src/input/subtitles.c

   1 /*****************************************************************************
   2  * subtitles.c : subtitles detection
   3  *****************************************************************************
   4  * Copyright (C) 2003-2009 the VideoLAN team
   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
  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., 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 <vlc_common.h>
  35 #include <vlc_charset.h>
  36 #include <vlc_url.h>
  37
  38 #ifdef HAVE_DIRENT_H
  39 #   include <dirent.h>
  40 #endif
  41
  42 #ifdef HAVE_UNISTD_H
  43 #   include <unistd.h>
  44 #endif
  45
  46 #include <sys/stat.h>
  47
  48 #include <ctype.h> /* isalnum */
  49
  50 #include "input_internal.h"
  51
  52 /**
  53  * We are not going to autodetect more subtitle files than this.
  54  */
  55 #define MAX_SUBTITLE_FILES 128
  56
  57 /**
  58  * The possible extensions for subtitle files we support
  59  */
  60 static const char const sub_exts[][6] = {
  61     "idx", "sub",  "srt",
  62     "ssa", "ass",  "smi",
  63     "utf", "utf8", "utf-8",
  64     "txt", "rt",   "aqt",
  65     "usf", "jss",  "cdg",
  66     "psb", "mpsub","mpl2",
  67     "pjs", "dks",
  68     ""
  69 };
  70
  71 static void strcpy_trim( char *d, const char *s )
  72 {
  73     /* skip leading whitespace */
  74     while( *s && !isalnum(*s) )
  75     {
  76         s++;
  77     }
  78     for(;;)
  79     {
  80         /* copy word */
  81         while( *s && isalnum(*s) )
  82         {
  83             *d = tolower(*s);
  84             s++; d++;
  85         }
  86         if( *s == 0 ) break;
  87         /* trim excess whitespace */
  88         while( *s && !isalnum(*s) )
  89         {
  90             s++;
  91         }
  92         if( *s == 0 ) break;
  93         *d++ = ' ';
  94     }
  95     *d = 0;
  96 }
  97
  98 static void strcpy_strip_ext( char *d, const char *s )
  99 {
 100     const char *tmp = strrchr(s, '.');
 101     if( !tmp )
 102     {
 103         strcpy(d, s);
 104         return;
 105     }
 106     else
 107         strlcpy(d, s, tmp - s + 1 );
 108     while( *d )
 109     {
 110         *d = tolower(*d);
 111         d++;
 112     }
 113 }
 114
 115 static void strcpy_get_ext( char *d, const char *s )
 116 {
 117     const char *tmp = strrchr(s, '.');
 118     if( !tmp )
 119         strcpy(d, "");
 120     else
 121         strcpy( d, tmp + 1 );
 122 }
 123
 124 static int whiteonly( const char *s )
 125 {
 126     while( *s )
 127     {
 128         if( isalnum( *s ) )
 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 #ifndef UNDER_CE
 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%c",
 222                   psz_subdir[0] == '.' ? psz_dir : "",
 223                   psz_subdir,
 224                   psz_subdir[strlen(psz_subdir) - 1] == DIR_SEP_CHAR ?
 225                                            '\0' : DIR_SEP_CHAR ) == -1 )
 226             break;
 227     }
 228     subdirs[i] = NULL;
 229
 230     return subdirs;
 231 }
 232
 233
 234 /**
 235  * Detect subtitle files.
 236  *
 237  * When called this function will split up the psz_name string into a
 238  * directory, filename and extension. It then opens the directory
 239  * in which the file resides and tries to find possible matches of
 240  * subtitles files.
 241  *
 242  * \ingroup Demux
 243  * \param p_this the calling \ref input_thread_t
 244  * \param psz_path a list of subdirectories (separated by a ',') to look in.
 245  * \param psz_name the complete filename to base the search on.
 246  * \return a NULL terminated array of filenames with detected possible subtitles.
 247  * The array contains max MAX_SUBTITLE_FILES items and you need to free it after use.
 248  */
 249 char **subtitles_Detect( input_thread_t *p_this, char *psz_path,
 250                          const char *psz_name_org )
 251 {
 252     int i_fuzzy;
 253     int j, i_result2, i_sub_count, i_fname_len;
 254     char *f_dir = NULL, *f_fname = NULL, *f_fname_noext = NULL, *f_fname_trim = NULL;
 255     char *tmp = NULL;
 256
 257     char **subdirs; /* list of subdirectories to look in */
 258
 259     vlc_subfn_t *result = NULL; /* unsorted results */
 260     char **result2; /* sorted results */
 261
 262     if( !psz_name_org )
 263         return NULL;
 264
 265     if( !strncmp( psz_name_org, "file://", 7 ) )
 266     {
 267         psz_name_org += 7;
 268         if( !strncmp( psz_name_org, "localhost", 9 ) )
 269             psz_name_org += 9;
 270     }
 271     char *psz_fname = decode_URI_duplicate( psz_name_org );
 272     if( !psz_fname )
 273         return NULL;
 274
 275     /* extract filename & dirname from psz_fname */
 276     tmp = strrchr( psz_fname, DIR_SEP_CHAR );
 277     if( tmp )
 278     {
 279         const int i_dirlen = strlen(psz_fname)-strlen(tmp)+1; /* include the separator */
 280         f_fname = strdup( &tmp[1] );    /* skip the separator */
 281         f_dir = strndup( psz_fname, i_dirlen );
 282     }
 283     else
 284     {
 285 #if defined (HAVE_UNISTD_H) && !defined (UNDER_CE)
 286         /* Get the current working directory */
 287         char *psz_cwd = getcwd( NULL, 0 );
 288 #else
 289         char *psz_cwd = NULL;
 290 #endif
 291         if( !psz_cwd )
 292         {
 293             free( psz_fname );
 294             return NULL;
 295         }
 296
 297         f_fname = strdup( psz_fname );
 298         if( asprintf( &f_dir, "%s%c", psz_cwd, DIR_SEP_CHAR ) == -1 )
 299             f_dir = NULL; /* Assure that function will return in next test */
 300         free( psz_cwd );
 301     }
 302     if( !f_fname || !f_dir )
 303     {
 304         free( f_fname );
 305         free( f_dir );
 306         free( psz_fname );
 307         return NULL;
 308     }
 309
 310     i_fname_len = strlen( f_fname );
 311
 312     f_fname_noext = malloc(i_fname_len + 1);
 313     f_fname_trim = malloc(i_fname_len + 1 );
 314     if( !f_fname_noext || !f_fname_trim )
 315     {
 316         free( f_fname );
 317         free( f_dir );
 318         free( f_fname_noext );
 319         free( f_fname_trim );
 320         free( psz_fname );
 321         return NULL;
 322     }
 323
 324     strcpy_strip_ext( f_fname_noext, f_fname );
 325     strcpy_trim( f_fname_trim, f_fname_noext );
 326
 327     i_fuzzy = var_GetInteger( p_this, "sub-autodetect-fuzzy" );
 328
 329     result = calloc( MAX_SUBTITLE_FILES+1, sizeof(vlc_subfn_t) ); /* We check it later (simplify code) */
 330     subdirs = paths_to_list( f_dir, psz_path );
 331     for( j = -1, i_sub_count = 0; (j == -1) || ( j >= 0 && subdirs != NULL && subdirs[j] != NULL ); j++ )
 332     {
 333         const char *psz_dir = j < 0 ? f_dir : subdirs[j];
 334         char **ppsz_dir_content;
 335         int i_dir_content;
 336
 337         if( psz_dir == NULL || ( j >= 0 && !strcmp( psz_dir, f_dir ) ) )
 338             continue;
 339
 340         /* parse psz_src dir */
 341         i_dir_content = utf8_scandir( psz_dir, &ppsz_dir_content,
 342                                       subtitles_Filter, NULL );
 343         if( i_dir_content < 0 )
 344             continue;
 345
 346         msg_Dbg( p_this, "looking for a subtitle file in %s", psz_dir );
 347         for( int a = 0; a < i_dir_content && i_sub_count < MAX_SUBTITLE_FILES ; a++ )
 348         {
 349             char *psz_name = ppsz_dir_content[a];
 350             char tmp_fname_noext[strlen( psz_name ) + 1];
 351             char tmp_fname_trim[strlen( psz_name ) + 1];
 352             char tmp_fname_ext[strlen( psz_name ) + 1];
 353
 354             int i_prio;
 355
 356             if( psz_name == NULL || psz_name[0] == '.' )
 357                 continue;
 358
 359             /* retrieve various parts of the filename */
 360             strcpy_strip_ext( tmp_fname_noext, psz_name );
 361             strcpy_get_ext( tmp_fname_ext, psz_name );
 362             strcpy_trim( tmp_fname_trim, tmp_fname_noext );
 363
 364             i_prio = SUB_PRIORITY_NONE;
 365             if( i_prio == SUB_PRIORITY_NONE && !strcmp( tmp_fname_trim, f_fname_trim ) )
 366             {
 367                 /* matches the movie name exactly */
 368                 i_prio = SUB_PRIORITY_MATCH_ALL;
 369             }
 370             if( i_prio == SUB_PRIORITY_NONE &&
 371                 ( tmp = strstr( tmp_fname_trim, f_fname_trim ) ) )
 372             {
 373                 /* contains the movie name */
 374                 tmp += strlen( f_fname_trim );
 375                 if( whiteonly( tmp ) )
 376                 {
 377                     /* chars in front of the movie name */
 378                     i_prio = SUB_PRIORITY_MATCH_RIGHT;
 379                 }
 380                 else
 381                 {
 382                     /* chars after (and possibly in front of)
 383                      * the movie name */
 384                     i_prio = SUB_PRIORITY_MATCH_LEFT;
 385                 }
 386             }
 387             if( i_prio == SUB_PRIORITY_NONE &&
 388                 j == 0 )
 389             {
 390                 /* doesn't contain the movie name, prefer files in f_dir over subdirs */
 391                 i_prio = SUB_PRIORITY_MATCH_NONE;
 392             }
 393             if( i_prio >= i_fuzzy )
 394             {
 395                 char psz_path[strlen( psz_dir ) + strlen( psz_name ) + 1];
 396                 struct stat st;
 397
 398                 sprintf( psz_path, "%s%s", psz_dir, psz_name );
 399                 if( !strcmp( psz_path, psz_fname ) )
 400                     continue;
 401
 402                 if( !utf8_stat( psz_path, &st ) && S_ISREG( st.st_mode ) && result )
 403                 {
 404                     msg_Dbg( p_this,
 405                             "autodetected subtitle: %s with priority %d",
 406                             psz_path, i_prio );
 407                     result[i_sub_count].priority = i_prio;
 408                     result[i_sub_count].psz_fname = strdup( psz_path );
 409                     result[i_sub_count].psz_ext = strdup(tmp_fname_ext);
 410                     i_sub_count++;
 411                 }
 412                 else
 413                 {
 414                     msg_Dbg( p_this, "stat failed (autodetecting subtitle: %s with priority %d)",
 415                              psz_path, i_prio );
 416                 }
 417             }
 418         }
 419         if( ppsz_dir_content )
 420         {
 421             for( int a = 0; a < i_dir_content; a++ )
 422                 free( ppsz_dir_content[a] );
 423             free( ppsz_dir_content );
 424         }
 425     }
 426     if( subdirs )
 427     {
 428         for( j = 0; subdirs[j]; j++ )
 429             free( subdirs[j] );
 430         free( subdirs );
 431     }
 432     free( f_fname );
 433     free( f_dir );
 434     free( f_fname_trim );
 435     free( f_fname_noext );
 436     free( psz_fname );
 437
 438     if( !result )
 439         return NULL;
 440
 441     qsort( result, i_sub_count, sizeof(vlc_subfn_t), compare_sub_priority );
 442
 443     result2 = calloc( i_sub_count + 1, sizeof(char*) );
 444
 445     for( j = 0, i_result2 = 0; j < i_sub_count && result2 != NULL; j++ )
 446     {
 447         bool b_reject = false;
 448
 449         if( !result[j].psz_fname || !result[j].psz_ext ) /* memory out */
 450             break;
 451
 452         if( !strcasecmp( result[j].psz_ext, "sub" ) )
 453         {
 454             int i;
 455             for( i = 0; i < i_sub_count; i++ )
 456             {
 457                 if( result[i].psz_fname && result[i].psz_ext &&
 458                     !strncasecmp( result[j].psz_fname, result[i].psz_fname,
 459                                   strlen( result[j].psz_fname) - 3 ) &&
 460                     !strcasecmp( result[i].psz_ext, "idx" ) )
 461                     break;
 462             }
 463             if( i < i_sub_count )
 464                 b_reject = true;
 465         }
 466         else if( !strcasecmp( result[j].psz_ext, "cdg" ) )
 467         {
 468             if( result[j].priority < SUB_PRIORITY_MATCH_ALL )
 469                 b_reject = true;
 470         }
 471
 472         /* */
 473         if( !b_reject )
 474             result2[i_result2++] = strdup( result[j].psz_fname );
 475     }
 476
 477     for( j = 0; j < i_sub_count; j++ )
 478     {
 479         free( result[j].psz_fname );
 480         free( result[j].psz_ext );
 481     }
 482     free( result );
 483
 484     return result2;
 485 }
 486