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 search
150 * pattern and we want to see if the next one starts with the remainder of
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
159 /* (1) previous partial match */
160 if (previous_partial_synch_match && _is_id3v2_second_synch_byte(buffer[0]))
163 if (previous_partial_match >= 0 && previous_partial_match < buffer_size) {
164 const int pattern_offset = buffer_size - previous_partial_match;
166 if (memcmp(buffer, pattern + pattern_offset, 3 - pattern_offset) == 0)
167 return buffer_offset - buffer_size + previous_partial_match;
170 /* (2) pattern contained in current buffer */
172 while ((p = memchr(p, 'I', buffer_size - (p - buffer)))) {
173 if (buffer_size - (p - buffer) < 3)
176 if (memcmp(p, pattern, 3) == 0)
177 return buffer_offset + (p - buffer);
182 p = memchr(buffer, 255, buffer_size);
184 first_synch_byte = p - buffer;
186 first_synch_byte = -1;
188 /* Here we have to loop because there could be several of the first
189 * (11111111) byte, and we want to check all such instances until we find
190 * a full match (11111111 111) or hit the end of the buffer. */
191 while (first_synch_byte >= 0) {
192 /* if this *is not* at the end of the buffer */
193 if (first_synch_byte < buffer_size - 1) {
194 if(_is_id3v2_second_synch_byte(buffer[first_synch_byte + 1]))
195 /* We've found the frame synch pattern. */
198 /* We found 11111111 at the end of the current buffer indicating a
199 * partial match of the synch pattern. The find() below should
200 * return -1 and break out of the loop. */
201 previous_partial_synch_match = 1;
204 /* Check in the rest of the buffer. */
205 p = memchr(p + 1, 255, buffer_size - (p + 1 - buffer));
207 first_synch_byte = p - buffer;
209 first_synch_byte = -1;
212 /* (3) partial match */
213 if (buffer[nread - 1] == pattern[1])
214 previous_partial_match = nread - 1;
215 else if (memcmp(&buffer[nread - 2], pattern, 2) == 0)
216 previous_partial_match = nread - 2;
217 buffer_offset += buffer_size;
219 if ((nread = read(fd, &buffer, sizeof(buffer))) == -1)
227 _get_id3v2_frame_header_size(unsigned int version)
242 _parse_id3v2_frame_header(char *data, unsigned int version, struct id3v2_frame_header *fh)
248 memcpy(fh->frame_id, data, 3);
250 fh->frame_size = _to_uint(data + 3, 3);
252 fh->data_length_indicator = 0;
255 memcpy(fh->frame_id, data, 4);
256 fh->frame_size = _to_uint(data + 4, 4);
257 fh->compression = data[9] & 0x40;
258 fh->data_length_indicator = 0;
262 memcpy(fh->frame_id, data, 4);
263 fh->frame_size = _to_uint(data + 4, 4);
264 fh->compression = data[9] & 0x4;
265 fh->data_length_indicator = data[9] & 0x1;
271 _get_id3v2_frame_info(const char *frame_data, unsigned int frame_size, struct lms_string_size *s, lms_charset_conv_t *cs_conv)
273 if (frame_size > s->len)
274 s->str = realloc(s->str, sizeof(char) * (frame_size + 1));
275 memcpy(s->str, frame_data, frame_size);
276 s->str[frame_size] = '\0';
279 lms_charset_conv(cs_conv, &s->str, &s->len);
283 _get_id3v1_genre(unsigned int genre, struct lms_string_size *out)
285 if (genre < ID3V1_NUM_GENRES) {
286 unsigned int size, base;
288 base = id3v1_genres_offsets[genre];
289 size = id3v1_genres_offsets[genre + 1] - base;
290 out->str = malloc(size);
292 memcpy(out->str, id3v1_genres_mem + base, size);
300 _parse_id3v1_genre(const char *str_genre, struct lms_string_size *out)
302 return _get_id3v1_genre(atoi(str_genre), out);
306 _get_id3v2_genre(const char *frame_data, unsigned int frame_size, struct lms_string_size *out, lms_charset_conv_t *cs_conv)
309 struct lms_string_size genre = {0};
311 _get_id3v2_frame_info(frame_data, frame_size, &genre, cs_conv);
319 is_number = (genre.len != 0 && genre.str[0] != '(');
321 for (i = 0; i < genre.len; ++i) {
322 if (!isdigit(genre.str[i])) {
329 if (is_number && _parse_id3v1_genre(genre.str, out) == 0) {
330 /* id3v1 genre found */
335 /* ID3v2.3 "content type" can contain a ID3v1 genre number in parenthesis at
336 * the beginning of the field. If this is all that the field contains, do a
337 * translation from that number to the name and return that. If there is a
338 * string folloing the ID3v1 genre number, that is considered to be
339 * authoritative and we return that instead. Or finally, the field may
340 * simply be free text, in which case we just return the value. */
342 if (genre.len > 1 && genre.str[0] == '(') {
343 char *closing = NULL;
345 if (genre.str[genre.len - 1] == ')') {
346 closing = strchr(genre.str, ')');
347 if (closing == genre.str + genre.len - 1) {
348 /* ) is the last character and only appears once in the string */
349 /* get the id3v1 genre enclosed by parentheses */
350 if (_parse_id3v1_genre(genre.str + 1, out) == 0) {
357 /* get the string followed by the id3v1 genre */
359 closing = strchr(genre.str, ')');
362 out->str = strdup(closing + 1);
363 out->len = genre.len - (closing + 1 - genre.str);
374 _parse_id3v2_frame(struct id3v2_frame_header *fh, const char *frame_data, struct id3_info *info, lms_charset_conv_t **cs_convs)
376 lms_charset_conv_t *cs_conv = NULL;
377 unsigned int text_encoding, frame_size;
378 static const int artist_priorities[] = { 3, 4, 2, 1 };
381 fprintf(stderr, "frame id = %.4s frame size = %d text encoding = %d\n", fh->frame_id, fh->frame_size, frame_data[0]);
390 text_encoding = frame_data[0];
392 /* skip first byte - text encoding */
394 frame_size = fh->frame_size - 1;
396 if (text_encoding >= 0 && text_encoding < ID3_NUM_ENCODINGS) {
397 if (text_encoding == ID3_ENCODING_UTF16) {
398 if (memcmp(frame_data, "\xfe\xff", 2) == 0)
399 text_encoding = ID3_ENCODING_UTF16BE;
401 text_encoding = ID3_ENCODING_UTF16LE;
405 cs_conv = cs_convs[text_encoding];
408 /* ID3v2.2 used 3 bytes for the frame id, so let's check it */
409 if (memcmp(fh->frame_id, "TIT2", 4) == 0 ||
410 memcmp(fh->frame_id, "TT2", 3) == 0)
411 _get_id3v2_frame_info(frame_data, frame_size, &info->title, cs_conv);
412 else if (memcmp(fh->frame_id, "TP", 2) == 0) {
414 struct lms_string_size artist = {0};
416 if (memcmp(fh->frame_id, "TPE", 3) == 0) {
417 /* this check shouldn't be needed, but let's make sure */
418 if (fh->frame_id[3] >= '1' && fh->frame_id[3] <= '4')
419 index = fh->frame_id[3] - '1';
422 /* ignore TPA, TPB */
423 if (fh->frame_id[2] >= '1' && fh->frame_id[2] <= '4')
424 index = fh->frame_id[2] - '1';
428 artist_priorities[index] > info->cur_artist_priority) {
429 info->cur_artist_priority = artist_priorities[index];
430 _get_id3v2_frame_info(frame_data, frame_size, &artist, cs_conv);
431 lms_string_size_strip_and_free(&artist);
433 if (info->artist.str)
434 free(info->artist.str);
435 info->artist = artist;
440 else if (memcmp(fh->frame_id, "TAL", 3) == 0)
441 _get_id3v2_frame_info(frame_data, frame_size, &info->album, cs_conv);
443 else if (memcmp(fh->frame_id, "TCO", 3) == 0)
444 _get_id3v2_genre(frame_data, frame_size, &info->genre, cs_conv);
445 else if (memcmp(fh->frame_id, "TRCK", 4) == 0 ||
446 memcmp(fh->frame_id, "TRK", 3) == 0) {
447 struct lms_string_size trackno = {0};
448 _get_id3v2_frame_info(frame_data, frame_size, &trackno, cs_conv);
449 info->trackno = atoi(trackno.str);
455 _parse_id3v2(int fd, long id3v2_offset, struct id3_info *info, lms_charset_conv_t **cs_convs)
457 char header_data[10], frame_header_data[10];
458 unsigned int tag_size, major_version, frame_data_pos, frame_data_length, frame_header_size;
459 int extended_header, footer_present;
460 struct id3v2_frame_header fh;
463 lseek(fd, id3v2_offset, SEEK_SET);
466 if (read(fd, header_data, ID3V2_HEADER_SIZE) != ID3V2_HEADER_SIZE)
469 tag_size = _to_uint(header_data + 6, 4);
474 major_version = header_data[3];
477 frame_data_length = tag_size;
479 /* check for extended header */
480 extended_header = header_data[5] & 0x20; /* bit 6 */
481 if (extended_header) {
482 /* skip extended header */
483 unsigned int extended_header_size;
484 char extended_header_data[4];
486 if (read(fd, extended_header_data, 4) != 4)
488 extended_header_size = _to_uint(extended_header_data, 4);
489 lseek(fd, extended_header_size - 4, SEEK_CUR);
490 frame_data_pos += extended_header_size;
491 frame_data_length -= extended_header_size;
494 footer_present = header_data[5] & 0x8; /* bit 4 */
495 if (footer_present && frame_data_length > ID3V2_FOOTER_SIZE)
496 frame_data_length -= ID3V2_FOOTER_SIZE;
498 frame_header_size = _get_id3v2_frame_header_size(major_version);
499 while (frame_data_pos < frame_data_length - frame_header_size) {
500 nread = read(fd, frame_header_data, frame_header_size);
504 if (nread != frame_header_size)
507 if (frame_header_data[0] == 0)
510 _parse_id3v2_frame_header(frame_header_data, major_version, &fh);
514 if (!fh.compression &&
515 fh.frame_id[0] == 'T' &&
516 memcmp(fh.frame_id, "TXXX", 4) != 0) {
519 if (fh.data_length_indicator)
520 lseek(fd, 4, SEEK_CUR);
522 frame_data = malloc(sizeof(char) * fh.frame_size);
523 if (read(fd, frame_data, fh.frame_size) != fh.frame_size) {
528 _parse_id3v2_frame(&fh, frame_data, info, cs_convs);
532 if (fh.data_length_indicator)
533 lseek(fd, fh.frame_size + 4, SEEK_CUR);
535 lseek(fd, fh.frame_size, SEEK_CUR);
538 frame_data_pos += fh.frame_size + frame_header_size;
545 _parse_id3v1(int fd, struct id3_info *info, lms_charset_conv_t *cs_conv)
547 struct id3v1_tag tag;
548 if (read(fd, &tag, sizeof(struct id3v1_tag)) == -1)
551 info->title.str = strndup(tag.title, 30);
552 info->title.len = strlen(info->title.str);
553 lms_charset_conv(cs_conv, &info->title.str, &info->title.len);
554 info->artist.str = strndup(tag.artist, 30);
555 info->artist.len = strlen(info->artist.str);
556 lms_charset_conv(cs_conv, &info->artist.str, &info->artist.len);
557 info->album.str = strndup(tag.album, 30);
558 info->album.len = strlen(info->album.str);
559 lms_charset_conv(cs_conv, &info->album.str, &info->album.len);
560 _get_id3v1_genre(tag.genre, &info->genre);
561 if (tag.comments[28] == 0 && tag.comments[29] != 0)
562 info->trackno = (unsigned char) tag.comments[29];
568 _match(struct plugin *p, const char *path, int len, int base)
572 i = lms_which_extension(path, len, _exts, LMS_ARRAY_SIZE(_exts));
576 return (void*)(i + 1);
580 _parse(struct plugin *plugin, struct lms_context *ctxt, const struct lms_file_info *finfo, void *match)
582 struct id3_info info = {{0}, {0}, {0}, {0}, 0, -1};
583 struct lms_audio_info audio_info = {0, {0}, {0}, {0}, {0}, 0, 0, 0};
587 fd = open(finfo->path, O_RDONLY);
593 id3v2_offset = _find_id3v2(fd);
594 if (id3v2_offset >= 0) {
596 fprintf(stderr, "id3v2 tag found in file %s with offset %ld\n",
597 finfo->path, id3v2_offset);
599 if (_parse_id3v2(fd, id3v2_offset, &info, plugin->cs_convs) != 0) {
607 fprintf(stderr, "id3v2 tag not found in file %s. trying id3v1\n", finfo->path);
609 /* check for id3v1 tag */
610 if (lseek(fd, -128, SEEK_END) == -1) {
615 if (read(fd, &tag, 3) == -1) {
620 if (memcmp(tag, "TAG", 3) == 0) {
622 fprintf(stderr, "id3v1 tag found in file %s\n", finfo->path);
624 if (_parse_id3v1(fd, &info, ctxt->cs_conv) != 0) {
631 lms_string_size_strip_and_free(&info.title);
632 lms_string_size_strip_and_free(&info.artist);
633 lms_string_size_strip_and_free(&info.album);
634 lms_string_size_strip_and_free(&info.genre);
636 if (!info.title.str) {
638 ext_idx = ((int)match) - 1;
639 info.title.len = finfo->path_len - finfo->base - _exts[ext_idx].len;
640 info.title.str = malloc((info.title.len + 1) * sizeof(char));
641 memcpy(info.title.str, finfo->path + finfo->base, info.title.len);
642 info.title.str[info.title.len] = '\0';
643 lms_charset_conv(ctxt->cs_conv, &info.title.str, &info.title.len);
647 fprintf(stderr, "file %s info\n", finfo->path);
648 fprintf(stderr, "\ttitle='%s'\n", info.title.str);
649 fprintf(stderr, "\tartist='%s'\n", info.artist.str);
650 fprintf(stderr, "\talbum='%s'\n", info.album.str);
651 fprintf(stderr, "\tgenre='%s'\n", info.genre.str);
652 fprintf(stderr, "\ttrack number='%d'\n", info.trackno);
655 audio_info.id = finfo->id;
656 audio_info.title = info.title;
657 audio_info.artist = info.artist;
658 audio_info.album = info.album;
659 audio_info.genre = info.genre;
660 audio_info.trackno = info.trackno;
661 r = lms_db_audio_add(plugin->audio_db, &audio_info);
664 posix_fadvise(fd, 0, 0, POSIX_FADV_DONTNEED);
668 free(info.title.str);
670 free(info.artist.str);
672 free(info.album.str);
674 free(info.genre.str);
680 _setup(struct plugin *plugin, struct lms_context *ctxt)
683 const char *id3_encodings[ID3_NUM_ENCODINGS] = {
691 plugin->audio_db = lms_db_audio_new(ctxt->db);
692 if (!plugin->audio_db)
695 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
696 /* do not create charset conv for UTF-8 encoding */
697 if (!id3_encodings[i]) {
698 plugin->cs_convs[i] = NULL;
701 plugin->cs_convs[i] = lms_charset_conv_new_full(0, 0);
702 if (!plugin->cs_convs[i])
704 lms_charset_conv_add(plugin->cs_convs[i], id3_encodings[i]);
711 _start(struct plugin *plugin, struct lms_context *ctxt)
713 return lms_db_audio_start(plugin->audio_db);
717 _finish(struct plugin *plugin, struct lms_context *ctxt)
721 if (plugin->audio_db)
722 lms_db_audio_free(plugin->audio_db);
724 for (i = 0; i < ID3_NUM_ENCODINGS; ++i) {
725 if (plugin->cs_convs[i])
726 lms_charset_conv_free(plugin->cs_convs[i]);
733 _close(struct plugin *plugin)
739 API struct lms_plugin *
740 lms_plugin_open(void)
742 struct plugin *plugin;
744 plugin = (struct plugin *)malloc(sizeof(*plugin));
745 plugin->plugin.name = _name;
746 plugin->plugin.match = (lms_plugin_match_fn_t)_match;
747 plugin->plugin.parse = (lms_plugin_parse_fn_t)_parse;
748 plugin->plugin.close = (lms_plugin_close_fn_t)_close;
749 plugin->plugin.setup = (lms_plugin_setup_fn_t)_setup;
750 plugin->plugin.start = (lms_plugin_start_fn_t)_start;
751 plugin->plugin.finish = (lms_plugin_finish_fn_t)_finish;
753 return (struct lms_plugin *)plugin;