Fix FS#11956. Call parse_replaygain() with correct parameters in ID3 tag parsing.
[kugel-rb.git] / apps / metadata / id3tags.c
blob3cbebc12e3fd087ff224e976a1c2218a15a9a9bd
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 /* Use bufferpos to hold current position in entry->id3v2buf. */
270 bufferpos = tag - entry->id3v2buf;
272 if(entry->id3version >= ID3_VER_2_4) {
273 /* In version 2.4 and up, there are no parentheses, and the genre frame
274 is a list of strings, either numbers or text. */
276 /* Is it a number? */
277 if(isdigit(tag[0])) {
278 entry->genre_string = id3_get_num_genre(atoi( tag ));
279 return bufferpos;
280 } else {
281 entry->genre_string = tag;
282 return bufferpos + strlen(tag) + 1;
284 } else {
285 if( tag[0] == '(' && tag[1] != '(' ) {
286 entry->genre_string = id3_get_num_genre(atoi( tag + 1 ));
287 return bufferpos;
289 else {
290 entry->genre_string = tag;
291 return bufferpos + strlen(tag) + 1;
296 #ifdef HAVE_ALBUMART
297 /* parse embed albumart */
298 static int parsealbumart( struct mp3entry* entry, char* tag, int bufferpos )
300 entry->embed_albumart = false;
302 /* we currently don't support unsynchronizing albumart */
303 if (entry->albumart.type == AA_TYPE_UNSYNC)
304 return bufferpos;
306 entry->albumart.type = AA_TYPE_UNKNOWN;
308 char *start = tag;
309 /* skip text encoding */
310 tag += 1;
312 if (memcmp(tag, "image/", 6) == 0)
314 /* ID3 v2.3+ */
315 tag += 6;
316 if (strcmp(tag, "jpeg") == 0)
318 entry->albumart.type = AA_TYPE_JPG;
319 tag += 5;
321 else if (strcmp(tag, "png") == 0)
323 entry->albumart.type = AA_TYPE_PNG;
324 tag += 4;
327 else
329 /* ID3 v2.2 */
330 if (memcmp(tag, "JPG", 3) == 0)
331 entry->albumart.type = AA_TYPE_JPG;
332 else if (memcmp(tag, "PNG", 3) == 0)
333 entry->albumart.type = AA_TYPE_PNG;
334 tag += 3;
337 if (entry->albumart.type != AA_TYPE_UNKNOWN)
339 /* skip picture type */
340 tag += 1;
341 /* skip description */
342 tag = strchr(tag, '\0') + 1;
343 /* fixup offset&size for image data */
344 entry->albumart.pos += tag - start;
345 entry->albumart.size -= tag - start;
346 entry->embed_albumart = true;
348 /* return bufferpos as we didn't store anything in id3v2buf */
349 return bufferpos;
351 #endif
353 /* parse user defined text, looking for album artist and replaygain
354 * information.
356 static int parseuser( struct mp3entry* entry, char* tag, int bufferpos )
358 char* value = NULL;
359 int desc_len = strlen(tag);
360 int length = 0;
362 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos) {
363 /* At least part of the value was read, so we can safely try to
364 * parse it */
365 value = tag + desc_len + 1;
367 if (!strcasecmp(tag, "ALBUM ARTIST")) {
368 length = strlen(value) + 1;
369 strlcpy(tag, value, length);
370 entry->albumartist = tag;
371 #if CONFIG_CODEC == SWCODEC
372 } else {
373 /* Calculate residual buffer size in bytes which can be used by
374 * parse_replaygain() to save the string representation of
375 * replaygain data.*/
376 length = sizeof(entry->id3v2buf) - (tag - entry->id3v2buf);
378 /* Call parse_replaygain(), returns length in bytes used by the
379 * string representation of replaygain data. */
380 length = parse_replaygain(tag, value, entry, tag, length);
381 #endif
385 return tag - entry->id3v2buf + length;
388 #if CONFIG_CODEC == SWCODEC
389 /* parse RVA2 binary data and convert to replaygain information. */
390 static int parserva2( struct mp3entry* entry, char* tag, int bufferpos )
392 int desc_len = strlen(tag);
393 int start_pos = tag - entry->id3v2buf;
394 int end_pos = start_pos + desc_len + 5;
395 int value_len = 0;
396 unsigned char* value = tag + desc_len + 1;
398 /* Only parse RVA2 replaygain tags if tag version == 2.4 and channel
399 * type is master volume.
401 if (entry->id3version == ID3_VER_2_4 && end_pos < bufferpos
402 && *value++ == 1) {
403 long gain = 0;
404 long peak = 0;
405 long peakbits;
406 long peakbytes;
407 bool album = false;
409 /* The RVA2 specification is unclear on some things (id string and
410 * peak volume), but this matches how Quod Libet use them.
413 gain = (int16_t) ((value[0] << 8) | value[1]);
414 value += 2;
415 peakbits = *value++;
416 peakbytes = (peakbits + 7) / 8;
418 /* Only use the topmost 24 bits for peak volume */
419 if (peakbytes > 3) {
420 peakbytes = 3;
423 /* Make sure the peak bits were read */
424 if (end_pos + peakbytes < bufferpos) {
425 long shift = ((8 - (peakbits & 7)) & 7) + (3 - peakbytes) * 8;
427 for ( ; peakbytes; peakbytes--) {
428 peak <<= 8;
429 peak += *value++;
432 peak <<= shift;
434 if (peakbits > 24) {
435 peak += *value >> (8 - shift);
439 if (strcasecmp(tag, "album") == 0) {
440 album = true;
441 } else if (strcasecmp(tag, "track") != 0) {
442 /* Only accept non-track values if we don't have any previous
443 * value.
445 if (entry->track_gain != 0) {
446 return start_pos;
450 value_len = parse_replaygain_int(album, gain, peak * 2, entry,
451 tag, sizeof(entry->id3v2buf) - start_pos);
454 return start_pos + value_len;
456 #endif
458 static int parsembtid( struct mp3entry* entry, char* tag, int bufferpos )
460 char* value = NULL;
461 int desc_len = strlen(tag);
462 /*DEBUGF("MBID len: %d\n", desc_len);*/
463 /* Musicbrainz track IDs are always 36 chars long */
464 const size_t mbtid_len = 36;
466 if ((tag - entry->id3v2buf + desc_len + 2) < bufferpos)
468 value = tag + desc_len + 1;
470 if (strcasecmp(tag, "http://musicbrainz.org") == 0)
472 if (mbtid_len == strlen(value))
474 entry->mb_track_id = value;
475 return bufferpos + mbtid_len + 1;
480 return bufferpos;
483 static const struct tag_resolver taglist[] = {
484 { "TPE1", 4, offsetof(struct mp3entry, artist), NULL, false },
485 { "TP1", 3, offsetof(struct mp3entry, artist), NULL, false },
486 { "TIT2", 4, offsetof(struct mp3entry, title), NULL, false },
487 { "TT2", 3, offsetof(struct mp3entry, title), NULL, false },
488 { "TALB", 4, offsetof(struct mp3entry, album), NULL, false },
489 { "TAL", 3, offsetof(struct mp3entry, album), NULL, false },
490 { "TRK", 3, offsetof(struct mp3entry, track_string), &parsetracknum, false },
491 { "TPOS", 4, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
492 { "TPA", 3, offsetof(struct mp3entry, disc_string), &parsediscnum, false },
493 { "TRCK", 4, offsetof(struct mp3entry, track_string), &parsetracknum, false },
494 { "TDRC", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
495 { "TYER", 4, offsetof(struct mp3entry, year_string), &parseyearnum, false },
496 { "TYE", 3, offsetof(struct mp3entry, year_string), &parseyearnum, false },
497 { "TCOM", 4, offsetof(struct mp3entry, composer), NULL, false },
498 { "TCM", 3, offsetof(struct mp3entry, composer), NULL, false },
499 { "TPE2", 4, offsetof(struct mp3entry, albumartist), NULL, false },
500 { "TP2", 3, offsetof(struct mp3entry, albumartist), NULL, false },
501 { "TIT1", 4, offsetof(struct mp3entry, grouping), NULL, false },
502 { "TT1", 3, offsetof(struct mp3entry, grouping), NULL, false },
503 { "COMM", 4, offsetof(struct mp3entry, comment), NULL, false },
504 { "COM", 3, offsetof(struct mp3entry, comment), NULL, false },
505 { "TCON", 4, offsetof(struct mp3entry, genre_string), &parsegenre, false },
506 { "TCO", 3, offsetof(struct mp3entry, genre_string), &parsegenre, false },
507 #ifdef HAVE_ALBUMART
508 { "APIC", 4, 0, &parsealbumart, true },
509 { "PIC", 3, 0, &parsealbumart, true },
510 #endif
511 { "TXXX", 4, 0, &parseuser, false },
512 #if CONFIG_CODEC == SWCODEC
513 { "RVA2", 4, 0, &parserva2, true },
514 #endif
515 { "UFID", 4, 0, &parsembtid, false },
518 #define TAGLIST_SIZE ((int)ARRAYLEN(taglist))
520 /* Get the length of an ID3 string in the given encoding. Returns the length
521 * in bytes, including end nil, or -1 if the encoding is unknown.
523 static int unicode_len(char encoding, const void* string)
525 int len = 0;
527 if (encoding == 0x01 || encoding == 0x02) {
528 char first;
529 const char *s = string;
530 /* string might be unaligned, so using short* can crash on ARM and SH1 */
531 do {
532 first = *s++;
533 } while ((first | *s++) != 0);
535 len = s - (const char*) string;
536 } else {
537 len = strlen((char*) string) + 1;
540 return len;
543 /* Checks to see if the passed in string is a 16-bit wide Unicode v2
544 string. If it is, we convert it to a UTF-8 string. If it's not unicode,
545 we convert from the default codepage */
546 static int unicode_munge(char* string, char* utf8buf, int *len) {
547 long tmp;
548 bool le = false;
549 int i = 0;
550 unsigned char *str = (unsigned char *)string;
551 int templen = 0;
552 unsigned char* utf8 = (unsigned char *)utf8buf;
554 switch (str[0]) {
555 case 0x00: /* Type 0x00 is ordinary ISO 8859-1 */
556 str++;
557 (*len)--;
558 utf8 = iso_decode(str, utf8, -1, *len);
559 *utf8 = 0;
560 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
561 break;
563 case 0x01: /* Unicode with or without BOM */
564 case 0x02:
565 (*len)--;
566 str++;
568 /* Handle frames with more than one string
569 (needed for TXXX frames).*/
570 do {
571 tmp = bytes2int(0, 0, str[0], str[1]);
573 /* Now check if there is a BOM
574 (zero-width non-breaking space, 0xfeff)
575 and if it is in little or big endian format */
576 if(tmp == 0xfffe) { /* Little endian? */
577 le = true;
578 str += 2;
579 (*len)-=2;
580 } else if(tmp == 0xfeff) { /* Big endian? */
581 str += 2;
582 (*len)-=2;
583 } else
584 /* If there is no BOM (which is a specification violation),
585 let's try to guess it. If one of the bytes is 0x00, it is
586 probably the most significant one. */
587 if(str[1] == 0)
588 le = true;
590 do {
591 if(le)
592 utf8 = utf16LEdecode(str, utf8, 1);
593 else
594 utf8 = utf16BEdecode(str, utf8, 1);
596 str+=2;
597 i += 2;
598 } while((str[0] || str[1]) && (i < *len));
600 *utf8++ = 0; /* Terminate the string */
601 templen += (strlen(&utf8buf[templen]) + 1);
602 str += 2;
603 i+=2;
604 } while(i < *len);
605 *len = templen - 1;
606 break;
608 case 0x03: /* UTF-8 encoded string */
609 for(i=0; i < *len; i++)
610 utf8[i] = str[i+1];
611 (*len)--;
612 break;
614 default: /* Plain old string */
615 utf8 = iso_decode(str, utf8, -1, *len);
616 *utf8 = 0;
617 *len = (unsigned long)utf8 - (unsigned long)utf8buf;
618 break;
620 return 0;
624 * Sets the title of an MP3 entry based on its ID3v1 tag.
626 * Arguments: file - the MP3 file to scen for a ID3v1 tag
627 * entry - the entry to set the title in
629 * Returns: true if a title was found and created, else false
631 bool setid3v1title(int fd, struct mp3entry *entry)
633 unsigned char buffer[128];
634 static const char offsets[] = {3, 33, 63, 97, 93, 125, 127};
635 int i, j;
636 unsigned char* utf8;
638 if (-1 == lseek(fd, -128, SEEK_END))
639 return false;
641 if (read(fd, buffer, sizeof buffer) != sizeof buffer)
642 return false;
644 if (strncmp((char *)buffer, "TAG", 3))
645 return false;
647 entry->id3v1len = 128;
648 entry->id3version = ID3_VER_1_0;
650 for (i=0; i < (int)sizeof offsets; i++) {
651 unsigned char* ptr = (unsigned char *)buffer + offsets[i];
653 switch(i) {
654 case 0:
655 case 1:
656 case 2:
657 /* kill trailing space in strings */
658 for (j=29; j && (ptr[j]==0 || ptr[j]==' '); j--)
659 ptr[j] = 0;
660 /* convert string to utf8 */
661 utf8 = (unsigned char *)entry->id3v1buf[i];
662 utf8 = iso_decode(ptr, utf8, -1, 30);
663 /* make sure string is terminated */
664 *utf8 = 0;
665 break;
667 case 3:
668 /* kill trailing space in strings */
669 for (j=27; j && (ptr[j]==0 || ptr[j]==' '); j--)
670 ptr[j] = 0;
671 /* convert string to utf8 */
672 utf8 = (unsigned char *)entry->id3v1buf[3];
673 utf8 = iso_decode(ptr, utf8, -1, 28);
674 /* make sure string is terminated */
675 *utf8 = 0;
676 break;
678 case 4:
679 ptr[4] = 0;
680 entry->year = atoi((char *)ptr);
681 break;
683 case 5:
684 /* id3v1.1 uses last two bytes of comment field for track
685 number: first must be 0 and second is track num */
686 if (!ptr[0] && ptr[1]) {
687 entry->tracknum = ptr[1];
688 entry->id3version = ID3_VER_1_1;
690 break;
692 case 6:
693 /* genre */
694 entry->genre_string = id3_get_num_genre(ptr[0]);
695 break;
699 entry->title = entry->id3v1buf[0];
700 entry->artist = entry->id3v1buf[1];
701 entry->album = entry->id3v1buf[2];
702 entry->comment = entry->id3v1buf[3];
704 return true;
709 * Sets the title of an MP3 entry based on its ID3v2 tag.
711 * Arguments: file - the MP3 file to scan for a ID3v2 tag
712 * entry - the entry to set the title in
714 * Returns: true if a title was found and created, else false
716 void setid3v2title(int fd, struct mp3entry *entry)
718 int minframesize;
719 int size;
720 long bufferpos = 0, totframelen, framelen;
721 char header[10];
722 char tmp[4];
723 unsigned char version;
724 char *buffer = entry->id3v2buf;
725 int bytesread = 0;
726 int buffersize = sizeof(entry->id3v2buf);
727 unsigned char global_flags;
728 int flags;
729 bool global_unsynch = false;
730 bool unsynch = false;
731 int i, j;
732 int rc;
733 #if CONFIG_CODEC == SWCODEC
734 bool itunes_gapless = false;
735 #endif
737 global_ff_found = false;
739 /* Bail out if the tag is shorter than 10 bytes */
740 if(entry->id3v2len < 10)
741 return;
743 /* Read the ID3 tag version from the header */
744 lseek(fd, 0, SEEK_SET);
745 if(10 != read(fd, header, 10))
746 return;
748 /* Get the total ID3 tag size */
749 size = entry->id3v2len - 10;
751 version = header[3];
752 switch ( version ) {
753 case 2:
754 version = ID3_VER_2_2;
755 minframesize = 8;
756 break;
758 case 3:
759 version = ID3_VER_2_3;
760 minframesize = 12;
761 break;
763 case 4:
764 version = ID3_VER_2_4;
765 minframesize = 12;
766 break;
768 default:
769 /* unsupported id3 version */
770 return;
772 entry->id3version = version;
773 entry->tracknum = entry->year = entry->discnum = 0;
774 entry->title = entry->artist = entry->album = NULL; /* FIXME incomplete */
776 global_flags = header[5];
778 /* Skip the extended header if it is present */
779 if(global_flags & 0x40) {
780 if(version == ID3_VER_2_3) {
781 if(10 != read(fd, header, 10))
782 return;
783 /* The 2.3 extended header size doesn't include the header size
784 field itself. Also, it is not unsynched. */
785 framelen =
786 bytes2int(header[0], header[1], header[2], header[3]) + 4;
788 /* Skip the rest of the header */
789 lseek(fd, framelen - 10, SEEK_CUR);
792 if(version >= ID3_VER_2_4) {
793 if(4 != read(fd, header, 4))
794 return;
796 /* The 2.4 extended header size does include the entire header,
797 so here we can just skip it. This header is unsynched. */
798 framelen = unsync(header[0], header[1],
799 header[2], header[3]);
801 lseek(fd, framelen - 4, SEEK_CUR);
805 /* Is unsynchronization applied? */
806 if(global_flags & 0x80) {
807 global_unsynch = true;
811 * We must have at least minframesize bytes left for the
812 * remaining frames to be interesting
814 while (size >= minframesize && bufferpos < buffersize - 1) {
815 flags = 0;
817 /* Read frame header and check length */
818 if(version >= ID3_VER_2_3) {
819 if(global_unsynch && version <= ID3_VER_2_3)
820 rc = read_unsynched(fd, header, 10);
821 else
822 rc = read(fd, header, 10);
823 if(rc != 10)
824 return;
825 /* Adjust for the 10 bytes we read */
826 size -= 10;
828 flags = bytes2int(0, 0, header[8], header[9]);
830 if (version >= ID3_VER_2_4) {
831 framelen = unsync(header[4], header[5],
832 header[6], header[7]);
833 } else {
834 /* version .3 files don't use synchsafe ints for
835 * size */
836 framelen = bytes2int(header[4], header[5],
837 header[6], header[7]);
839 } else {
840 if(6 != read(fd, header, 6))
841 return;
842 /* Adjust for the 6 bytes we read */
843 size -= 6;
845 framelen = bytes2int(0, header[3], header[4], header[5]);
848 logf("framelen = %ld, flags = 0x%04x", framelen, flags);
849 if(framelen == 0){
850 if (header[0] == 0 && header[1] == 0 && header[2] == 0)
851 return;
852 else
853 continue;
856 unsynch = false;
858 if(flags)
860 if (version >= ID3_VER_2_4) {
861 if(flags & 0x0040) { /* Grouping identity */
862 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
863 framelen--;
865 } else {
866 if(flags & 0x0020) { /* Grouping identity */
867 lseek(fd, 1, SEEK_CUR); /* Skip 1 byte */
868 framelen--;
872 if(flags & 0x000c) /* Compression or encryption */
874 /* Skip it */
875 size -= framelen;
876 lseek(fd, framelen, SEEK_CUR);
877 continue;
880 if(flags & 0x0002) /* Unsynchronization */
881 unsynch = true;
883 if (version >= ID3_VER_2_4) {
884 if(flags & 0x0001) { /* Data length indicator */
885 if(4 != read(fd, tmp, 4))
886 return;
888 /* We don't need the data length */
889 framelen -= 4;
894 if (framelen == 0)
895 continue;
897 if (framelen < 0)
898 return;
900 /* Keep track of the remaining frame size */
901 totframelen = framelen;
903 /* If the frame is larger than the remaining buffer space we try
904 to read as much as would fit in the buffer */
905 if(framelen >= buffersize - bufferpos)
906 framelen = buffersize - bufferpos - 1;
908 /* Limit the maximum length of an id3 data item to ID3V2_MAX_ITEM_SIZE
909 bytes. This reduces the chance that the available buffer is filled
910 by single metadata items like large comments. */
911 if (ID3V2_MAX_ITEM_SIZE < framelen)
912 framelen = ID3V2_MAX_ITEM_SIZE;
914 logf("id3v2 frame: %.4s", header);
916 /* Check for certain frame headers
918 'size' is the amount of frame bytes remaining. We decrement it by
919 the amount of bytes we read. If we fail to read as many bytes as
920 we expect, we assume that we can't read from this file, and bail
921 out.
923 For each frame. we will iterate over the list of supported tags,
924 and read the tag into entry's buffer. All tags will be kept as
925 strings, for cases where a number won't do, e.g., YEAR: "circa
926 1765", "1790/1977" (composed/performed), "28 Feb 1969" TRACK:
927 "1/12", "1 of 12", GENRE: "Freeform genre name" Text is more
928 flexible, and as the main use of id3 data is to display it,
929 converting it to an int just means reconverting to display it, at a
930 runtime cost.
932 For tags that the current code does convert to ints, a post
933 processing function will be called via a pointer to function. */
935 for (i=0; i<TAGLIST_SIZE; i++) {
936 const struct tag_resolver* tr = &taglist[i];
937 char** ptag = tr->offset ? (char**) (((char*)entry) + tr->offset)
938 : NULL;
939 char* tag;
941 /* Only ID3_VER_2_2 uses frames with three-character names. */
942 if (((version == ID3_VER_2_2) && (tr->tag_length != 3))
943 || ((version > ID3_VER_2_2) && (tr->tag_length != 4))) {
944 continue;
947 if( !memcmp( header, tr->tag, tr->tag_length ) ) {
949 /* found a tag matching one in tagList, and not yet filled */
950 tag = buffer + bufferpos;
952 if(global_unsynch && version <= ID3_VER_2_3)
953 bytesread = read_unsynched(fd, tag, framelen);
954 else
955 bytesread = read(fd, tag, framelen);
957 if( bytesread != framelen )
958 return;
960 size -= bytesread;
962 if(unsynch || (global_unsynch && version >= ID3_VER_2_4))
963 bytesread = unsynchronize_frame(tag, bytesread);
965 /* the COMM frame has a 3 char field to hold an ISO-639-1
966 * language string and an optional short description;
967 * remove them so unicode_munge can work correctly
970 if((tr->tag_length == 4 && !memcmp( header, "COMM", 4)) ||
971 (tr->tag_length == 3 && !memcmp( header, "COM", 3))) {
972 int offset;
973 if(!strncmp(tag+4, "iTun", 4)) {
974 #if CONFIG_CODEC == SWCODEC
975 /* check for iTunes gapless information */
976 if(!strncmp(tag+4, "iTunSMPB", 8))
977 itunes_gapless = true;
978 else
979 #endif
980 /* ignore other with iTunes tags */
981 break;
984 offset = 3 + unicode_len(*tag, tag + 4);
985 if(bytesread > offset) {
986 bytesread -= offset;
987 memmove(tag + 1, tag + 1 + offset, bytesread - 1);
991 /* Attempt to parse Unicode string only if the tag contents
992 aren't binary */
993 if(!tr->binary) {
994 /* UTF-8 could potentially be 3 times larger */
995 /* so we need to create a new buffer */
996 char utf8buf[(3 * bytesread) + 1];
998 unicode_munge( tag, utf8buf, &bytesread );
1000 if(bytesread >= buffersize - bufferpos)
1001 bytesread = buffersize - bufferpos - 1;
1003 for (j = 0; j < bytesread; j++)
1004 tag[j] = utf8buf[j];
1006 /* remove trailing spaces */
1007 while ( bytesread > 0 && isspace(tag[bytesread-1]))
1008 bytesread--;
1011 tag[bytesread] = 0;
1012 bufferpos += bytesread + 1;
1014 #if CONFIG_CODEC == SWCODEC
1015 /* parse the tag if it contains iTunes gapless info */
1016 if (itunes_gapless)
1018 itunes_gapless = false;
1019 entry->lead_trim = get_itunes_int32(tag, 1);
1020 entry->tail_trim = get_itunes_int32(tag, 2);
1022 #endif
1024 /* Note that parser functions sometimes set *ptag to NULL, so
1025 * the "!*ptag" check here doesn't always have the desired
1026 * effect. Should the parser functions (parsegenre in
1027 * particular) be updated to handle the case of being called
1028 * multiple times, or should the "*ptag" check be removed?
1030 if (ptag && !*ptag)
1031 *ptag = tag;
1033 #ifdef HAVE_ALBUMART
1034 /* albumart */
1035 if ((!entry->embed_albumart) &&
1036 ((tr->tag_length == 4 && !memcmp( header, "APIC", 4)) ||
1037 (tr->tag_length == 3 && !memcmp( header, "PIC" , 3))))
1039 if (unsynch || (global_unsynch && version <= ID3_VER_2_3))
1040 entry->albumart.type = AA_TYPE_UNSYNC;
1041 else
1043 entry->albumart.pos = lseek(fd, 0, SEEK_CUR) - framelen;
1044 entry->albumart.size = totframelen;
1045 entry->albumart.type = AA_TYPE_UNKNOWN;
1048 #endif
1049 if( tr->ppFunc )
1050 bufferpos = tr->ppFunc(entry, tag, bufferpos);
1052 /* Seek to the next frame */
1053 if(framelen < totframelen)
1054 lseek(fd, totframelen - framelen, SEEK_CUR);
1055 break;
1059 if( i == TAGLIST_SIZE ) {
1060 /* no tag in tagList was found, or it was a repeat.
1061 skip it using the total size */
1063 if(global_unsynch && version <= ID3_VER_2_3) {
1064 size -= skip_unsynched(fd, totframelen);
1065 } else {
1066 size -= totframelen;
1067 if( lseek(fd, totframelen, SEEK_CUR) == -1 )
1068 return;
1075 * Calculates the size of the ID3v2 tag.
1077 * Arguments: file - the file to search for a tag.
1079 * Returns: the size of the tag or 0 if none was found
1081 int getid3v2len(int fd)
1083 char buf[6];
1084 int offset;
1086 /* Make sure file has a ID3 tag */
1087 if((-1 == lseek(fd, 0, SEEK_SET)) ||
1088 (read(fd, buf, 6) != 6) ||
1089 (strncmp(buf, "ID3", strlen("ID3")) != 0))
1090 offset = 0;
1092 /* Now check what the ID3v2 size field says */
1093 else
1094 if(read(fd, buf, 4) != 4)
1095 offset = 0;
1096 else
1097 offset = unsync(buf[0], buf[1], buf[2], buf[3]) + 10;
1099 logf("ID3V2 Length: 0x%x", offset);
1100 return offset;
1103 #ifdef DEBUG_STANDALONE
1105 char *secs2str(int ms)
1107 static char buffer[32];
1108 int secs = ms/1000;
1109 ms %= 1000;
1110 snprintf(buffer, sizeof(buffer), "%d:%02d.%d", secs/60, secs%60, ms/100);
1111 return buffer;
1114 int main(int argc, char **argv)
1116 int i;
1117 for(i=1; i<argc; i++) {
1118 struct mp3entry mp3;
1119 mp3.album = "Bogus";
1120 if(mp3info(&mp3, argv[i], false)) {
1121 printf("Failed to get %s\n", argv[i]);
1122 return 0;
1125 printf("****** File: %s\n"
1126 " Title: %s\n"
1127 " Artist: %s\n"
1128 " Album: %s\n"
1129 " Genre: %s (%d) \n"
1130 " Composer: %s\n"
1131 " Year: %s (%d)\n"
1132 " Track: %s (%d)\n"
1133 " Length: %s / %d s\n"
1134 " Bitrate: %d\n"
1135 " Frequency: %d\n",
1136 argv[i],
1137 mp3.title?mp3.title:"<blank>",
1138 mp3.artist?mp3.artist:"<blank>",
1139 mp3.album?mp3.album:"<blank>",
1140 mp3.genre_string?mp3.genre_string:"<blank>",
1141 mp3.genre,
1142 mp3.composer?mp3.composer:"<blank>",
1143 mp3.year_string?mp3.year_string:"<blank>",
1144 mp3.year,
1145 mp3.track_string?mp3.track_string:"<blank>",
1146 mp3.tracknum,
1147 secs2str(mp3.length),
1148 mp3.length/1000,
1149 mp3.bitrate,
1150 mp3.frequency);
1153 return 0;
1156 #endif