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 1.10 2003/11/19 13:25:48 hartman 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 <stdlib.h>                                      /* malloc(), free() */
  29 #include <string.h>                                    /* memcpy(), memset() */
  30
  31 #include <vlc/vlc.h>
  32 #include <vlc/vout.h>
  33 #include <vlc/decoder.h>
  34 #include <osd.h>
  35 #include <codecs.h>
  36
  37 #if defined(HAVE_ICONV)
  38 #include <iconv.h>
  39 #endif
  40
  41 #include "charset.h"
  42
  43 /*****************************************************************************
  44  * decoder_sys_t : decoder descriptor
  45  *****************************************************************************/
  46 struct decoder_sys_t
  47 {
  48     int                 i_align;          /* Subtitles alignment on the vout */
  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
  65 #define DEFAULT_NAME "System Default"
  66
  67 /*****************************************************************************
  68  * Module descriptor.
  69  *****************************************************************************/
  70 #if defined(HAVE_ICONV)
  71 static char *ppsz_encodings[] = { DEFAULT_NAME, "ASCII", "UTF-8", "",
  72     "ISO-8859-1", "CP1252", "MacRoman", "MacIceland","ISO-8859-15", "",
  73     "ISO-8859-2", "CP1250", "MacCentralEurope", "MacCroatian", "MacRomania", "",
  74     "ISO-8859-5", "CP1251", "MacCyrillic", "MacUkraine", "KOI8-R", "KOI8-U", "KOI8-RU", "",
  75     "ISO-8859-6", "CP1256", "MacArabic", "",
  76     "ISO-8859-7", "CP1253", "MacGreek", "",
  77     "ISO-8859-8", "CP1255", "MacHebrew", "",
  78     "ISO-8859-9", "CP1254", "MacTurkish", "",
  79     "ISO-8859-13", "CP1257", "",
  80     "ISO-2022-JP", "ISO-2022-JP-1", "ISO-2022-JP-2", "EUC-JP", "SHIFT_JIS", "",
  81     "ISO-2022-CN", "ISO-2022-CN-EXT", "EUC-CN", "EUC-TW", "BIG5", "BIG5-HKSCS", "",
  82     "ISO-2022-KR", "EUC-KR", "",
  83     "MacThai", "KOI8-T", "",
  84     "ISO-8859-3", "ISO-8859-4", "ISO-8859-10", "ISO-8859-14", "ISO-8859-16", "",
  85     "CP850", "CP862", "CP866", "CP874", "CP932", "CP949", "CP950", "CP1133", "CP1258", "",
  86     "Macintosh", "",
  87     "UTF-7", "UTF-16", "UTF-16BE", "UTF-16LE", "UTF-32", "UTF-32BE", "UTF-32LE",
  88     "C99", "JAVA", "UCS-2", "UCS-2BE", "UCS-2LE", "UCS-4", "UCS-4BE", "UCS-4LE", "",
  89     "HZ", "GBK", "GB18030", "JOHAB", "ARMSCII-8",
  90     "Georgian-Academy", "Georgian-PS", "TIS-620", "MuleLao-1", "VISCII", "TCVN",
  91     "HPROMAN8", "NEXTSTEP" };
  92 #endif
  93
  94 static int  pi_justification[] = { 0, 1, 2 };
  95 static char *ppsz_justification_text[] = {N_("Center"),N_("Left"),N_("Right")};
  96
  97 #define ENCODING_TEXT N_("Subtitles text encoding")
  98 #define ENCODING_LONGTEXT N_("Change the encoding used in text subtitles")
  99 #define ALIGN_TEXT N_("Subtitles justification")
 100 #define ALIGN_LONGTEXT N_("Change the justification of substitles")
 101
 102 vlc_module_begin();
 103     set_description( _("text subtitles decoder") );
 104     set_capability( "decoder", 50 );
 105     set_callbacks( OpenDecoder, CloseDecoder );
 106
 107     add_category_hint( N_("Subtitles"), NULL, VLC_FALSE );
 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", "UTF-8", 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     var_Create( p_dec, "subsdec-encoding",
 152                 VLC_VAR_STRING | VLC_VAR_DOINHERIT );
 153     var_Get( p_dec, "subsdec-encoding", &val );
 154     if( !strcmp( val.psz_string, DEFAULT_NAME ) )
 155     {
 156         char *psz_charset =(char*)malloc( 100 );
 157         vlc_current_charset( &psz_charset );
 158         p_sys->iconv_handle = iconv_open( "UTF-8", psz_charset );
 159         msg_Dbg( p_dec, "Using character encoding: %s", psz_charset );
 160         free( psz_charset );
 161     }
 162     else if( val.psz_string )
 163     {
 164         msg_Dbg( p_dec, "Using character encoding: %s", val.psz_string );
 165         p_sys->iconv_handle = iconv_open( "UTF-8", val.psz_string );
 166     }
 167
 168     if( p_sys->iconv_handle == (iconv_t)-1 )
 169     {
 170         msg_Warn( p_dec, "Unable to do requested conversion" );
 171     }
 172
 173     if( val.psz_string ) free( val.psz_string );
 174 #else
 175
 176     msg_Dbg( p_dec, "No iconv support available" );
 177 #endif
 178
 179 #if 0
 180     if( p_demux_data )
 181         msg_Dbg( p_dec, p_demux_data->psz_header );
 182 #endif
 183
 184     return VLC_SUCCESS;
 185 }
 186
 187 /****************************************************************************
 188  * DecodeBlock: the whole thing
 189  ****************************************************************************
 190  * This function must be fed with complete subtitles units.
 191  ****************************************************************************/
 192 static void DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 193 {
 194     vout_thread_t *p_vout;
 195
 196     if( !pp_block || *pp_block == NULL )
 197     {
 198         return;
 199     }
 200
 201     /* Here we are dealing with text subtitles */
 202     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 203     if( !p_vout )
 204     {
 205         msg_Warn( p_dec, "couldn't find a video output, trashing subtitle" );
 206         return;
 207     }
 208
 209     ParseText( p_dec, *pp_block, p_vout );
 210     vlc_object_release( p_vout );
 211
 212     block_Release( *pp_block );
 213     *pp_block = NULL;
 214 }
 215
 216 /*****************************************************************************
 217  * CloseDecoder: clean up the decoder
 218  *****************************************************************************/
 219 static void CloseDecoder( vlc_object_t *p_this )
 220 {
 221     decoder_t *p_dec = (decoder_t *)p_this;
 222     decoder_sys_t *p_sys = p_dec->p_sys;
 223     vout_thread_t *p_vout;
 224
 225     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 226     if( p_vout != NULL && p_vout->p_subpicture != NULL )
 227     {
 228         subpicture_t *p_subpic;
 229         int          i_subpic;
 230
 231         for( i_subpic = 0; i_subpic < VOUT_MAX_SUBPICTURES; i_subpic++ )
 232         {
 233             p_subpic = &p_vout->p_subpicture[i_subpic];
 234
 235             if( p_subpic != NULL &&
 236               ( p_subpic->i_status == RESERVED_SUBPICTURE
 237                 || p_subpic->i_status == READY_SUBPICTURE ) )
 238             {
 239                 vout_DestroySubPicture( p_vout, p_subpic );
 240             }
 241         }
 242     }
 243     if( p_vout ) vlc_object_release( p_vout );
 244
 245 #if defined(HAVE_ICONV)
 246     if( p_sys->iconv_handle != (iconv_t)-1 )
 247     {
 248         iconv_close( p_sys->iconv_handle );
 249     }
 250 #endif
 251
 252     free( p_sys );
 253 }
 254
 255 /*****************************************************************************
 256  * ParseText: parse an text subtitle packet and send it to the video output
 257  *****************************************************************************/
 258 static void ParseText( decoder_t *p_dec, block_t *p_block,
 259                        vout_thread_t *p_vout )
 260 {
 261     decoder_sys_t *p_sys = p_dec->p_sys;
 262     char *psz_subtitle;
 263     int i_align_h, i_align_v;
 264
 265     /* We cannot display a subpicture with no date */
 266     if( p_block->i_pts == 0 )
 267     {
 268         msg_Warn( p_dec, "subtitle without a date" );
 269         return;
 270     }
 271
 272     /* Check validity of packet data */
 273     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 274     {
 275         msg_Warn( p_dec, "empty subtitle" );
 276         return;
 277     }
 278
 279     /* Should be resiliant against bad subtitles */
 280     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 281
 282     i_align_h = p_sys->i_align ? 20 : 0;
 283     i_align_v = 10;
 284
 285 #if defined(HAVE_ICONV)
 286     if( p_sys->iconv_handle != (iconv_t)-1 )
 287     {
 288         char *psz_new_subtitle;
 289         char *psz_convert_buffer_out;
 290         char *psz_convert_buffer_in;
 291         size_t ret, inbytes_left, outbytes_left;
 292
 293         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 294         psz_convert_buffer_out = psz_new_subtitle;
 295         psz_convert_buffer_in = psz_subtitle;
 296         inbytes_left = strlen( psz_subtitle );
 297         outbytes_left = 6 * inbytes_left;
 298         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 299                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 300         *psz_convert_buffer_out = '\0';
 301
 302         if( inbytes_left )
 303         {
 304             msg_Warn( p_dec, "Failed to convert subtitle encoding, dropping subtitle" );
 305             free( psz_subtitle );
 306             return;
 307         }
 308         else
 309         {
 310             free( psz_subtitle );
 311             psz_subtitle = psz_new_subtitle;
 312         }
 313     }
 314 #endif
 315
 316     if( p_dec->p_fifo->i_fourcc == VLC_FOURCC('s','s','a',' ') )
 317     {
 318         /* Decode SSA strings */
 319         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR, MarginV, Effect, Text */
 320         char *psz_new_subtitle;
 321         char *psz_buffer_sub;
 322         int         i_comma;
 323         int         i_text;
 324
 325         psz_buffer_sub = psz_subtitle;
 326         for( ;; )
 327         {
 328             i_comma = 0;
 329             while( i_comma < 8 &&
 330                 *psz_buffer_sub != '\0' )
 331             {
 332                 if( *psz_buffer_sub == ',' )
 333                 {
 334                     i_comma++;
 335                 }
 336                 psz_buffer_sub++;
 337             }
 338             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 339             i_text = 0;
 340             while( psz_buffer_sub[0] != '\0' )
 341             {
 342                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] =='n' || psz_buffer_sub[1] =='N' ) )
 343                 {
 344                     psz_new_subtitle[i_text] = '\n';
 345                     i_text++;
 346                     psz_buffer_sub += 2;
 347                 }
 348                 else if( psz_buffer_sub[0] == '{' && psz_buffer_sub[1] == '\\' )
 349                 {
 350                     /* SSA control code */
 351                     while( psz_buffer_sub[0] != '\0' && psz_buffer_sub[0] != '}' )
 352                     {
 353                         psz_buffer_sub++;
 354                     }
 355                     psz_buffer_sub++;
 356                 }
 357                 else
 358                 {
 359                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 360                     i_text++;
 361                     psz_buffer_sub++;
 362                 }
 363             }
 364             psz_new_subtitle[i_text] = '\0';
 365             free( psz_subtitle );
 366             psz_subtitle = psz_new_subtitle;
 367             break;
 368         }
 369     }
 370
 371     vout_ShowTextAbsolute( p_vout, psz_subtitle, NULL,
 372                            OSD_ALIGN_BOTTOM | p_sys->i_align,
 373                            i_align_h, i_align_v,
 374                            p_block->i_pts, p_block->i_dts );
 375
 376     free( psz_subtitle );
 377 }