add plugins categories file to tarball. Should fix FS#7567.
[Rockbox.git] / firmware / id3.c
blob851aa83a5a20cf239f882170b1afbb76f886181f
1 /***************************************************************************
2 * __________ __ ___.
3 * Open \______ \ ____ ____ | | _\_ |__ _______ ___
4 * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
5 * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
6 * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
7 * \/ \/ \/ \/ \/
8 * $Id$
10 * Copyright (C) 2002 by Daniel Stenberg
12 * All files in this archive are subject to the GNU General Public License.
13 * See the file COPYING in the source tree root for full license agreement.
15 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
16 * KIND, either express or implied.
18 ****************************************************************************/
20 * Parts of this code has been stolen from the Ample project and was written
21 * by David H�deman. It has since been extended and enhanced pretty much by
22 * all sorts of friendly Rockbox people.
26 /* tagResolver and associated code copyright 2003 Thomas Paul Diffenbach
29 #include <stdio.h>
30 #include <stdlib.h>
31 #include <string.h>
32 #include <errno.h>
33 #include <stdbool.h>
34 #include <stddef.h>
35 #include <ctype.h>
36 #include "config.h"
37 #include "file.h"
38 #include "debug.h"
39 #include "atoi.h"
41 #include "id3.h"
42 #include "mp3data.h"
43 #include "system.h"
44 #include "replaygain.h"
45 #include "rbunicode.h"
47 /** Database of audio formats **/
48 const struct afmt_entry audio_formats[AFMT_NUM_CODECS] =
50 /* Unknown file format */
51 [AFMT_UNKNOWN] =
52 AFMT_ENTRY("???", NULL, NULL, NULL ),
54 /* MPEG Audio layer 1 */
55 [AFMT_MPA_L1] =
56 AFMT_ENTRY("MP1", "mpa", NULL, "mp1\0" ),
57 /* MPEG Audio layer 2 */
58 [AFMT_MPA_L2] =
59 AFMT_ENTRY("MP2", "mpa", NULL, "mpa\0mp2\0" ),
60 /* MPEG Audio layer 3 */
61 [AFMT_MPA_L3] =
62 AFMT_ENTRY("MP3", "mpa", "mp3_enc", "mp3\0" ),
64 #if CONFIG_CODEC == SWCODEC
65 /* Audio Interchange File Format */
66 [AFMT_AIFF] =
67 AFMT_ENTRY("AIFF", "aiff", "aiff_enc", "aiff\0aif\0"),
68 /* Uncompressed PCM in a WAV file */
69 [AFMT_PCM_WAV] =
70 AFMT_ENTRY("WAV", "wav", "wav_enc", "wav\0" ),
71 /* Ogg Vorbis */
72 [AFMT_OGG_VORBIS] =
73 AFMT_ENTRY("Ogg", "vorbis", NULL, "ogg\0" ),
74 /* FLAC */
75 [AFMT_FLAC] =
76 AFMT_ENTRY("FLAC", "flac", NULL, "flac\0" ),
77 /* Musepack */
78 [AFMT_MPC] =
79 AFMT_ENTRY("MPC", "mpc", NULL, "mpc\0" ),
80 /* A/52 (aka AC3) audio */
81 [AFMT_A52] =
82 AFMT_ENTRY("AC3", "a52", NULL, "a52\0ac3\0" ),
83 /* WavPack */
84 [AFMT_WAVPACK] =
85 AFMT_ENTRY("WV", "wavpack", "wavpack_enc", "wv\0" ),
86 /* Apple Lossless Audio Codec */
87 [AFMT_ALAC] =
88 AFMT_ENTRY("ALAC", "alac", NULL, "m4a\0m4b\0" ),
89 /* Advanced Audio Coding in M4A container */
90 [AFMT_AAC] =
91 AFMT_ENTRY("AAC", "aac", NULL, "mp4\0" ),
92 /* Shorten */
93 [AFMT_SHN] =
94 AFMT_ENTRY("SHN", "shorten", NULL, "shn\0" ),
95 /* SID File Format */
96 [AFMT_SID] =
97 AFMT_ENTRY("SID", "sid", NULL, "sid\0" ),
98 /* ADX File Format */
99 [AFMT_ADX] =
100 AFMT_ENTRY("ADX", "adx", NULL, "adx\0" ),
101 /* NESM (NES Sound Format) */
102 [AFMT_NSF] =
103 AFMT_ENTRY("NSF", "nsf", NULL, "nsf\0nsfe\0" ),
104 /* Speex File Format */
105 [AFMT_SPEEX] =
106 AFMT_ENTRY("Speex","speex", NULL, "spx\0" ),
107 /* SPC700 Save State */
108 [AFMT_SPC] =
109 AFMT_ENTRY("SPC", "spc", NULL, "spc\0" ),
110 /* APE (Monkey's Audio) */
111 [AFMT_APE] =
112 AFMT_ENTRY("APE", "ape", NULL, "ape\0mac\0" ),
113 /* WMA (WMAV1/V2 in ASF) */
114 [AFMT_WMA] =
115 AFMT_ENTRY("WMA", "wma", NULL, "wma\0wmv\0asf\0" ),
116 #endif
119 #if CONFIG_CODEC == SWCODEC && defined (HAVE_RECORDING)
120 /* get REC_FORMAT_* corresponding AFMT_* */
121 const int rec_format_afmt[REC_NUM_FORMATS] =
123 /* give AFMT_UNKNOWN by default */
124 [0 ... REC_NUM_FORMATS-1] = AFMT_UNKNOWN,
125 /* add new entries below this line */
126 [REC_FORMAT_AIFF] = AFMT_AIFF,
127 [REC_FORMAT_MPA_L3] = AFMT_MPA_L3,
128 [REC_FORMAT_WAVPACK] = AFMT_WAVPACK,
129 [REC_FORMAT_PCM_WAV] = AFMT_PCM_WAV,
132 /* get AFMT_* corresponding REC_FORMAT_* */
133 const int afmt_rec_format[AFMT_NUM_CODECS] =
135 /* give -1 by default */
136 [0 ... AFMT_NUM_CODECS-1] = -1,
137 /* add new entries below this line */
138 [AFMT_AIFF] = REC_FORMAT_AIFF,
139 [AFMT_MPA_L3] = REC_FORMAT_MPA_L3,
140 [AFMT_WAVPACK] = REC_FORMAT_WAVPACK,
141 [AFMT_PCM_WAV] = REC_FORMAT_PCM_WAV,
143 #endif /* CONFIG_CODEC == SWCODEC && defined (HAVE_RECORDING) */
144 /****/
146 unsigned long unsync(unsigned long b0,
147 unsigned long b1,
148 unsigned long b2,
149 unsigned long b3)
151 return (((long)(b0 & 0x7F) << (3*7)) |
152 ((long)(b1 & 0x7F) << (2*7)) |
153 ((long)(b2 & 0x7F) << (1*7)) |
154 ((long)(b3 & 0x7F) << (0*7)));
157 static const char* const genres[] = {
158 "Blues", "Classic Rock", "Country", "Dance", "Disco", "Funk", "Grunge",
159 "Hip-Hop", "Jazz", "Metal", "New Age", "Oldies", "Other", "Pop", "R&B",
160 "Rap", "Reggae", "Rock", "Techno", "Industrial", "Alternative", "Ska",
161 "Death Metal", "Pranks", "Soundtrack", "Euro-Techno", "Ambient", "Trip-Hop",
162 "Vocal", "Jazz+Funk", "Fusion", "Trance", "Classical", "Instrumental",
163 "Acid", "House", "Game", "Sound Clip", "Gospel", "Noise", "AlternRock",
164 "Bass", "Soul", "Punk", "Space", "Meditative", "Instrumental Pop",
165 "Instrumental Rock", "Ethnic", "Gothic", "Darkwave", "Techno-Industrial",
166 "Electronic", "Pop-Folk", "Eurodance", "Dream", "Southern Rock", "Comedy",
167 "Cult", "Gangsta", "Top 40", "Christian Rap", "Pop/Funk", "Jungle",
168 "Native American", "Cabaret", "New Wave", "Psychadelic", "Rave",
169 "Showtunes", "Trailer", "Lo-Fi", "Tribal", "Acid Punk", "Acid Jazz",
170 "Polka", "Retro", "Musical", "Rock & Roll", "Hard Rock",
172 /* winamp extensions */
173 "Folk", "Folk-Rock", "National Folk", "Swing", "Fast Fusion", "Bebob",
174 "Latin", "Revival", "Celtic", "Bluegrass", "Avantgarde", "Gothic Rock",
175 "Progressive Rock", "Psychedelic Rock", "Symphonic Rock", "Slow Rock",
176 "Big Band", "Chorus", "Easy Listening", "Acoustic", "Humour", "Speech",
177 "Chanson", "Opera", "Chamber Music", "Sonata", "Symphony", "Booty Bass",
178 "Primus", "Porn Groove", "Satire", "Slow Jam", "Club", "Tango", "Samba",
179 "Folklore", "Ballad", "Power Ballad", "Rhythmic Soul", "Freestyle",
180 "Duet", "Punk Rock", "Drum Solo", "A capella", "Euro-House", "Dance Hall",
181 "Goa", "Drum & Bass", "Club-House", "Hardcore", "Terror", "Indie",
182 "BritPop", "Negerpunk", "Polsk Punk", "Beat", "Christian Gangsta Rap",
183 "Heavy Metal", "Black Metal", "Crossover", "Contemporary Christian",
184 "Christian Rock", "Merengue", "Salsa", "Thrash Metal", "Anime", "Jpop",
185 "Synthpop"
188 char* id3_get_num_genre(const unsigned int genre_num)
190 if (genre_num < sizeof(genres)/sizeof(char*))
191 return (char*)genres[genre_num];
192 return NULL;
195 char* id3_get_codec(const struct mp3entry* id3)
197 if (id3->codectype < AFMT_NUM_CODECS) {
198 return (char*)audio_formats[id3->codectype].label;
199 } else {
200 return NULL;
205 HOW TO ADD ADDITIONAL ID3 VERSION 2 TAGS
206 Code and comments by Thomas Paul Diffenbach
208 To add another ID3v2 Tag, do the following:
209 1. add a char* named for the tag to struct mp3entry in id3.h,
210 (I (tpd) prefer to use char* rather than ints, even for what seems like
211 numerical values, for cases where a number won't do, e.g.,
212 YEAR: "circa 1765", "1790/1977" (composed/performed), "28 Feb 1969"
213 TRACK: "1/12", "1 of 12", GENRE: "Freeform genre name"
214 Text is more flexible, and as the main use of id3 data is to
215 display it, converting it to an int just means reconverting to
216 display it, at a runtime cost.)
218 2. If any special processing beyond copying the tag value from the Id3
219 block to the struct mp3entry is rrequired (such as converting to an
220 int), write a function to perform this special processing.
222 This function's prototype must match that of
223 typedef tagPostProcessFunc, that is it must be:
224 int func( struct mp3entry*, char* tag, int bufferpos )
225 the first argument is a pointer to the current mp3entry structure the
226 second argument is a pointer to the null terminated string value of the
227 tag found the third argument is the offset of the next free byte in the
228 mp3entry's buffer your function should return the corrected offset; if
229 you don't lengthen or shorten the tag string, you can return the third
230 argument unchanged.
232 Unless you have a good reason no to, make the function static.
233 TO JUST COPY THE TAG NO SPECIAL PROCESSING FUNCTION IS NEEDED.
235 3. add one or more entries to the tagList array, using the format:
236 char* ID3 Tag symbolic name -- see the ID3 specification for these,
237 sizeof() that name minus 1,
238 offsetof( struct mp3entry, variable_name_in_struct_mp3entry ),
239 pointer to your special processing function or NULL
240 if you need no special processing
241 flag indicating if this tag is binary or textual
242 Many ID3 symbolic names come in more than one form. You can add both
243 forms, each referencing the same variable in struct mp3entry.
244 If both forms are present, the last found will be used.
245 Note that the offset can be zero, in which case no entry will be set
246 in the mp3entry struct; the frame is still read into the buffer and
247 the special processing function is called (several times, if there
248 are several frames with the same name).
250 4. Alternately, use the TAG_LIST_ENTRY macro with
251 ID3 tag symbolic name,
252 variable in struct mp3entry,
253 special processing function address
255 5. Add code to wps-display.c function get_tag to assign a printf-like
256 format specifier for the tag */
258 /* Structure for ID3 Tag extraction information */
259 struct tag_resolver {
260 const char* tag;
261 int tag_length;
262 size_t offset;
263 int (*ppFunc)(struct mp3entry*, char* tag, int bufferpos);
264 bool binary;
267 static bool global_ff_found;
269 static int unsynchronize(char* tag, int len, bool *ff_found)
271 int i;
272 unsigned char c;
273 unsigned char *rp, *wp;
275 wp = rp = (unsigned char *)tag;
277 rp = (unsigned char *)tag;
278 for(i = 0;i < len;i++) {
279 /* Read the next byte and write it back, but don't increment the
280 write pointer */
281 c = *rp++;
282 *wp = c;
283 if(*ff_found) {
284 /* Increment the write pointer if it isn't an unsynch pattern */
285 if(c != 0)
286 wp++;
287 *ff_found = false;
288 } else {
289 if(c == 0xff)
290 *ff_found = true;
291 wp++;
294 return (long)wp - (long)tag;
297 static int unsynchronize_frame(char* tag, int len)
299 bool ff_found = false;
301 return unsynchronize(tag, len, &ff_found);
304 static int read_unsynched(int fd, void *buf, int len)
306 int i;
307 int rc;
308 int remaining = len;
309 char *wp;
310 char *rp;
312 wp = buf;
314 while(remaining) {
315 rp = wp;
316 rc = read(fd, rp, remaining);
317 if(rc <= 0)
318 return rc;
320 i = unsynchronize(wp, remaining, &global_ff_found);
321 remaining -= i;
322 wp += i;
325 return len;
328 static int skip_unsynched(int fd, int len)
330 int rc;
331 int remaining = len;
332 int rlen;
333 char buf[32];
335 while(remaining) {
336 rlen = MIN(sizeof(buf), (unsigned int)remaining);
337 rc = read(fd, buf, rlen);
338 if(rc <= 0)
339 return rc;
341 remaining -= unsynchronize(buf, rlen, &global_ff_found);
344 return len;
347 /* parse numeric value from string */
348 static int parsetracknum( struct mp3entry* entry, char* tag, int bufferpos )
350 entry->tracknum = atoi( tag );
351 return bufferpos;
354 /* parse numeric value from string */
355 static int parsediscnum( struct mp3entry* entry, char* tag, int bufferpos )
357 entry->discnum = atoi( tag );
358 return bufferpos;
361 /* parse numeric value from string */
362 static int parseyearnum( struct mp3entry* entry, char* tag, int bufferpos )
364 entry->year = atoi( tag );
365 return bufferpos;
368 /* parse numeric genre from string, version 2.2 and 2.3 */
369 static int parsegenre( struct mp3entry* entry, char* tag, int bufferpos )
371 if(entry->id3version >= ID3_VER_2_4) {
372 /* In version 2.4 and up, there are no parentheses, and the genre frame
373 is a list of strings, either numbers or text. */
375 /* Is it a number? */
376 if(isdigit(tag[0])) {
377 entry->genre_string = id3_get_num_genre(atoi( tag ));
378 return tag - entry->id3v2buf;
379 } else {
380 entry->genre_string = tag;
381 return bufferpos;
383 } else {
384 if( tag[0] == '(' && tag[1] != '(' ) {
385 entry->genre_string = id3_get_num_genre(atoi( tag + 1 ));
386 return tag - entry->id3v2buf;
388 else {
389 entry->genre_string = tag;
390 return bufferpos;
395 #if CONFIG_CODEC == SWCODEC
396 /* parse user defined text, looking for replaygain information. */
397 static int parseuser( struct mp3entry* entry, char* tag, int bufferpos )
399 char* value = NULL;
400 int desc_len = strlen(tag);
401 int value_len = 0;
403 /* Note: for ID3v2.4, parse_replaygain will not overwrite replaygain
404 values already parsed from RVA2 tags */
405 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos) {
406 /* At least part of the value was read, so we can safely try to
407 * parse it
409 value = tag + desc_len + 1;
410 value_len = parse_replaygain(tag, value, entry, tag,
411 bufferpos - (tag - entry->id3v2buf));
414 if (value_len) {
415 bufferpos = tag - entry->id3v2buf + value_len;
416 } else {
417 bufferpos = tag - entry->id3v2buf;
420 return bufferpos;
423 /* parse RVA2 binary data and convert to replaygain information. */
424 static int parserva2( struct mp3entry* entry, char* tag, int bufferpos )
426 char* value = NULL;
427 int desc_len = strlen(tag);
428 int value_len = 0;
430 /* Only parse RVA2 replaygain tags if tag version == 2.4 */
431 if (entry->id3version == ID3_VER_2_4 &&
432 (tag - entry->id3v2buf + desc_len + 2) < bufferpos) {
433 value = tag + desc_len + 1;
434 value_len = parse_replaygain_rva(tag, value, entry, tag,
435 bufferpos - (tag - entry->id3v2buf));
438 if (value_len) {
439 bufferpos = tag - entry->id3v2buf + value_len;
440 } else {
441 bufferpos = tag - entry->id3v2buf;
444 return bufferpos;
446 #endif
448 static const struct tag_resolver taglist[] = {
449 { "TPE1", 4, offsetof(struct mp3entry, artist), NULL, false },
450 { "TP1", 3, offsetof(struct mp3entry, artist), NULL, false },
451 { "TIT2", 4, offsetof(struct mp3entry, title), NULL, false },
452 { "TT2", 3, offsetof(struct mp3entry, title), NULL, false },
453 { "TALB", 4, offsetof(struct mp3entry, album), NULL, false },
454 { "TAL", 3, offsetof(struct mp3entry, album), NULL, false },
455 { "TRK", 3, offsetof(struct mp3entry, track_string), &parsetracknum, false },
456 { "TPOS", 4, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
457 { "TRCK", 4, offsetof(struct mp3entry, track_string), &parsetracknum, false },
458 { "TDRC", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
459 { "TYER", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
460 { "TYE", 3, offsetof(struct mp3entry, year_string), &parseyearnum, false },
461 { "TCOM", 4, offsetof(struct mp3entry, composer), NULL, false },
462 { "TPE2", 4, offsetof(struct mp3entry, albumartist), NULL, false },
463 { "TP2", 3, offsetof(struct mp3entry, albumartist), NULL, false },
464 { "TIT1", 4, offsetof(struct mp3entry, grouping), NULL, false },
465 { "TT1", 3, offsetof(struct mp3entry, grouping), NULL, false },
466 { "COMM", 4, offsetof(struct mp3entry, comment), NULL, false },
467 { "TCON", 4, offsetof(struct mp3entry, genre_string), &parsegenre, false },
468 { "TCO", 3, offsetof(struct mp3entry, genre_string), &parsegenre, false },
469 #if CONFIG_CODEC == SWCODEC
470 { "TXXX", 4, 0, &parseuser, false },
471 { "RVA2", 4, 0, &parserva2, true },
472 #endif
475 #define TAGLIST_SIZE ((int)(sizeof(taglist) / sizeof(taglist[0])))
477 /* Checks to see if the passed in string is a 16-bit wide Unicode v2
478 string. If it is, we convert it to a UTF-8 string. If it's not unicode,
479 we convert from the default codepage */
480 static int unicode_munge(char* string, char* utf8buf, int *len) {
481 long tmp;
482 bool le = false;
483 int i = 0;
484 unsigned char *str = (unsigned char *)string;
485 int templen = 0;
486 unsigned char* utf8 = (unsigned char *)utf8buf;
488 switch (str[0]) {
489 case 0x00: /* Type 0x00 is ordinary ISO 8859-1 */
490 str++;
491 (*len)--;
492 utf8 = iso_decode(str, utf8, -1, *len);
493 *utf8 = 0;
494 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
495 break;
497 case 0x01: /* Unicode with or without BOM */
498 case 0x02:
499 (*len)--;
500 str++;
502 /* Handle frames with more than one string
503 (needed for TXXX frames).*/
504 do {
505 tmp = bytes2int(0, 0, str[0], str[1]);
507 /* Now check if there is a BOM
508 (zero-width non-breaking space, 0xfeff)
509 and if it is in little or big endian format */
510 if(tmp == 0xfffe) { /* Little endian? */
511 le = true;
512 str += 2;
513 (*len)-=2;
514 } else if(tmp == 0xfeff) { /* Big endian? */
515 str += 2;
516 (*len)-=2;
517 } else
518 /* If there is no BOM (which is a specification violation),
519 let's try to guess it. If one of the bytes is 0x00, it is
520 probably the most significant one. */
521 if(str[1] == 0)
522 le = true;
524 do {
525 if(le)
526 utf8 = utf16LEdecode(str, utf8, 1);
527 else
528 utf8 = utf16BEdecode(str, utf8, 1);
530 str+=2;
531 i += 2;
532 } while((str[0] || str[1]) && (i < *len));
534 *utf8++ = 0; /* Terminate the string */
535 templen += (strlen(&utf8buf[templen]) + 1);
536 str += 2;
537 i+=2;
538 } while(i < *len);
539 *len = templen - 1;
540 break;
542 case 0x03: /* UTF-8 encoded string */
543 for(i=0; i < *len; i++)
544 utf8[i] = str[i+1];
545 (*len)--;
546 break;
548 default: /* Plain old string */
549 utf8 = iso_decode(str, utf8, -1, *len);
550 *utf8 = 0;
551 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
552 break;
554 return 0;
558 * Sets the title of an MP3 entry based on its ID3v1 tag.
560 * Arguments: file - the MP3 file to scen for a ID3v1 tag
561 * entry - the entry to set the title in
563 * Returns: true if a title was found and created, else false
565 static bool setid3v1title(int fd, struct mp3entry *entry)
567 unsigned char buffer[128];
568 static const char offsets[] = {3, 33, 63, 97, 93, 125, 127};
569 int i, j;
570 unsigned char* utf8;
572 if (-1 == lseek(fd, -128, SEEK_END))
573 return false;
575 if (read(fd, buffer, sizeof buffer) != sizeof buffer)
576 return false;
578 if (strncmp((char *)buffer, "TAG", 3))
579 return false;
581 entry->id3v1len = 128;
582 entry->id3version = ID3_VER_1_0;
584 for (i=0; i < (int)sizeof offsets; i++) {
585 unsigned char* ptr = (unsigned char *)buffer + offsets[i];
587 switch(i) {
588 case 0:
589 case 1:
590 case 2:
591 /* kill trailing space in strings */
592 for (j=29; j && (ptr[j]==0 || ptr[j]==' '); j--)
593 ptr[j] = 0;
594 /* convert string to utf8 */
595 utf8 = (unsigned char *)entry->id3v1buf[i];
596 utf8 = iso_decode(ptr, utf8, -1, 30);
597 /* make sure string is terminated */
598 *utf8 = 0;
599 break;
601 case 3:
602 /* kill trailing space in strings */
603 for (j=27; j && (ptr[j]==0 || ptr[j]==' '); j--)
604 ptr[j] = 0;
605 /* convert string to utf8 */
606 utf8 = (unsigned char *)entry->id3v1buf[3];
607 utf8 = iso_decode(ptr, utf8, -1, 28);
608 /* make sure string is terminated */
609 *utf8 = 0;
610 break;
612 case 4:
613 ptr[4] = 0;
614 entry->year = atoi((char *)ptr);
615 break;
617 case 5:
618 /* id3v1.1 uses last two bytes of comment field for track
619 number: first must be 0 and second is track num */
620 if (!ptr[0] && ptr[1]) {
621 entry->tracknum = ptr[1];
622 entry->id3version = ID3_VER_1_1;
624 break;
626 case 6:
627 /* genre */
628 entry->genre_string = id3_get_num_genre(ptr[0]);
629 break;
633 entry->title = entry->id3v1buf[0];
634 entry->artist = entry->id3v1buf[1];
635 entry->album = entry->id3v1buf[2];
636 entry->comment = entry->id3v1buf[3];
638 return true;
643 * Sets the title of an MP3 entry based on its ID3v2 tag.
645 * Arguments: file - the MP3 file to scan for a ID3v2 tag
646 * entry - the entry to set the title in
648 * Returns: true if a title was found and created, else false
650 static void setid3v2title(int fd, struct mp3entry *entry)
652 int minframesize;
653 int size;
654 long bufferpos = 0, totframelen, framelen;
655 char header[10];
656 char tmp[4];
657 unsigned char version;
658 char *buffer = entry->id3v2buf;
659 int bytesread = 0;
660 int buffersize = sizeof(entry->id3v2buf);
661 unsigned char global_flags;
662 int flags;
663 int skip;
664 bool global_unsynch = false;
665 bool unsynch = false;
666 int data_length_ind;
667 int i, j;
668 int rc;
670 global_ff_found = false;
672 /* Bail out if the tag is shorter than 10 bytes */
673 if(entry->id3v2len < 10)
674 return;
676 /* Read the ID3 tag version from the header */
677 lseek(fd, 0, SEEK_SET);
678 if(10 != read(fd, header, 10))
679 return;
681 /* Get the total ID3 tag size */
682 size = entry->id3v2len - 10;
684 version = header[3];
685 switch ( version ) {
686 case 2:
687 version = ID3_VER_2_2;
688 minframesize = 8;
689 break;
691 case 3:
692 version = ID3_VER_2_3;
693 minframesize = 12;
694 break;
696 case 4:
697 version = ID3_VER_2_4;
698 minframesize = 12;
699 break;
701 default:
702 /* unsupported id3 version */
703 return;
705 entry->id3version = version;
706 entry->tracknum = entry->year = entry->discnum = 0;
707 entry->title = entry->artist = entry->album = NULL; /* FIXME incomplete */
709 global_flags = header[5];
711 /* Skip the extended header if it is present */
712 if(global_flags & 0x40) {
713 if(version == ID3_VER_2_3) {
714 if(10 != read(fd, header, 10))
715 return;
716 /* The 2.3 extended header size doesn't include the header size
717 field itself. Also, it is not unsynched. */
718 framelen =
719 bytes2int(header[0], header[1], header[2], header[3]) + 4;
721 /* Skip the rest of the header */
722 lseek(fd, framelen - 10, SEEK_CUR);
725 if(version >= ID3_VER_2_4) {
726 if(4 != read(fd, header, 4))
727 return;
729 /* The 2.4 extended header size does include the entire header,
730 so here we can just skip it. This header is unsynched. */
731 framelen = unsync(header[0], header[1],
732 header[2], header[3]);
734 lseek(fd, framelen - 4, SEEK_CUR);
738 /* Is unsynchronization applied? */
739 if(global_flags & 0x80) {
740 global_unsynch = true;
744 * We must have at least minframesize bytes left for the
745 * remaining frames to be interesting
747 while (size >= minframesize && bufferpos < buffersize - 1) {
748 flags = 0;
750 /* Read frame header and check length */
751 if(version >= ID3_VER_2_3) {
752 if(global_unsynch && version <= ID3_VER_2_3)
753 rc = read_unsynched(fd, header, 10);
754 else
755 rc = read(fd, header, 10);
756 if(rc != 10)
757 return;
758 /* Adjust for the 10 bytes we read */
759 size -= 10;
761 flags = bytes2int(0, 0, header[8], header[9]);
763 if (version >= ID3_VER_2_4) {
764 framelen = unsync(header[4], header[5],
765 header[6], header[7]);
766 } else {
767 /* version .3 files don't use synchsafe ints for
768 * size */
769 framelen = bytes2int(header[4], header[5],
770 header[6], header[7]);
772 } else {
773 if(6 != read(fd, header, 6))
774 return;
775 /* Adjust for the 6 bytes we read */
776 size -= 6;
778 framelen = bytes2int(0, header[3], header[4], header[5]);
781 /* Keep track of the total size */
782 totframelen = framelen;
784 DEBUGF("framelen = %ld\n", framelen);
785 if(framelen == 0){
786 if (header[0] == 0 && header[1] == 0 && header[2] == 0)
787 return;
788 else
789 continue;
792 unsynch = false;
793 data_length_ind = 0;
795 if(flags)
797 skip = 0;
799 if (version >= ID3_VER_2_4) {
800 if(flags & 0x0040) { /* Grouping identity */
801 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
802 framelen--;
804 } else {
805 if(flags & 0x0020) { /* Grouping identity */
806 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
807 framelen--;
811 if(flags & 0x000c) /* Compression or encryption */
813 /* Skip it using the total size in case
814 it was truncated */
815 size -= totframelen;
816 lseek(fd, totframelen, SEEK_CUR);
817 continue;
820 if(flags & 0x0002) /* Unsynchronization */
821 unsynch = true;
823 if (version >= ID3_VER_2_4) {
824 if(flags & 0x0001) { /* Data length indicator */
825 if(4 != read(fd, tmp, 4))
826 return;
828 data_length_ind = unsync(tmp[0], tmp[1], tmp[2], tmp[3]);
829 framelen -= 4;
834 /* If the frame is larger than the remaining buffer space we try
835 to read as much as would fit in the buffer */
836 if(framelen >= buffersize - bufferpos)
837 framelen = buffersize - bufferpos - 1;
839 DEBUGF("id3v2 frame: %.4s\n", header);
841 /* Check for certain frame headers
843 'size' is the amount of frame bytes remaining. We decrement it by
844 the amount of bytes we read. If we fail to read as many bytes as
845 we expect, we assume that we can't read from this file, and bail
846 out.
848 For each frame. we will iterate over the list of supported tags,
849 and read the tag into entry's buffer. All tags will be kept as
850 strings, for cases where a number won't do, e.g., YEAR: "circa
851 1765", "1790/1977" (composed/performed), "28 Feb 1969" TRACK:
852 "1/12", "1 of 12", GENRE: "Freeform genre name" Text is more
853 flexible, and as the main use of id3 data is to display it,
854 converting it to an int just means reconverting to display it, at a
855 runtime cost.
857 For tags that the current code does convert to ints, a post
858 processing function will be called via a pointer to function. */
860 for (i=0; i<TAGLIST_SIZE; i++) {
861 const struct tag_resolver* tr = &taglist[i];
862 char** ptag = tr->offset ? (char**) (((char*)entry) + tr->offset)
863 : NULL;
864 char* tag;
865 int comm_offset=0;
867 /* Only ID3_VER_2_2 uses frames with three-character names. */
868 if (((version == ID3_VER_2_2) && (tr->tag_length != 3))
869 || ((version > ID3_VER_2_2) && (tr->tag_length != 4))) {
870 continue;
873 /* Note that parser functions sometimes set *ptag to NULL, so
874 * the "!*ptag" check here doesn't always have the desired
875 * effect. Should the parser functions (parsegenre in
876 * particular) be updated to handle the case of being called
877 * multiple times, or should the "*ptag" check be removed?
879 if( (!ptag || !*ptag) && !memcmp( header, tr->tag, tr->tag_length ) ) {
881 /* found a tag matching one in tagList, and not yet filled */
882 tag = buffer + bufferpos;
884 if(global_unsynch && version <= ID3_VER_2_3)
885 bytesread = read_unsynched(fd, tag, framelen);
886 else
887 bytesread = read(fd, tag, framelen);
889 if( bytesread != framelen )
890 return;
892 size -= bytesread;
894 if(unsynch || (global_unsynch && version >= ID3_VER_2_4))
895 bytesread = unsynchronize_frame(tag, bytesread);
897 /* the COMM frame has a 3 char field to hold an ISO-639-1
898 * language string and an optional short description;
899 * remove them so unicode_munge can work correctly
902 if(!memcmp( header, "COMM", 4 )) {
903 comm_offset = 3 + strlen(tag+4) + 1;
904 if(bytesread>comm_offset) {
905 bytesread-=comm_offset;
906 memmove(tag+1, tag+comm_offset+1, bytesread-1);
910 /* Attempt to parse Unicode string only if the tag contents
911 aren't binary */
912 if(!tr->binary) {
913 /* UTF-8 could potentially be 3 times larger */
914 /* so we need to create a new buffer */
915 char utf8buf[(3 * bytesread) + 1];
917 unicode_munge( tag, utf8buf, &bytesread );
919 if(bytesread >= buffersize - bufferpos)
920 bytesread = buffersize - bufferpos - 1;
922 for (j = 0; j < bytesread; j++)
923 tag[j] = utf8buf[j];
925 /* remove trailing spaces */
926 while ( bytesread > 0 && isspace(tag[bytesread-1]))
927 bytesread--;
930 tag[bytesread] = 0;
931 bufferpos += bytesread + 1;
933 if (ptag)
934 *ptag = tag;
936 if( tr->ppFunc )
937 bufferpos = tr->ppFunc(entry, tag, bufferpos);
939 /* Seek to the next frame */
940 if(framelen < totframelen)
941 lseek(fd, totframelen - framelen, SEEK_CUR);
942 break;
946 if( i == TAGLIST_SIZE ) {
947 /* no tag in tagList was found, or it was a repeat.
948 skip it using the total size */
950 if(global_unsynch && version <= ID3_VER_2_3) {
951 size -= skip_unsynched(fd, totframelen);
952 } else {
953 if(data_length_ind)
954 totframelen = data_length_ind;
956 size -= totframelen;
957 if( lseek(fd, totframelen, SEEK_CUR) == -1 )
958 return;
965 * Calculates the size of the ID3v2 tag.
967 * Arguments: file - the file to search for a tag.
969 * Returns: the size of the tag or 0 if none was found
971 int getid3v2len(int fd)
973 char buf[6];
974 int offset;
976 /* Make sure file has a ID3 tag */
977 if((-1 == lseek(fd, 0, SEEK_SET)) ||
978 (read(fd, buf, 6) != 6) ||
979 (strncmp(buf, "ID3", strlen("ID3")) != 0))
980 offset = 0;
982 /* Now check what the ID3v2 size field says */
983 else
984 if(read(fd, buf, 4) != 4)
985 offset = 0;
986 else
987 offset = unsync(buf[0], buf[1], buf[2], buf[3]) + 10;
989 DEBUGF("ID3V2 Length: 0x%x\n", offset);
990 return offset;
994 * Calculates the length (in milliseconds) of an MP3 file.
996 * Modified to only use integers.
998 * Arguments: file - the file to calculate the length upon
999 * entry - the entry to update with the length
1001 * Returns: the song length in milliseconds,
1002 * 0 means that it couldn't be calculated
1004 static int getsonglength(int fd, struct mp3entry *entry)
1006 unsigned long filetime = 0;
1007 struct mp3info info;
1008 long bytecount;
1010 /* Start searching after ID3v2 header */
1011 if(-1 == lseek(fd, entry->id3v2len, SEEK_SET))
1012 return 0;
1014 bytecount = get_mp3file_info(fd, &info);
1016 DEBUGF("Space between ID3V2 tag and first audio frame: 0x%lx bytes\n",
1017 bytecount);
1019 if(bytecount < 0)
1020 return -1;
1022 bytecount += entry->id3v2len;
1024 /* Validate byte count, in case the file has been edited without
1025 * updating the header.
1027 if (info.byte_count)
1029 const unsigned long expected = entry->filesize - entry->id3v1len
1030 - entry->id3v2len;
1031 const unsigned long diff = MAX(10240, info.byte_count / 20);
1033 if ((info.byte_count > expected + diff)
1034 || (info.byte_count < expected - diff))
1036 DEBUGF("Note: info.byte_count differs from expected value by "
1037 "%ld bytes\n", labs((long) (expected - info.byte_count)));
1038 info.byte_count = 0;
1039 info.frame_count = 0;
1040 info.file_time = 0;
1041 info.enc_padding = 0;
1043 /* Even if the bitrate was based on "known bad" values, it
1044 * should still be better for VBR files than using the bitrate
1045 * of the first audio frame.
1050 entry->bitrate = info.bitrate;
1051 entry->frequency = info.frequency;
1052 entry->version = info.version;
1053 entry->layer = info.layer;
1054 switch(entry->layer) {
1055 #if CONFIG_CODEC==SWCODEC
1056 case 0:
1057 entry->codectype=AFMT_MPA_L1;
1058 break;
1059 #endif
1060 case 1:
1061 entry->codectype=AFMT_MPA_L2;
1062 break;
1063 case 2:
1064 entry->codectype=AFMT_MPA_L3;
1065 break;
1068 /* If the file time hasn't been established, this may be a fixed
1069 rate MP3, so just use the default formula */
1071 filetime = info.file_time;
1073 if(filetime == 0)
1075 /* Prevent a division by zero */
1076 if (info.bitrate < 8)
1077 filetime = 0;
1078 else
1079 filetime = (entry->filesize - bytecount) / (info.bitrate / 8);
1080 /* bitrate is in kbps so this delivers milliseconds. Doing bitrate / 8
1081 * instead of filesize * 8 is exact, because mpeg audio bitrates are
1082 * always multiples of 8, and it avoids overflows. */
1085 entry->frame_count = info.frame_count;
1087 entry->vbr = info.is_vbr;
1088 entry->has_toc = info.has_toc;
1090 #if CONFIG_CODEC==SWCODEC
1091 entry->lead_trim = info.enc_delay;
1092 entry->tail_trim = info.enc_padding;
1093 #endif
1095 memcpy(entry->toc, info.toc, sizeof(info.toc));
1097 entry->vbr_header_pos = info.vbr_header_pos;
1099 /* Update the seek point for the first playable frame */
1100 entry->first_frame_offset = bytecount;
1101 DEBUGF("First frame is at %lx\n", entry->first_frame_offset);
1103 return filetime;
1107 * Checks all relevant information (such as ID3v1 tag, ID3v2 tag, length etc)
1108 * about an MP3 file and updates it's entry accordingly.
1110 Note, that this returns true for successful, false for error! */
1111 bool get_mp3_metadata(int fd, struct mp3entry *entry, const char *filename, bool v1first)
1113 int v1found = false;
1115 #if CONFIG_CODEC != SWCODEC
1116 memset(entry, 0, sizeof(struct mp3entry));
1117 #endif
1119 strncpy(entry->path, filename, sizeof(entry->path));
1121 entry->title = NULL;
1122 entry->filesize = filesize(fd);
1123 entry->id3v2len = getid3v2len(fd);
1124 entry->tracknum = 0;
1125 entry->discnum = 0;
1127 if(v1first)
1128 v1found = setid3v1title(fd, entry);
1130 if (!v1found && entry->id3v2len)
1131 setid3v2title(fd, entry);
1132 entry->length = getsonglength(fd, entry);
1134 /* Subtract the meta information from the file size to get
1135 the true size of the MP3 stream */
1136 entry->filesize -= entry->first_frame_offset;
1138 /* only seek to end of file if no id3v2 tags were found,
1139 and we already haven't looked for a v1 tag */
1140 if (!v1first && !entry->id3v2len) {
1141 setid3v1title(fd, entry);
1144 if(!entry->length || (entry->filesize < 8 ))
1145 /* no song length or less than 8 bytes is hereby considered to be an
1146 invalid mp3 and won't be played by us! */
1147 return false;
1149 return true;
1152 /* Note, that this returns false for successful, true for error! */
1153 bool mp3info(struct mp3entry *entry, const char *filename, bool v1first)
1155 int fd;
1156 bool result;
1158 fd = open(filename, O_RDONLY);
1159 if (fd < 0)
1160 return true;
1162 result = !get_mp3_metadata(fd, entry, filename, v1first);
1164 close(fd);
1166 return result;
1169 void adjust_mp3entry(struct mp3entry *entry, void *dest, void *orig)
1171 long offset;
1172 if (orig > dest)
1173 offset = - ((size_t)orig - (size_t)dest);
1174 else
1175 offset = (size_t)dest - (size_t)orig;
1177 if (entry->title)
1178 entry->title += offset;
1179 if (entry->artist)
1180 entry->artist += offset;
1181 if (entry->album)
1182 entry->album += offset;
1183 if (entry->genre_string)
1184 entry->genre_string += offset;
1185 if (entry->track_string)
1186 entry->track_string += offset;
1187 if (entry->disc_string)
1188 entry->disc_string += offset;
1189 if (entry->year_string)
1190 entry->year_string += offset;
1191 if (entry->composer)
1192 entry->composer += offset;
1193 if (entry->comment)
1194 entry->comment += offset;
1195 if (entry->albumartist)
1196 entry->albumartist += offset;
1197 if (entry->grouping)
1198 entry->grouping += offset;
1199 #if CONFIG_CODEC == SWCODEC
1200 if (entry->track_gain_string)
1201 entry->track_gain_string += offset;
1202 if (entry->album_gain_string)
1203 entry->album_gain_string += offset;
1204 #endif
1207 void copy_mp3entry(struct mp3entry *dest, struct mp3entry *orig)
1209 memcpy(dest, orig, sizeof(struct mp3entry));
1210 adjust_mp3entry(dest, dest, orig);
1213 #ifdef DEBUG_STANDALONE
1215 char *secs2str(int ms)
1217 static char buffer[32];
1218 int secs = ms/1000;
1219 ms %= 1000;
1220 snprintf(buffer, sizeof(buffer), "%d:%02d.%d", secs/60, secs%60, ms/100);
1221 return buffer;
1224 int main(int argc, char **argv)
1226 int i;
1227 for(i=1; i<argc; i++) {
1228 struct mp3entry mp3;
1229 mp3.album = "Bogus";
1230 if(mp3info(&mp3, argv[i], false)) {
1231 printf("Failed to get %s\n", argv[i]);
1232 return 0;
1235 printf("****** File: %s\n"
1236 " Title: %s\n"
1237 " Artist: %s\n"
1238 " Album: %s\n"
1239 " Genre: %s (%d) \n"
1240 " Composer: %s\n"
1241 " Year: %s (%d)\n"
1242 " Track: %s (%d)\n"
1243 " Length: %s / %d s\n"
1244 " Bitrate: %d\n"
1245 " Frequency: %d\n",
1246 argv[i],
1247 mp3.title?mp3.title:"<blank>",
1248 mp3.artist?mp3.artist:"<blank>",
1249 mp3.album?mp3.album:"<blank>",
1250 mp3.genre_string?mp3.genre_string:"<blank>",
1251 mp3.genre,
1252 mp3.composer?mp3.composer:"<blank>",
1253 mp3.year_string?mp3.year_string:"<blank>",
1254 mp3.year,
1255 mp3.track_string?mp3.track_string:"<blank>",
1256 mp3.tracknum,
1257 secs2str(mp3.length),
1258 mp3.length/1000,
1259 mp3.bitrate,
1260 mp3.frequency);
1263 return 0;
1266 #endif