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 dectect subtitle files.
  28  */
  29
  30 #ifdef HAVE_CONFIG_H
  31 # include "config.h"
  32 #endif
  33
  34 #include <ctype.h> /* isalnum() */
  35 #ifdef HAVE_UNISTD_H
  36 #   include <unistd.h>
  37 #endif
  38 #include <sys/stat.h>
  39
  40 #include <vlc_common.h>
  41 #include <vlc_fs.h>
  42 #include <vlc_url.h>
  43
  44 #include "input_internal.h"
  45
  46 /**
  47  * We are not going to autodetect more subtitle files than this.
  48  */
  49 #define MAX_SUBTITLE_FILES 128
  50
  51 /**
  52  * The possible extensions for subtitle files we support
  53  */
  54 static const char const sub_exts[][6] = {
  55     "idx", "sub",  "srt",
  56     "ssa", "ass",  "smi",
  57     "utf", "utf8", "utf-8",
  58     "rt",   "aqt", "txt",
  59     "usf", "jss",  "cdg",
  60     "psb", "mpsub","mpl2",
  61     "pjs", "dks", "stl",
  62     ""
  63 };
  64
  65 static void strcpy_trim( char *d, const char *s )
  66 {
  67     unsigned char c;
  68
  69     /* skip leading whitespace */
  70     while( ((c = *s) != '\0') && !isalnum(c) )
  71     {
  72         s++;
  73     }
  74     for(;;)
  75     {
  76         /* copy word */
  77         while( ((c = *s) != '\0') && isalnum(c) )
  78         {
  79             *d = tolower(c);
  80             s++; d++;
  81         }
  82         if( *s == 0 ) break;
  83         /* trim excess whitespace */
  84         while( ((c = *s) != '\0') && !isalnum(c) )
  85         {
  86             s++;
  87         }
  88         if( *s == 0 ) break;
  89         *d++ = ' ';
  90     }
  91     *d = 0;
  92 }
  93
  94 static void strcpy_strip_ext( char *d, const char *s )
  95 {
  96     unsigned char c;
  97
  98     const char *tmp = strrchr(s, '.');
  99     if( !tmp )
 100     {
 101         strcpy(d, s);
 102         return;
 103     }
 104     else
 105         strlcpy(d, s, tmp - s + 1 );
 106     while( (c = *d) != '\0' )
 107     {
 108         *d = tolower(c);
 109         d++;
 110     }
 111 }
 112
 113 static void strcpy_get_ext( char *d, const char *s )
 114 {
 115     const char *tmp = strrchr(s, '.');
 116     if( !tmp )
 117         strcpy(d, "");
 118     else
 119         strcpy( d, tmp + 1 );
 120 }
 121
 122 static int whiteonly( const char *s )
 123 {
 124     unsigned char c;
 125
 126     while( (c = *s) != '\0' )
 127     {
 128         if( isalnum( c ) )
 129             return 0;
 130         s++;
 131     }
 132     return 1;
 133 }
 134
 135 enum
 136 {
 137     SUB_PRIORITY_NONE        = 0,
 138     SUB_PRIORITY_MATCH_NONE  = 1,
 139     SUB_PRIORITY_MATCH_RIGHT = 2,
 140     SUB_PRIORITY_MATCH_LEFT  = 3,
 141     SUB_PRIORITY_MATCH_ALL   = 4,
 142 };
 143 typedef struct
 144 {
 145     int priority;
 146     char *psz_fname;
 147     char *psz_ext;
 148 } vlc_subfn_t;
 149
 150 static int compare_sub_priority( const void *a, const void *b )
 151 {
 152     const vlc_subfn_t *p0 = a;
 153     const vlc_subfn_t *p1 = b;
 154
 155     if( p0->priority > p1->priority )
 156         return -1;
 157
 158     if( p0->priority < p1->priority )
 159         return 1;
 160
 161 #ifdef HAVE_STRCOLL
 162     return strcoll( p0->psz_fname, p1->psz_fname);
 163 #else
 164     return strcmp( p0->psz_fname, p1->psz_fname);
 165 #endif
 166 }
 167
 168 /*
 169  * Check if a file ends with a subtitle extension
 170  */
 171 int subtitles_Filter( const char *psz_dir_content )
 172 {
 173     const char *tmp = strrchr( psz_dir_content, '.');
 174
 175     if( !tmp )
 176         return 0;
 177     tmp++;
 178
 179     for( int i = 0; sub_exts[i][0]; i++ )
 180         if( strcasecmp( sub_exts[i], tmp ) == 0 )
 181             return 1;
 182     return 0;
 183 }
 184
 185
 186 /**
 187  * Convert a list of paths separated by ',' to a char**
 188  */
 189 static char **paths_to_list( const char *psz_dir, char *psz_path )
 190 {
 191     unsigned int i, k, i_nb_subdirs;
 192     char **subdirs; /* list of subdirectories to look in */
 193     char *psz_parser = psz_path;
 194
 195     if( !psz_dir || !psz_path )
 196         return NULL;
 197
 198     for( k = 0, i_nb_subdirs = 1; psz_path[k] != '\0'; k++ )
 199     {
 200         if( psz_path[k] == ',' )
 201             i_nb_subdirs++;
 202     }
 203
 204     subdirs = calloc( i_nb_subdirs + 1, sizeof(char*) );
 205     if( !subdirs )
 206         return NULL;
 207
 208     for( i = 0; psz_parser && *psz_parser != '\0' ; )
 209     {
 210         char *psz_subdir = psz_parser;
 211         psz_parser = strchr( psz_subdir, ',' );
 212         if( psz_parser )
 213         {
 214             *psz_parser++ = '\0';
 215             while( *psz_parser == ' ' )
 216                 psz_parser++;
 217         }
 218         if( *psz_subdir == '\0' )
 219             continue;
 220
 221         if( asprintf( &subdirs[i++], "%s%s",
 222                   psz_subdir[0] == '.' ? psz_dir : "",
 223                   psz_subdir ) == -1 )
 224             break;
 225     }
 226     subdirs[i] = NULL;
 227
 228     return subdirs;
 229 }
 230
 231
 232 /**
 233  * Detect subtitle files.
 234  *
 235  * When called this function will split up the psz_name string into a
 236  * directory, filename and extension. It then opens the directory
 237  * in which the file resides and tries to find possible matches of
 238  * subtitles files.
 239  *
 240  * \ingroup Demux
 241  * \param p_this the calling \ref input_thread_t
 242  * \param psz_path a list of subdirectories (separated by a ',') to look in.
 243  * \param psz_name the complete filename to base the search on.
 244  * \return a NULL terminated array of filenames with detected possible subtitles.
 245  * The array contains max MAX_SUBTITLE_FILES items and you need to free it after use.
 246  */
 247 char **subtitles_Detect( input_thread_t *p_this, char *psz_path,
 248                          const char *psz_name_org )
 249 {
 250     int i_fuzzy = var_GetInteger( p_this, "sub-autodetect-fuzzy" );
 251     if ( i_fuzzy == 0 )
 252         return NULL;
 253     int j, i_result2, i_sub_count, i_fname_len;
 254     char *f_fname_noext = NULL, *f_fname_trim = NULL;
 255
 256     char **subdirs; /* list of subdirectories to look in */
 257
 258     vlc_subfn_t *result = NULL; /* unsorted results */
 259     char **result2; /* sorted results */
 260
 261     if( !psz_name_org )
 262         return NULL;
 263
 264     char *psz_fname = make_path( psz_name_org );
 265     if( !psz_fname )
 266         return NULL;
 267
 268     /* extract filename & dirname from psz_fname */
 269     char *f_dir = strdup( psz_fname );
 270     if( f_dir == NULL )
 271     {
 272         free( psz_fname );
 273         return NULL;
 274     }
 275
 276     const char *f_fname = strrchr( psz_fname, DIR_SEP_CHAR );
 277     if( !f_fname )
 278     {
 279         free( f_dir );
 280         free( psz_fname );
 281         return NULL;
 282     }
 283     f_fname++; /* Skip the '/' */
 284     f_dir[f_fname - psz_fname] = 0; /* keep dir separator in f_dir */
 285
 286     i_fname_len = strlen( f_fname );
 287
 288     f_fname_noext = malloc(i_fname_len + 1);
 289     f_fname_trim = malloc(i_fname_len + 1 );
 290     if( !f_fname_noext || !f_fname_trim )
 291     {
 292         free( f_dir );
 293         free( f_fname_noext );
 294         free( f_fname_trim );
 295         free( psz_fname );
 296         return NULL;
 297     }
 298
 299     strcpy_strip_ext( f_fname_noext, f_fname );
 300     strcpy_trim( f_fname_trim, f_fname_noext );
 301
 302     result = calloc( MAX_SUBTITLE_FILES+1, sizeof(vlc_subfn_t) ); /* We check it later (simplify code) */
 303     subdirs = paths_to_list( f_dir, psz_path );
 304     for( j = -1, i_sub_count = 0; (j == -1) || ( j >= 0 && subdirs != NULL && subdirs[j] != NULL ); j++ )
 305     {
 306         const char *psz_dir = (j < 0) ? f_dir : subdirs[j];
 307         if( psz_dir == NULL || ( j >= 0 && !strcmp( psz_dir, f_dir ) ) )
 308             continue;
 309
 310         /* parse psz_src dir */
 311         DIR *dir = vlc_opendir( psz_dir );
 312         if( dir == NULL )
 313             continue;
 314
 315         msg_Dbg( p_this, "looking for a subtitle file in %s", psz_dir );
 316
 317         char *psz_name;
 318         while( (psz_name = vlc_readdir( dir )) && i_sub_count < MAX_SUBTITLE_FILES )
 319         {
 320             if( psz_name[0] == '.' || !subtitles_Filter( psz_name ) )
 321             {
 322                 free( psz_name );
 323                 continue;
 324             }
 325
 326             char tmp_fname_noext[strlen( psz_name ) + 1];
 327             char tmp_fname_trim[strlen( psz_name ) + 1];
 328             char tmp_fname_ext[strlen( psz_name ) + 1];
 329             const char *tmp;
 330             int i_prio = SUB_PRIORITY_NONE;
 331
 332             /* retrieve various parts of the filename */
 333             strcpy_strip_ext( tmp_fname_noext, psz_name );
 334             strcpy_get_ext( tmp_fname_ext, psz_name );
 335             strcpy_trim( tmp_fname_trim, tmp_fname_noext );
 336
 337             if( !strcmp( tmp_fname_trim, f_fname_trim ) )
 338             {
 339                 /* matches the movie name exactly */
 340                 i_prio = SUB_PRIORITY_MATCH_ALL;
 341             }
 342             else if( (tmp = strstr( tmp_fname_trim, f_fname_trim )) )
 343             {
 344                 /* contains the movie name */
 345                 tmp += strlen( f_fname_trim );
 346                 if( whiteonly( tmp ) )
 347                 {
 348                     /* chars in front of the movie name */
 349                     i_prio = SUB_PRIORITY_MATCH_RIGHT;
 350                 }
 351                 else
 352                 {
 353                     /* chars after (and possibly in front of)
 354                      * the movie name */
 355                     i_prio = SUB_PRIORITY_MATCH_LEFT;
 356                 }
 357             }
 358             else if( j == -1 )
 359             {
 360                 /* doesn't contain the movie name, prefer files in f_dir over subdirs */
 361                 i_prio = SUB_PRIORITY_MATCH_NONE;
 362             }
 363             if( i_prio >= i_fuzzy )
 364             {
 365                 char psz_path[strlen( psz_dir ) + strlen( psz_name ) + 2];
 366                 struct stat st;
 367
 368                 sprintf( psz_path, "%s"DIR_SEP"%s", psz_dir, psz_name );
 369                 if( !strcmp( psz_path, psz_fname ) )
 370                 {
 371                     free( psz_name );
 372                     continue;
 373                 }
 374
 375                 if( !vlc_stat( psz_path, &st ) && S_ISREG( st.st_mode ) && result )
 376                 {
 377                     msg_Dbg( p_this,
 378                             "autodetected subtitle: %s with priority %d",
 379                             psz_path, i_prio );
 380                     result[i_sub_count].priority = i_prio;
 381                     result[i_sub_count].psz_fname = strdup( psz_path );
 382                     result[i_sub_count].psz_ext = strdup(tmp_fname_ext);
 383                     i_sub_count++;
 384                 }
 385                 else
 386                 {
 387                     msg_Dbg( p_this, "stat failed (autodetecting subtitle: %s with priority %d)",
 388                              psz_path, i_prio );
 389                 }
 390             }
 391             free( psz_name );
 392         }
 393         closedir( dir );
 394     }
 395     if( subdirs )
 396     {
 397         for( j = 0; subdirs[j]; j++ )
 398             free( subdirs[j] );
 399         free( subdirs );
 400     }
 401     free( f_dir );
 402     free( f_fname_trim );
 403     free( f_fname_noext );
 404     free( psz_fname );
 405
 406     if( !result )
 407         return NULL;
 408
 409     qsort( result, i_sub_count, sizeof(vlc_subfn_t), compare_sub_priority );
 410
 411     result2 = calloc( i_sub_count + 1, sizeof(char*) );
 412
 413     for( j = 0, i_result2 = 0; j < i_sub_count && result2 != NULL; j++ )
 414     {
 415         bool b_reject = false;
 416
 417         if( !result[j].psz_fname || !result[j].psz_ext ) /* memory out */
 418             break;
 419
 420         if( !strcasecmp( result[j].psz_ext, "sub" ) )
 421         {
 422             int i;
 423             for( i = 0; i < i_sub_count; i++ )
 424             {
 425                 if( result[i].psz_fname && result[i].psz_ext &&
 426                     !strncasecmp( result[j].psz_fname, result[i].psz_fname,
 427                                   strlen( result[j].psz_fname) - 3 ) &&
 428                     !strcasecmp( result[i].psz_ext, "idx" ) )
 429                     break;
 430             }
 431             if( i < i_sub_count )
 432                 b_reject = true;
 433         }
 434         else if( !strcasecmp( result[j].psz_ext, "cdg" ) )
 435         {
 436             if( result[j].priority < SUB_PRIORITY_MATCH_ALL )
 437                 b_reject = true;
 438         }
 439
 440         /* */
 441         if( !b_reject )
 442             result2[i_result2++] = strdup( result[j].psz_fname );
 443     }
 444
 445     for( j = 0; j < i_sub_count; j++ )
 446     {
 447         free( result[j].psz_fname );
 448         free( result[j].psz_ext );
 449     }
 450     free( result );
 451
 452     return result2;
 453 }
 454