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     int                 i_subpic_channel;    /* Subpic channel for subtitles */
  47
  48     vout_thread_t       *p_vout;                           /* last vout used */
  49
  50 #if defined(HAVE_ICONV)
  51     iconv_t             iconv_handle;            /* handle to iconv instance */
  52 #endif
  53 };
  54
  55 /*****************************************************************************
  56  * Local prototypes
  57  *****************************************************************************/
  58 static int  OpenDecoder   ( vlc_object_t * );
  59 static void CloseDecoder  ( vlc_object_t * );
  60
  61 static void DecodeBlock   ( decoder_t *, block_t ** );
  62
  63 static void ParseText     ( decoder_t *, block_t *, vout_thread_t * );
  64 static void StripTags     ( char * );
  65
  66 #define DEFAULT_NAME "System Default"
  67
  68 /*****************************************************************************
  69  * Module descriptor.
  70  *****************************************************************************/
  71 #if defined(HAVE_ICONV)
  72 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  73     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  74     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  75     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  76     "ISO-8859-6", "CP1256", "MacArabic", "",
  77     "ISO-8859-7", "CP1253", "MacGreek", "",
  78     "ISO-8859-8", "CP1255", "MacHebrew", "",
  79     "ISO-8859-9", "CP1254", "MacTurkish", "",
  80     "ISO-8859-13", "CP1257", "",
  81     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  82     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  83     "ISO-2022-KR", "EUC-KR", "",
  84     "MacThai", "KOI8-T", "",
  85     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  86     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  87     "Macintosh", "",
  88     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  89     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  90     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  91     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  92     "HPROMAN8", "NEXTSTEP" };
  93 #endif
  94
  95 static int  pi_justification[] = { 0, 1, 2 };
  96 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  97
  98 #define ENCODING_TEXT N_("Subtitles text encoding")
  99 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
 100 #define ALIGN_TEXT N_("Subtitles justification")
 101 #define ALIGN_LONGTEXT N_("Set the justification of subtitles")
 102
 103 vlc_module_begin();
 104     set_description( _("text subtitles decoder") );
 105     set_capability( "decoder", 50 );
 106     set_callbacks( OpenDecoder, CloseDecoder );
 107
 108     add_integer( "subsdec-align", 0, NULL, ALIGN_TEXT, ALIGN_LONGTEXT,
 109                  VLC_TRUE );
 110         change_integer_list( pi_justification, ppsz_justification_text, 0 );
 111 #if defined(HAVE_ICONV)
 112     add_string( "subsdec-encoding", DEFAULT_NAME, NULL,
 113                 ENCODING_TEXT, ENCODING_LONGTEXT, VLC_FALSE );
 114         change_string_list( ppsz_encodings, 0, 0 );
 115 #endif
 116 vlc_module_end();
 117
 118 /*****************************************************************************
 119  * OpenDecoder: probe the decoder and return score
 120  *****************************************************************************
 121  * Tries to launch a decoder and return score so that the interface is able
 122  * to chose.
 123  *****************************************************************************/
 124 static int OpenDecoder( vlc_object_t *p_this )
 125 {
 126     decoder_t     *p_dec = (decoder_t*)p_this;
 127     decoder_sys_t *p_sys;
 128     vlc_value_t val;
 129
 130     if( p_dec->fmt_in.i_codec != VLC_FOURCC('s','u','b','t') &&
 131         p_dec->fmt_in.i_codec != VLC_FOURCC('s','s','a',' ') )
 132     {
 133         return VLC_EGENERIC;
 134     }
 135
 136     p_dec->pf_decode_sub = DecodeBlock;
 137
 138     /* Allocate the memory needed to store the decoder's structure */
 139     if( ( p_dec->p_sys = p_sys =
 140           (decoder_sys_t *)malloc(sizeof(decoder_sys_t)) ) == NULL )
 141     {
 142         msg_Err( p_dec, "out of memory" );
 143         return VLC_EGENERIC;
 144     }
 145
 146     var_Create( p_dec, "subsdec-align", VLC_VAR_INTEGER | VLC_VAR_DOINHERIT );
 147     var_Get( p_dec, "subsdec-align", &val );
 148     p_sys->i_align = val.i_int;
 149
 150 #if defined(HAVE_ICONV)
 151     if( p_dec->fmt_in.subs.psz_encoding && *p_dec->fmt_in.subs.psz_encoding )
 152     {
 153         msg_Dbg( p_dec, "using character encoding: %s",
 154                  p_dec->fmt_in.subs.psz_encoding );
 155         p_sys->iconv_handle = iconv_open( "UTF-8",
 156                                           p_dec->fmt_in.subs.psz_encoding );
 157     }
 158     else
 159     {
 160         var_Create( p_dec, "subsdec-encoding",
 161                     VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 162         var_Get( p_dec, "subsdec-encoding", &val );
 163         if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 164         {
 165             char *psz_charset =(char*)malloc( 100 );
 166             vlc_current_charset( &psz_charset );
 167             p_sys->iconv_handle = iconv_open( "UTF-8", psz_charset );
 168             msg_Dbg( p_dec, "using character encoding: %s", psz_charset );
 169             free( psz_charset );
 170         }
 171         else if( val.psz_string )
 172         {
 173             msg_Dbg( p_dec, "using character encoding: %s", val.psz_string );
 174             p_sys->iconv_handle = iconv_open( "UTF-8", val.psz_string );
 175         }
 176
 177         if( p_sys->iconv_handle == (iconv_t)-1 )
 178         {
 179             msg_Warn( p_dec, "unable to do requested conversion" );
 180         }
 181
 182         if( val.psz_string ) free( val.psz_string );
 183     }
 184 #else
 185
 186     msg_Dbg( p_dec, "no iconv support available" );
 187 #endif
 188
 189     p_dec->p_sys->p_vout = NULL;
 190
 191     return VLC_SUCCESS;
 192 }
 193
 194 /****************************************************************************
 195  * DecodeBlock: the whole thing
 196  ****************************************************************************
 197  * This function must be fed with complete subtitles units.
 198  ****************************************************************************/
 199 static void DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 200 {
 201     vout_thread_t *p_vout;
 202
 203     if( !pp_block || *pp_block == NULL )
 204     {
 205         return;
 206     }
 207
 208     /* Here we are dealing with text subtitles */
 209     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 210     if( p_vout )
 211     {
 212         if( p_dec->p_sys->p_vout != p_vout )
 213         {
 214             p_dec->p_sys->i_subpic_channel = vout_RegisterOSDChannel( p_vout );
 215         }
 216         ParseText( p_dec, *pp_block, p_vout );
 217         vlc_object_release( p_vout );
 218     }
 219     else
 220     {
 221         msg_Warn( p_dec, "couldn't find a video output, trashing subtitle" );
 222     }
 223     p_dec->p_sys->p_vout = p_vout;
 224
 225     block_Release( *pp_block );
 226     *pp_block = NULL;
 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     vout_thread_t *p_vout;
 237
 238     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 239     if( p_vout != NULL && p_vout->p_subpicture != NULL )
 240     {
 241         subpicture_t *p_subpic;
 242         int          i_subpic;
 243
 244         for( i_subpic = 0; i_subpic < VOUT_MAX_SUBPICTURES; i_subpic++ )
 245         {
 246             p_subpic = &p_vout->p_subpicture[i_subpic];
 247
 248             if( p_subpic != NULL &&
 249               ( p_subpic->i_status == RESERVED_SUBPICTURE
 250                 || p_subpic->i_status == READY_SUBPICTURE ) )
 251             {
 252                 vout_DestroySubPicture( p_vout, p_subpic );
 253             }
 254         }
 255     }
 256     if( p_vout ) vlc_object_release( p_vout );
 257
 258 #if defined(HAVE_ICONV)
 259     if( p_sys->iconv_handle != (iconv_t)-1 )
 260     {
 261         iconv_close( p_sys->iconv_handle );
 262     }
 263 #endif
 264
 265     free( p_sys );
 266 }
 267
 268 /*****************************************************************************
 269  * ParseText: parse an text subtitle packet and send it to the video output
 270  *****************************************************************************/
 271 static void ParseText( decoder_t *p_dec, block_t *p_block,
 272                        vout_thread_t *p_vout )
 273 {
 274     decoder_sys_t *p_sys = p_dec->p_sys;
 275     char *psz_subtitle;
 276     int i_align_h, i_align_v;
 277
 278     /* We cannot display a subpicture with no date */
 279     if( p_block->i_pts == 0 )
 280     {
 281         msg_Warn( p_dec, "subtitle without a date" );
 282         return;
 283     }
 284
 285     /* Check validity of packet data */
 286     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 287     {
 288         msg_Warn( p_dec, "empty subtitle" );
 289         return;
 290     }
 291
 292     /* Should be resiliant against bad subtitles */
 293     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 294
 295     i_align_h = p_sys->i_align ? 20 : 0;
 296     i_align_v = 10;
 297
 298 #if defined(HAVE_ICONV)
 299     if( p_sys->iconv_handle != (iconv_t)-1 )
 300     {
 301         char *psz_new_subtitle;
 302         char *psz_convert_buffer_out;
 303         char *psz_convert_buffer_in;
 304         size_t ret, inbytes_left, outbytes_left;
 305
 306         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 307         psz_convert_buffer_out = psz_new_subtitle;
 308         psz_convert_buffer_in = psz_subtitle;
 309         inbytes_left = strlen( psz_subtitle );
 310         outbytes_left = 6 * inbytes_left;
 311         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 312                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 313         *psz_convert_buffer_out = '\0';
 314
 315         if( inbytes_left )
 316         {
 317             msg_Warn( p_dec, "Failed to convert subtitle encoding, dropping subtitle.\nTry setting a different character-encoding for the subtitle." );
 318             free( psz_subtitle );
 319             return;
 320         }
 321         else
 322         {
 323             free( psz_subtitle );
 324             psz_subtitle = psz_new_subtitle;
 325         }
 326     }
 327 #endif
 328
 329     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 330     {
 331         /* Decode SSA strings */
 332         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR, MarginV, Effect, Text */
 333         char *psz_new_subtitle;
 334         char *psz_buffer_sub;
 335         int         i_comma;
 336         int         i_text;
 337
 338         psz_buffer_sub = psz_subtitle;
 339         for( ;; )
 340         {
 341             i_comma = 0;
 342             while( i_comma < 8 &&
 343                 *psz_buffer_sub != '\0' )
 344             {
 345                 if( *psz_buffer_sub == ',' )
 346                 {
 347                     i_comma++;
 348                 }
 349                 psz_buffer_sub++;
 350             }
 351             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 352             i_text = 0;
 353             while( psz_buffer_sub[0] != '\0' )
 354             {
 355                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] =='n' || psz_buffer_sub[1] =='N' ) )
 356                 {
 357                     psz_new_subtitle[i_text] = '\n';
 358                     i_text++;
 359                     psz_buffer_sub += 2;
 360                 }
 361                 else if( psz_buffer_sub[0] == '{' && psz_buffer_sub[1] == '\\' )
 362                 {
 363                     /* SSA control code */
 364                     while( psz_buffer_sub[0] != '\0' && psz_buffer_sub[0] != '}' )
 365                     {
 366                         psz_buffer_sub++;
 367                     }
 368                     psz_buffer_sub++;
 369                 }
 370                 else
 371                 {
 372                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 373                     i_text++;
 374                     psz_buffer_sub++;
 375                 }
 376             }
 377             psz_new_subtitle[i_text] = '\0';
 378             free( psz_subtitle );
 379             psz_subtitle = psz_new_subtitle;
 380             break;
 381         }
 382     }
 383     StripTags( psz_subtitle );
 384     vout_ShowTextAbsolute( p_vout, p_sys->i_subpic_channel, psz_subtitle, NULL,
 385         OSD_ALIGN_BOTTOM | p_sys->i_align, i_align_h,
 386         i_align_v, p_block->i_pts,
 387         p_block->i_length ? p_block->i_pts + p_block->i_length : 0 );
 388
 389     free( psz_subtitle );
 390 }
 391
 392 static void StripTags( char *psz_text )
 393 {
 394     int i_left_moves = 0;
 395     vlc_bool_t b_inside_tag = VLC_FALSE;
 396     int i = 0;
 397     int i_tag_start = -1;
 398     while( psz_text[ i ] )
 399     {
 400         if( !b_inside_tag )
 401         {
 402             if( psz_text[ i ] == '<' )
 403             {
 404                 b_inside_tag = VLC_TRUE;
 405                 i_tag_start = i;
 406             }
 407             psz_text[ i - i_left_moves ] = psz_text[ i ];
 408         }
 409         else
 410         {
 411             if( ( psz_text[ i ] == ' ' ) ||
 412                 ( psz_text[ i ] == '\t' ) ||
 413                 ( psz_text[ i ] == '\n' ) ||
 414                 ( psz_text[ i ] == '\r' ) )
 415             {
 416                 b_inside_tag = VLC_FALSE;
 417                 i_tag_start = -1;
 418             }
 419             else if( psz_text[ i ] == '>' )
 420             {
 421                 i_left_moves += i - i_tag_start + 1;
 422                 i_tag_start = -1;
 423                 b_inside_tag = VLC_FALSE;
 424             }
 425             else
 426             {
 427                 psz_text[ i - i_left_moves ] = psz_text[ i ];
 428             }
 429         }
 430         i++;
 431     }
 432     psz_text[ i - i_left_moves ] = '\0';
 433 }