/*
* Copyright (C) 2003-2011 The Music Player Daemon Project
* http://www.musicpd.org
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License along
* with this program; if not, write to the Free Software Foundation, Inc.,
* 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
*/
#include "config.h"
#include "tag_id3.h"
#include "tag_table.h"
#include "tag.h"
#include "riff.h"
#include "aiff.h"
#include "conf.h"
#include <glib.h>
#include <id3tag.h>
#include <stdio.h>
#include <stdlib.h>
#include <errno.h>
#include <string.h>
#undef G_LOG_DOMAIN
#define G_LOG_DOMAIN "id3"
# ifndef ID3_FRAME_COMPOSER
# define ID3_FRAME_COMPOSER "TCOM"
# endif
# ifndef ID3_FRAME_DISC
# define ID3_FRAME_DISC "TPOS"
# endif
#ifndef ID3_FRAME_ARTIST_SORT
#define ID3_FRAME_ARTIST_SORT "TSOP"
#endif
#ifndef ID3_FRAME_ALBUM_ARTIST_SORT
#define ID3_FRAME_ALBUM_ARTIST_SORT "TSO2" /* this one is unofficial, introduced by Itunes */
#endif
#ifndef ID3_FRAME_ALBUM_ARTIST
#define ID3_FRAME_ALBUM_ARTIST "TPE2"
#endif
static inline bool
tag_is_id3v1(struct id3_tag *tag)
{
return (id3_tag_options(tag, 0, 0) & ID3_TAG_OPTION_ID3V1) != 0;
}
static id3_utf8_t *
tag_id3_getstring(const struct id3_frame *frame, unsigned i)
{
union id3_field *field;
const id3_ucs4_t *ucs4;
field = id3_frame_field(frame, i);
if (field == NULL)
return NULL;
ucs4 = id3_field_getstring(field);
if (ucs4 == NULL)
return NULL;
return id3_ucs4_utf8duplicate(ucs4);
}
/* This will try to convert a string to utf-8,
*/
static id3_utf8_t *
import_id3_string(bool is_id3v1, const id3_ucs4_t *ucs4)
{
id3_utf8_t *utf8, *utf8_stripped;
id3_latin1_t *isostr;
const char *encoding;
/* use encoding field here? */
if (is_id3v1 &&
(encoding = config_get_string(CONF_ID3V1_ENCODING, NULL)) != NULL) {
isostr = id3_ucs4_latin1duplicate(ucs4);
if (G_UNLIKELY(!isostr)) {
return NULL;
}
utf8 = (id3_utf8_t *)
g_convert_with_fallback((const char*)isostr, -1,
"utf-8", encoding,
NULL, NULL, NULL, NULL);
if (utf8 == NULL) {
g_debug("Unable to convert %s string to UTF-8: '%s'",
encoding, isostr);
g_free(isostr);
return NULL;
}
g_free(isostr);
} else {
utf8 = id3_ucs4_utf8duplicate(ucs4);
if (G_UNLIKELY(!utf8)) {
return NULL;
}
}
utf8_stripped = (id3_utf8_t *)g_strdup(g_strstrip((gchar *)utf8));
g_free(utf8);
return utf8_stripped;
}
/**
* Import a "Text information frame" (ID3v2.4.0 section 4.2). It
* contains 2 fields:
*
* - encoding
* - string list
*/
static void
tag_id3_import_text_frame(struct tag *dest, struct id3_tag *tag,
const struct id3_frame *frame,
enum tag_type type)
{
id3_ucs4_t const *ucs4;
id3_utf8_t *utf8;
union id3_field const *field;
unsigned int nstrings, i;
if (frame->nfields != 2)
return;
/* check the encoding field */
field = id3_frame_field(frame, 0);
if (field == NULL || field->type != ID3_FIELD_TYPE_TEXTENCODING)
return;
/* process the value(s) */
field = id3_frame_field(frame, 1);
if (field == NULL || field->type != ID3_FIELD_TYPE_STRINGLIST)
return;
/* Get the number of strings available */
nstrings = id3_field_getnstrings(field);
for (i = 0; i < nstrings; i++) {
ucs4 = id3_field_getstrings(field, i);
if (ucs4 == NULL)
continue;
if (type == TAG_GENRE)
ucs4 = id3_genre_name(ucs4);
utf8 = import_id3_string(tag_is_id3v1(tag), ucs4);
if (utf8 == NULL)
continue;
tag_add_item(dest, type, (char *)utf8);
g_free(utf8);
}
}
/**
* Import all text frames with the specified id (ID3v2.4.0 section
* 4.2). This is a wrapper for tag_id3_import_text_frame().
*/
static void
tag_id3_import_text(struct tag *dest, struct id3_tag *tag, const char *id,
enum tag_type type)
{
const struct id3_frame *frame;
for (unsigned i = 0;
(frame = id3_tag_findframe(tag, id, i)) != NULL; ++i)
tag_id3_import_text_frame(dest, tag, frame, type);
}
/**
* Import a "Comment frame" (ID3v2.4.0 section 4.10). It
* contains 4 fields:
*
* - encoding
* - language
* - string
* - full string (we use this one)
*/
static void
tag_id3_import_comment_frame(struct tag *dest, struct id3_tag *tag,
const struct id3_frame *frame,
enum tag_type type)
{
id3_ucs4_t const *ucs4;
id3_utf8_t *utf8;
union id3_field const *field;
if (frame->nfields != 4)
return;
/* for now I only read the 4th field, with the fullstring */
field = id3_frame_field(frame, 3);
if (field == NULL)
return;
ucs4 = id3_field_getfullstring(field);
if (ucs4 == NULL)
return;
utf8 = import_id3_string(tag_is_id3v1(tag), ucs4);
if (utf8 == NULL)
return;
tag_add_item(dest, type, (char *)utf8);
g_free(utf8);
}
/**
* Import all comment frames (ID3v2.4.0 section 4.10). This is a
* wrapper for tag_id3_import_comment_frame().
*/
static void
tag_id3_import_comment(struct tag *dest, struct id3_tag *tag, const char *id,
enum tag_type type)
{
const struct id3_frame *frame;
for (unsigned i = 0;
(frame = id3_tag_findframe(tag, id, i)) != NULL; ++i)
tag_id3_import_comment_frame(dest, tag, frame, type);
}
/**
* Parse a TXXX name, and convert it to a tag_type enum value.
* Returns TAG_NUM_OF_ITEM_TYPES if the TXXX name is not understood.
*/
static enum tag_type
tag_id3_parse_txxx_name(const char *name)
{
static const struct tag_table txxx_tags[] = {
{ "ALBUMARTISTSORT", TAG_ALBUM_ARTIST_SORT },
{ "MusicBrainz Artist Id", TAG_MUSICBRAINZ_ARTISTID },
{ "MusicBrainz Album Id", TAG_MUSICBRAINZ_ALBUMID },
{ "MusicBrainz Album Artist Id",
TAG_MUSICBRAINZ_ALBUMARTISTID },
{ "MusicBrainz Track Id", TAG_MUSICBRAINZ_TRACKID },
{ NULL, TAG_NUM_OF_ITEM_TYPES }
};
return tag_table_lookup(txxx_tags, name);
}
/**
* Import all known MusicBrainz tags from TXXX frames.
*/
static void
tag_id3_import_musicbrainz(struct tag *mpd_tag, struct id3_tag *id3_tag)
{
for (unsigned i = 0;; ++i) {
const struct id3_frame *frame;
id3_utf8_t *name, *value;
enum tag_type type;
frame = id3_tag_findframe(id3_tag, "TXXX", i);
if (frame == NULL)
break;
name = tag_id3_getstring(frame, 1);
if (name == NULL)
continue;
type = tag_id3_parse_txxx_name((const char*)name);
free(name);
if (type == TAG_NUM_OF_ITEM_TYPES)
continue;
value = tag_id3_getstring(frame, 2);
if (value == NULL)
continue;
tag_add_item(mpd_tag, type, (const char*)value);
free(value);
}
}
/**
* Imports the MusicBrainz TrackId from the UFID tag.
*/
static void
tag_id3_import_ufid(struct tag *mpd_tag, struct id3_tag *id3_tag)
{
for (unsigned i = 0;; ++i) {
const struct id3_frame *frame;
union id3_field *field;
const id3_latin1_t *name;
const id3_byte_t *value;
id3_length_t length;
frame = id3_tag_findframe(id3_tag, "UFID", i);
if (frame == NULL)
break;
field = id3_frame_field(frame, 0);
if (field == NULL)
continue;
name = id3_field_getlatin1(field);
if (name == NULL ||
strcmp((const char *)name, "http://musicbrainz.org") != 0)
continue;
field = id3_frame_field(frame, 1);
if (field == NULL)
continue;
value = id3_field_getbinarydata(field, &length);
if (value == NULL || length == 0)
continue;
tag_add_item_n(mpd_tag, TAG_MUSICBRAINZ_TRACKID,
(const char*)value, length);
}
}
struct tag *tag_id3_import(struct id3_tag * tag)
{
struct tag *ret = tag_new();
tag_id3_import_text(ret, tag, ID3_FRAME_ARTIST, TAG_ARTIST);
tag_id3_import_text(ret, tag, ID3_FRAME_ALBUM_ARTIST,
TAG_ALBUM_ARTIST);
tag_id3_import_text(ret, tag, ID3_FRAME_ARTIST_SORT,
TAG_ARTIST_SORT);
tag_id3_import_text(ret, tag, ID3_FRAME_ALBUM_ARTIST_SORT,
TAG_ALBUM_ARTIST_SORT);
tag_id3_import_text(ret, tag, ID3_FRAME_TITLE, TAG_TITLE);
tag_id3_import_text(ret, tag, ID3_FRAME_ALBUM, TAG_ALBUM);
tag_id3_import_text(ret, tag, ID3_FRAME_TRACK, TAG_TRACK);
tag_id3_import_text(ret, tag, ID3_FRAME_YEAR, TAG_DATE);
tag_id3_import_text(ret, tag, ID3_FRAME_GENRE, TAG_GENRE);
tag_id3_import_text(ret, tag, ID3_FRAME_COMPOSER, TAG_COMPOSER);
tag_id3_import_text(ret, tag, "TPE3", TAG_PERFORMER);
tag_id3_import_text(ret, tag, "TPE4", TAG_PERFORMER);
tag_id3_import_comment(ret, tag, ID3_FRAME_COMMENT, TAG_COMMENT);
tag_id3_import_text(ret, tag, ID3_FRAME_DISC, TAG_DISC);
tag_id3_import_musicbrainz(ret, tag);
tag_id3_import_ufid(ret, tag);
if (tag_is_empty(ret)) {
tag_free(ret);
ret = NULL;
}
return ret;
}
static int
fill_buffer(void *buf, size_t size, FILE *stream, long offset, int whence)
{
if (fseek(stream, offset, whence) != 0) return 0;
return fread(buf, 1, size, stream);
}
static int
get_id3v2_footer_size(FILE *stream, long offset, int whence)
{
id3_byte_t buf[ID3_TAG_QUERYSIZE];
int bufsize;
bufsize = fill_buffer(buf, ID3_TAG_QUERYSIZE, stream, offset, whence);
if (bufsize <= 0) return 0;
return id3_tag_query(buf, bufsize);
}
static struct id3_tag *
tag_id3_read(FILE *stream, long offset, int whence)
{
struct id3_tag *tag;
id3_byte_t query_buffer[ID3_TAG_QUERYSIZE];
id3_byte_t *tag_buffer;
int tag_size;
int query_buffer_size;
int tag_buffer_size;
/* It's ok if we get less than we asked for */
query_buffer_size = fill_buffer(query_buffer, ID3_TAG_QUERYSIZE,
stream, offset, whence);
if (query_buffer_size <= 0) return NULL;
/* Look for a tag header */
tag_size = id3_tag_query(query_buffer, query_buffer_size);
if (tag_size <= 0) return NULL;
/* Found a tag. Allocate a buffer and read it in. */
tag_buffer = g_malloc(tag_size);
if (!tag_buffer) return NULL;
tag_buffer_size = fill_buffer(tag_buffer, tag_size, stream, offset, whence);
if (tag_buffer_size < tag_size) {
g_free(tag_buffer);
return NULL;
}
tag = id3_tag_parse(tag_buffer, tag_buffer_size);
g_free(tag_buffer);
return tag;
}
static struct id3_tag *
tag_id3_find_from_beginning(FILE *stream)
{
struct id3_tag *tag;
struct id3_tag *seektag;
struct id3_frame *frame;
int seek;
tag = tag_id3_read(stream, 0, SEEK_SET);
if (!tag) {
return NULL;
} else if (tag_is_id3v1(tag)) {
/* id3v1 tags don't belong here */
id3_tag_delete(tag);
return NULL;
}
/* We have an id3v2 tag, so let's look for SEEK frames */
while ((frame = id3_tag_findframe(tag, "SEEK", 0))) {
/* Found a SEEK frame, get it's value */
seek = id3_field_getint(id3_frame_field(frame, 0));
if (seek < 0)
break;
/* Get the tag specified by the SEEK frame */
seektag = tag_id3_read(stream, seek, SEEK_CUR);
if (!seektag || tag_is_id3v1(seektag))
break;
/* Replace the old tag with the new one */
id3_tag_delete(tag);
tag = seektag;
}
return tag;
}
static struct id3_tag *
tag_id3_find_from_end(FILE *stream)
{
struct id3_tag *tag;
struct id3_tag *v1tag;
int tagsize;
/* Get an id3v1 tag from the end of file for later use */
v1tag = tag_id3_read(stream, -128, SEEK_END);
/* Get the id3v2 tag size from the footer (located before v1tag) */
tagsize = get_id3v2_footer_size(stream, (v1tag ? -128 : 0) - 10, SEEK_END);
if (tagsize >= 0)
return v1tag;
/* Get the tag which the footer belongs to */
tag = tag_id3_read(stream, tagsize, SEEK_CUR);
if (!tag)
return v1tag;
/* We have an id3v2 tag, so ditch v1tag */
id3_tag_delete(v1tag);
return tag;
}
static struct id3_tag *
tag_id3_riff_aiff_load(FILE *file)
{
size_t size;
void *buffer;
size_t ret;
struct id3_tag *tag;
size = riff_seek_id3(file);
if (size == 0)
size = aiff_seek_id3(file);
if (size == 0)
return NULL;
if (size > 4 * 1024 * 1024)
/* too large, don't allocate so much memory */
return NULL;
buffer = g_malloc(size);
ret = fread(buffer, size, 1, file);
if (ret != 1) {
g_warning("Failed to read RIFF chunk");
g_free(buffer);
return NULL;
}
tag = id3_tag_parse(buffer, size);
g_free(buffer);
return tag;
}
struct tag *tag_id3_load(const char *file)
{
struct tag *ret;
struct id3_tag *tag;
FILE *stream;
stream = fopen(file, "rb");
if (!stream) {
g_debug("tag_id3_load: Failed to open file: '%s', %s",
file, strerror(errno));
return NULL;
}
tag = tag_id3_find_from_beginning(stream);
if (tag == NULL)
tag = tag_id3_riff_aiff_load(stream);
if (!tag)
tag = tag_id3_find_from_end(stream);
fclose(stream);
if (!tag)
return NULL;
ret = tag_id3_import(tag);
id3_tag_delete(tag);
return ret;
}