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: subsdec.c,v 1.16 2004/01/25 18:20:12 bigben Exp $
   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
  62 #define DEFAULT_NAME "System Default"
  63
  64 /*****************************************************************************
  65  * Module descriptor.
  66  *****************************************************************************/
  67 #if defined(HAVE_ICONV)
  68 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  69     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  70     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  71     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  72     "ISO-8859-6", "CP1256", "MacArabic", "",
  73     "ISO-8859-7", "CP1253", "MacGreek", "",
  74     "ISO-8859-8", "CP1255", "MacHebrew", "",
  75     "ISO-8859-9", "CP1254", "MacTurkish", "",
  76     "ISO-8859-13", "CP1257", "",
  77     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  78     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  79     "ISO-2022-KR", "EUC-KR", "",
  80     "MacThai", "KOI8-T", "",
  81     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  82     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  83     "Macintosh", "",
  84     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  85     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  86     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  87     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  88     "HPROMAN8", "NEXTSTEP" };
  89 #endif
  90
  91 static int  pi_justification[] = { 0, 1, 2 };
  92 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  93
  94 #define ENCODING_TEXT N_("Subtitles text encoding")
  95 #define ENCODING_LONGTEXT N_("Set the encoding used in text subtitles")
  96 #define ALIGN_TEXT N_("Subtitles justification")
  97 #define ALIGN_LONGTEXT N_("Set the justification of substitles")
  98
  99 vlc_module_begin();
 100     set_description( _("text subtitles decoder") );
 101     set_capability( "decoder", 50 );
 102     set_callbacks( OpenDecoder, CloseDecoder );
 103
 104     add_category_hint( N_("Subtitles"), NULL, VLC_FALSE );
 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     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 = 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 = iconv_open( "UTF-8", val.psz_string );
 163     }
 164
 165     if( p_sys->iconv_handle == (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 #else
 172
 173     msg_Dbg( p_dec, "no iconv support available" );
 174 #endif
 175
 176 #if 0
 177     if( p_demux_data )
 178         msg_Dbg( p_dec, p_demux_data->psz_header );
 179 #endif
 180
 181     return VLC_SUCCESS;
 182 }
 183
 184 /****************************************************************************
 185  * DecodeBlock: the whole thing
 186  ****************************************************************************
 187  * This function must be fed with complete subtitles units.
 188  ****************************************************************************/
 189 static void DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 190 {
 191     vout_thread_t *p_vout;
 192
 193     if( !pp_block || *pp_block == NULL )
 194     {
 195         return;
 196     }
 197
 198     /* Here we are dealing with text subtitles */
 199     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 200     if( p_vout )
 201     {
 202         ParseText( p_dec, *pp_block, p_vout );
 203         vlc_object_release( p_vout );
 204     }
 205     else
 206     {
 207         msg_Warn( p_dec, "couldn't find a video output, trashing subtitle" );
 208     }
 209
 210     block_Release( *pp_block );
 211     *pp_block = NULL;
 212 }
 213
 214 /*****************************************************************************
 215  * CloseDecoder: clean up the decoder
 216  *****************************************************************************/
 217 static void CloseDecoder( vlc_object_t *p_this )
 218 {
 219     decoder_t *p_dec = (decoder_t *)p_this;
 220     decoder_sys_t *p_sys = p_dec->p_sys;
 221     vout_thread_t *p_vout;
 222
 223     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 224     if( p_vout != NULL && p_vout->p_subpicture != NULL )
 225     {
 226         subpicture_t *p_subpic;
 227         int          i_subpic;
 228
 229         for( i_subpic = 0; i_subpic < VOUT_MAX_SUBPICTURES; i_subpic++ )
 230         {
 231             p_subpic = &p_vout->p_subpicture[i_subpic];
 232
 233             if( p_subpic != NULL &&
 234               ( p_subpic->i_status == RESERVED_SUBPICTURE
 235                 || p_subpic->i_status == READY_SUBPICTURE ) )
 236             {
 237                 vout_DestroySubPicture( p_vout, p_subpic );
 238             }
 239         }
 240     }
 241     if( p_vout ) vlc_object_release( p_vout );
 242
 243 #if defined(HAVE_ICONV)
 244     if( p_sys->iconv_handle != (iconv_t)-1 )
 245     {
 246         iconv_close( p_sys->iconv_handle );
 247     }
 248 #endif
 249
 250     free( p_sys );
 251 }
 252
 253 /*****************************************************************************
 254  * ParseText: parse an text subtitle packet and send it to the video output
 255  *****************************************************************************/
 256 static void ParseText( decoder_t *p_dec, block_t *p_block,
 257                        vout_thread_t *p_vout )
 258 {
 259     decoder_sys_t *p_sys = p_dec->p_sys;
 260     char *psz_subtitle;
 261     int i_align_h, i_align_v;
 262
 263     /* We cannot display a subpicture with no date */
 264     if( p_block->i_pts == 0 )
 265     {
 266         msg_Warn( p_dec, "subtitle without a date" );
 267         return;
 268     }
 269
 270     /* Check validity of packet data */
 271     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 272     {
 273         msg_Warn( p_dec, "empty subtitle" );
 274         return;
 275     }
 276
 277     /* Should be resiliant against bad subtitles */
 278     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 279
 280     i_align_h = p_sys->i_align ? 20 : 0;
 281     i_align_v = 10;
 282
 283 #if defined(HAVE_ICONV)
 284     if( p_sys->iconv_handle != (iconv_t)-1 )
 285     {
 286         char *psz_new_subtitle;
 287         char *psz_convert_buffer_out;
 288         char *psz_convert_buffer_in;
 289         size_t ret, inbytes_left, outbytes_left;
 290
 291         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 292         psz_convert_buffer_out = psz_new_subtitle;
 293         psz_convert_buffer_in = psz_subtitle;
 294         inbytes_left = strlen( psz_subtitle );
 295         outbytes_left = 6 * inbytes_left;
 296         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 297                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 298         *psz_convert_buffer_out = '\0';
 299
 300         if( inbytes_left )
 301         {
 302             msg_Warn( p_dec, "Failed to convert subtitle encoding, dropping subtitle.\nTry setting a different character-encoding for the subtitle." );
 303             free( psz_subtitle );
 304             return;
 305         }
 306         else
 307         {
 308             free( psz_subtitle );
 309             psz_subtitle = psz_new_subtitle;
 310         }
 311     }
 312 #endif
 313
 314     if( p_dec->fmt_in.i_codec == VLC_FOURCC('s','s','a',' ') )
 315     {
 316         /* Decode SSA strings */
 317         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR, MarginV, Effect, Text */
 318         char *psz_new_subtitle;
 319         char *psz_buffer_sub;
 320         int         i_comma;
 321         int         i_text;
 322
 323         psz_buffer_sub = psz_subtitle;
 324         for( ;; )
 325         {
 326             i_comma = 0;
 327             while( i_comma < 8 &&
 328                 *psz_buffer_sub != '\0' )
 329             {
 330                 if( *psz_buffer_sub == ',' )
 331                 {
 332                     i_comma++;
 333                 }
 334                 psz_buffer_sub++;
 335             }
 336             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 337             i_text = 0;
 338             while( psz_buffer_sub[0] != '\0' )
 339             {
 340                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] =='n' || psz_buffer_sub[1] =='N' ) )
 341                 {
 342                     psz_new_subtitle[i_text] = '\n';
 343                     i_text++;
 344                     psz_buffer_sub += 2;
 345                 }
 346                 else if( psz_buffer_sub[0] == '{' && psz_buffer_sub[1] == '\\' )
 347                 {
 348                     /* SSA control code */
 349                     while( psz_buffer_sub[0] != '\0' && psz_buffer_sub[0] != '}' )
 350                     {
 351                         psz_buffer_sub++;
 352                     }
 353                     psz_buffer_sub++;
 354                 }
 355                 else
 356                 {
 357                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 358                     i_text++;
 359                     psz_buffer_sub++;
 360                 }
 361             }
 362             psz_new_subtitle[i_text] = '\0';
 363             free( psz_subtitle );
 364             psz_subtitle = psz_new_subtitle;
 365             break;
 366         }
 367     }
 368
 369     vout_ShowTextAbsolute( p_vout, psz_subtitle, NULL,
 370                            OSD_ALIGN_BOTTOM | p_sys->i_align,
 371                            i_align_h, i_align_v,
 372                            p_block->i_pts, p_block->i_dts );
 373
 374     free( psz_subtitle );
 375 }