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