src/input/subtitles.c

   1 /*****************************************************************************
   2  * subtitles.c : subtitles detection
   3  *****************************************************************************
   4  * Copyright (C) 2003-2009 VLC authors and VideoLAN
   5  * $Id$
   6  *
   7  * Authors: Derk-Jan Hartman <hartman at videolan.org>
   8  * This is adapted code from the GPL'ed MPlayer (http://mplayerhq.hu)
   9  *
  10  * This program is free software; you can redistribute it and/or modify it
  11  * under the terms of the GNU Lesser General Public License as published by
  12  * the Free Software Foundation; either version 2.1 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 Lesser General Public License for more details.
  19  *
  20  * You should have received a copy of the GNU Lesser General Public License
  21  * along with this program; if not, write to the Free Software Foundation,
  22  * Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
  23  *****************************************************************************/
  24
  25 /**
  26  *  \file
  27  *  This file contains functions to detect subtitle files.
  28  */
  29
  30 #ifdef HAVE_CONFIG_H
  31 # include "config.h"
  32 #endif
  33
  34 #include <ctype.h> /* isalnum() */
  35 #include <unistd.h>
  36 #include <sys/stat.h>
  37
  38 #include <vlc_common.h>
  39 #include <vlc_fs.h>
  40 #include <vlc_url.h>
  41
  42 #include "input_internal.h"
  43
  44 /**
  45  * The possible extensions for subtitle files we support
  46  */
  47 static const char *const sub_exts[] = { SLAVE_SPU_EXTENSIONS, "" };
  48
  49 static void strcpy_trim( char *d, const char *s )
  50 {
  51     unsigned char c;
  52
  53     /* skip leading whitespace */
  54     while( ((c = *s) != '\0') && !isalnum(c) )
  55     {
  56         s++;
  57     }
  58     for(;;)
  59     {
  60         /* copy word */
  61         while( ((c = *s) != '\0') && isalnum(c) )
  62         {
  63             *d = tolower(c);
  64             s++; d++;
  65         }
  66         if( *s == 0 ) break;
  67         /* trim excess whitespace */
  68         while( ((c = *s) != '\0') && !isalnum(c) )
  69         {
  70             s++;
  71         }
  72         if( *s == 0 ) break;
  73         *d++ = ' ';
  74     }
  75     *d = 0;
  76 }
  77
  78 static void strcpy_strip_ext( char *d, const char *s )
  79 {
  80     unsigned char c;
  81
  82     const char *tmp = strrchr(s, '.');
  83     if( !tmp )
  84     {
  85         strcpy(d, s);
  86         return;
  87     }
  88     else
  89         strlcpy(d, s, tmp - s + 1 );
  90     while( (c = *d) != '\0' )
  91     {
  92         *d = tolower(c);
  93         d++;
  94     }
  95 }
  96
  97 static void strcpy_get_ext( char *d, const char *s )
  98 {
  99     const char *tmp = strrchr(s, '.');
 100     if( !tmp )
 101         strcpy(d, "");
 102     else
 103         strcpy( d, tmp + 1 );
 104 }
 105
 106 static int whiteonly( const char *s )
 107 {
 108     unsigned char c;
 109
 110     while( (c = *s) != '\0' )
 111     {
 112         if( isalnum( c ) )
 113             return 0;
 114         s++;
 115     }
 116     return 1;
 117 }
 118
 119 static int slave_strcmp( const void *a, const void *b )
 120 {
 121     const input_item_slave_t *p_slave0 = *((const input_item_slave_t **) a);
 122     const input_item_slave_t *p_slave1 = *((const input_item_slave_t **) b);
 123
 124     if( p_slave0 == NULL || p_slave1 == NULL )
 125         return 0;
 126
 127     /* We can compare these uris since they come from the file system */
 128 #ifdef HAVE_STRCOLL
 129     return strcoll( p_slave0->psz_uri, p_slave1->psz_uri );
 130 #else
 131     return strcmp( p_slave0->psz_uri, p_slave1->psz_uri );
 132 #endif
 133 }
 134
 135 /*
 136  * Check if a file ends with a subtitle extension
 137  */
 138 int subtitles_Filter( const char *psz_dir_content )
 139 {
 140     const char *tmp = strrchr( psz_dir_content, '.');
 141
 142     if( !tmp )
 143         return 0;
 144     tmp++;
 145
 146     for( int i = 0; sub_exts[i][0]; i++ )
 147         if( strcasecmp( sub_exts[i], tmp ) == 0 )
 148             return 1;
 149     return 0;
 150 }
 151
 152
 153 /**
 154  * Convert a list of paths separated by ',' to a char**
 155  */
 156 static char **paths_to_list( const char *psz_dir, char *psz_path )
 157 {
 158     unsigned int i, k, i_nb_subdirs;
 159     char **subdirs; /* list of subdirectories to look in */
 160     char *psz_parser = psz_path;
 161
 162     if( !psz_dir || !psz_path )
 163         return NULL;
 164
 165     for( k = 0, i_nb_subdirs = 1; psz_path[k] != '\0'; k++ )
 166     {
 167         if( psz_path[k] == ',' )
 168             i_nb_subdirs++;
 169     }
 170
 171     subdirs = calloc( i_nb_subdirs + 1, sizeof(char*) );
 172     if( !subdirs )
 173         return NULL;
 174
 175     for( i = 0; psz_parser && *psz_parser != '\0' ; )
 176     {
 177         char *psz_subdir = psz_parser;
 178         psz_parser = strchr( psz_subdir, ',' );
 179         if( psz_parser )
 180         {
 181             *psz_parser++ = '\0';
 182             while( *psz_parser == ' ' )
 183                 psz_parser++;
 184         }
 185         if( *psz_subdir == '\0' )
 186             continue;
 187
 188         if( asprintf( &subdirs[i++], "%s%s",
 189                   psz_subdir[0] == '.' ? psz_dir : "",
 190                   psz_subdir ) == -1 )
 191             break;
 192     }
 193     subdirs[i] = NULL;
 194
 195     return subdirs;
 196 }
 197
 198 /**
 199  * Detect subtitle files.
 200  *
 201  * When called this function will split up the psz_name string into a
 202  * directory, filename and extension. It then opens the directory
 203  * in which the file resides and tries to find possible matches of
 204  * subtitles files.
 205  *
 206  * \ingroup Demux
 207  * \param p_this the calling \ref input_thread_t
 208  * \param psz_path a list of subdirectories (separated by a ',') to look in.
 209  * \param psz_name_org the complete filename to base the search on.
 210  * \param pp_slaves an initialized input item slave list to append detected subtitles to
 211  * \param p_slaves pointer to the size of the slave list
 212  * \return VLC_SUCCESS if ok
 213  */
 214 int subtitles_Detect( input_thread_t *p_this, char *psz_path, const char *psz_name_org,
 215                       input_item_slave_t ***ppp_slaves, int *p_slaves )
 216 {
 217     int i_fuzzy = var_GetInteger( p_this, "sub-autodetect-fuzzy" );
 218     if ( i_fuzzy == 0 )
 219         return VLC_EGENERIC;
 220     int i_fname_len;
 221     input_item_slave_t **pp_slaves = *ppp_slaves;
 222     int i_slaves = *p_slaves;
 223     char *f_fname_noext = NULL, *f_fname_trim = NULL;
 224     char **subdirs; /* list of subdirectories to look in */
 225
 226     if( !psz_name_org )
 227         return VLC_EGENERIC;
 228
 229     char *psz_fname = vlc_uri2path( psz_name_org );
 230     if( !psz_fname )
 231         return VLC_EGENERIC;
 232
 233     /* extract filename & dirname from psz_fname */
 234     char *f_dir = strdup( psz_fname );
 235     if( f_dir == 0 )
 236     {
 237         free( psz_fname );
 238         return VLC_ENOMEM;
 239     }
 240
 241     const char *f_fname = strrchr( psz_fname, DIR_SEP_CHAR );
 242     if( !f_fname )
 243     {
 244         free( f_dir );
 245         free( psz_fname );
 246         return VLC_EGENERIC;
 247     }
 248     f_fname++; /* Skip the '/' */
 249     f_dir[f_fname - psz_fname] = 0; /* keep dir separator in f_dir */
 250
 251     i_fname_len = strlen( f_fname );
 252
 253     f_fname_noext = malloc(i_fname_len + 1);
 254     f_fname_trim = malloc(i_fname_len + 1 );
 255     if( !f_fname_noext || !f_fname_trim )
 256     {
 257         free( f_dir );
 258         free( f_fname_noext );
 259         free( f_fname_trim );
 260         free( psz_fname );
 261         return VLC_ENOMEM;
 262     }
 263
 264     strcpy_strip_ext( f_fname_noext, f_fname );
 265     strcpy_trim( f_fname_trim, f_fname_noext );
 266
 267     subdirs = paths_to_list( f_dir, psz_path );
 268     for( int j = -1; (j == -1) || ( j >= 0 && subdirs != NULL && subdirs[j] != NULL ); j++ )
 269     {
 270         const char *psz_dir = (j < 0) ? f_dir : subdirs[j];
 271         if( psz_dir == NULL || ( j >= 0 && !strcmp( psz_dir, f_dir ) ) )
 272             continue;
 273
 274         /* parse psz_src dir */
 275         DIR *dir = vlc_opendir( psz_dir );
 276         if( dir == NULL )
 277             continue;
 278
 279         msg_Dbg( p_this, "looking for a subtitle file in %s", psz_dir );
 280
 281         const char *psz_name;
 282         while( (psz_name = vlc_readdir( dir )) )
 283         {
 284             if( psz_name[0] == '.' || !subtitles_Filter( psz_name ) )
 285                 continue;
 286
 287             char tmp_fname_noext[strlen( psz_name ) + 1];
 288             char tmp_fname_trim[strlen( psz_name ) + 1];
 289             char tmp_fname_ext[strlen( psz_name ) + 1];
 290             const char *tmp;
 291             int i_prio = 0;
 292
 293             /* retrieve various parts of the filename */
 294             strcpy_strip_ext( tmp_fname_noext, psz_name );
 295             strcpy_get_ext( tmp_fname_ext, psz_name );
 296             strcpy_trim( tmp_fname_trim, tmp_fname_noext );
 297
 298             if( !strcmp( tmp_fname_trim, f_fname_trim ) )
 299             {
 300                 /* matches the movie name exactly */
 301                 i_prio = SLAVE_PRIORITY_MATCH_ALL;
 302             }
 303             else if( (tmp = strstr( tmp_fname_trim, f_fname_trim )) )
 304             {
 305                 /* contains the movie name */
 306                 tmp += strlen( f_fname_trim );
 307                 if( whiteonly( tmp ) )
 308                 {
 309                     /* chars in front of the movie name */
 310                     i_prio = SLAVE_PRIORITY_MATCH_RIGHT;
 311                 }
 312                 else
 313                 {
 314                     /* chars after (and possibly in front of)
 315                      * the movie name */
 316                     i_prio = SLAVE_PRIORITY_MATCH_LEFT;
 317                 }
 318             }
 319             else if( j == -1 )
 320             {
 321                 /* doesn't contain the movie name, prefer files in f_dir over subdirs */
 322                 i_prio = SLAVE_PRIORITY_MATCH_NONE;
 323             }
 324             if( i_prio >= i_fuzzy )
 325             {
 326                 struct stat st;
 327                 char *path;
 328
 329                 size_t i_len = strlen( psz_dir );
 330                 const char *psz_format;
 331                 if ( i_len == 0 )
 332                     continue;
 333                 if( psz_dir[i_len - 1] == DIR_SEP_CHAR )
 334                     psz_format = "%s%s";
 335                 else
 336                     psz_format = "%s"DIR_SEP"%s";
 337
 338                 if( asprintf( &path, psz_format, psz_dir, psz_name ) < 0 )
 339                     continue;
 340
 341                 if( strcmp( path, psz_fname )
 342                  && vlc_stat( path, &st ) == 0
 343                  && S_ISREG( st.st_mode ) )
 344                 {
 345                     msg_Dbg( p_this,
 346                             "autodetected subtitle: %s with priority %d",
 347                             path, i_prio );
 348                     char *psz_uri = vlc_path2uri( path, NULL );
 349                     input_item_slave_t *p_sub = psz_uri != NULL ?
 350                         input_item_slave_New( psz_uri, SLAVE_TYPE_SPU, i_prio )
 351                         : NULL;
 352                     if( p_sub )
 353                     {
 354                         p_sub->b_forced = true;
 355                         TAB_APPEND(i_slaves, pp_slaves, p_sub);
 356                     }
 357                     free( psz_uri );
 358                 }
 359                 free( path );
 360             }
 361         }
 362         closedir( dir );
 363     }
 364     if( subdirs )
 365     {
 366         for( size_t j = 0; subdirs[j] != NULL; j++ )
 367             free( subdirs[j] );
 368         free( subdirs );
 369     }
 370     free( f_dir );
 371     free( f_fname_trim );
 372     free( f_fname_noext );
 373     free( psz_fname );
 374
 375     for( int i = 0; i < i_slaves; i++ )
 376     {
 377         input_item_slave_t *p_sub = pp_slaves[i];
 378
 379         bool b_reject = false;
 380         char *psz_ext = strrchr( p_sub->psz_uri, '.' );
 381         if( !psz_ext )
 382             continue;
 383         psz_ext++;
 384
 385         if( !strcasecmp( psz_ext, "sub" ) )
 386         {
 387             for( int j = 0; j < i_slaves; j++ )
 388             {
 389                 input_item_slave_t *p_sub_inner = pp_slaves[j];
 390
 391                 /* A slave can be null if it's already rejected */
 392                 if( p_sub_inner == NULL )
 393                     continue;
 394
 395                 /* check that the filenames without extension match */
 396                 if( strncasecmp( p_sub->psz_uri, p_sub_inner->psz_uri,
 397                     strlen( p_sub->psz_uri ) - 3 ) )
 398                     continue;
 399
 400                 char *psz_ext_inner = strrchr( p_sub_inner->psz_uri, '.' );
 401                 if( !psz_ext_inner )
 402                     continue;
 403                 psz_ext_inner++;
 404
 405                 /* check that we have an idx file */
 406                 if( !strcasecmp( psz_ext_inner, "idx" ) )
 407                 {
 408                     b_reject = true;
 409                     break;
 410                 }
 411             }
 412         }
 413         else if( !strcasecmp( psz_ext, "cdg" ) )
 414         {
 415             if( p_sub->i_priority < SLAVE_PRIORITY_MATCH_ALL )
 416                 b_reject = true;
 417         }
 418         if( b_reject )
 419         {
 420             pp_slaves[i] = NULL;
 421             input_item_slave_Delete( p_sub );
 422         }
 423     }
 424
 425     /* Sort alphabetically */
 426     if( i_slaves > 0 )
 427         qsort( pp_slaves, i_slaves, sizeof (input_item_slave_t*), slave_strcmp );
 428
 429     *ppp_slaves = pp_slaves; /* in case of realloc */
 430     *p_slaves = i_slaves;
 431     return VLC_SUCCESS;
 432 }