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 static const char *id3v1_genres[] = {
196 "Christian Gangsta Rap",
200 "Contemporary Christian",
209 #define ID3V1_NUM_GENRES (sizeof(id3v1_genres) / sizeof(char *))
212 struct lms_string_size title;
213 struct lms_string_size artist;
214 struct lms_string_size album;
215 struct lms_string_size genre;
216 unsigned char trackno;
217 int cur_artist_priority;
220 struct id3v2_frame_header {
222 unsigned int frame_size;
224 int data_length_indicator;
234 } __attribute__((packed));
237 struct lms_plugin plugin;
238 lms_db_audio_t *audio_db;
239 lms_charset_conv_t *cs_convs[ID3_NUM_ENCODINGS];
242 static const char _name[] = "id3";
243 static const struct lms_string_size _exts[] = {
244 LMS_STATIC_STRING_SIZE(".mp3"),
245 LMS_STATIC_STRING_SIZE(".aac")
249 _to_uint(const char *data, int data_size)
251 unsigned int sum = 0;
252 unsigned int last, i;
254 last = data_size > 4 ? 3 : data_size - 1;
256 for (i = 0; i <= last; i++)
257 sum |= ((unsigned char) data[i]) << ((last - i) * 8);
263 _is_id3v2_second_synch_byte(unsigned char byte)
267 if ((byte & 0xE0) == 0xE0)
275 long buffer_offset = 0;
277 int buffer_size = sizeof(buffer);
278 const char pattern[] = "ID3";
281 /* These variables are used to keep track of a partial match that happens at
282 * the end of a buffer. */
283 int previous_partial_match = -1;
284 int previous_partial_synch_match = 0;
285 int first_synch_byte;
287 /* Start the search at the beginning of the file. */
288 lseek(fd, 0, SEEK_SET);
290 if ((nread = read(fd, &buffer, buffer_size)) != buffer_size)
293 /* check if pattern is in the beggining of the file */
294 if (memcmp(buffer, pattern, 3) == 0)
297 /* This loop is the crux of the find method. There are three cases that we
298 * want to account for:
299 * (1) The previously searched buffer contained a partial match of the search
300 * pattern and we want to see if the next one starts with the remainder of
303 * (2) The search pattern is wholly contained within the current buffer.
305 * (3) The current buffer ends with a partial match of the pattern. We will
306 * note this for use in the next iteration, where we will check for the rest
309 /* (1) previous partial match */
310 if (previous_partial_synch_match && _is_id3v2_second_synch_byte(buffer[0]))
313 if (previous_partial_match >= 0 && previous_partial_match < buffer_size) {
314 const int pattern_offset = buffer_size - previous_partial_match;
316 if (memcmp(buffer, pattern + pattern_offset, 3 - pattern_offset) == 0)
317 return buffer_offset - buffer_size + previous_partial_match;
320 /* (2) pattern contained in current buffer */
322 while ((p = memchr(p, 'I', buffer_size - (p - buffer)))) {
323 if (buffer_size - (p - buffer) < 3)
326 if (memcmp(p, pattern, 3) == 0)
327 return buffer_offset + (p - buffer);
332 p = memchr(buffer, 255, buffer_size);
334 first_synch_byte = p - buffer;
336 first_synch_byte = -1;
338 /* Here we have to loop because there could be several of the first
339 * (11111111) byte, and we want to check all such instances until we find
340 * a full match (11111111 111) or hit the end of the buffer. */
341 while (first_synch_byte >= 0) {
342 /* if this *is not* at the end of the buffer */
343 if (first_synch_byte < buffer_size - 1) {
344 if(_is_id3v2_second_synch_byte(buffer[first_synch_byte + 1]))
345 /* We've found the frame synch pattern. */
348 /* We found 11111111 at the end of the current buffer indicating a
349 * partial match of the synch pattern. The find() below should
350 * return -1 and break out of the loop. */
351 previous_partial_synch_match = 1;
354 /* Check in the rest of the buffer. */
355 p = memchr(p + 1, 255, buffer_size - (p + 1 - buffer));
357 first_synch_byte = p - buffer;
359 first_synch_byte = -1;
362 /* (3) partial match */
363 if (buffer[nread - 1] == pattern[1])
364 previous_partial_match = nread - 1;
365 else if (memcmp(&buffer[nread - 2], pattern, 2) == 0)
366 previous_partial_match = nread - 2;
367 buffer_offset += buffer_size;
369 if ((nread = read(fd, &buffer, sizeof(buffer))) == -1)
377 _get_id3v2_frame_header_size(unsigned int version)
392 _parse_id3v2_frame_header(char *data, unsigned int version, struct id3v2_frame_header *fh)
398 memcpy(fh->frame_id, data, 3);
400 fh->frame_size = _to_uint(data + 3, 3);
402 fh->data_length_indicator = 0;
405 memcpy(fh->frame_id, data, 4);
406 fh->frame_size = _to_uint(data + 4, 4);
407 fh->compression = data[9] & 0x40;
408 fh->data_length_indicator = 0;
412 memcpy(fh->frame_id, data, 4);
413 fh->frame_size = _to_uint(data + 4, 4);
414 fh->compression = data[9] & 0x4;
415 fh->data_length_indicator = data[9] & 0x1;
421 _get_id3v2_frame_info(const char *frame_data, unsigned int frame_size, struct lms_string_size *s, lms_charset_conv_t *cs_conv)
423 if (frame_size > s->len)
424 s->str = realloc(s->str, sizeof(char) * (frame_size + 1));
425 memcpy(s->str, frame_data, frame_size);
426 s->str[frame_size] = '\0';
429 lms_charset_conv(cs_conv, &s->str, &s->len);
433 _get_id3v1_genre(int genre, struct lms_string_size *out)
435 if (genre >= 0 && genre < ID3V1_NUM_GENRES) {
436 out->str = strdup(id3v1_genres[genre]);
437 out->len = strlen(out->str);
444 _parse_id3v1_genre(const char *str_genre, struct lms_string_size *out)
446 return _get_id3v1_genre(atoi(str_genre), out);
450 _get_id3v2_genre(const char *frame_data, unsigned int frame_size, struct lms_string_size *out, lms_charset_conv_t *cs_conv)
453 struct lms_string_size genre = {0};
455 _get_id3v2_frame_info(frame_data, frame_size, &genre, cs_conv);
463 is_number = (genre.len != 0 && genre.str[0] != '(');
465 for (i = 0; i < genre.len; ++i) {
466 if (!isdigit(genre.str[i])) {
473 if (is_number && _parse_id3v1_genre(genre.str, out) == 0) {
474 /* id3v1 genre found */
479 /* ID3v2.3 "content type" can contain a ID3v1 genre number in parenthesis at
480 * the beginning of the field. If this is all that the field contains, do a
481 * translation from that number to the name and return that. If there is a
482 * string folloing the ID3v1 genre number, that is considered to be
483 * authoritative and we return that instead. Or finally, the field may
484 * simply be free text, in which case we just return the value. */
486 if (genre.len > 1 && genre.str[0] == '(') {
487 char *closing = NULL;
489 if (genre.str[genre.len - 1] == ')') {
490 closing = strchr(genre.str, ')');
491 if (closing == genre.str + genre.len - 1) {
492 /* ) is the last character and only appears once in the string */
493 /* get the id3v1 genre enclosed by parentheses */
494 if (_parse_id3v1_genre(genre.str + 1, out) == 0) {
501 /* get the string followed by the id3v1 genre */
503 closing = strchr(genre.str, ')');
506 out->str = strdup(closing + 1);
507 out->len = genre.len - (closing + 1 - genre.str);
518 _parse_id3v2_frame(struct id3v2_frame_header *fh, const char *frame_data, struct id3_info *info, lms_charset_conv_t **cs_convs)
520 lms_charset_conv_t *cs_conv = NULL;
521 unsigned int text_encoding, frame_size;
522 static const int artist_priorities[] = { 3, 4, 2, 1 };
525 fprintf(stderr, "frame id = %.4s frame size = %d text encoding = %d\n", fh->frame_id, fh->frame_size, frame_data[0]);
534 text_encoding = frame_data[0];
536 /* skip first byte - text encoding */
538 frame_size = fh->frame_size - 1;
540 if (text_encoding >= 0 && text_encoding < ID3_NUM_ENCODINGS) {
541 if (text_encoding == ID3_ENCODING_UTF16) {
542 if (memcmp(frame_data, "\xfe\xff", 2) == 0)
543 text_encoding = ID3_ENCODING_UTF16BE;
545 text_encoding = ID3_ENCODING_UTF16LE;
549 cs_conv = cs_convs[text_encoding];
552 /* ID3v2.2 used 3 bytes for the frame id, so let's check it */
553 if (memcmp(fh->frame_id, "TIT2", 4) == 0 ||
554 memcmp(fh->frame_id, "TT2", 3) == 0)
555 _get_id3v2_frame_info(frame_data, frame_size, &info->title, cs_conv);
556 else if (memcmp(fh->frame_id, "TP", 2) == 0) {
558 struct lms_string_size artist = {0};
560 if (memcmp(fh->frame_id, "TPE", 3) == 0) {
561 /* this check shouldn't be needed, but let's make sure */
562 if (fh->frame_id[3] >= '1' && fh->frame_id[3] <= '4')
563 index = fh->frame_id[3] - '1';
566 /* ignore TPA, TPB */
567 if (fh->frame_id[2] >= '1' && fh->frame_id[2] <= '4')
568 index = fh->frame_id[2] - '1';
572 artist_priorities[index] > info->cur_artist_priority) {
573 info->cur_artist_priority = artist_priorities[index];
574 _get_id3v2_frame_info(frame_data, frame_size, &artist, cs_conv);
575 lms_string_size_strip_and_free(&artist);
577 if (info->artist.str)
578 free(info->artist.str);
579 info->artist = artist;
584 else if (memcmp(fh->frame_id, "TAL", 3) == 0)
585 _get_id3v2_frame_info(frame_data, frame_size, &info->album, cs_conv);
587 else if (memcmp(fh->frame_id, "TCO", 3) == 0)
588 _get_id3v2_genre(frame_data, frame_size, &info->genre, cs_conv);
589 else if (memcmp(fh->frame_id, "TRCK", 4) == 0 ||
590 memcmp(fh->frame_id, "TRK", 3) == 0) {
591 struct lms_string_size trackno = {0};
592 _get_id3v2_frame_info(frame_data, frame_size, &trackno, cs_conv);
593 info->trackno = atoi(trackno.str);
599 _parse_id3v2(int fd, long id3v2_offset, struct id3_info *info, lms_charset_conv_t **cs_convs)
601 char header_data[10], frame_header_data[10];
602 unsigned int tag_size, major_version, frame_data_pos, frame_data_length, frame_header_size;
603 int extended_header, footer_present;
604 struct id3v2_frame_header fh;
607 lseek(fd, id3v2_offset, SEEK_SET);
610 if (read(fd, header_data, ID3V2_HEADER_SIZE) != ID3V2_HEADER_SIZE)
613 tag_size = _to_uint(header_data + 6, 4);
618 major_version = header_data[3];
621 frame_data_length = tag_size;
623 /* check for extended header */
624 extended_header = header_data[5] & 0x20; /* bit 6 */
625 if (extended_header) {
626 /* skip extended header */
627 unsigned int extended_header_size;
628 char extended_header_data[4];
630 if (read(fd, extended_header_data, 4) != 4)
632 extended_header_size = _to_uint(extended_header_data, 4);
633 lseek(fd, extended_header_size - 4, SEEK_CUR);
634 frame_data_pos += extended_header_size;
635 frame_data_length -= extended_header_size;
638 footer_present = header_data[5] & 0x8; /* bit 4 */
639 if (footer_present && frame_data_length > ID3V2_FOOTER_SIZE)
640 frame_data_length -= ID3V2_FOOTER_SIZE;
642 frame_header_size = _get_id3v2_frame_header_size(major_version);
643 while (frame_data_pos < frame_data_length - frame_header_size) {
644 nread = read(fd, frame_header_data, frame_header_size);
648 if (nread != frame_header_size)
651 if (frame_header_data[0] == 0)
654 _parse_id3v2_frame_header(frame_header_data, major_version, &fh);
658 if (!fh.compression &&
659 fh.frame_id[0] == 'T' &&
660 memcmp(fh.frame_id, "TXXX", 4) != 0) {
663 if (fh.data_length_indicator)
664 lseek(fd, 4, SEEK_CUR);
666 frame_data = malloc(sizeof(char) * fh.frame_size);
667 if (read(fd, frame_data, fh.frame_size) != fh.frame_size) {
672 _parse_id3v2_frame(&fh, frame_data, info, cs_convs);
676 if (fh.data_length_indicator)
677 lseek(fd, fh.frame_size + 4, SEEK_CUR);
679 lseek(fd, fh.frame_size, SEEK_CUR);
682 frame_data_pos += fh.frame_size + frame_header_size;
689 _parse_id3v1(int fd, struct id3_info *info, lms_charset_conv_t *cs_conv)
691 struct id3v1_tag tag;
692 if (read(fd, &tag, sizeof(struct id3v1_tag)) == -1)
695 info->title.str = strndup(tag.title, 30);
696 info->title.len = strlen(info->title.str);
697 lms_charset_conv(cs_conv, &info->title.str, &info->title.len);
698 info->artist.str = strndup(tag.artist, 30);
699 info->artist.len = strlen(info->artist.str);
700 lms_charset_conv(cs_conv, &info->artist.str, &info->artist.len);
701 info->album.str = strndup(tag.album, 30);
702 info->album.len = strlen(info->album.str);
703 lms_charset_conv(cs_conv, &info->album.str, &info->album.len);
704 _get_id3v1_genre(tag.genre, &info->genre);
705 if (tag.comments[28] == 0 && tag.comments[29] != 0)
706 info->trackno = (unsigned char) tag.comments[29];
712 _match(struct plugin *p, const char *path, int len, int base)
716 i = lms_which_extension(path, len, _exts, LMS_ARRAY_SIZE(_exts));
720 return (void*)(i + 1);
724 _parse(struct plugin *plugin, struct lms_context *ctxt, const struct lms_file_info *finfo, void *match)
726 struct id3_info info = {{0}, {0}, {0}, {0}, 0, -1};
727 struct lms_audio_info audio_info = {0, {0}, {0}, {0}, {0}, 0, 0, 0};
731 fd = open(finfo->path, O_RDONLY);
737 id3v2_offset = _find_id3v2(fd);
738 if (id3v2_offset >= 0) {
740 fprintf(stderr, "id3v2 tag found in file %s with offset %ld\n",
741 finfo->path, id3v2_offset);
743 if (_parse_id3v2(fd, id3v2_offset, &info, plugin->cs_convs) != 0) {
751 fprintf(stderr, "id3v2 tag not found in file %s. trying id3v1\n", finfo->path);
753 /* check for id3v1 tag */
754 if (lseek(fd, -128, SEEK_END) == -1) {
759 if (read(fd, &tag, 3) == -1) {
764 if (memcmp(tag, "TAG", 3) == 0) {
766 fprintf(stderr, "id3v1 tag found in file %s\n", finfo->path);
768 if (_parse_id3v1(fd, &info, ctxt->cs_conv) != 0) {
775 lms_string_size_strip_and_free(&info.title);
776 lms_string_size_strip_and_free(&info.artist);
777 lms_string_size_strip_and_free(&info.album);
778 lms_string_size_strip_and_free(&info.genre);
780 if (!info.title.str) {
782 ext_idx = ((int)match) - 1;
783 info.title.len = finfo->path_len - finfo->base - _exts[ext_idx].len;
784 info.title.str = malloc((info.title.len + 1) * sizeof(char));
785 memcpy(info.title.str, finfo->path + finfo->base, info.title.len);
786 info.title.str[info.title.len] = '\0';
787 lms_charset_conv(ctxt->cs_conv, &info.title.str, &info.title.len);
791 fprintf(stderr, "file %s info\n", finfo->path);
792 fprintf(stderr, "\ttitle='%s'\n", info.title.str);
793 fprintf(stderr, "\tartist='%s'\n", info.artist.str);
794 fprintf(stderr, "\talbum='%s'\n", info.album.str);
795 fprintf(stderr, "\tgenre='%s'\n", info.genre.str);
796 fprintf(stderr, "\ttrack number='%d'\n", info.trackno);
799 audio_info.id = finfo->id;
800 audio_info.title = info.title;
801 audio_info.artist = info.artist;
802 audio_info.album = info.album;
803 audio_info.genre = info.genre;
804 audio_info.trackno = info.trackno;
805 r = lms_db_audio_add(plugin->audio_db, &audio_info);
808 posix_fadvise(fd, 0, 0, POSIX_FADV_DONTNEED);
812 free(info.title.str);
814 free(info.artist.str);
816 free(info.album.str);
818 free(info.genre.str);
824 _setup(struct plugin *plugin, struct lms_context *ctxt)
827 const char *id3_encodings[ID3_NUM_ENCODINGS] = {
835 plugin->audio_db = lms_db_audio_new(ctxt->db);
836 if (!plugin->audio_db)
839 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
840 /* do not create charset conv for UTF-8 encoding */
841 if (!id3_encodings[i]) {
842 plugin->cs_convs[i] = NULL;
845 plugin->cs_convs[i] = lms_charset_conv_new_full(0, 0);
846 if (!plugin->cs_convs[i])
848 lms_charset_conv_add(plugin->cs_convs[i], id3_encodings[i]);
855 _start(struct plugin *plugin, struct lms_context *ctxt)
857 return lms_db_audio_start(plugin->audio_db);
861 _finish(struct plugin *plugin, struct lms_context *ctxt)
865 if (plugin->audio_db)
866 lms_db_audio_free(plugin->audio_db);
868 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
869 if (plugin->cs_convs[i])
870 lms_charset_conv_free(plugin->cs_convs[i]);
877 _close(struct plugin *plugin)
883 API struct lms_plugin *
884 lms_plugin_open(void)
886 struct plugin *plugin;
888 plugin = (struct plugin *)malloc(sizeof(*plugin));
889 plugin->plugin.name = _name;
890 plugin->plugin.match = (lms_plugin_match_fn_t)_match;
891 plugin->plugin.parse = (lms_plugin_parse_fn_t)_parse;
892 plugin->plugin.close = (lms_plugin_close_fn_t)_close;
893 plugin->plugin.setup = (lms_plugin_setup_fn_t)_setup;
894 plugin->plugin.start = (lms_plugin_start_fn_t)_start;
895 plugin->plugin.finish = (lms_plugin_finish_fn_t)_finish;
897 return (struct lms_plugin *)plugin;