2 * Copyright (C) 2008 by INdT
4 * This program is free software; you can redistribute it and/or
5 * modify it under the terms of the GNU Lesser General Public License
6 * as published by the Free Software Foundation; either version 2
7 * of the License, or (at your option) any later version.
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
14 * You should have received a copy of the GNU Lesser General Public License
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA 02111-1307, USA.
18 * @author Andre Moreira Magalhaes <andre.magalhaes@openbossa.org>
32 #define _XOPEN_SOURCE 600
33 #include <lightmediascanner_plugin.h>
34 #include <lightmediascanner_db.h>
35 #include <lightmediascanner_charset_conv.h>
36 #include <sys/types.h>
45 #define ID3V2_HEADER_SIZE 10
46 #define ID3V2_FOOTER_SIZE 10
49 ID3_ENCODING_LATIN1 = 0,
56 #define ID3_NUM_ENCODINGS ID3_ENCODING_LAST
59 #include "id3v1_genres.c"
62 struct lms_string_size title;
63 struct lms_string_size artist;
64 struct lms_string_size album;
65 struct lms_string_size genre;
66 unsigned char trackno;
67 int cur_artist_priority;
70 struct id3v2_frame_header {
72 unsigned int frame_size;
74 int data_length_indicator;
84 } __attribute__((packed));
87 struct lms_plugin plugin;
88 lms_db_audio_t *audio_db;
89 lms_charset_conv_t *cs_convs[ID3_NUM_ENCODINGS];
92 static const char _name[] = "id3";
93 static const struct lms_string_size _exts[] = {
94 LMS_STATIC_STRING_SIZE(".mp3"),
95 LMS_STATIC_STRING_SIZE(".aac")
99 _to_uint(const char *data, int data_size)
101 unsigned int sum = 0;
102 unsigned int last, i;
104 last = data_size > 4 ? 3 : data_size - 1;
106 for (i = 0; i <= last; i++)
107 sum |= ((unsigned char) data[i]) << ((last - i) * 8);
113 _is_id3v2_second_synch_byte(unsigned char byte)
117 if ((byte & 0xE0) == 0xE0)
125 long buffer_offset = 0;
127 int buffer_size = sizeof(buffer);
128 const char pattern[] = "ID3";
131 /* These variables are used to keep track of a partial match that happens at
132 * the end of a buffer. */
133 int previous_partial_match = -1;
134 int previous_partial_synch_match = 0;
135 int first_synch_byte;
137 /* Start the search at the beginning of the file. */
138 lseek(fd, 0, SEEK_SET);
140 if ((nread = read(fd, &buffer, buffer_size)) != buffer_size)
143 /* check if pattern is in the beggining of the file */
144 if (memcmp(buffer, pattern, 3) == 0)
147 /* This loop is the crux of the find method. There are three cases that we
148 * want to account for:
149 * (1) The previously searched buffer contained a partial match of the
150 * search pattern and we want to see if the next one starts with the
151 * remainder of that pattern.
153 * (2) The search pattern is wholly contained within the current buffer.
155 * (3) The current buffer ends with a partial match of the pattern. We will
156 * note this for use in the next iteration, where we will check for the rest
160 /* (1) previous partial match */
161 if (previous_partial_synch_match &&
162 _is_id3v2_second_synch_byte(buffer[0]))
165 if (previous_partial_match >= 0 &&
166 previous_partial_match < buffer_size) {
167 const int pat_offset = buffer_size - previous_partial_match;
169 if (memcmp(buffer, pattern + pat_offset, 3 - pat_offset) == 0)
170 return buffer_offset - buffer_size + previous_partial_match;
173 /* (2) pattern contained in current buffer */
175 while ((p = memchr(p, 'I', buffer_size - (p - buffer)))) {
176 if (buffer_size - (p - buffer) < 3)
179 if (memcmp(p, pattern, 3) == 0)
180 return buffer_offset + (p - buffer);
185 p = memchr(buffer, 255, buffer_size);
187 first_synch_byte = p - buffer;
189 first_synch_byte = -1;
191 /* Here we have to loop because there could be several of the first
192 * (11111111) byte, and we want to check all such instances until we
193 * find a full match (11111111 111) or hit the end of the buffer.
195 while (first_synch_byte >= 0) {
196 /* if this *is not* at the end of the buffer */
197 if (first_synch_byte < buffer_size - 1) {
198 if(_is_id3v2_second_synch_byte(buffer[first_synch_byte + 1]))
199 /* We've found the frame synch pattern. */
202 /* We found 11111111 at the end of the current buffer
203 * indicating a partial match of the synch pattern.
204 * The find() below should return -1 and break out of
207 previous_partial_synch_match = 1;
210 /* Check in the rest of the buffer. */
211 p = memchr(p + 1, 255, buffer_size - (p + 1 - buffer));
213 first_synch_byte = p - buffer;
215 first_synch_byte = -1;
218 /* (3) partial match */
219 if (buffer[nread - 1] == pattern[1])
220 previous_partial_match = nread - 1;
221 else if (memcmp(&buffer[nread - 2], pattern, 2) == 0)
222 previous_partial_match = nread - 2;
223 buffer_offset += buffer_size;
225 if ((nread = read(fd, &buffer, sizeof(buffer))) == -1)
233 _get_id3v2_frame_header_size(unsigned int version)
248 _parse_id3v2_frame_header(char *data, unsigned int version, struct id3v2_frame_header *fh)
254 memcpy(fh->frame_id, data, 3);
256 fh->frame_size = _to_uint(data + 3, 3);
258 fh->data_length_indicator = 0;
261 memcpy(fh->frame_id, data, 4);
262 fh->frame_size = _to_uint(data + 4, 4);
263 fh->compression = data[9] & 0x40;
264 fh->data_length_indicator = 0;
268 memcpy(fh->frame_id, data, 4);
269 fh->frame_size = _to_uint(data + 4, 4);
270 fh->compression = data[9] & 0x4;
271 fh->data_length_indicator = data[9] & 0x1;
277 _get_id3v2_frame_info(const char *frame_data, unsigned int frame_size, struct lms_string_size *s, lms_charset_conv_t *cs_conv, int strip)
279 if (frame_size > s->len)
280 s->str = realloc(s->str, sizeof(char) * (frame_size + 1));
281 memcpy(s->str, frame_data, frame_size);
282 s->str[frame_size] = '\0';
285 lms_charset_conv(cs_conv, &s->str, &s->len);
287 lms_string_size_strip_and_free(s);
291 _get_id3v1_genre(unsigned int genre, struct lms_string_size *out)
293 if (genre < ID3V1_NUM_GENRES) {
294 unsigned int size, base, len;
296 base = id3v1_genres_offsets[genre];
297 size = id3v1_genres_offsets[genre + 1] - base;
300 if (len > out->len) {
301 char *p = realloc(out->str, size);
308 memcpy(out->str, id3v1_genres_mem + base, size);
316 _parse_id3v1_genre(const char *str_genre, struct lms_string_size *out)
318 return _get_id3v1_genre(atoi(str_genre), out);
322 _get_id3v2_genre(const char *frame_data, unsigned int frame_size, struct lms_string_size *out, lms_charset_conv_t *cs_conv)
325 struct lms_string_size genre = {0};
327 _get_id3v2_frame_info(frame_data, frame_size, &genre, cs_conv, 1);
332 is_number = (genre.len != 0 && genre.str[0] != '(');
334 for (i = 0; i < genre.len; ++i) {
335 if (!isdigit(genre.str[i])) {
342 if (is_number && _parse_id3v1_genre(genre.str, out) == 0) {
343 /* id3v1 genre found */
348 /* ID3v2.3 "content type" can contain a ID3v1 genre number in parenthesis at
349 * the beginning of the field. If this is all that the field contains, do a
350 * translation from that number to the name and return that. If there is a
351 * string folloing the ID3v1 genre number, that is considered to be
352 * authoritative and we return that instead. Or finally, the field may
353 * simply be free text, in which case we just return the value. */
355 if (genre.len > 1 && genre.str[0] == '(') {
356 char *closing = NULL;
358 if (genre.str[genre.len - 1] == ')') {
359 closing = strchr(genre.str, ')');
360 if (closing == genre.str + genre.len - 1) {
361 /* ) is the last character and only appears once in the
362 * string get the id3v1 genre enclosed by parentheses
364 if (_parse_id3v1_genre(genre.str + 1, out) == 0) {
371 /* get the string followed by the id3v1 genre */
373 closing = strchr(genre.str, ')');
377 out->len = genre.len - (closing - genre.str);
378 out->str = genre.str;
379 memmove(out->str, closing, out->len + 1); /* includes '\0' */
380 lms_string_size_strip_and_free(out);
390 _parse_id3v2_frame(struct id3v2_frame_header *fh, const char *frame_data, struct id3_info *info, lms_charset_conv_t **cs_convs)
392 lms_charset_conv_t *cs_conv = NULL;
393 unsigned int text_encoding, frame_size;
394 static const int artist_priorities[] = { 3, 4, 2, 1 };
397 fprintf(stderr, "frame id = %.4s frame size = %d text encoding = %d\n",
398 fh->frame_id, fh->frame_size, frame_data[0]);
407 text_encoding = frame_data[0];
409 /* skip first byte - text encoding */
411 frame_size = fh->frame_size - 1;
413 if (text_encoding >= 0 && text_encoding < ID3_NUM_ENCODINGS) {
414 if (text_encoding == ID3_ENCODING_UTF16) {
415 if (memcmp(frame_data, "\xfe\xff", 2) == 0)
416 text_encoding = ID3_ENCODING_UTF16BE;
418 text_encoding = ID3_ENCODING_UTF16LE;
422 cs_conv = cs_convs[text_encoding];
425 /* ID3v2.2 used 3 bytes for the frame id, so let's check it */
426 if (memcmp(fh->frame_id, "TIT2", 4) == 0 ||
427 memcmp(fh->frame_id, "TT2", 3) == 0)
428 _get_id3v2_frame_info(frame_data, frame_size, &info->title, cs_conv, 1);
429 else if (memcmp(fh->frame_id, "TP", 2) == 0) {
432 if (memcmp(fh->frame_id, "TPE", 3) == 0) {
433 /* this check shouldn't be needed, but let's make sure */
434 if (fh->frame_id[3] >= '1' && fh->frame_id[3] <= '4')
435 index = fh->frame_id[3] - '1';
438 /* ignore TPA, TPB */
439 if (fh->frame_id[2] >= '1' && fh->frame_id[2] <= '4')
440 index = fh->frame_id[2] - '1';
444 artist_priorities[index] > info->cur_artist_priority) {
445 struct lms_string_size artist = {0};
447 _get_id3v2_frame_info(frame_data, frame_size, &artist, cs_conv, 1);
449 if (info->artist.str)
450 free(info->artist.str);
451 info->artist = artist;
452 info->cur_artist_priority = artist_priorities[index];
457 else if (memcmp(fh->frame_id, "TAL", 3) == 0)
458 _get_id3v2_frame_info(frame_data, frame_size, &info->album, cs_conv, 1);
460 else if (memcmp(fh->frame_id, "TCO", 3) == 0)
461 _get_id3v2_genre(frame_data, frame_size, &info->genre, cs_conv);
462 else if (memcmp(fh->frame_id, "TRCK", 4) == 0 ||
463 memcmp(fh->frame_id, "TRK", 3) == 0) {
464 struct lms_string_size trackno = {0};
465 _get_id3v2_frame_info(frame_data, frame_size, &trackno, cs_conv, 0);
466 info->trackno = atoi(trackno.str);
472 _parse_id3v2(int fd, long id3v2_offset, struct id3_info *info, lms_charset_conv_t **cs_convs)
474 char header_data[10], frame_header_data[10];
475 unsigned int tag_size, major_version, frame_data_pos, frame_data_length, frame_header_size;
476 int extended_header, footer_present;
477 struct id3v2_frame_header fh;
480 lseek(fd, id3v2_offset, SEEK_SET);
483 if (read(fd, header_data, ID3V2_HEADER_SIZE) != ID3V2_HEADER_SIZE)
486 tag_size = _to_uint(header_data + 6, 4);
491 major_version = header_data[3];
494 frame_data_length = tag_size;
496 /* check for extended header */
497 extended_header = header_data[5] & 0x20; /* bit 6 */
498 if (extended_header) {
499 /* skip extended header */
500 unsigned int extended_header_size;
501 char extended_header_data[4];
503 if (read(fd, extended_header_data, 4) != 4)
505 extended_header_size = _to_uint(extended_header_data, 4);
506 lseek(fd, extended_header_size - 4, SEEK_CUR);
507 frame_data_pos += extended_header_size;
508 frame_data_length -= extended_header_size;
511 footer_present = header_data[5] & 0x8; /* bit 4 */
512 if (footer_present && frame_data_length > ID3V2_FOOTER_SIZE)
513 frame_data_length -= ID3V2_FOOTER_SIZE;
515 frame_header_size = _get_id3v2_frame_header_size(major_version);
516 while (frame_data_pos < frame_data_length - frame_header_size) {
517 nread = read(fd, frame_header_data, frame_header_size);
521 if (nread != frame_header_size)
524 if (frame_header_data[0] == 0)
527 _parse_id3v2_frame_header(frame_header_data, major_version, &fh);
531 if (!fh.compression &&
532 fh.frame_id[0] == 'T' &&
533 memcmp(fh.frame_id, "TXXX", 4) != 0) {
536 if (fh.data_length_indicator)
537 lseek(fd, 4, SEEK_CUR);
539 frame_data = malloc(sizeof(char) * fh.frame_size);
540 if (read(fd, frame_data, fh.frame_size) != fh.frame_size) {
545 _parse_id3v2_frame(&fh, frame_data, info, cs_convs);
549 if (fh.data_length_indicator)
550 lseek(fd, fh.frame_size + 4, SEEK_CUR);
552 lseek(fd, fh.frame_size, SEEK_CUR);
555 frame_data_pos += fh.frame_size + frame_header_size;
562 _parse_id3v1(int fd, struct id3_info *info, lms_charset_conv_t *cs_conv)
564 struct id3v1_tag tag;
565 if (read(fd, &tag, sizeof(struct id3v1_tag)) == -1)
568 info->title.str = strndup(tag.title, 30);
569 info->title.len = strlen(info->title.str);
570 lms_charset_conv(cs_conv, &info->title.str, &info->title.len);
571 info->artist.str = strndup(tag.artist, 30);
572 info->artist.len = strlen(info->artist.str);
573 lms_charset_conv(cs_conv, &info->artist.str, &info->artist.len);
574 info->album.str = strndup(tag.album, 30);
575 info->album.len = strlen(info->album.str);
576 lms_charset_conv(cs_conv, &info->album.str, &info->album.len);
577 _get_id3v1_genre(tag.genre, &info->genre);
578 if (tag.comments[28] == 0 && tag.comments[29] != 0)
579 info->trackno = (unsigned char) tag.comments[29];
585 _match(struct plugin *p, const char *path, int len, int base)
589 i = lms_which_extension(path, len, _exts, LMS_ARRAY_SIZE(_exts));
593 return (void*)(i + 1);
597 _parse(struct plugin *plugin, struct lms_context *ctxt, const struct lms_file_info *finfo, void *match)
599 struct id3_info info = {{0}, {0}, {0}, {0}, 0, -1};
600 struct lms_audio_info audio_info = {0, {0}, {0}, {0}, {0}, 0, 0, 0};
604 fd = open(finfo->path, O_RDONLY);
610 id3v2_offset = _find_id3v2(fd);
611 if (id3v2_offset >= 0) {
613 fprintf(stderr, "id3v2 tag found in file %s with offset %ld\n",
614 finfo->path, id3v2_offset);
616 if (_parse_id3v2(fd, id3v2_offset, &info, plugin->cs_convs) != 0) {
624 fprintf(stderr, "id3v2 tag not found in file %s. trying id3v1\n",
627 /* check for id3v1 tag */
628 if (lseek(fd, -128, SEEK_END) == -1) {
633 if (read(fd, &tag, 3) == -1) {
638 if (memcmp(tag, "TAG", 3) == 0) {
640 fprintf(stderr, "id3v1 tag found in file %s\n", finfo->path);
642 if (_parse_id3v1(fd, &info, ctxt->cs_conv) != 0) {
649 if (!info.title.str) {
651 ext_idx = ((int)match) - 1;
652 info.title.len = finfo->path_len - finfo->base - _exts[ext_idx].len;
653 info.title.str = malloc((info.title.len + 1) * sizeof(char));
654 memcpy(info.title.str, finfo->path + finfo->base, info.title.len);
655 info.title.str[info.title.len] = '\0';
656 lms_charset_conv(ctxt->cs_conv, &info.title.str, &info.title.len);
660 fprintf(stderr, "file %s info\n", finfo->path);
661 fprintf(stderr, "\ttitle='%s'\n", info.title.str);
662 fprintf(stderr, "\tartist='%s'\n", info.artist.str);
663 fprintf(stderr, "\talbum='%s'\n", info.album.str);
664 fprintf(stderr, "\tgenre='%s'\n", info.genre.str);
665 fprintf(stderr, "\ttrack number='%d'\n", info.trackno);
668 audio_info.id = finfo->id;
669 audio_info.title = info.title;
670 audio_info.artist = info.artist;
671 audio_info.album = info.album;
672 audio_info.genre = info.genre;
673 audio_info.trackno = info.trackno;
674 r = lms_db_audio_add(plugin->audio_db, &audio_info);
677 posix_fadvise(fd, 0, 0, POSIX_FADV_DONTNEED);
681 free(info.title.str);
683 free(info.artist.str);
685 free(info.album.str);
687 free(info.genre.str);
693 _setup(struct plugin *plugin, struct lms_context *ctxt)
696 const char *id3_encodings[ID3_NUM_ENCODINGS] = {
704 plugin->audio_db = lms_db_audio_new(ctxt->db);
705 if (!plugin->audio_db)
708 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
709 /* do not create charset conv for UTF-8 encoding */
710 if (!id3_encodings[i]) {
711 plugin->cs_convs[i] = NULL;
714 plugin->cs_convs[i] = lms_charset_conv_new_full(0, 0);
715 if (!plugin->cs_convs[i])
717 lms_charset_conv_add(plugin->cs_convs[i], id3_encodings[i]);
724 _start(struct plugin *plugin, struct lms_context *ctxt)
726 return lms_db_audio_start(plugin->audio_db);
730 _finish(struct plugin *plugin, struct lms_context *ctxt)
734 if (plugin->audio_db)
735 lms_db_audio_free(plugin->audio_db);
737 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
738 if (plugin->cs_convs[i])
739 lms_charset_conv_free(plugin->cs_convs[i]);
746 _close(struct plugin *plugin)
752 API struct lms_plugin *
753 lms_plugin_open(void)
755 struct plugin *plugin;
757 plugin = (struct plugin *)malloc(sizeof(*plugin));
758 plugin->plugin.name = _name;
759 plugin->plugin.match = (lms_plugin_match_fn_t)_match;
760 plugin->plugin.parse = (lms_plugin_parse_fn_t)_parse;
761 plugin->plugin.close = (lms_plugin_close_fn_t)_close;
762 plugin->plugin.setup = (lms_plugin_setup_fn_t)_setup;
763 plugin->plugin.start = (lms_plugin_start_fn_t)_start;
764 plugin->plugin.finish = (lms_plugin_finish_fn_t)_finish;
766 return (struct lms_plugin *)plugin;