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 1.7 2003/11/16 21:07:30 gbazin 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         free( psz_charset );
 160     }
 161     else
 162     {
 163         p_sys->iconv_handle = iconv_open( "UTF-8", val.psz_string );
 164     }
 165
 166     if( p_sys->iconv_handle == (iconv_t)-1 )
 167     {
 168         msg_Warn( p_dec, "Unable to do requested conversion" );
 169     }
 170
 171     if( val.psz_string ) free( val.psz_string );
 172 #else
 173
 174     msg_Dbg( p_dec, "No iconv support available" );
 175 #endif
 176
 177 #if 0
 178     if( p_demux_data )
 179         msg_Dbg( p_dec, p_demux_data->psz_header );
 180 #endif
 181
 182     return VLC_SUCCESS;
 183 }
 184
 185 /****************************************************************************
 186  * DecodeBlock: the whole thing
 187  ****************************************************************************
 188  * This function must be fed with complete subtitles units.
 189  ****************************************************************************/
 190 static void DecodeBlock( decoder_t *p_dec, block_t **pp_block )
 191 {
 192     vout_thread_t *p_vout;
 193
 194     /* Here we are dealing with text subtitles */
 195     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 196     if( !p_vout )
 197     {
 198         msg_Warn( p_dec, "couldn't find a video output, trashing subtitle" );
 199     }
 200
 201     ParseText( p_dec, *pp_block, p_vout );
 202     vlc_object_release( p_vout );
 203 }
 204
 205 /*****************************************************************************
 206  * CloseDecoder: clean up the decoder
 207  *****************************************************************************/
 208 static void CloseDecoder( vlc_object_t *p_this )
 209 {
 210     decoder_t *p_dec = (decoder_t *)p_this;
 211     decoder_sys_t *p_sys = p_dec->p_sys;
 212     vout_thread_t *p_vout;
 213
 214     p_vout = vlc_object_find( p_dec, VLC_OBJECT_VOUT, FIND_ANYWHERE );
 215     if( p_vout != NULL && p_vout->p_subpicture != NULL )
 216     {
 217         subpicture_t *p_subpic;
 218         int          i_subpic;
 219
 220         for( i_subpic = 0; i_subpic < VOUT_MAX_SUBPICTURES; i_subpic++ )
 221         {
 222             p_subpic = &p_vout->p_subpicture[i_subpic];
 223
 224             if( p_subpic != NULL &&
 225               ( p_subpic->i_status == RESERVED_SUBPICTURE
 226                 || p_subpic->i_status == READY_SUBPICTURE ) )
 227             {
 228                 vout_DestroySubPicture( p_vout, p_subpic );
 229             }
 230         }
 231     }
 232     if( p_vout ) vlc_object_release( p_vout );
 233
 234 #if defined(HAVE_ICONV)
 235     if( p_sys->iconv_handle != (iconv_t)-1 )
 236     {
 237         iconv_close( p_sys->iconv_handle );
 238     }
 239 #endif
 240
 241     free( p_sys );
 242 }
 243
 244 /*****************************************************************************
 245  * ParseText: parse an text subtitle packet and send it to the video output
 246  *****************************************************************************/
 247 static void ParseText( decoder_t *p_dec, block_t *p_block,
 248                        vout_thread_t *p_vout )
 249 {
 250     decoder_sys_t *p_sys = p_dec->p_sys;
 251     char *psz_subtitle;
 252     int i_align_h, i_align_v;
 253
 254     /* We cannot display a subpicture with no date */
 255     if( p_block->i_pts == 0 )
 256     {
 257         msg_Warn( p_dec, "subtitle without a date" );
 258         return;
 259     }
 260
 261     /* Check validity of packet data */
 262     if( p_block->i_buffer <= 1 ||  p_block->p_buffer[0] == '\0' )
 263     {
 264         msg_Warn( p_dec, "empty subtitle" );
 265         return;
 266     }
 267
 268     /* Should be resiliant against bad subtitles */
 269     psz_subtitle = strndup( p_block->p_buffer, p_block->i_buffer );
 270
 271     i_align_h = p_sys->i_align ? 20 : 0;
 272     i_align_v = 10;
 273
 274 #if defined(HAVE_ICONV)
 275     if( p_sys->iconv_handle != (iconv_t)-1 )
 276     {
 277         char *psz_new_subtitle;
 278         char *psz_convert_buffer_out;
 279         const char *psz_convert_buffer_in;
 280         size_t ret, inbytes_left, outbytes_left;
 281
 282         psz_new_subtitle = malloc( 6 * strlen( psz_subtitle ) );
 283         psz_convert_buffer_out = psz_new_subtitle;
 284         psz_convert_buffer_in = psz_subtitle;
 285         inbytes_left = strlen( psz_subtitle );
 286         outbytes_left = 6 * inbytes_left;
 287         ret = iconv( p_sys->iconv_handle, &psz_convert_buffer_in,
 288                      &inbytes_left, &psz_convert_buffer_out, &outbytes_left );
 289         *psz_convert_buffer_out = '\0';
 290
 291         if( inbytes_left )
 292         {
 293             msg_Warn( p_dec, "Something fishy happened during conversion" );
 294         }
 295         else
 296         {
 297             free( psz_subtitle );
 298             psz_subtitle = psz_new_subtitle;
 299         }
 300     }
 301 #endif
 302
 303     if( p_dec->p_fifo->i_fourcc == VLC_FOURCC('s','s','a',' ') )
 304     {
 305         /* Decode SSA strings */
 306         /* We expect: ReadOrder, Layer, Style, Name, MarginL, MarginR, MarginV, Effect, Text */
 307         char *psz_new_subtitle;
 308         char *psz_buffer_sub;
 309         int         i_comma;
 310         int         i_text;
 311
 312         psz_buffer_sub = psz_subtitle;
 313         for( ;; )
 314         {
 315             i_comma = 0;
 316             while( i_comma < 8 &&
 317                 *psz_buffer_sub != '\0' )
 318             {
 319                 if( *psz_buffer_sub == ',' )
 320                 {
 321                     i_comma++;
 322                 }
 323                 psz_buffer_sub++;
 324             }
 325             psz_new_subtitle = malloc( strlen( psz_buffer_sub ) + 1);
 326             i_text = 0;
 327             while( psz_buffer_sub[0] != '\0' )
 328             {
 329                 if( psz_buffer_sub[0] == '\\' && ( psz_buffer_sub[1] =='n' || psz_buffer_sub[1] =='N' ) )
 330                 {
 331                     psz_new_subtitle[i_text] = '\n';
 332                     i_text++;
 333                     psz_buffer_sub += 2;
 334                 }
 335                 else if( psz_buffer_sub[0] == '{' && psz_buffer_sub[1] == '\\' )
 336                 {
 337                     /* SSA control code */
 338                     while( psz_buffer_sub[0] != '\0' && psz_buffer_sub[0] != '}' )
 339                     {
 340                         psz_buffer_sub++;
 341                     }
 342                     psz_buffer_sub++;
 343                 }
 344                 else
 345                 {
 346                     psz_new_subtitle[i_text] = psz_buffer_sub[0];
 347                     i_text++;
 348                     psz_buffer_sub++;
 349                 }
 350             }
 351             psz_new_subtitle[i_text] = '\0';
 352             free( psz_subtitle );
 353             psz_subtitle = psz_new_subtitle;
 354             break;
 355         }
 356     }
 357
 358     vout_ShowTextAbsolute( p_vout, psz_subtitle, NULL,
 359                            OSD_ALIGN_BOTTOM | p_sys->i_align,
 360                            i_align_h, i_align_v,
 361                            p_block->i_pts, p_block->i_dts );
 362
 363     free( psz_subtitle );
 364 }