X-Git-Url: https://git.sesse.net/?a=blobdiff_plain;f=src%2Finput%2Fsubtitles.c;h=2ca105ff08618b16702eea71c662e485639ebcd2;hb=fbbde4b28ac2e8a957b23507de411be088a80214;hp=875fa7429e99440007f20c3d92397e40e20be382;hpb=fed15a8567eb171ddd273f7bac2eba5842862cf4;p=vlc diff --git a/src/input/subtitles.c b/src/input/subtitles.c index 875fa7429e..2ca105ff08 100644 --- a/src/input/subtitles.c +++ b/src/input/subtitles.c @@ -1,7 +1,7 @@ /***************************************************************************** - * subtitles.c + * subtitles.c : subtitles detection ***************************************************************************** - * Copyright (C) 2003-2004 the VideoLAN team + * Copyright (C) 2003-2009 the VideoLAN team * $Id$ * * Authors: Derk-Jan Hartman @@ -27,48 +27,48 @@ * This file contains functions to dectect subtitle files. */ -#include -#include -#include -#include "charset.h" +#ifdef HAVE_CONFIG_H +# include "config.h" +#endif + +#include +#include +#include #ifdef HAVE_DIRENT_H # include #endif -#ifdef HAVE_LIMITS_H -# include -#endif - #ifdef HAVE_UNISTD_H # include #endif -#include +#include -/** - * What's between a directory and a filename? - */ -#if defined( WIN32 ) - #define DIRECTORY_SEPARATOR '\\' -#else - #define DIRECTORY_SEPARATOR '/' -#endif +#include /* isalnum */ + +#include "input_internal.h" /** * We are not going to autodetect more subtitle files than this. */ #define MAX_SUBTITLE_FILES 128 - /** - * The possible extentions for subtitle files we support + * The possible extensions for subtitle files we support */ -static const char * sub_exts[] = { "utf", "utf8", "utf-8", "sub", "srt", "smi", "txt", "ssa", "idx", NULL}; -/* extensions from unsupported types */ -/* rt, aqt, jss, js, ass */ - -static void strcpy_trim( char *d, char *s ) +static const char const sub_exts[][6] = { + "idx", "sub", "srt", + "ssa", "ass", "smi", + "utf", "utf8", "utf-8", + "txt", "rt", "aqt", + "usf", "jss", "cdg", + "psb", "mpsub","mpl2", + "pjs", "dks", + "" +}; + +static void strcpy_trim( char *d, const char *s ) { /* skip leading whitespace */ while( *s && !isalnum(*s) ) @@ -95,19 +95,16 @@ static void strcpy_trim( char *d, char *s ) *d = 0; } -static void strcpy_strip_ext( char *d, char *s ) +static void strcpy_strip_ext( char *d, const char *s ) { - char *tmp = strrchr(s, '.'); + const char *tmp = strrchr(s, '.'); if( !tmp ) { strcpy(d, s); return; } else - { - strncpy(d, s, tmp - s); - d[tmp - s] = 0; - } + strlcpy(d, s, tmp - s + 1 ); while( *d ) { *d = tolower(*d); @@ -115,143 +112,121 @@ static void strcpy_strip_ext( char *d, char *s ) } } -static void strcpy_get_ext( char *d, char *s ) +static void strcpy_get_ext( char *d, const char *s ) { - char *tmp = strrchr(s, '.'); + const char *tmp = strrchr(s, '.'); if( !tmp ) - { strcpy(d, ""); - return; - } else strcpy( d, tmp + 1 ); + else + strcpy( d, tmp + 1 ); } -static int whiteonly( char *s ) +static int whiteonly( const char *s ) { - while ( *s ) - { - if( isalnum( *s ) ) return 0; + while( *s ) + { + if( isalnum( *s ) ) + return 0; s++; - } - return 1; + } + return 1; } -typedef struct _subfn +enum +{ + SUB_PRIORITY_NONE = 0, + SUB_PRIORITY_MATCH_NONE = 1, + SUB_PRIORITY_MATCH_RIGHT = 2, + SUB_PRIORITY_MATCH_LEFT = 3, + SUB_PRIORITY_MATCH_ALL = 4, +}; +typedef struct { int priority; char *psz_fname; char *psz_ext; -} subfn; +} vlc_subfn_t; static int compare_sub_priority( const void *a, const void *b ) { - if (((subfn*)a)->priority > ((subfn*)b)->priority) - { + const vlc_subfn_t *p0 = a; + const vlc_subfn_t *p1 = b; + + if( p0->priority > p1->priority ) return -1; - } - if (((subfn*)a)->priority < ((subfn*)b)->priority) - { + if( p0->priority < p1->priority ) return 1; - } #ifndef UNDER_CE - return strcoll(((subfn*)a)->psz_fname, ((subfn*)b)->psz_fname); + return strcoll( p0->psz_fname, p1->psz_fname); #else - return strcmp(((subfn*)a)->psz_fname, ((subfn*)b)->psz_fname); + return strcmp( p0->psz_fname, p1->psz_fname); #endif } -/* Utility function for scandir */ -static int Filter( const struct dirent *p_dir_content ) +/* + * Check if a file ends with a subtitle extension + */ +int subtitles_Filter( const char *psz_dir_content ) { - int i; - char *tmp = NULL; + const char *tmp = strrchr( psz_dir_content, '.'); - 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; + return 0; + tmp++; + + for( int i = 0; sub_exts[i][0]; i++ ) + if( strcasecmp( sub_exts[i], tmp ) == 0 ) + return 1; + return 0; } /** * Convert a list of paths separated by ',' to a char** */ -static char **paths_to_list( char *psz_dir, char *psz_path ) +static char **paths_to_list( const char *psz_dir, char *psz_path ) { unsigned int i, k, i_nb_subdirs; char **subdirs; /* list of subdirectories to look in */ + char *psz_parser = psz_path; - if( !psz_dir ) return NULL; - if( !psz_path ) return NULL; + if( !psz_dir || !psz_path ) + return NULL; - i_nb_subdirs = 1; - for( k = 0; k < strlen( psz_path ); k++ ) + for( k = 0, i_nb_subdirs = 1; psz_path[k] != '\0'; k++ ) { if( psz_path[k] == ',' ) - { i_nb_subdirs++; - } } - if( i_nb_subdirs > 0 ) - { - char *psz_parser = NULL, *psz_temp = NULL; + subdirs = calloc( i_nb_subdirs + 1, sizeof(char*) ); + if( !subdirs ) + return NULL; - subdirs = (char**)malloc( sizeof(char*) * ( i_nb_subdirs + 1 ) ); - memset( subdirs, 0, sizeof(char*) * ( i_nb_subdirs + 1 ) ); - i = 0; - psz_parser = psz_path; - while( psz_parser && *psz_parser ) + for( i = 0; psz_parser && *psz_parser != '\0' ; ) + { + char *psz_subdir = psz_parser; + psz_parser = strchr( psz_subdir, ',' ); + if( psz_parser ) { - char *psz_subdir; - psz_subdir = psz_parser; - psz_parser = strchr( psz_subdir, ',' ); - if( psz_parser ) - { - *psz_parser = '\0'; + *psz_parser++ = '\0'; + while( *psz_parser == ' ' ) psz_parser++; - while( *psz_parser == ' ' ) - { - psz_parser++; - } - } - if( strlen( psz_subdir ) > 0 ) - { - psz_temp = (char *)malloc( strlen(psz_dir) - + strlen(psz_subdir) + 2 ); - if( psz_temp ) - { - sprintf( psz_temp, "%s%s%c", - psz_subdir[0] == '.' ? psz_dir : "", - psz_subdir, - psz_subdir[strlen(psz_subdir) - 1] == - DIRECTORY_SEPARATOR ? '\0' : DIRECTORY_SEPARATOR ); - subdirs[i] = psz_temp; - i++; - } - } } - subdirs[i] = NULL; - } - else - { - subdirs = NULL; + if( *psz_subdir == '\0' ) + continue; + + if( asprintf( &subdirs[i++], "%s%s%c", + psz_subdir[0] == '.' ? psz_dir : "", + psz_subdir, + psz_subdir[strlen(psz_subdir) - 1] == DIR_SEP_CHAR ? + '\0' : DIR_SEP_CHAR ) == -1 ) + break; } + subdirs[i] = NULL; + return subdirs; } @@ -272,210 +247,234 @@ static char **paths_to_list( char *psz_dir, char *psz_path ) * 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 *psz_name ) + const char *psz_name_org ) { - vlc_value_t fuzzy; - int j, i_result2, i_dir_content, i_sub_count = 0, i_fname_len = 0; + int i_fuzzy; + int j, i_result2, i_sub_count, i_fname_len; char *f_dir = NULL, *f_fname = NULL, *f_fname_noext = NULL, *f_fname_trim = NULL; char *tmp = NULL; - 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 */ + char **subdirs; /* list of subdirectories to look in */ - subfn *result = NULL; /* unsorted results */ + vlc_subfn_t *result = NULL; /* unsorted results */ char **result2; /* sorted results */ - char *psz_fname_original = strdup( psz_name ); - char *psz_fname = psz_fname_original; + if( !psz_name_org ) + return NULL; - if( psz_fname == NULL ) return NULL; - - if( !strncmp( psz_fname, "file://", 7 ) ) - { - psz_fname += 7; - } + char *psz_fname = make_path( psz_name_org ); + if( !psz_fname ) + return NULL; /* extract filename & dirname from psz_fname */ - tmp = strrchr( psz_fname, DIRECTORY_SEPARATOR ); + tmp = strrchr( psz_fname, DIR_SEP_CHAR ); if( tmp ) { - int dirlen = 0; - - f_fname = malloc( strlen(tmp) ); - if( f_fname ) - strcpy( f_fname, tmp+1 ); // we skip the seperator, so it will still fit in the allocated space - dirlen = strlen(psz_fname) - strlen(tmp) + 1; // add the seperator - f_dir = malloc( dirlen + 1 ); - if( f_dir ) - { - strncpy( f_dir, psz_fname, dirlen ); - f_dir[dirlen] = 0; - } + const int i_dirlen = strlen(psz_fname)-strlen(tmp)+1; /* include the separator */ + f_fname = strdup( &tmp[1] ); /* skip the separator */ + f_dir = strndup( psz_fname, i_dirlen ); } else { +#if defined (HAVE_UNISTD_H) && !defined (UNDER_CE) /* Get the current working directory */ -#ifdef HAVE_UNISTD_H - f_dir = getcwd( NULL, 0 ); + char *psz_cwd = getcwd( NULL, 0 ); +#else + char *psz_cwd = NULL; #endif - if( f_dir == NULL ) + if( !psz_cwd ) { - if( psz_fname_original ) free( psz_fname_original ); + free( psz_fname ); return NULL; } + f_fname = strdup( psz_fname ); + if( asprintf( &f_dir, "%s%c", psz_cwd, DIR_SEP_CHAR ) == -1 ) + f_dir = NULL; /* Assure that function will return in next test */ + free( psz_cwd ); + } + if( !f_fname || !f_dir ) + { + free( f_fname ); + free( f_dir ); + free( psz_fname ); + return NULL; } i_fname_len = strlen( f_fname ); + f_fname_noext = malloc(i_fname_len + 1); f_fname_trim = malloc(i_fname_len + 1 ); + if( !f_fname_noext || !f_fname_trim ) + { + free( f_fname ); + free( f_dir ); + free( f_fname_noext ); + free( f_fname_trim ); + free( psz_fname ); + return NULL; + } strcpy_strip_ext( f_fname_noext, f_fname ); strcpy_trim( f_fname_trim, f_fname_noext ); - result = (subfn*)malloc( sizeof(subfn) * MAX_SUBTITLE_FILES ); - if( result ) - 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 ); - subdirs = tmp_subdirs; + i_fuzzy = var_GetInteger( p_this, "sub-autodetect-fuzzy" ); - for( j = -1; (j == -1) || ( (j >= 0) && (subdirs != NULL) && (*subdirs != NULL) ); - j++) + result = calloc( MAX_SUBTITLE_FILES+1, sizeof(vlc_subfn_t) ); /* We check it later (simplify code) */ + subdirs = paths_to_list( f_dir, psz_path ); + for( j = -1, i_sub_count = 0; (j == -1) || ( j >= 0 && subdirs != NULL && subdirs[j] != NULL ); j++ ) { - pp_dir_content = NULL; - i_dir_content = 0; + const char *psz_dir = j < 0 ? f_dir : subdirs[j]; + char **ppsz_dir_content; + int i_dir_content; - if( j < 0 && f_dir == NULL ) + if( psz_dir == NULL || ( j >= 0 && !strcmp( psz_dir, f_dir ) ) ) continue; - /* parse psz_src dir */ - if( ( i_dir_content = scandir( j < 0 ? f_dir : *subdirs, &pp_dir_content, Filter, - NULL ) ) != -1 ) + /* parse psz_src dir */ + i_dir_content = vlc_scandir( psz_dir, &ppsz_dir_content, + subtitles_Filter, NULL ); + if( i_dir_content < 0 ) + continue; + + msg_Dbg( p_this, "looking for a subtitle file in %s", psz_dir ); + for( int a = 0; a < i_dir_content && i_sub_count < MAX_SUBTITLE_FILES ; a++ ) { - int a; + char *psz_name = ppsz_dir_content[a]; + char tmp_fname_noext[strlen( psz_name ) + 1]; + char tmp_fname_trim[strlen( psz_name ) + 1]; + char tmp_fname_ext[strlen( psz_name ) + 1]; - msg_Dbg( p_this, "looking for a subtitle file in %s", j < 0 ? f_dir : *subdirs ); - 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 ); + int i_prio; - LocaleFree( psz_inUTF8 ); + if( psz_name == NULL || psz_name[0] == '.' ) + continue; - /* retrieve various parts of the filename */ - strcpy_strip_ext( tmp_fname_noext, p_fixed_name ); - strcpy_get_ext( tmp_fname_ext, p_fixed_name ); - strcpy_trim( tmp_fname_trim, tmp_fname_noext ); + /* retrieve various parts of the filename */ + strcpy_strip_ext( tmp_fname_noext, psz_name ); + strcpy_get_ext( tmp_fname_ext, psz_name ); + strcpy_trim( tmp_fname_trim, tmp_fname_noext ); - if( !i_prio && !strcmp( tmp_fname_trim, f_fname_trim ) ) + i_prio = SUB_PRIORITY_NONE; + if( i_prio == SUB_PRIORITY_NONE && !strcmp( tmp_fname_trim, f_fname_trim ) ) + { + /* matches the movie name exactly */ + i_prio = SUB_PRIORITY_MATCH_ALL; + } + if( i_prio == SUB_PRIORITY_NONE && + ( tmp = strstr( tmp_fname_trim, f_fname_trim ) ) ) + { + /* contains the movie name */ + tmp += strlen( f_fname_trim ); + if( whiteonly( tmp ) ) { - /* matches the movie name exactly */ - i_prio = 4; + /* chars in front of the movie name */ + i_prio = SUB_PRIORITY_MATCH_RIGHT; } - if( !i_prio && - ( tmp = strstr( tmp_fname_trim, f_fname_trim ) ) ) + else { - /* contains the movie name */ - tmp += strlen( f_fname_trim ); - 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; - } + /* chars after (and possibly in front of) + * the movie name */ + i_prio = SUB_PRIORITY_MATCH_LEFT; } - if( !i_prio ) + } + if( i_prio == SUB_PRIORITY_NONE && + j == 0 ) + { + /* doesn't contain the movie name, prefer files in f_dir over subdirs */ + i_prio = SUB_PRIORITY_MATCH_NONE; + } + if( i_prio >= i_fuzzy ) + { + char psz_path[strlen( psz_dir ) + strlen( psz_name ) + 1]; + struct stat st; + + sprintf( psz_path, "%s%s", psz_dir, psz_name ); + if( !strcmp( psz_path, psz_fname ) ) + continue; + + if( !vlc_stat( psz_path, &st ) && S_ISREG( st.st_mode ) && result ) { - /* doesn't contain the movie name */ - if( j == 0 ) i_prio = 1; + msg_Dbg( p_this, + "autodetected subtitle: %s with priority %d", + psz_path, i_prio ); + result[i_sub_count].priority = i_prio; + result[i_sub_count].psz_fname = strdup( psz_path ); + result[i_sub_count].psz_ext = strdup(tmp_fname_ext); + i_sub_count++; } - if( i_prio >= fuzzy.i_int ) + else { - FILE *f; - char *psz_UTF8_path; - char *psz_locale_path; - - asprintf( &psz_UTF8_path, "%s%s", j < 0 ? f_dir : *subdirs, p_fixed_name ); - psz_locale_path = ToLocale( psz_UTF8_path ); - msg_Dbg( p_this, "autodetected subtitle: %s with priority %d", p_fixed_name, i_prio ); - if( ( f = fopen( psz_locale_path, "rt" ) ) ) - { - fclose( f ); - LocaleFree( psz_locale_path ); - result[i_sub_count].priority = i_prio; - result[i_sub_count].psz_fname = psz_UTF8_path; - result[i_sub_count].psz_ext = strdup(tmp_fname_ext); - i_sub_count++; - } - else - { - if( psz_UTF8_path ) free( psz_UTF8_path ); - LocaleFree( psz_locale_path ); - } + msg_Dbg( p_this, "stat failed (autodetecting subtitle: %s with priority %d)", + psz_path, i_prio ); } - if( i_sub_count >= MAX_SUBTITLE_FILES ) break; - if( p_fixed_name ) free( p_fixed_name ); } - for( a = 0; a < i_dir_content; a++ ) - if( pp_dir_content[a] ) free( pp_dir_content[a] ); - if( pp_dir_content ) free( pp_dir_content ); } - if( j >= 0 ) if( *subdirs ) free( *subdirs++ ); + if( ppsz_dir_content ) + { + for( int a = 0; a < i_dir_content; a++ ) + free( ppsz_dir_content[a] ); + free( ppsz_dir_content ); + } + } + if( subdirs ) + { + for( j = 0; subdirs[j]; j++ ) + free( subdirs[j] ); + free( subdirs ); } + free( f_fname ); + free( f_dir ); + free( f_fname_trim ); + free( f_fname_noext ); + free( psz_fname ); - if( tmp_subdirs ) free( tmp_subdirs ); - if( f_fname_trim ) free( f_fname_trim ); - if( f_fname_noext ) free( f_fname_noext ); - if( f_fname ) free( f_fname ); - if( f_dir ) free( f_dir ); + if( !result ) + return NULL; - qsort( result, i_sub_count, sizeof( subfn ), compare_sub_priority ); + qsort( result, i_sub_count, sizeof(vlc_subfn_t), compare_sub_priority ); - result2 = (char**)malloc( sizeof(char*) * ( i_sub_count + 1 ) ); - if( result2 ) - memset( result2, 0, sizeof(char*) * ( i_sub_count + 1 ) ); - i_result2 = 0; + result2 = calloc( i_sub_count + 1, sizeof(char*) ); - for( j = 0; j < i_sub_count; j++ ) + for( j = 0, i_result2 = 0; j < i_sub_count && result2 != NULL; j++ ) { - if( result[j].psz_ext && !strcasecmp( result[j].psz_ext, "sub" ) ) + bool b_reject = false; + + if( !result[j].psz_fname || !result[j].psz_ext ) /* memory out */ + break; + + if( !strcasecmp( result[j].psz_ext, "sub" ) ) { int i; for( i = 0; i < i_sub_count; i++ ) { - if( result[i].psz_fname && result[j].psz_fname && - !strncasecmp( result[j].psz_fname, result[i].psz_fname, sizeof( result[j].psz_fname) - 4 ) && + if( result[i].psz_fname && result[i].psz_ext && + !strncasecmp( result[j].psz_fname, result[i].psz_fname, + strlen( result[j].psz_fname) - 3 ) && !strcasecmp( result[i].psz_ext, "idx" ) ) break; } - if( i >= i_sub_count ) - { - result2[i_result2] = result[j].psz_fname; - i_result2++; - } + if( i < i_sub_count ) + b_reject = true; } - else + else if( !strcasecmp( result[j].psz_ext, "cdg" ) ) { - result2[i_result2] = result[j].psz_fname; - i_result2++; + if( result[j].priority < SUB_PRIORITY_MATCH_ALL ) + b_reject = true; } + + /* */ + if( !b_reject ) + result2[i_result2++] = strdup( result[j].psz_fname ); + } + + for( j = 0; j < i_sub_count; j++ ) + { + free( result[j].psz_fname ); + free( result[j].psz_ext ); } + free( result ); - if( psz_fname_original ) free( psz_fname_original ); - if( result ) free( result ); return result2; } +