summaryrefslogtreecommitdiff
path: root/apps/metadata/id3tags.c
diff options
context:
space:
mode:
Diffstat (limited to 'apps/metadata/id3tags.c')
-rw-r--r--apps/metadata/id3tags.c1065
1 files changed, 1065 insertions, 0 deletions
diff --git a/apps/metadata/id3tags.c b/apps/metadata/id3tags.c
new file mode 100644
index 0000000000..2f736d7b9b
--- /dev/null
+++ b/apps/metadata/id3tags.c
@@ -0,0 +1,1065 @@
1/***************************************************************************
2 * __________ __ ___.
3 * Open \______ \ ____ ____ | | _\_ |__ _______ ___
4 * Source | _// _ \_/ ___\| |/ /| __ \ / _ \ \/ /
5 * Jukebox | | ( <_> ) \___| < | \_\ ( <_> > < <
6 * Firmware |____|_ /\____/ \___ >__|_ \|___ /\____/__/\_ \
7 * \/ \/ \/ \/ \/
8 * $Id$
9 *
10 * Copyright (C) 2002 by Daniel Stenberg
11 *
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.
16 *
17 * This software is distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY
18 * KIND, either express or implied.
19 *
20 ****************************************************************************/
21/*
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.
25 *
26 */
27
28 /* tagResolver and associated code copyright 2003 Thomas Paul Diffenbach
29 */
30
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 "config.h"
39#include "file.h"
40#include "logf.h"
41#include "system.h"
42#include "replaygain.h"
43#include "rbunicode.h"
44
45#include "metadata.h"
46#include "mp3data.h"
47#if CONFIG_CODEC == SWCODEC
48#include "metadata_common.h"
49#endif
50#include "metadata_parsers.h"
51
52static unsigned long unsync(unsigned long b0,
53 unsigned long b1,
54 unsigned long b2,
55 unsigned long b3)
56{
57 return (((long)(b0 & 0x7F) << (3*7)) |
58 ((long)(b1 & 0x7F) << (2*7)) |
59 ((long)(b2 & 0x7F) << (1*7)) |
60 ((long)(b3 & 0x7F) << (0*7)));
61}
62
63static const char* const genres[] = {
64 "Blues", "Classic Rock", "Country", "Dance", "Disco", "Funk", "Grunge",
65 "Hip-Hop", "Jazz", "Metal", "New Age", "Oldies", "Other", "Pop", "R&B",
66 "Rap", "Reggae", "Rock", "Techno", "Industrial", "Alternative", "Ska",
67 "Death Metal", "Pranks", "Soundtrack", "Euro-Techno", "Ambient", "Trip-Hop",
68 "Vocal", "Jazz+Funk", "Fusion", "Trance", "Classical", "Instrumental",
69 "Acid", "House", "Game", "Sound Clip", "Gospel", "Noise", "AlternRock",
70 "Bass", "Soul", "Punk", "Space", "Meditative", "Instrumental Pop",
71 "Instrumental Rock", "Ethnic", "Gothic", "Darkwave", "Techno-Industrial",
72 "Electronic", "Pop-Folk", "Eurodance", "Dream", "Southern Rock", "Comedy",
73 "Cult", "Gangsta", "Top 40", "Christian Rap", "Pop/Funk", "Jungle",
74 "Native American", "Cabaret", "New Wave", "Psychadelic", "Rave",
75 "Showtunes", "Trailer", "Lo-Fi", "Tribal", "Acid Punk", "Acid Jazz",
76 "Polka", "Retro", "Musical", "Rock & Roll", "Hard Rock",
77
78 /* winamp extensions */
79 "Folk", "Folk-Rock", "National Folk", "Swing", "Fast Fusion", "Bebob",
80 "Latin", "Revival", "Celtic", "Bluegrass", "Avantgarde", "Gothic Rock",
81 "Progressive Rock", "Psychedelic Rock", "Symphonic Rock", "Slow Rock",
82 "Big Band", "Chorus", "Easy Listening", "Acoustic", "Humour", "Speech",
83 "Chanson", "Opera", "Chamber Music", "Sonata", "Symphony", "Booty Bass",
84 "Primus", "Porn Groove", "Satire", "Slow Jam", "Club", "Tango", "Samba",
85 "Folklore", "Ballad", "Power Ballad", "Rhythmic Soul", "Freestyle",
86 "Duet", "Punk Rock", "Drum Solo", "A capella", "Euro-House", "Dance Hall",
87 "Goa", "Drum & Bass", "Club-House", "Hardcore", "Terror", "Indie",
88 "BritPop", "Negerpunk", "Polsk Punk", "Beat", "Christian Gangsta Rap",
89 "Heavy Metal", "Black Metal", "Crossover", "Contemporary Christian",
90 "Christian Rock", "Merengue", "Salsa", "Thrash Metal", "Anime", "Jpop",
91 "Synthpop"
92};
93
94char* id3_get_num_genre(unsigned int genre_num)
95{
96 if (genre_num < ARRAYLEN(genres))
97 return (char*)genres[genre_num];
98 return NULL;
99}
100
101/*
102 HOW TO ADD ADDITIONAL ID3 VERSION 2 TAGS
103 Code and comments by Thomas Paul Diffenbach
104
105 To add another ID3v2 Tag, do the following:
106 1. add a char* named for the tag to struct mp3entry in id3.h,
107 (I (tpd) prefer to use char* rather than ints, even for what seems like
108 numerical values, for cases where a number won't do, e.g.,
109 YEAR: "circa 1765", "1790/1977" (composed/performed), "28 Feb 1969"
110 TRACK: "1/12", "1 of 12", GENRE: "Freeform genre name"
111 Text is more flexible, and as the main use of id3 data is to
112 display it, converting it to an int just means reconverting to
113 display it, at a runtime cost.)
114
115 2. If any special processing beyond copying the tag value from the Id3
116 block to the struct mp3entry is rrequired (such as converting to an
117 int), write a function to perform this special processing.
118
119 This function's prototype must match that of
120 typedef tagPostProcessFunc, that is it must be:
121 int func( struct mp3entry*, char* tag, int bufferpos )
122 the first argument is a pointer to the current mp3entry structure the
123 second argument is a pointer to the null terminated string value of the
124 tag found the third argument is the offset of the next free byte in the
125 mp3entry's buffer your function should return the corrected offset; if
126 you don't lengthen or shorten the tag string, you can return the third
127 argument unchanged.
128
129 Unless you have a good reason no to, make the function static.
130 TO JUST COPY THE TAG NO SPECIAL PROCESSING FUNCTION IS NEEDED.
131
132 3. add one or more entries to the tagList array, using the format:
133 char* ID3 Tag symbolic name -- see the ID3 specification for these,
134 sizeof() that name minus 1,
135 offsetof( struct mp3entry, variable_name_in_struct_mp3entry ),
136 pointer to your special processing function or NULL
137 if you need no special processing
138 flag indicating if this tag is binary or textual
139 Many ID3 symbolic names come in more than one form. You can add both
140 forms, each referencing the same variable in struct mp3entry.
141 If both forms are present, the last found will be used.
142 Note that the offset can be zero, in which case no entry will be set
143 in the mp3entry struct; the frame is still read into the buffer and
144 the special processing function is called (several times, if there
145 are several frames with the same name).
146
147 4. Alternately, use the TAG_LIST_ENTRY macro with
148 ID3 tag symbolic name,
149 variable in struct mp3entry,
150 special processing function address
151
152 5. Add code to wps-display.c function get_tag to assign a printf-like
153 format specifier for the tag */
154
155/* Structure for ID3 Tag extraction information */
156struct tag_resolver {
157 const char* tag;
158 int tag_length;
159 size_t offset;
160 int (*ppFunc)(struct mp3entry*, char* tag, int bufferpos);
161 bool binary;
162};
163
164static bool global_ff_found;
165
166static int unsynchronize(char* tag, int len, bool *ff_found)
167{
168 int i;
169 unsigned char c;
170 unsigned char *rp, *wp;
171
172 wp = rp = (unsigned char *)tag;
173
174 rp = (unsigned char *)tag;
175 for(i = 0;i < len;i++) {
176 /* Read the next byte and write it back, but don't increment the
177 write pointer */
178 c = *rp++;
179 *wp = c;
180 if(*ff_found) {
181 /* Increment the write pointer if it isn't an unsynch pattern */
182 if(c != 0)
183 wp++;
184 *ff_found = false;
185 } else {
186 if(c == 0xff)
187 *ff_found = true;
188 wp++;
189 }
190 }
191 return (long)wp - (long)tag;
192}
193
194static int unsynchronize_frame(char* tag, int len)
195{
196 bool ff_found = false;
197
198 return unsynchronize(tag, len, &ff_found);
199}
200
201static int read_unsynched(int fd, void *buf, int len)
202{
203 int i;
204 int rc;
205 int remaining = len;
206 char *wp;
207 char *rp;
208
209 wp = buf;
210
211 while(remaining) {
212 rp = wp;
213 rc = read(fd, rp, remaining);
214 if(rc <= 0)
215 return rc;
216
217 i = unsynchronize(wp, remaining, &global_ff_found);
218 remaining -= i;
219 wp += i;
220 }
221
222 return len;
223}
224
225static int skip_unsynched(int fd, int len)
226{
227 int rc;
228 int remaining = len;
229 int rlen;
230 char buf[32];
231
232 while(remaining) {
233 rlen = MIN(sizeof(buf), (unsigned int)remaining);
234 rc = read(fd, buf, rlen);
235 if(rc <= 0)
236 return rc;
237
238 remaining -= unsynchronize(buf, rlen, &global_ff_found);
239 }
240
241 return len;
242}
243
244/* parse numeric value from string */
245static int parsetracknum( struct mp3entry* entry, char* tag, int bufferpos )
246{
247 entry->tracknum = atoi( tag );
248 return bufferpos;
249}
250
251/* parse numeric value from string */
252static int parsediscnum( struct mp3entry* entry, char* tag, int bufferpos )
253{
254 entry->discnum = atoi( tag );
255 return bufferpos;
256}
257
258/* parse numeric value from string */
259static int parseyearnum( struct mp3entry* entry, char* tag, int bufferpos )
260{
261 entry->year = atoi( tag );
262 return bufferpos;
263}
264
265/* parse numeric genre from string, version 2.2 and 2.3 */
266static int parsegenre( struct mp3entry* entry, char* tag, int bufferpos )
267{
268 if(entry->id3version >= ID3_VER_2_4) {
269 /* In version 2.4 and up, there are no parentheses, and the genre frame
270 is a list of strings, either numbers or text. */
271
272 /* Is it a number? */
273 if(isdigit(tag[0])) {
274 entry->genre_string = id3_get_num_genre(atoi( tag ));
275 return tag - entry->id3v2buf;
276 } else {
277 entry->genre_string = tag;
278 return bufferpos;
279 }
280 } else {
281 if( tag[0] == '(' && tag[1] != '(' ) {
282 entry->genre_string = id3_get_num_genre(atoi( tag + 1 ));
283 return tag - entry->id3v2buf;
284 }
285 else {
286 entry->genre_string = tag;
287 return bufferpos;
288 }
289 }
290}
291
292/* parse user defined text, looking for album artist and replaygain
293 * information.
294 */
295static int parseuser( struct mp3entry* entry, char* tag, int bufferpos )
296{
297 char* value = NULL;
298 int desc_len = strlen(tag);
299 int value_len = 0;
300
301 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos) {
302 /* At least part of the value was read, so we can safely try to
303 * parse it
304 */
305 value = tag + desc_len + 1;
306 value_len = bufferpos - (tag - entry->id3v2buf);
307
308 if (!strcasecmp(tag, "ALBUM ARTIST")) {
309 strlcpy(tag, value, value_len);
310 entry->albumartist = tag;
311#if CONFIG_CODEC == SWCODEC
312 } else {
313 value_len = parse_replaygain(tag, value, entry, tag,
314 value_len);
315#endif
316 }
317 }
318
319 return tag - entry->id3v2buf + value_len;
320}
321
322#if CONFIG_CODEC == SWCODEC
323/* parse RVA2 binary data and convert to replaygain information. */
324static int parserva2( struct mp3entry* entry, char* tag, int bufferpos )
325{
326 int desc_len = strlen(tag);
327 int start_pos = tag - entry->id3v2buf;
328 int end_pos = start_pos + desc_len + 5;
329 int value_len = 0;
330 unsigned char* value = tag + desc_len + 1;
331
332 /* Only parse RVA2 replaygain tags if tag version == 2.4 and channel
333 * type is master volume.
334 */
335 if (entry->id3version == ID3_VER_2_4 && end_pos < bufferpos
336 && *value++ == 1) {
337 long gain = 0;
338 long peak = 0;
339 long peakbits;
340 long peakbytes;
341 bool album = false;
342
343 /* The RVA2 specification is unclear on some things (id string and
344 * peak volume), but this matches how Quod Libet use them.
345 */
346
347 gain = (int16_t) ((value[0] << 8) | value[1]);
348 value += 2;
349 peakbits = *value++;
350 peakbytes = (peakbits + 7) / 8;
351
352 /* Only use the topmost 24 bits for peak volume */
353 if (peakbytes > 3) {
354 peakbytes = 3;
355 }
356
357 /* Make sure the peak bits were read */
358 if (end_pos + peakbytes < bufferpos) {
359 long shift = ((8 - (peakbits & 7)) & 7) + (3 - peakbytes) * 8;
360
361 for ( ; peakbytes; peakbytes--) {
362 peak <<= 8;
363 peak += *value++;
364 }
365
366 peak <<= shift;
367
368 if (peakbits > 24) {
369 peak += *value >> (8 - shift);
370 }
371 }
372
373 if (strcasecmp(tag, "album") == 0) {
374 album = true;
375 } else if (strcasecmp(tag, "track") != 0) {
376 /* Only accept non-track values if we don't have any previous
377 * value.
378 */
379 if (entry->track_gain != 0) {
380 return start_pos;
381 }
382 }
383
384 value_len = parse_replaygain_int(album, gain, peak * 2, entry,
385 tag, sizeof(entry->id3v2buf) - start_pos);
386 }
387
388 return start_pos + value_len;
389}
390#endif
391
392static int parsembtid( struct mp3entry* entry, char* tag, int bufferpos )
393{
394 char* value = NULL;
395 int desc_len = strlen(tag);
396 /*DEBUGF("MBID len: %d\n", desc_len);*/
397 /* Musicbrainz track IDs are always 36 chars long */
398 const size_t mbtid_len = 36;
399
400 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos)
401 {
402 value = tag + desc_len + 1;
403
404 if (strcasecmp(tag, "http://musicbrainz.org") == 0)
405 {
406 if (mbtid_len == strlen(value))
407 {
408 entry->mb_track_id = value;
409 return bufferpos + mbtid_len + 1;
410 }
411 }
412 }
413
414 return bufferpos;
415}
416
417static const struct tag_resolver taglist[] = {
418 { "TPE1", 4, offsetof(struct mp3entry, artist), NULL, false },
419 { "TP1", 3, offsetof(struct mp3entry, artist), NULL, false },
420 { "TIT2", 4, offsetof(struct mp3entry, title), NULL, false },
421 { "TT2", 3, offsetof(struct mp3entry, title), NULL, false },
422 { "TALB", 4, offsetof(struct mp3entry, album), NULL, false },
423 { "TAL", 3, offsetof(struct mp3entry, album), NULL, false },
424 { "TRK", 3, offsetof(struct mp3entry, track_string), &parsetracknum, false },
425 { "TPOS", 4, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
426 { "TRCK", 4, offsetof(struct mp3entry, track_string), &parsetracknum, false },
427 { "TDRC", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
428 { "TYER", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
429 { "TYE", 3, offsetof(struct mp3entry, year_string), &parseyearnum, false },
430 { "TCOM", 4, offsetof(struct mp3entry, composer), NULL, false },
431 { "TPE2", 4, offsetof(struct mp3entry, albumartist), NULL, false },
432 { "TP2", 3, offsetof(struct mp3entry, albumartist), NULL, false },
433 { "TIT1", 4, offsetof(struct mp3entry, grouping), NULL, false },
434 { "TT1", 3, offsetof(struct mp3entry, grouping), NULL, false },
435 { "COMM", 4, offsetof(struct mp3entry, comment), NULL, false },
436 { "COM", 3, offsetof(struct mp3entry, comment), NULL, false },
437 { "TCON", 4, offsetof(struct mp3entry, genre_string), &parsegenre, false },
438 { "TCO", 3, offsetof(struct mp3entry, genre_string), &parsegenre, false },
439 { "TXXX", 4, 0, &parseuser, false },
440#if CONFIG_CODEC == SWCODEC
441 { "RVA2", 4, 0, &parserva2, true },
442#endif
443 { "UFID", 4, 0, &parsembtid, false },
444};
445
446#define TAGLIST_SIZE ((int)ARRAYLEN(taglist))
447
448/* Get the length of an ID3 string in the given encoding. Returns the length
449 * in bytes, including end nil, or -1 if the encoding is unknown.
450 */
451static int unicode_len(char encoding, const void* string)
452{
453 int len = 0;
454
455 if (encoding == 0x01 || encoding == 0x02) {
456 char first;
457 const char *s = string;
458 /* string might be unaligned, so using short* can crash on ARM and SH1 */
459 do {
460 first = *s++;
461 } while ((first | *s++) != 0);
462
463 len = s - (const char*) string;
464 } else {
465 len = strlen((char*) string) + 1;
466 }
467
468 return len;
469}
470
471/* Checks to see if the passed in string is a 16-bit wide Unicode v2
472 string. If it is, we convert it to a UTF-8 string. If it's not unicode,
473 we convert from the default codepage */
474static int unicode_munge(char* string, char* utf8buf, int *len) {
475 long tmp;
476 bool le = false;
477 int i = 0;
478 unsigned char *str = (unsigned char *)string;
479 int templen = 0;
480 unsigned char* utf8 = (unsigned char *)utf8buf;
481
482 switch (str[0]) {
483 case 0x00: /* Type 0x00 is ordinary ISO 8859-1 */
484 str++;
485 (*len)--;
486 utf8 = iso_decode(str, utf8, -1, *len);
487 *utf8 = 0;
488 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
489 break;
490
491 case 0x01: /* Unicode with or without BOM */
492 case 0x02:
493 (*len)--;
494 str++;
495
496 /* Handle frames with more than one string
497 (needed for TXXX frames).*/
498 do {
499 tmp = bytes2int(0, 0, str[0], str[1]);
500
501 /* Now check if there is a BOM
502 (zero-width non-breaking space, 0xfeff)
503 and if it is in little or big endian format */
504 if(tmp == 0xfffe) { /* Little endian? */
505 le = true;
506 str += 2;
507 (*len)-=2;
508 } else if(tmp == 0xfeff) { /* Big endian? */
509 str += 2;
510 (*len)-=2;
511 } else
512 /* If there is no BOM (which is a specification violation),
513 let's try to guess it. If one of the bytes is 0x00, it is
514 probably the most significant one. */
515 if(str[1] == 0)
516 le = true;
517
518 do {
519 if(le)
520 utf8 = utf16LEdecode(str, utf8, 1);
521 else
522 utf8 = utf16BEdecode(str, utf8, 1);
523
524 str+=2;
525 i += 2;
526 } while((str[0] || str[1]) && (i < *len));
527
528 *utf8++ = 0; /* Terminate the string */
529 templen += (strlen(&utf8buf[templen]) + 1);
530 str += 2;
531 i+=2;
532 } while(i < *len);
533 *len = templen - 1;
534 break;
535
536 case 0x03: /* UTF-8 encoded string */
537 for(i=0; i < *len; i++)
538 utf8[i] = str[i+1];
539 (*len)--;
540 break;
541
542 default: /* Plain old string */
543 utf8 = iso_decode(str, utf8, -1, *len);
544 *utf8 = 0;
545 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
546 break;
547 }
548 return 0;
549}
550
551/*
552 * Sets the title of an MP3 entry based on its ID3v1 tag.
553 *
554 * Arguments: file - the MP3 file to scen for a ID3v1 tag
555 * entry - the entry to set the title in
556 *
557 * Returns: true if a title was found and created, else false
558 */
559bool setid3v1title(int fd, struct mp3entry *entry)
560{
561 unsigned char buffer[128];
562 static const char offsets[] = {3, 33, 63, 97, 93, 125, 127};
563 int i, j;
564 unsigned char* utf8;
565
566 if (-1 == lseek(fd, -128, SEEK_END))
567 return false;
568
569 if (read(fd, buffer, sizeof buffer) != sizeof buffer)
570 return false;
571
572 if (strncmp((char *)buffer, "TAG", 3))
573 return false;
574
575 entry->id3v1len = 128;
576 entry->id3version = ID3_VER_1_0;
577
578 for (i=0; i < (int)sizeof offsets; i++) {
579 unsigned char* ptr = (unsigned char *)buffer + offsets[i];
580
581 switch(i) {
582 case 0:
583 case 1:
584 case 2:
585 /* kill trailing space in strings */
586 for (j=29; j && (ptr[j]==0 || ptr[j]==' '); j--)
587 ptr[j] = 0;
588 /* convert string to utf8 */
589 utf8 = (unsigned char *)entry->id3v1buf[i];
590 utf8 = iso_decode(ptr, utf8, -1, 30);
591 /* make sure string is terminated */
592 *utf8 = 0;
593 break;
594
595 case 3:
596 /* kill trailing space in strings */
597 for (j=27; j && (ptr[j]==0 || ptr[j]==' '); j--)
598 ptr[j] = 0;
599 /* convert string to utf8 */
600 utf8 = (unsigned char *)entry->id3v1buf[3];
601 utf8 = iso_decode(ptr, utf8, -1, 28);
602 /* make sure string is terminated */
603 *utf8 = 0;
604 break;
605
606 case 4:
607 ptr[4] = 0;
608 entry->year = atoi((char *)ptr);
609 break;
610
611 case 5:
612 /* id3v1.1 uses last two bytes of comment field for track
613 number: first must be 0 and second is track num */
614 if (!ptr[0] && ptr[1]) {
615 entry->tracknum = ptr[1];
616 entry->id3version = ID3_VER_1_1;
617 }
618 break;
619
620 case 6:
621 /* genre */
622 entry->genre_string = id3_get_num_genre(ptr[0]);
623 break;
624 }
625 }
626
627 entry->title = entry->id3v1buf[0];
628 entry->artist = entry->id3v1buf[1];
629 entry->album = entry->id3v1buf[2];
630 entry->comment = entry->id3v1buf[3];
631
632 return true;
633}
634
635
636/*
637 * Sets the title of an MP3 entry based on its ID3v2 tag.
638 *
639 * Arguments: file - the MP3 file to scan for a ID3v2 tag
640 * entry - the entry to set the title in
641 *
642 * Returns: true if a title was found and created, else false
643 */
644void setid3v2title(int fd, struct mp3entry *entry)
645{
646 int minframesize;
647 int size;
648 long bufferpos = 0, totframelen, framelen;
649 char header[10];
650 char tmp[4];
651 unsigned char version;
652 char *buffer = entry->id3v2buf;
653 int bytesread = 0;
654 int buffersize = sizeof(entry->id3v2buf);
655 unsigned char global_flags;
656 int flags;
657 int skip;
658 bool global_unsynch = false;
659 bool unsynch = false;
660 int i, j;
661 int rc;
662#if CONFIG_CODEC == SWCODEC
663 bool itunes_gapless = false;
664#endif
665
666 global_ff_found = false;
667
668 /* Bail out if the tag is shorter than 10 bytes */
669 if(entry->id3v2len < 10)
670 return;
671
672 /* Read the ID3 tag version from the header */
673 lseek(fd, 0, SEEK_SET);
674 if(10 != read(fd, header, 10))
675 return;
676
677 /* Get the total ID3 tag size */
678 size = entry->id3v2len - 10;
679
680 version = header[3];
681 switch ( version ) {
682 case 2:
683 version = ID3_VER_2_2;
684 minframesize = 8;
685 break;
686
687 case 3:
688 version = ID3_VER_2_3;
689 minframesize = 12;
690 break;
691
692 case 4:
693 version = ID3_VER_2_4;
694 minframesize = 12;
695 break;
696
697 default:
698 /* unsupported id3 version */
699 return;
700 }
701 entry->id3version = version;
702 entry->tracknum = entry->year = entry->discnum = 0;
703 entry->title = entry->artist = entry->album = NULL; /* FIXME incomplete */
704
705 global_flags = header[5];
706
707 /* Skip the extended header if it is present */
708 if(global_flags & 0x40) {
709 if(version == ID3_VER_2_3) {
710 if(10 != read(fd, header, 10))
711 return;
712 /* The 2.3 extended header size doesn't include the header size
713 field itself. Also, it is not unsynched. */
714 framelen =
715 bytes2int(header[0], header[1], header[2], header[3]) + 4;
716
717 /* Skip the rest of the header */
718 lseek(fd, framelen - 10, SEEK_CUR);
719 }
720
721 if(version >= ID3_VER_2_4) {
722 if(4 != read(fd, header, 4))
723 return;
724
725 /* The 2.4 extended header size does include the entire header,
726 so here we can just skip it. This header is unsynched. */
727 framelen = unsync(header[0], header[1],
728 header[2], header[3]);
729
730 lseek(fd, framelen - 4, SEEK_CUR);
731 }
732 }
733
734 /* Is unsynchronization applied? */
735 if(global_flags & 0x80) {
736 global_unsynch = true;
737 }
738
739 /*
740 * We must have at least minframesize bytes left for the
741 * remaining frames to be interesting
742 */
743 while (size >= minframesize && bufferpos < buffersize - 1) {
744 flags = 0;
745
746 /* Read frame header and check length */
747 if(version >= ID3_VER_2_3) {
748 if(global_unsynch && version <= ID3_VER_2_3)
749 rc = read_unsynched(fd, header, 10);
750 else
751 rc = read(fd, header, 10);
752 if(rc != 10)
753 return;
754 /* Adjust for the 10 bytes we read */
755 size -= 10;
756
757 flags = bytes2int(0, 0, header[8], header[9]);
758
759 if (version >= ID3_VER_2_4) {
760 framelen = unsync(header[4], header[5],
761 header[6], header[7]);
762 } else {
763 /* version .3 files don't use synchsafe ints for
764 * size */
765 framelen = bytes2int(header[4], header[5],
766 header[6], header[7]);
767 }
768 } else {
769 if(6 != read(fd, header, 6))
770 return;
771 /* Adjust for the 6 bytes we read */
772 size -= 6;
773
774 framelen = bytes2int(0, header[3], header[4], header[5]);
775 }
776
777 logf("framelen = %ld, flags = 0x%04x", framelen, flags);
778 if(framelen == 0){
779 if (header[0] == 0 && header[1] == 0 && header[2] == 0)
780 return;
781 else
782 continue;
783 }
784
785 unsynch = false;
786
787 if(flags)
788 {
789 skip = 0;
790
791 if (version >= ID3_VER_2_4) {
792 if(flags & 0x0040) { /* Grouping identity */
793 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
794 framelen--;
795 }
796 } else {
797 if(flags & 0x0020) { /* Grouping identity */
798 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
799 framelen--;
800 }
801 }
802
803 if(flags & 0x000c) /* Compression or encryption */
804 {
805 /* Skip it */
806 size -= framelen;
807 lseek(fd, framelen, SEEK_CUR);
808 continue;
809 }
810
811 if(flags & 0x0002) /* Unsynchronization */
812 unsynch = true;
813
814 if (version >= ID3_VER_2_4) {
815 if(flags & 0x0001) { /* Data length indicator */
816 if(4 != read(fd, tmp, 4))
817 return;
818
819 /* We don't need the data length */
820 framelen -= 4;
821 }
822 }
823 }
824
825 if (framelen == 0)
826 continue;
827
828 if (framelen < 0)
829 return;
830
831 /* Keep track of the remaining frame size */
832 totframelen = framelen;
833
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;
838
839 logf("id3v2 frame: %.4s", header);
840
841 /* Check for certain frame headers
842
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.
847
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.
856
857 For tags that the current code does convert to ints, a post
858 processing function will be called via a pointer to function. */
859
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
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;
870 }
871
872 if( !memcmp( header, tr->tag, tr->tag_length ) ) {
873
874 /* found a tag matching one in tagList, and not yet filled */
875 tag = buffer + bufferpos;
876
877 if(global_unsynch && version <= ID3_VER_2_3)
878 bytesread = read_unsynched(fd, tag, framelen);
879 else
880 bytesread = read(fd, tag, framelen);
881
882 if( bytesread != framelen )
883 return;
884
885 size -= bytesread;
886
887 if(unsynch || (global_unsynch && version >= ID3_VER_2_4))
888 bytesread = unsynchronize_frame(tag, bytesread);
889
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
893 */
894
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;
907 }
908
909 offset = 3 + unicode_len(*tag, tag + 4);
910 if(bytesread > offset) {
911 bytesread -= offset;
912 memmove(tag + 1, tag + 1 + offset, bytesread - 1);
913 }
914 }
915
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];
922
923 unicode_munge( tag, utf8buf, &bytesread );
924
925 if(bytesread >= buffersize - bufferpos)
926 bytesread = buffersize - bufferpos - 1;
927
928 for (j = 0; j < bytesread; j++)
929 tag[j] = utf8buf[j];
930
931 /* remove trailing spaces */
932 while ( bytesread > 0 && isspace(tag[bytesread-1]))
933 bytesread--;
934 }
935
936 tag[bytesread] = 0;
937 bufferpos += bytesread + 1;
938
939#if CONFIG_CODEC == SWCODEC
940 /* parse the tag if it contains iTunes gapless info */
941 if (itunes_gapless)
942 {
943 itunes_gapless = false;
944 entry->lead_trim = get_itunes_int32(tag, 1);
945 entry->tail_trim = get_itunes_int32(tag, 2);
946 }
947#endif
948
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?
954 */
955 if (ptag && !*ptag)
956 *ptag = tag;
957
958 if( tr->ppFunc )
959 bufferpos = tr->ppFunc(entry, tag, bufferpos);
960
961 /* Seek to the next frame */
962 if(framelen < totframelen)
963 lseek(fd, totframelen - framelen, SEEK_CUR);
964 break;
965 }
966 }
967
968 if( i == TAGLIST_SIZE ) {
969 /* no tag in tagList was found, or it was a repeat.
970 skip it using the total size */
971
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;
978 }
979 }
980 }
981}
982
983/*
984 * Calculates the size of the ID3v2 tag.
985 *
986 * Arguments: file - the file to search for a tag.
987 *
988 * Returns: the size of the tag or 0 if none was found
989 */
990int getid3v2len(int fd)
991{
992 char buf[6];
993 int offset;
994
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;
1000
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;
1007
1008 logf("ID3V2 Length: 0x%x", offset);
1009 return offset;
1010}
1011
1012#ifdef DEBUG_STANDALONE
1013
1014char *secs2str(int ms)
1015{
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;
1021}
1022
1023int main(int argc, char **argv)
1024{
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;
1032 }
1033
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);
1060 }
1061
1062 return 0;
1063}
1064
1065#endif