Use MEM_ALIGN_ATTR in pitch detector plugin.
[kugel-rb.git] / apps / metadata / id3tags.c
blobf3ddd73077ff2ecd3a489b67437921ecbfdc05aa
1 /***************************************************************************
2 * __________ __ ___.
3 * Open \______ \ ____ ____ | | _\_ |__ _______ ___
4 * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
5 * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
6 * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
7 * \/ \/ \/ \/ \/
8 * $Id$
10 * Copyright (C) 2002 by Daniel Stenberg
12 * This program is free software; you can redistribute it and/or
13 * modify it under the terms of the GNU General Public License
14 * as published by the Free Software Foundation; either version 2
15 * of the License, or (at your option) any later version.
17 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
18 * KIND, either express or implied.
20 ****************************************************************************/
22 * Parts of this code has been stolen from the Ample project and was written
23 * by David H�deman. It has since been extended and enhanced pretty much by
24 * all sorts of friendly Rockbox people.
28 /* tagResolver and associated code copyright 2003 Thomas Paul Diffenbach
31 #include <stdio.h>
32 #include <stdlib.h>
33 #include <string.h>
34 #include <errno.h>
35 #include <stdbool.h>
36 #include <stddef.h>
37 #include <ctype.h>
38 #include "string-extra.h"
39 #include "config.h"
40 #include "file.h"
41 #include "logf.h"
42 #include "system.h"
43 #include "replaygain.h"
44 #include "rbunicode.h"
46 #include "metadata.h"
47 #include "mp3data.h"
48 #if CONFIG_CODEC == SWCODEC
49 #include "metadata_common.h"
50 #endif
51 #include "metadata_parsers.h"
53 static unsigned long unsync(unsigned long b0,
54 unsigned long b1,
55 unsigned long b2,
56 unsigned long b3)
58 return (((long)(b0 & 0x7F) << (3*7)) |
59 ((long)(b1 & 0x7F) << (2*7)) |
60 ((long)(b2 & 0x7F) << (1*7)) |
61 ((long)(b3 & 0x7F) << (0*7)));
64 static const char* const genres[] = {
65 "Blues", "Classic Rock", "Country", "Dance", "Disco", "Funk", "Grunge",
66 "Hip-Hop", "Jazz", "Metal", "New Age", "Oldies", "Other", "Pop", "R&B",
67 "Rap", "Reggae", "Rock", "Techno", "Industrial", "Alternative", "Ska",
68 "Death Metal", "Pranks", "Soundtrack", "Euro-Techno", "Ambient", "Trip-Hop",
69 "Vocal", "Jazz+Funk", "Fusion", "Trance", "Classical", "Instrumental",
70 "Acid", "House", "Game", "Sound Clip", "Gospel", "Noise", "AlternRock",
71 "Bass", "Soul", "Punk", "Space", "Meditative", "Instrumental Pop",
72 "Instrumental Rock", "Ethnic", "Gothic", "Darkwave", "Techno-Industrial",
73 "Electronic", "Pop-Folk", "Eurodance", "Dream", "Southern Rock", "Comedy",
74 "Cult", "Gangsta", "Top 40", "Christian Rap", "Pop/Funk", "Jungle",
75 "Native American", "Cabaret", "New Wave", "Psychadelic", "Rave",
76 "Showtunes", "Trailer", "Lo-Fi", "Tribal", "Acid Punk", "Acid Jazz",
77 "Polka", "Retro", "Musical", "Rock & Roll", "Hard Rock",
79 /* winamp extensions */
80 "Folk", "Folk-Rock", "National Folk", "Swing", "Fast Fusion", "Bebob",
81 "Latin", "Revival", "Celtic", "Bluegrass", "Avantgarde", "Gothic Rock",
82 "Progressive Rock", "Psychedelic Rock", "Symphonic Rock", "Slow Rock",
83 "Big Band", "Chorus", "Easy Listening", "Acoustic", "Humour", "Speech",
84 "Chanson", "Opera", "Chamber Music", "Sonata", "Symphony", "Booty Bass",
85 "Primus", "Porn Groove", "Satire", "Slow Jam", "Club", "Tango", "Samba",
86 "Folklore", "Ballad", "Power Ballad", "Rhythmic Soul", "Freestyle",
87 "Duet", "Punk Rock", "Drum Solo", "A capella", "Euro-House", "Dance Hall",
88 "Goa", "Drum & Bass", "Club-House", "Hardcore", "Terror", "Indie",
89 "BritPop", "Negerpunk", "Polsk Punk", "Beat", "Christian Gangsta Rap",
90 "Heavy Metal", "Black Metal", "Crossover", "Contemporary Christian",
91 "Christian Rock", "Merengue", "Salsa", "Thrash Metal", "Anime", "Jpop",
92 "Synthpop"
95 char* id3_get_num_genre(unsigned int genre_num)
97 if (genre_num < ARRAYLEN(genres))
98 return (char*)genres[genre_num];
99 return NULL;
103 HOW TO ADD ADDITIONAL ID3 VERSION 2 TAGS
104 Code and comments by Thomas Paul Diffenbach
106 To add another ID3v2 Tag, do the following:
107 1. add a char* named for the tag to struct mp3entry in id3.h,
108 (I (tpd) prefer to use char* rather than ints, even for what seems like
109 numerical values, for cases where a number won't do, e.g.,
110 YEAR: "circa 1765", "1790/1977" (composed/performed), "28 Feb 1969"
111 TRACK: "1/12", "1 of 12", GENRE: "Freeform genre name"
112 Text is more flexible, and as the main use of id3 data is to
113 display it, converting it to an int just means reconverting to
114 display it, at a runtime cost.)
116 2. If any special processing beyond copying the tag value from the Id3
117 block to the struct mp3entry is rrequired (such as converting to an
118 int), write a function to perform this special processing.
120 This function's prototype must match that of
121 typedef tagPostProcessFunc, that is it must be:
122 int func( struct mp3entry*, char* tag, int bufferpos )
123 the first argument is a pointer to the current mp3entry structure the
124 second argument is a pointer to the null terminated string value of the
125 tag found the third argument is the offset of the next free byte in the
126 mp3entry's buffer your function should return the corrected offset; if
127 you don't lengthen or shorten the tag string, you can return the third
128 argument unchanged.
130 Unless you have a good reason no to, make the function static.
131 TO JUST COPY THE TAG NO SPECIAL PROCESSING FUNCTION IS NEEDED.
133 3. add one or more entries to the tagList array, using the format:
134 char* ID3 Tag symbolic name -- see the ID3 specification for these,
135 sizeof() that name minus 1,
136 offsetof( struct mp3entry, variable_name_in_struct_mp3entry ),
137 pointer to your special processing function or NULL
138 if you need no special processing
139 flag indicating if this tag is binary or textual
140 Many ID3 symbolic names come in more than one form. You can add both
141 forms, each referencing the same variable in struct mp3entry.
142 If both forms are present, the last found will be used.
143 Note that the offset can be zero, in which case no entry will be set
144 in the mp3entry struct; the frame is still read into the buffer and
145 the special processing function is called (several times, if there
146 are several frames with the same name).
148 4. Alternately, use the TAG_LIST_ENTRY macro with
149 ID3 tag symbolic name,
150 variable in struct mp3entry,
151 special processing function address
153 5. Add code to wps-display.c function get_tag to assign a printf-like
154 format specifier for the tag */
156 /* Structure for ID3 Tag extraction information */
157 struct tag_resolver {
158 const char* tag;
159 int tag_length;
160 size_t offset;
161 int (*ppFunc)(struct mp3entry*, char* tag, int bufferpos);
162 bool binary;
165 static bool global_ff_found;
167 static int unsynchronize(char* tag, int len, bool *ff_found)
169 int i;
170 unsigned char c;
171 unsigned char *rp, *wp;
173 wp = rp = (unsigned char *)tag;
175 rp = (unsigned char *)tag;
176 for(i = 0;i < len;i++) {
177 /* Read the next byte and write it back, but don't increment the
178 write pointer */
179 c = *rp++;
180 *wp = c;
181 if(*ff_found) {
182 /* Increment the write pointer if it isn't an unsynch pattern */
183 if(c != 0)
184 wp++;
185 *ff_found = false;
186 } else {
187 if(c == 0xff)
188 *ff_found = true;
189 wp++;
192 return (long)wp - (long)tag;
195 static int unsynchronize_frame(char* tag, int len)
197 bool ff_found = false;
199 return unsynchronize(tag, len, &ff_found);
202 static int read_unsynched(int fd, void *buf, int len)
204 int i;
205 int rc;
206 int remaining = len;
207 char *wp;
208 char *rp;
210 wp = buf;
212 while(remaining) {
213 rp = wp;
214 rc = read(fd, rp, remaining);
215 if(rc <= 0)
216 return rc;
218 i = unsynchronize(wp, remaining, &global_ff_found);
219 remaining -= i;
220 wp += i;
223 return len;
226 static int skip_unsynched(int fd, int len)
228 int rc;
229 int remaining = len;
230 int rlen;
231 char buf[32];
233 while(remaining) {
234 rlen = MIN(sizeof(buf), (unsigned int)remaining);
235 rc = read(fd, buf, rlen);
236 if(rc <= 0)
237 return rc;
239 remaining -= unsynchronize(buf, rlen, &global_ff_found);
242 return len;
245 /* parse numeric value from string */
246 static int parsetracknum( struct mp3entry* entry, char* tag, int bufferpos )
248 entry->tracknum = atoi( tag );
249 return bufferpos;
252 /* parse numeric value from string */
253 static int parsediscnum( struct mp3entry* entry, char* tag, int bufferpos )
255 entry->discnum = atoi( tag );
256 return bufferpos;
259 /* parse numeric value from string */
260 static int parseyearnum( struct mp3entry* entry, char* tag, int bufferpos )
262 entry->year = atoi( tag );
263 return bufferpos;
266 /* parse numeric genre from string, version 2.2 and 2.3 */
267 static int parsegenre( struct mp3entry* entry, char* tag, int bufferpos )
269 if(entry->id3version >= ID3_VER_2_4) {
270 /* In version 2.4 and up, there are no parentheses, and the genre frame
271 is a list of strings, either numbers or text. */
273 /* Is it a number? */
274 if(isdigit(tag[0])) {
275 entry->genre_string = id3_get_num_genre(atoi( tag ));
276 return tag - entry->id3v2buf;
277 } else {
278 entry->genre_string = tag;
279 return bufferpos;
281 } else {
282 if( tag[0] == '(' && tag[1] != '(' ) {
283 entry->genre_string = id3_get_num_genre(atoi( tag + 1 ));
284 return tag - entry->id3v2buf;
286 else {
287 entry->genre_string = tag;
288 return bufferpos;
293 /* parse user defined text, looking for album artist and replaygain
294 * information.
296 static int parseuser( struct mp3entry* entry, char* tag, int bufferpos )
298 char* value = NULL;
299 int desc_len = strlen(tag);
300 int value_len = 0;
302 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos) {
303 /* At least part of the value was read, so we can safely try to
304 * parse it
306 value = tag + desc_len + 1;
307 value_len = bufferpos - (tag - entry->id3v2buf);
309 if (!strcasecmp(tag, "ALBUM ARTIST")) {
310 strlcpy(tag, value, value_len);
311 entry->albumartist = tag;
312 #if CONFIG_CODEC == SWCODEC
313 } else {
314 value_len = parse_replaygain(tag, value, entry, tag,
315 value_len);
316 #endif
320 return tag - entry->id3v2buf + value_len;
323 #if CONFIG_CODEC == SWCODEC
324 /* parse RVA2 binary data and convert to replaygain information. */
325 static int parserva2( struct mp3entry* entry, char* tag, int bufferpos )
327 int desc_len = strlen(tag);
328 int start_pos = tag - entry->id3v2buf;
329 int end_pos = start_pos + desc_len + 5;
330 int value_len = 0;
331 unsigned char* value = tag + desc_len + 1;
333 /* Only parse RVA2 replaygain tags if tag version == 2.4 and channel
334 * type is master volume.
336 if (entry->id3version == ID3_VER_2_4 && end_pos < bufferpos
337 && *value++ == 1) {
338 long gain = 0;
339 long peak = 0;
340 long peakbits;
341 long peakbytes;
342 bool album = false;
344 /* The RVA2 specification is unclear on some things (id string and
345 * peak volume), but this matches how Quod Libet use them.
348 gain = (int16_t) ((value[0] << 8) | value[1]);
349 value += 2;
350 peakbits = *value++;
351 peakbytes = (peakbits + 7) / 8;
353 /* Only use the topmost 24 bits for peak volume */
354 if (peakbytes > 3) {
355 peakbytes = 3;
358 /* Make sure the peak bits were read */
359 if (end_pos + peakbytes < bufferpos) {
360 long shift = ((8 - (peakbits & 7)) & 7) + (3 - peakbytes) * 8;
362 for ( ; peakbytes; peakbytes--) {
363 peak <<= 8;
364 peak += *value++;
367 peak <<= shift;
369 if (peakbits > 24) {
370 peak += *value >> (8 - shift);
374 if (strcasecmp(tag, "album") == 0) {
375 album = true;
376 } else if (strcasecmp(tag, "track") != 0) {
377 /* Only accept non-track values if we don't have any previous
378 * value.
380 if (entry->track_gain != 0) {
381 return start_pos;
385 value_len = parse_replaygain_int(album, gain, peak * 2, entry,
386 tag, sizeof(entry->id3v2buf) - start_pos);
389 return start_pos + value_len;
391 #endif
393 static int parsembtid( struct mp3entry* entry, char* tag, int bufferpos )
395 char* value = NULL;
396 int desc_len = strlen(tag);
397 /*DEBUGF("MBID len: %d\n", desc_len);*/
398 /* Musicbrainz track IDs are always 36 chars long */
399 const size_t mbtid_len = 36;
401 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos)
403 value = tag + desc_len + 1;
405 if (strcasecmp(tag, "http://musicbrainz.org") == 0)
407 if (mbtid_len == strlen(value))
409 entry->mb_track_id = value;
410 return bufferpos + mbtid_len + 1;
415 return bufferpos;
418 static const struct tag_resolver taglist[] = {
419 { "TPE1", 4, offsetof(struct mp3entry, artist), NULL, false },
420 { "TP1", 3, offsetof(struct mp3entry, artist), NULL, false },
421 { "TIT2", 4, offsetof(struct mp3entry, title), NULL, false },
422 { "TT2", 3, offsetof(struct mp3entry, title), NULL, false },
423 { "TALB", 4, offsetof(struct mp3entry, album), NULL, false },
424 { "TAL", 3, offsetof(struct mp3entry, album), NULL, false },
425 { "TRK", 3, offsetof(struct mp3entry, track_string), &parsetracknum, false },
426 { "TPOS", 4, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
427 { "TPA", 3, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
428 { "TRCK", 4, offsetof(struct mp3entry, track_string), &parsetracknum, false },
429 { "TDRC", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
430 { "TYER", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
431 { "TYE", 3, offsetof(struct mp3entry, year_string), &parseyearnum, false },
432 { "TCOM", 4, offsetof(struct mp3entry, composer), NULL, false },
433 { "TCM", 3, offsetof(struct mp3entry, composer), NULL, false },
434 { "TPE2", 4, offsetof(struct mp3entry, albumartist), NULL, false },
435 { "TP2", 3, offsetof(struct mp3entry, albumartist), NULL, false },
436 { "TIT1", 4, offsetof(struct mp3entry, grouping), NULL, false },
437 { "TT1", 3, offsetof(struct mp3entry, grouping), NULL, false },
438 { "COMM", 4, offsetof(struct mp3entry, comment), NULL, false },
439 { "COM", 3, offsetof(struct mp3entry, comment), NULL, false },
440 { "TCON", 4, offsetof(struct mp3entry, genre_string), &parsegenre, false },
441 { "TCO", 3, offsetof(struct mp3entry, genre_string), &parsegenre, false },
442 { "TXXX", 4, 0, &parseuser, false },
443 #if CONFIG_CODEC == SWCODEC
444 { "RVA2", 4, 0, &parserva2, true },
445 #endif
446 { "UFID", 4, 0, &parsembtid, false },
449 #define TAGLIST_SIZE ((int)ARRAYLEN(taglist))
451 /* Get the length of an ID3 string in the given encoding. Returns the length
452 * in bytes, including end nil, or -1 if the encoding is unknown.
454 static int unicode_len(char encoding, const void* string)
456 int len = 0;
458 if (encoding == 0x01 || encoding == 0x02) {
459 char first;
460 const char *s = string;
461 /* string might be unaligned, so using short* can crash on ARM and SH1 */
462 do {
463 first = *s++;
464 } while ((first | *s++) != 0);
466 len = s - (const char*) string;
467 } else {
468 len = strlen((char*) string) + 1;
471 return len;
474 /* Checks to see if the passed in string is a 16-bit wide Unicode v2
475 string. If it is, we convert it to a UTF-8 string. If it's not unicode,
476 we convert from the default codepage */
477 static int unicode_munge(char* string, char* utf8buf, int *len) {
478 long tmp;
479 bool le = false;
480 int i = 0;
481 unsigned char *str = (unsigned char *)string;
482 int templen = 0;
483 unsigned char* utf8 = (unsigned char *)utf8buf;
485 switch (str[0]) {
486 case 0x00: /* Type 0x00 is ordinary ISO 8859-1 */
487 str++;
488 (*len)--;
489 utf8 = iso_decode(str, utf8, -1, *len);
490 *utf8 = 0;
491 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
492 break;
494 case 0x01: /* Unicode with or without BOM */
495 case 0x02:
496 (*len)--;
497 str++;
499 /* Handle frames with more than one string
500 (needed for TXXX frames).*/
501 do {
502 tmp = bytes2int(0, 0, str[0], str[1]);
504 /* Now check if there is a BOM
505 (zero-width non-breaking space, 0xfeff)
506 and if it is in little or big endian format */
507 if(tmp == 0xfffe) { /* Little endian? */
508 le = true;
509 str += 2;
510 (*len)-=2;
511 } else if(tmp == 0xfeff) { /* Big endian? */
512 str += 2;
513 (*len)-=2;
514 } else
515 /* If there is no BOM (which is a specification violation),
516 let's try to guess it. If one of the bytes is 0x00, it is
517 probably the most significant one. */
518 if(str[1] == 0)
519 le = true;
521 do {
522 if(le)
523 utf8 = utf16LEdecode(str, utf8, 1);
524 else
525 utf8 = utf16BEdecode(str, utf8, 1);
527 str+=2;
528 i += 2;
529 } while((str[0] || str[1]) && (i < *len));
531 *utf8++ = 0; /* Terminate the string */
532 templen += (strlen(&utf8buf[templen]) + 1);
533 str += 2;
534 i+=2;
535 } while(i < *len);
536 *len = templen - 1;
537 break;
539 case 0x03: /* UTF-8 encoded string */
540 for(i=0; i < *len; i++)
541 utf8[i] = str[i+1];
542 (*len)--;
543 break;
545 default: /* Plain old string */
546 utf8 = iso_decode(str, utf8, -1, *len);
547 *utf8 = 0;
548 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
549 break;
551 return 0;
555 * Sets the title of an MP3 entry based on its ID3v1 tag.
557 * Arguments: file - the MP3 file to scen for a ID3v1 tag
558 * entry - the entry to set the title in
560 * Returns: true if a title was found and created, else false
562 bool setid3v1title(int fd, struct mp3entry *entry)
564 unsigned char buffer[128];
565 static const char offsets[] = {3, 33, 63, 97, 93, 125, 127};
566 int i, j;
567 unsigned char* utf8;
569 if (-1 == lseek(fd, -128, SEEK_END))
570 return false;
572 if (read(fd, buffer, sizeof buffer) != sizeof buffer)
573 return false;
575 if (strncmp((char *)buffer, "TAG", 3))
576 return false;
578 entry->id3v1len = 128;
579 entry->id3version = ID3_VER_1_0;
581 for (i=0; i < (int)sizeof offsets; i++) {
582 unsigned char* ptr = (unsigned char *)buffer + offsets[i];
584 switch(i) {
585 case 0:
586 case 1:
587 case 2:
588 /* kill trailing space in strings */
589 for (j=29; j && (ptr[j]==0 || ptr[j]==' '); j--)
590 ptr[j] = 0;
591 /* convert string to utf8 */
592 utf8 = (unsigned char *)entry->id3v1buf[i];
593 utf8 = iso_decode(ptr, utf8, -1, 30);
594 /* make sure string is terminated */
595 *utf8 = 0;
596 break;
598 case 3:
599 /* kill trailing space in strings */
600 for (j=27; j && (ptr[j]==0 || ptr[j]==' '); j--)
601 ptr[j] = 0;
602 /* convert string to utf8 */
603 utf8 = (unsigned char *)entry->id3v1buf[3];
604 utf8 = iso_decode(ptr, utf8, -1, 28);
605 /* make sure string is terminated */
606 *utf8 = 0;
607 break;
609 case 4:
610 ptr[4] = 0;
611 entry->year = atoi((char *)ptr);
612 break;
614 case 5:
615 /* id3v1.1 uses last two bytes of comment field for track
616 number: first must be 0 and second is track num */
617 if (!ptr[0] && ptr[1]) {
618 entry->tracknum = ptr[1];
619 entry->id3version = ID3_VER_1_1;
621 break;
623 case 6:
624 /* genre */
625 entry->genre_string = id3_get_num_genre(ptr[0]);
626 break;
630 entry->title = entry->id3v1buf[0];
631 entry->artist = entry->id3v1buf[1];
632 entry->album = entry->id3v1buf[2];
633 entry->comment = entry->id3v1buf[3];
635 return true;
640 * Sets the title of an MP3 entry based on its ID3v2 tag.
642 * Arguments: file - the MP3 file to scan for a ID3v2 tag
643 * entry - the entry to set the title in
645 * Returns: true if a title was found and created, else false
647 void setid3v2title(int fd, struct mp3entry *entry)
649 int minframesize;
650 int size;
651 long bufferpos = 0, totframelen, framelen;
652 char header[10];
653 char tmp[4];
654 unsigned char version;
655 char *buffer = entry->id3v2buf;
656 int bytesread = 0;
657 int buffersize = sizeof(entry->id3v2buf);
658 unsigned char global_flags;
659 int flags;
660 bool global_unsynch = false;
661 bool unsynch = false;
662 int i, j;
663 int rc;
664 #if CONFIG_CODEC == SWCODEC
665 bool itunes_gapless = false;
666 #endif
668 global_ff_found = false;
670 /* Bail out if the tag is shorter than 10 bytes */
671 if(entry->id3v2len < 10)
672 return;
674 /* Read the ID3 tag version from the header */
675 lseek(fd, 0, SEEK_SET);
676 if(10 != read(fd, header, 10))
677 return;
679 /* Get the total ID3 tag size */
680 size = entry->id3v2len - 10;
682 version = header[3];
683 switch ( version ) {
684 case 2:
685 version = ID3_VER_2_2;
686 minframesize = 8;
687 break;
689 case 3:
690 version = ID3_VER_2_3;
691 minframesize = 12;
692 break;
694 case 4:
695 version = ID3_VER_2_4;
696 minframesize = 12;
697 break;
699 default:
700 /* unsupported id3 version */
701 return;
703 entry->id3version = version;
704 entry->tracknum = entry->year = entry->discnum = 0;
705 entry->title = entry->artist = entry->album = NULL; /* FIXME incomplete */
707 global_flags = header[5];
709 /* Skip the extended header if it is present */
710 if(global_flags & 0x40) {
711 if(version == ID3_VER_2_3) {
712 if(10 != read(fd, header, 10))
713 return;
714 /* The 2.3 extended header size doesn't include the header size
715 field itself. Also, it is not unsynched. */
716 framelen =
717 bytes2int(header[0], header[1], header[2], header[3]) + 4;
719 /* Skip the rest of the header */
720 lseek(fd, framelen - 10, SEEK_CUR);
723 if(version >= ID3_VER_2_4) {
724 if(4 != read(fd, header, 4))
725 return;
727 /* The 2.4 extended header size does include the entire header,
728 so here we can just skip it. This header is unsynched. */
729 framelen = unsync(header[0], header[1],
730 header[2], header[3]);
732 lseek(fd, framelen - 4, SEEK_CUR);
736 /* Is unsynchronization applied? */
737 if(global_flags & 0x80) {
738 global_unsynch = true;
742 * We must have at least minframesize bytes left for the
743 * remaining frames to be interesting
745 while (size >= minframesize && bufferpos < buffersize - 1) {
746 flags = 0;
748 /* Read frame header and check length */
749 if(version >= ID3_VER_2_3) {
750 if(global_unsynch && version <= ID3_VER_2_3)
751 rc = read_unsynched(fd, header, 10);
752 else
753 rc = read(fd, header, 10);
754 if(rc != 10)
755 return;
756 /* Adjust for the 10 bytes we read */
757 size -= 10;
759 flags = bytes2int(0, 0, header[8], header[9]);
761 if (version >= ID3_VER_2_4) {
762 framelen = unsync(header[4], header[5],
763 header[6], header[7]);
764 } else {
765 /* version .3 files don't use synchsafe ints for
766 * size */
767 framelen = bytes2int(header[4], header[5],
768 header[6], header[7]);
770 } else {
771 if(6 != read(fd, header, 6))
772 return;
773 /* Adjust for the 6 bytes we read */
774 size -= 6;
776 framelen = bytes2int(0, header[3], header[4], header[5]);
779 logf("framelen = %ld, flags = 0x%04x", framelen, flags);
780 if(framelen == 0){
781 if (header[0] == 0 && header[1] == 0 && header[2] == 0)
782 return;
783 else
784 continue;
787 unsynch = false;
789 if(flags)
791 if (version >= ID3_VER_2_4) {
792 if(flags & 0x0040) { /* Grouping identity */
793 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
794 framelen--;
796 } else {
797 if(flags & 0x0020) { /* Grouping identity */
798 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
799 framelen--;
803 if(flags & 0x000c) /* Compression or encryption */
805 /* Skip it */
806 size -= framelen;
807 lseek(fd, framelen, SEEK_CUR);
808 continue;
811 if(flags & 0x0002) /* Unsynchronization */
812 unsynch = true;
814 if (version >= ID3_VER_2_4) {
815 if(flags & 0x0001) { /* Data length indicator */
816 if(4 != read(fd, tmp, 4))
817 return;
819 /* We don't need the data length */
820 framelen -= 4;
825 if (framelen == 0)
826 continue;
828 if (framelen < 0)
829 return;
831 /* Keep track of the remaining frame size */
832 totframelen = framelen;
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 logf("id3v2 frame: %.4s", 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;
866 /* Only ID3_VER_2_2 uses frames with three-character names. */
867 if (((version == ID3_VER_2_2) && (tr->tag_length != 3))
868 || ((version > ID3_VER_2_2) && (tr->tag_length != 4))) {
869 continue;
872 if( !memcmp( header, tr->tag, tr->tag_length ) ) {
874 /* found a tag matching one in tagList, and not yet filled */
875 tag = buffer + bufferpos;
877 if(global_unsynch && version <= ID3_VER_2_3)
878 bytesread = read_unsynched(fd, tag, framelen);
879 else
880 bytesread = read(fd, tag, framelen);
882 if( bytesread != framelen )
883 return;
885 size -= bytesread;
887 if(unsynch || (global_unsynch && version >= ID3_VER_2_4))
888 bytesread = unsynchronize_frame(tag, bytesread);
890 /* the COMM frame has a 3 char field to hold an ISO-639-1
891 * language string and an optional short description;
892 * remove them so unicode_munge can work correctly
895 if((tr->tag_length == 4 && !memcmp( header, "COMM", 4)) ||
896 (tr->tag_length == 3 && !memcmp( header, "COM", 3))) {
897 int offset;
898 if(!strncmp(tag+4, "iTun", 4)) {
899 #if CONFIG_CODEC == SWCODEC
900 /* check for iTunes gapless information */
901 if(!strncmp(tag+4, "iTunSMPB", 8))
902 itunes_gapless = true;
903 else
904 #endif
905 /* ignore other with iTunes tags */
906 break;
909 offset = 3 + unicode_len(*tag, tag + 4);
910 if(bytesread > offset) {
911 bytesread -= offset;
912 memmove(tag + 1, tag + 1 + offset, bytesread - 1);
916 /* Attempt to parse Unicode string only if the tag contents
917 aren't binary */
918 if(!tr->binary) {
919 /* UTF-8 could potentially be 3 times larger */
920 /* so we need to create a new buffer */
921 char utf8buf[(3 * bytesread) + 1];
923 unicode_munge( tag, utf8buf, &bytesread );
925 if(bytesread >= buffersize - bufferpos)
926 bytesread = buffersize - bufferpos - 1;
928 for (j = 0; j < bytesread; j++)
929 tag[j] = utf8buf[j];
931 /* remove trailing spaces */
932 while ( bytesread > 0 && isspace(tag[bytesread-1]))
933 bytesread--;
936 tag[bytesread] = 0;
937 bufferpos += bytesread + 1;
939 #if CONFIG_CODEC == SWCODEC
940 /* parse the tag if it contains iTunes gapless info */
941 if (itunes_gapless)
943 itunes_gapless = false;
944 entry->lead_trim = get_itunes_int32(tag, 1);
945 entry->tail_trim = get_itunes_int32(tag, 2);
947 #endif
949 /* Note that parser functions sometimes set *ptag to NULL, so
950 * the "!*ptag" check here doesn't always have the desired
951 * effect. Should the parser functions (parsegenre in
952 * particular) be updated to handle the case of being called
953 * multiple times, or should the "*ptag" check be removed?
955 if (ptag && !*ptag)
956 *ptag = tag;
958 if( tr->ppFunc )
959 bufferpos = tr->ppFunc(entry, tag, bufferpos);
961 /* Seek to the next frame */
962 if(framelen < totframelen)
963 lseek(fd, totframelen - framelen, SEEK_CUR);
964 break;
968 if( i == TAGLIST_SIZE ) {
969 /* no tag in tagList was found, or it was a repeat.
970 skip it using the total size */
972 if(global_unsynch && version <= ID3_VER_2_3) {
973 size -= skip_unsynched(fd, totframelen);
974 } else {
975 size -= totframelen;
976 if( lseek(fd, totframelen, SEEK_CUR) == -1 )
977 return;
984 * Calculates the size of the ID3v2 tag.
986 * Arguments: file - the file to search for a tag.
988 * Returns: the size of the tag or 0 if none was found
990 int getid3v2len(int fd)
992 char buf[6];
993 int offset;
995 /* Make sure file has a ID3 tag */
996 if((-1 == lseek(fd, 0, SEEK_SET)) ||
997 (read(fd, buf, 6) != 6) ||
998 (strncmp(buf, "ID3", strlen("ID3")) != 0))
999 offset = 0;
1001 /* Now check what the ID3v2 size field says */
1002 else
1003 if(read(fd, buf, 4) != 4)
1004 offset = 0;
1005 else
1006 offset = unsync(buf[0], buf[1], buf[2], buf[3]) + 10;
1008 logf("ID3V2 Length: 0x%x", offset);
1009 return offset;
1012 #ifdef DEBUG_STANDALONE
1014 char *secs2str(int ms)
1016 static char buffer[32];
1017 int secs = ms/1000;
1018 ms %= 1000;
1019 snprintf(buffer, sizeof(buffer), "%d:%02d.%d", secs/60, secs%60, ms/100);
1020 return buffer;
1023 int main(int argc, char **argv)
1025 int i;
1026 for(i=1; i<argc; i++) {
1027 struct mp3entry mp3;
1028 mp3.album = "Bogus";
1029 if(mp3info(&mp3, argv[i], false)) {
1030 printf("Failed to get %s\n", argv[i]);
1031 return 0;
1034 printf("****** File: %s\n"
1035 " Title: %s\n"
1036 " Artist: %s\n"
1037 " Album: %s\n"
1038 " Genre: %s (%d) \n"
1039 " Composer: %s\n"
1040 " Year: %s (%d)\n"
1041 " Track: %s (%d)\n"
1042 " Length: %s / %d s\n"
1043 " Bitrate: %d\n"
1044 " Frequency: %d\n",
1045 argv[i],
1046 mp3.title?mp3.title:"<blank>",
1047 mp3.artist?mp3.artist:"<blank>",
1048 mp3.album?mp3.album:"<blank>",
1049 mp3.genre_string?mp3.genre_string:"<blank>",
1050 mp3.genre,
1051 mp3.composer?mp3.composer:"<blank>",
1052 mp3.year_string?mp3.year_string:"<blank>",
1053 mp3.year,
1054 mp3.track_string?mp3.track_string:"<blank>",
1055 mp3.tracknum,
1056 secs2str(mp3.length),
1057 mp3.length/1000,
1058 mp3.bitrate,
1059 mp3.frequency);
1062 return 0;
1065 #endif