Commit d6fb7fa7 authored by Derk-Jan Hartman's avatar Derk-Jan Hartman

* UTF-8 ified the subtitle autodetect core. (fixes #373)

* converted it to a scandir system

Please test this heavily on all your files and directories, because these routines are triggered on each opening of a file. If someone would like to review it for leaks and memory messups, I would be eternal grateful.
parent 506e6faa
...@@ -27,9 +27,11 @@ ...@@ -27,9 +27,11 @@
* This file contains functions to dectect subtitle files. * This file contains functions to dectect subtitle files.
*/ */
#include <stdlib.h> #include <stdlib.h>
#include <vlc/vlc.h> #include <vlc/vlc.h>
#include <vlc/input.h> #include <vlc/input.h>
#include "charset.h"
#ifdef HAVE_DIRENT_H #ifdef HAVE_DIRENT_H
# include <dirent.h> # include <dirent.h>
...@@ -58,7 +60,8 @@ ...@@ -58,7 +60,8 @@
static const char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "txt", "ssa", "idx", NULL}; static const char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "txt", "ssa", "idx", NULL};
/* extensions from unsupported types */ /* extensions from unsupported types */
/* rt, aqt, jss, js, ass */ /* rt, aqt, jss, js, ass */
static void strcpy_trim( char *d, char *s ) static void strcpy_trim( char *d, char *s )
{ {
/* skip leading whitespace */ /* skip leading whitespace */
...@@ -74,7 +77,7 @@ static void strcpy_trim( char *d, char *s ) ...@@ -74,7 +77,7 @@ static void strcpy_trim( char *d, char *s )
*d = tolower(*s); *d = tolower(*s);
s++; d++; s++; d++;
} }
if (*s == 0) break; if( *s == 0 ) break;
/* trim excess whitespace */ /* trim excess whitespace */
while( *s && !isalnum(*s) ) while( *s && !isalnum(*s) )
{ {
...@@ -151,6 +154,33 @@ static int compare_sub_priority( const void *a, const void *b ) ...@@ -151,6 +154,33 @@ static int compare_sub_priority( const void *a, const void *b )
#endif #endif
} }
/* Utility function for scandir */
static int Filter( const struct dirent *p_dir_content )
{
int i;
char *tmp;
if( p_dir_content == NULL || p_dir_content->d_name == NULL ) return VLC_FALSE;
/* does it end with a subtitle extension? */
tmp = strrchr( p_dir_content->d_name, '.');
if( !tmp )
{
return VLC_FALSE;
}
else
{
for( i = 0; sub_exts[i]; i++ )
{
if( strcmp( sub_exts[i], tmp+1 ) == 0 )
{
return VLC_TRUE;
}
}
}
return VLC_FALSE;
}
/** /**
* Convert a list of paths separated by ',' to a char** * Convert a list of paths separated by ',' to a char**
*/ */
...@@ -215,10 +245,11 @@ static char **paths_to_list( char *psz_dir, char *psz_path ) ...@@ -215,10 +245,11 @@ static char **paths_to_list( char *psz_dir, char *psz_path )
return subdirs; return subdirs;
} }
/** /**
* Detect subtitle files. * Detect subtitle files.
* *
* When called this function will split up the psz_fname string into a * When called this function will split up the psz_name string into a
* directory, filename and extension. It then opens the directory * directory, filename and extension. It then opens the directory
* in which the file resides and tries to find possible matches of * in which the file resides and tries to find possible matches of
* subtitles files. * subtitles files.
...@@ -226,202 +257,184 @@ static char **paths_to_list( char *psz_dir, char *psz_path ) ...@@ -226,202 +257,184 @@ static char **paths_to_list( char *psz_dir, char *psz_path )
* \ingroup Demux * \ingroup Demux
* \param p_this the calling \ref input_thread_t * \param p_this the calling \ref input_thread_t
* \param psz_path a list of subdirectories (separated by a ',') to look in. * \param psz_path a list of subdirectories (separated by a ',') to look in.
* \param psz_fname the complete filename to base the search on. * \param psz_name the complete filename to base the search on.
* \return a NULL terminated array of filenames with detected possible subtitles. * \return a NULL terminated array of filenames with detected possible subtitles.
* The array contains max MAX_SUBTITLE_FILES items and you need to free it after use. * The array contains max MAX_SUBTITLE_FILES items and you need to free it after use.
*/ */
char **subtitles_Detect( input_thread_t *p_this, char *psz_path, char **subtitles_Detect( input_thread_t *p_this, char *psz_path,
char *psz_fname ) char *psz_name )
{ {
/* variables to be used for derivatives of psz_fname */
char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim, *tmp;
/* variables to be used for derivatives FILE *f */
char *tmp_fname_noext, *tmp_fname_trim, *tmp_fname_ext, *tmpresult;
vlc_value_t fuzzy; vlc_value_t fuzzy;
int len, i, j, i_sub_count, i_result2; int j, i_result2, i_dir_content, i_sub_count = 0, i_fname_len = 0;
char *f_dir, *f_fname, *f_fname_noext, *f_fname_trim;
char *tmp;
char tmp_fname_noext[PATH_MAX];
char tmp_fname_trim[PATH_MAX];
char tmp_fname_ext[PATH_MAX];
struct dirent **pp_dir_content;
char **tmp_subdirs, **subdirs; /* list of subdirectories to look in */
subfn *result; /* unsorted results */ subfn *result; /* unsorted results */
char **result2; /* sorted results */ char **result2; /* sorted results */
char **tmp_subdirs, **subdirs; /* list of subdirectories to look in */
char *psz_fname_original = strdup( psz_name );
FILE *f; char *psz_fname = psz_fname_original;
DIR *d;
struct dirent *de;
if( !strncmp( psz_fname, "file://", 7 ) ) if( !strncmp( psz_fname, "file://", 7 ) )
{ {
psz_fname += 7; psz_fname += 7;
} }
i_sub_count = 0;
len = strlen( psz_fname ) > 256 ? strlen( psz_fname ) : 256;
f_dir = (char*)malloc(len);
f_fname = (char*)malloc(len);
f_fname_noext = (char*)malloc(len);
f_fname_trim = (char*)malloc(len);
tmp_fname_noext = (char*)malloc(len);
tmp_fname_trim = (char*)malloc(len);
tmp_fname_ext = (char*)malloc(len);
tmpresult = (char*)malloc(len);
result = (subfn*)malloc( sizeof(subfn) * MAX_SUBTITLE_FILES );
memset( result, 0, sizeof(subfn) * MAX_SUBTITLE_FILES );
/* extract filename & dirname from psz_fname */ /* extract filename & dirname from psz_fname */
tmp = strrchr( psz_fname, DIRECTORY_SEPARATOR ); tmp = strrchr( psz_fname, DIRECTORY_SEPARATOR );
if( tmp ) if( tmp )
{ {
int pos; int dirlen;
strncpy( f_fname, tmp + 1, len - 1 ); f_fname = malloc( strlen(tmp) );
f_fname[len - 1] = 0; strcpy( f_fname, tmp+1 ); // we skip the seperator, so it will still fit in the allocated space
pos = tmp - psz_fname + 1; dirlen = strlen(psz_fname) - strlen(tmp);
strncpy( f_dir, psz_fname, __MIN(pos,len-1) ); f_dir = malloc( dirlen + 1 );
f_dir[__MIN(pos,len-1)] = 0; strncpy( f_dir, psz_fname, dirlen + 1 );
f_dir[dirlen + 1] = 0;
} }
else else
{ {
strncpy( f_fname, psz_fname, len - 1 ); f_fname = strdup( psz_fname );
f_fname[len - 1] = 0;
strcpy( f_dir, "" );
} }
i_fname_len = strlen( f_fname );
f_fname_noext = malloc(i_fname_len + 1);
f_fname_trim = malloc(i_fname_len + 1 );
strcpy_strip_ext( f_fname_noext, f_fname ); strcpy_strip_ext( f_fname_noext, f_fname );
strcpy_trim( f_fname_trim, f_fname_noext ); strcpy_trim( f_fname_trim, f_fname_noext );
var_Get( p_this, "sub-autodetect-fuzzy", &fuzzy );
result = (subfn*)malloc( sizeof(subfn) * MAX_SUBTITLE_FILES );
memset( result, 0, sizeof(subfn) * MAX_SUBTITLE_FILES );
var_Get( p_this, "sub-autodetect-fuzzy", &fuzzy );
tmp_subdirs = paths_to_list( f_dir, psz_path ); tmp_subdirs = paths_to_list( f_dir, psz_path );
subdirs = tmp_subdirs; subdirs = tmp_subdirs;
for( j = -1; j == -1 || ( j >= 0 && subdirs != NULL && *subdirs != NULL ); for( j = -1; j == -1 || ( j >= 0 && subdirs != NULL && *subdirs != NULL );
j++) j++)
{ {
d = opendir( j < 0 ? f_dir : *subdirs ); pp_dir_content = NULL;
if( d ) i_dir_content = 0;
/* parse psz_src dir */
if( ( i_dir_content = scandir( j < 0 ? f_dir : *subdirs, &pp_dir_content, Filter,
NULL ) ) != -1 )
{ {
int b_found; int a;
msg_Dbg( p_this, "looking for a subtitle file in %s", msg_Dbg( p_this, "looking for a subtitle file in %s", j < 0 ? f_dir : *subdirs );
j < 0 ? f_dir : *subdirs );
while( ( de = readdir( d ) ) ) for( a = 0; a < i_dir_content; a++ )
{ {
int i_prio = 0;
struct dirent *p_dir_content = pp_dir_content[a];
char *psz_inUTF8 = FromLocale( p_dir_content->d_name );
char *p_fixed_name = vlc_fix_readdir_charset( p_this, psz_inUTF8 );
LocaleFree( psz_inUTF8 );
/* retrieve various parts of the filename */ /* retrieve various parts of the filename */
strcpy_strip_ext( tmp_fname_noext, de->d_name ); strcpy_strip_ext( tmp_fname_noext, p_fixed_name );
strcpy_get_ext( tmp_fname_ext, de->d_name ); strcpy_get_ext( tmp_fname_ext, p_fixed_name );
strcpy_trim( tmp_fname_trim, tmp_fname_noext ); strcpy_trim( tmp_fname_trim, tmp_fname_noext );
/* does it end with a subtitle extension? */ if( !i_prio && !strcmp( tmp_fname_trim, f_fname_trim ) )
b_found = 0;
for( i = 0; sub_exts[i]; i++ )
{ {
if( strcmp( sub_exts[i], tmp_fname_ext ) == 0 ) /* matches the movie name exactly */
{ i_prio = 4;
b_found = 1;
msg_Dbg( p_this, "found a possible subtitle: %s",
de->d_name );
break;
}
} }
if( !i_prio &&
/* we have a (likely) subtitle file */ ( tmp = strstr( tmp_fname_trim, f_fname_trim ) ) )
if( b_found )
{ {
int i_prio = 0; /* contains the movie name */
if( !i_prio && !strcmp( tmp_fname_trim, f_fname_trim ) ) tmp += strlen( f_fname_trim );
{ if( whiteonly( tmp ) )
/* matches the movie name exactly */
i_prio = 4;
}
if( !i_prio &&
( tmp = strstr( tmp_fname_trim, f_fname_trim ) ) )
{ {
/* contains the movie name */ /* chars in front of the movie name */
tmp += strlen( f_fname_trim ); i_prio = 2;
if( whiteonly( tmp ) )
{
/* chars in front of the movie name */
i_prio = 2;
}
else
{
/* chars after (and possibly in front of)
* the movie name */
i_prio = 3;
}
} }
if( !i_prio ) else
{ {
/* doesn't contain the movie name */ /* chars after (and possibly in front of)
if( j == 0 ) i_prio = 1; * the movie name */
i_prio = 3;
} }
}
if( i_prio >= fuzzy.i_int ) if( !i_prio )
{
/* doesn't contain the movie name */
if( j == 0 ) i_prio = 1;
}
if( i_prio >= fuzzy.i_int )
{
FILE *f;
char *tmpresult;
asprintf( &tmpresult, "%s%s", j < 0 ? f_dir : *subdirs, p_fixed_name );
msg_Dbg( p_this, "autodetected subtitle: %s with priority %d", p_fixed_name, i_prio );
if( ( f = fopen( tmpresult, "rt" ) ) )
{ {
sprintf( tmpresult, "%s%s", j == -1 ? f_dir : *subdirs, fclose( f );
de->d_name ); result[i_sub_count].priority = i_prio;
msg_Dbg( p_this, "autodetected subtitle: %s with " result[i_sub_count].psz_fname = tmpresult;
"priority %d", de->d_name, i_prio ); result[i_sub_count].psz_ext = strdup(tmp_fname_ext);
if( ( f = fopen( tmpresult, "rt" ) ) ) i_sub_count++;
{ } else free( tmpresult );
fclose( f );
result[i_sub_count].priority = i_prio;
result[i_sub_count].psz_fname = strdup(tmpresult);
result[i_sub_count].psz_ext = strdup(tmp_fname_ext);
i_sub_count++;
}
}
} }
if( i_sub_count >= MAX_SUBTITLE_FILES ) break; if( i_sub_count >= MAX_SUBTITLE_FILES ) break;
free( p_fixed_name );
} }
closedir( d );
} }
if( j >= 0 ) free( *subdirs++ ); if( j >= 0 ) free( *subdirs++ );
} }
if( tmp_subdirs ) free( tmp_subdirs ); if(tmp_subdirs) free(tmp_subdirs);
if(f_fname_trim) free(f_fname_trim);
free( f_dir ); if(f_fname_noext) free(f_fname_noext);
free( f_fname ); if(f_fname) free(f_fname);
free( f_fname_noext ); if(f_dir) free(f_dir);
free( f_fname_trim );
free( tmp_fname_noext );
free( tmp_fname_trim );
free( tmp_fname_ext );
free( tmpresult );
qsort( result, i_sub_count, sizeof( subfn ), compare_sub_priority ); qsort( result, i_sub_count, sizeof( subfn ), compare_sub_priority );
result2 = (char**)malloc( sizeof(char*) * ( i_sub_count + 1 ) ); result2 = (char**)malloc( sizeof(char*) * ( i_sub_count + 1 ) );
memset( result2, 0, sizeof(char*) * ( i_sub_count + 1 ) ); memset( result2, 0, sizeof(char*) * ( i_sub_count + 1 ) );
i_result2 = 0; i_result2 = 0;
for( i = 0; i < i_sub_count; i++ ) for( j = 0; j < i_sub_count; j++ )
{ {
if( result[i].psz_ext && !strcasecmp( result[i].psz_ext, "sub" ) ) if( result[j].psz_ext && !strcasecmp( result[j].psz_ext, "sub" ) )
{ {
int j; int i;
for( j = 0; j < i_sub_count; j++ ) for( i = 0; i < i_sub_count; i++ )
{ {
if( result[j].psz_fname && result[i].psz_fname && if( result[i].psz_fname && result[j].psz_fname &&
!strncasecmp( result[i].psz_fname, result[j].psz_fname, sizeof( result[i].psz_fname) - 4 ) && !strncasecmp( result[j].psz_fname, result[i].psz_fname, sizeof( result[j].psz_fname) - 4 ) &&
!strcasecmp( result[j].psz_ext, "idx" ) ) !strcasecmp( result[i].psz_ext, "idx" ) )
break; break;
} }
if( j >= i_sub_count ) if( i >= i_sub_count )
{ {
result2[i_result2] = result[i].psz_fname; result2[i_result2] = result[j].psz_fname;
i_result2++; i_result2++;
} }
} }
else else
{ {
result2[i_result2] = result[i].psz_fname; result2[i_result2] = result[j].psz_fname;
i_result2++; i_result2++;
} }
} }
free( psz_fname_original );
free( result ); free( result );
return result2; return result2;
} }
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment