: the whole thing
 191  ****************************************************************************
 192  * This function must be fed with complete subtitles units.
 193  ****************************************************************************/
 194 static void DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 195 {
 196     vout_thread_t *p_vout;
 197
 198     if( !pp_block || *pp_block == NULL )
 199     {
 200         return;
 201     }
 202
 203     /* Here we are dealing with text subtitles */
 204     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 205     if( p_vout )
 206     {
 207         ParseText( p_dec, *pp_block, p_vout );
 208         vlc_object_release( p_vout );
 209     }
 210     else
 211     {
 212         msg_Warn( p_dec, "couldn't find a video output, trashing subtitle" );
 213     }
 214
 215     block_Release( *pp_block );
 216     *pp_block = NULL;
 217 }
 218
 219 /*****************************************************************************
 220  * CloseDecoder: clean up the decoder
 221  *****************************************************************************/
 222 static void CloseDecoder( vlc_object_t *p_this )
 223 {
 224     decoder_t *p_dec = (decoder_t *)p_this;
 225     decoder_sys_t *p_sys = p_dec->p_sys;
 226     vout_thread_t *p_vout;
 227
 228     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 229     if( p_vout != NULL && p_vout->p_subpicture != NULL )
 230     {
 231         subpicture_t *p_subpic;
 232         int          i_subpic;
 233
 234         for( i_subpic = 0; i_subpic < VOUT_MAX_SUBPICTURES; i_subpic++ )
 235         {
 236             p_subpic = &p_vout->p_subpicture[i_subpic];
 237
 238             if( p_subpic != NULL &&
 239               ( p_subpic->i_status == RESERVED_SUBPICTURE
 240                 || p_subpic->i_status == READY_SUBPICTURE ) )
 241             {
 242                 vout_DestroySubPicture( p_vout, p_subpic );
 243             }
 244         }
 245     }
 246     if( p_vout ) vlc_object_release( p_vout );
 247
 248 #if defined(HAVE_ICONV)
 249     if( p_sys->iconv_handle != (iconv_t)-1 )
 250     {
 251         iconv_close( p_sys->iconv_handle );
 252     }
 253 #endif
 254
 255     free( p_sys );
 256 }
 257
 258 /*****************************************************************************
 259  * ParseText: parse an text subtitle packet and send it to the video output
 260  *****************************************************************************/
 261 static void ParseText( decoder_t *p_dec, block_t *p_block,
 262                        vout_thread_t *p_vout )
 263 {
 264     decoder_sys_t *p_sys = p_dec->p_sys;
 265     char *psz_subtitle;
 266     int i_align_h, i_align_v;
 267
 268     /* We cannot display a subpicture with no date */
 269     if( p_block->i_pts == 0 )
 270     {
 271         msg_Warn( p_dec, "subtitle without a date" );
 272         return;
 273     }
 274
 275     /* Check validity of packet data */
 276     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 277     {
 278         msg_Warn( p_dec, "empty subtitle" );
 279         return;
 280     }
 281
 282     /* Should be resiliant against bad subtitles */
 283     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 284
 285     i_align_h = p_sys->i_align ? 20 : 0;
 286     i_align_v = 10;
 287
 288 #if defined(HAVE_ICONV)
 289     if( p_sys->iconv_handle != (iconv_t)-1 )
 290     {
 291         char *psz_new_subtitle;
 292         char *psz_convert_buffer_out;
 293         char *psz_convert_buffer_in;
 294         size_t ret, inbytes_left, outbytes_left;
 295
 296         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 297         psz_convert_buffer_out = psz_new_subtitle;
 298         psz_convert_buffer_in = psz_subtitle;
 299         inbytes_left = strlen( psz_subtitle );
 300         outbytes_left = 6 * inbytes_left;
 301         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 302                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 303         *psz_convert_buffer_out = '\0';
 304
 305         if( inbytes_left )
 306         {
 307             msg_Warn( p_dec, "Failed to convert subtitle encoding, dropping subtitle.\nTry setting a different character-encoding for the subtitle." );
 308             free( psz_subtitle );
 309             return;
 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, MarginV, Effect, Text */
 323         char *psz_new_subtitle;
 324         char *psz_buffer_sub;
 325         int         i_comma;
 326         int         i_text;
 327
 328         psz_buffer_sub = psz_subtitle;
 329         for( ;; )
 330         {
 331             i_comma = 0;
 332             while( i_comma < 8 &&
 333                 *psz_buffer_sub != '\0' )
 334             {
 335                 if( *psz_buffer_sub == ',' )
 336                 {
 337                     i_comma++;
 338                 }
 339                 psz_buffer_sub++;
 340             }
 341             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 342             i_text = 0;
 343             while( psz_buffer_sub[0] != '\0' )
 344             {
 345                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] =='n' || psz_buffer_sub[1] =='N' ) )
 346                 {
 347                     psz_new_subtitle[i_text] = '\n';
 348                     i_text++;
 349                     psz_buffer_sub += 2;
 350                 }
 351                 else if( psz_buffer_sub[0] == '{' && psz_buffer_sub[1] == '\\' )
 352                 {
 353                     /* SSA control code */
 354                     while( psz_buffer_sub[0] != '\0' && psz_buffer_sub[0] != '}' )
 355                     {
 356                         psz_buffer_sub++;
 357                     }
 358                     psz_buffer_sub++;
 359                 }
 360                 else
 361                 {
 362                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 363                     i_text++;
 364                     psz_buffer_sub++;
 365                 }
 366             }
 367             psz_new_subtitle[i_text] = '\0';
 368             free( psz_subtitle );
 369             psz_subtitle = psz_new_subtitle;
 370             break;
 371         }
 372     }
 373     StripTags( psz_subtitle );
 374     vout_ShowTextAbsolute( p_vout, SUBT1_CHAN, psz_subtitle, NULL,
 375         OSD_ALIGN_BOTTOM | p_sys->i_align, i_align_h,
 376         i_align_v, p_block->i_pts,
 377         p_block->i_length ? p_block->i_pts + p_block->i_length : 0 );
 378
 379     free( psz_subtitle );
 380 }
 381
 382 static void StripTags( char *psz_text )
 383 {
 384     int i_left_moves = 0;
 385     vlc_bool_t b_inside_tag = VLC_FALSE;
 386     int i = 0;
 387     int i_tag_start = -1;
 388     while( psz_text[ i ] )
 389     {
 390         if( !b_inside_tag )
 391         {
 392             if( psz_text[ i ] == '<' )
 393             {
 394                 b_inside_tag = VLC_TRUE;
 395                 i_tag_start = i;
 396             }
 397             psz_text[ i - i_left_moves ] = psz_text[ i ];
 398         }
 399         else
 400         {
 401             if( ( psz_text[ i ] == ' ' ) ||
 402                 ( psz_text[ i ] == '\t' ) ||
 403                 ( psz_text[ i ] == '\n' ) ||
 404                 ( psz_text[ i ] == '\r' ) )
 405             {
 406                 b_inside_tag = VLC_FALSE;
 407                 i_tag_start = -1;
 408             }
 409             else if( psz_text[ i ] == '>' )
 410             {
 411                 i_left_moves += i - i_tag_start + 1;
 412                 i_tag_start = -1;
 413                 b_inside_tag = VLC_FALSE;
 414             }
 415             else
 416             {
 417                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 418             }
 419         }
 420         i++;
 421     }
 422     psz_text[ i - i_left_moves ] = '\0';
 423 }
 424