oma: PCM support
[libav.git] / libavformat / oma.c
1 /*
2 * Sony OpenMG (OMA) demuxer
3 *
4 * Copyright (c) 2008 Maxim Poliakovski
5 * 2008 Benjamin Larsson
6 * 2011 David Goldwich
7 *
8 * This file is part of Libav.
9 *
10 * Libav is free software; you can redistribute it and/or
11 * modify it under the terms of the GNU Lesser General Public
12 * License as published by the Free Software Foundation; either
13 * version 2.1 of the License, or (at your option) any later version.
14 *
15 * Libav is distributed in the hope that it will be useful,
16 * but WITHOUT ANY WARRANTY; without even the implied warranty of
17 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 * Lesser General Public License for more details.
19 *
20 * You should have received a copy of the GNU Lesser General Public
21 * License along with Libav; if not, write to the Free Software
22 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
23 */
24
25 /**
26 * @file
27 * This is a demuxer for Sony OpenMG Music files
28 *
29 * Known file extensions: ".oma", "aa3"
30 * The format of such files consists of three parts:
31 * - "ea3" header carrying overall info and metadata. Except for starting with
32 * "ea" instead of "ID", it's an ID3v2 header.
33 * - "EA3" header is a Sony-specific header containing information about
34 * the OpenMG file: codec type (usually ATRAC, can also be MP3 or WMA),
35 * codec specific info (packet size, sample rate, channels and so on)
36 * and DRM related info (file encryption, content id).
37 * - Sound data organized in packets follow the EA3 header
38 * (can be encrypted using the Sony DRM!).
39 *
40 * CODEC SUPPORT: Only ATRAC3 codec is currently supported!
41 */
42
43 #include "avformat.h"
44 #include "internal.h"
45 #include "libavutil/intreadwrite.h"
46 #include "libavutil/des.h"
47 #include "pcm.h"
48 #include "riff.h"
49 #include "id3v2.h"
50
51 #define EA3_HEADER_SIZE 96
52 #define ID3v2_EA3_MAGIC "ea3"
53 #define OMA_ENC_HEADER_SIZE 16
54
55 enum {
56 OMA_CODECID_ATRAC3 = 0,
57 OMA_CODECID_ATRAC3P = 1,
58 OMA_CODECID_MP3 = 3,
59 OMA_CODECID_LPCM = 4,
60 OMA_CODECID_WMA = 5,
61 };
62
63 static const AVCodecTag codec_oma_tags[] = {
64 { CODEC_ID_ATRAC3, OMA_CODECID_ATRAC3 },
65 { CODEC_ID_ATRAC3P, OMA_CODECID_ATRAC3P },
66 { CODEC_ID_MP3, OMA_CODECID_MP3 },
67 { CODEC_ID_PCM_S16BE, OMA_CODECID_LPCM },
68 };
69
70 static const uint64_t leaf_table[] = {
71 0xd79e8283acea4620, 0x7a9762f445afd0d8,
72 0x354d60a60b8c79f1, 0x584e1cde00b07aee,
73 0x1573cd93da7df623, 0x47f98d79620dd535
74 };
75
76 typedef struct OMAContext {
77 uint64_t content_start;
78 int encrypted;
79 uint16_t k_size;
80 uint16_t e_size;
81 uint16_t i_size;
82 uint16_t s_size;
83 uint32_t rid;
84 uint8_t r_val[24];
85 uint8_t n_val[24];
86 uint8_t m_val[8];
87 uint8_t s_val[8];
88 uint8_t sm_val[8];
89 uint8_t e_val[8];
90 uint8_t iv[8];
91 struct AVDES av_des;
92 } OMAContext;
93
94 static void hex_log(AVFormatContext *s, int level, const char *name, const uint8_t *value, int len)
95 {
96 char buf[33];
97 len = FFMIN(len, 16);
98 if (av_log_get_level() < level)
99 return;
100 ff_data_to_hex(buf, value, len, 1);
101 buf[len<<1] = '\0';
102 av_log(s, level, "%s: %s\n", name, buf);
103 }
104
105 static int kset(AVFormatContext *s, const uint8_t *r_val, const uint8_t *n_val, int len)
106 {
107 OMAContext *oc = s->priv_data;
108
109 if (!r_val && !n_val)
110 return -1;
111
112 len = FFMIN(len, 16);
113
114 /* use first 64 bits in the third round again */
115 if (r_val) {
116 if (r_val != oc->r_val) {
117 memset(oc->r_val, 0, 24);
118 memcpy(oc->r_val, r_val, len);
119 }
120 memcpy(&oc->r_val[16], r_val, 8);
121 }
122 if (n_val) {
123 if (n_val != oc->n_val) {
124 memset(oc->n_val, 0, 24);
125 memcpy(oc->n_val, n_val, len);
126 }
127 memcpy(&oc->n_val[16], n_val, 8);
128 }
129
130 return 0;
131 }
132
133 static int rprobe(AVFormatContext *s, uint8_t *enc_header, const uint8_t *r_val)
134 {
135 OMAContext *oc = s->priv_data;
136 unsigned int pos;
137 struct AVDES av_des;
138
139 if (!enc_header || !r_val)
140 return -1;
141
142 /* m_val */
143 av_des_init(&av_des, r_val, 192, 1);
144 av_des_crypt(&av_des, oc->m_val, &enc_header[48], 1, NULL, 1);
145
146 /* s_val */
147 av_des_init(&av_des, oc->m_val, 64, 0);
148 av_des_crypt(&av_des, oc->s_val, NULL, 1, NULL, 0);
149
150 /* sm_val */
151 pos = OMA_ENC_HEADER_SIZE + oc->k_size + oc->e_size;
152 av_des_init(&av_des, oc->s_val, 64, 0);
153 av_des_mac(&av_des, oc->sm_val, &enc_header[pos], (oc->i_size >> 3));
154
155 pos += oc->i_size;
156
157 return memcmp(&enc_header[pos], oc->sm_val, 8) ? -1 : 0;
158 }
159
160 static int nprobe(AVFormatContext *s, uint8_t *enc_header, const uint8_t *n_val)
161 {
162 OMAContext *oc = s->priv_data;
163 uint32_t pos, taglen, datalen;
164 struct AVDES av_des;
165
166 if (!enc_header || !n_val)
167 return -1;
168
169 pos = OMA_ENC_HEADER_SIZE + oc->k_size;
170 if (!memcmp(&enc_header[pos], "EKB ", 4))
171 pos += 32;
172
173 if (AV_RB32(&enc_header[pos]) != oc->rid)
174 av_log(s, AV_LOG_DEBUG, "Mismatching RID\n");
175
176 taglen = AV_RB32(&enc_header[pos+32]);
177 datalen = AV_RB32(&enc_header[pos+36]) >> 4;
178
179 pos += 44 + taglen;
180
181 av_des_init(&av_des, n_val, 192, 1);
182 while (datalen-- > 0) {
183 av_des_crypt(&av_des, oc->r_val, &enc_header[pos], 2, NULL, 1);
184 kset(s, oc->r_val, NULL, 16);
185 if (!rprobe(s, enc_header, oc->r_val))
186 return 0;
187 pos += 16;
188 }
189
190 return -1;
191 }
192
193 static int decrypt_init(AVFormatContext *s, ID3v2ExtraMeta *em, uint8_t *header)
194 {
195 OMAContext *oc = s->priv_data;
196 ID3v2ExtraMetaGEOB *geob = NULL;
197 uint8_t *gdata;
198
199 oc->encrypted = 1;
200 av_log(s, AV_LOG_INFO, "File is encrypted\n");
201
202 /* find GEOB metadata */
203 while (em) {
204 if (!strcmp(em->tag, "GEOB") &&
205 (geob = em->data) &&
206 (!strcmp(geob->description, "OMG_LSI") ||
207 !strcmp(geob->description, "OMG_BKLSI"))) {
208 break;
209 }
210 em = em->next;
211 }
212 if (!em) {
213 av_log(s, AV_LOG_ERROR, "No encryption header found\n");
214 return -1;
215 }
216
217 if (geob->datasize < 64) {
218 av_log(s, AV_LOG_ERROR, "Invalid GEOB data size: %u\n", geob->datasize);
219 return -1;
220 }
221
222 gdata = geob->data;
223
224 if (AV_RB16(gdata) != 1)
225 av_log(s, AV_LOG_WARNING, "Unknown version in encryption header\n");
226
227 oc->k_size = AV_RB16(&gdata[2]);
228 oc->e_size = AV_RB16(&gdata[4]);
229 oc->i_size = AV_RB16(&gdata[6]);
230 oc->s_size = AV_RB16(&gdata[8]);
231
232 if (memcmp(&gdata[OMA_ENC_HEADER_SIZE], "KEYRING ", 12)) {
233 av_log(s, AV_LOG_ERROR, "Invalid encryption header\n");
234 return -1;
235 }
236 oc->rid = AV_RB32(&gdata[OMA_ENC_HEADER_SIZE + 28]);
237 av_log(s, AV_LOG_DEBUG, "RID: %.8x\n", oc->rid);
238
239 memcpy(oc->iv, &header[0x58], 8);
240 hex_log(s, AV_LOG_DEBUG, "IV", oc->iv, 8);
241
242 hex_log(s, AV_LOG_DEBUG, "CBC-MAC", &gdata[OMA_ENC_HEADER_SIZE+oc->k_size+oc->e_size+oc->i_size], 8);
243
244 if (s->keylen > 0) {
245 kset(s, s->key, s->key, s->keylen);
246 }
247 if (!memcmp(oc->r_val, (const uint8_t[8]){0}, 8) ||
248 rprobe(s, gdata, oc->r_val) < 0 &&
249 nprobe(s, gdata, oc->n_val) < 0) {
250 int i;
251 for (i = 0; i < sizeof(leaf_table); i += 2) {
252 uint8_t buf[16];
253 AV_WL64(buf, leaf_table[i]);
254 AV_WL64(&buf[8], leaf_table[i+1]);
255 kset(s, buf, buf, 16);
256 if (!rprobe(s, gdata, oc->r_val) || !nprobe(s, gdata, oc->n_val))
257 break;
258 }
259 if (i >= sizeof(leaf_table)) {
260 av_log(s, AV_LOG_ERROR, "Invalid key\n");
261 return -1;
262 }
263 }
264
265 /* e_val */
266 av_des_init(&oc->av_des, oc->m_val, 64, 0);
267 av_des_crypt(&oc->av_des, oc->e_val, &gdata[OMA_ENC_HEADER_SIZE + 40], 1, NULL, 0);
268 hex_log(s, AV_LOG_DEBUG, "EK", oc->e_val, 8);
269
270 /* init e_val */
271 av_des_init(&oc->av_des, oc->e_val, 64, 1);
272
273 return 0;
274 }
275
276 static int oma_read_header(AVFormatContext *s,
277 AVFormatParameters *ap)
278 {
279 static const uint16_t srate_tab[6] = {320,441,480,882,960,0};
280 int ret, framesize, jsflag, samplerate;
281 uint32_t codec_params;
282 int16_t eid;
283 uint8_t buf[EA3_HEADER_SIZE];
284 uint8_t *edata;
285 AVStream *st;
286 ID3v2ExtraMeta *extra_meta = NULL;
287 OMAContext *oc = s->priv_data;
288
289 ff_id3v2_read_all(s, ID3v2_EA3_MAGIC, &extra_meta);
290 ret = avio_read(s->pb, buf, EA3_HEADER_SIZE);
291 if (ret < EA3_HEADER_SIZE)
292 return -1;
293
294 if (memcmp(buf, ((const uint8_t[]){'E', 'A', '3'}),3) || buf[4] != 0 || buf[5] != EA3_HEADER_SIZE) {
295 av_log(s, AV_LOG_ERROR, "Couldn't find the EA3 header !\n");
296 return -1;
297 }
298
299 oc->content_start = avio_tell(s->pb);
300
301 /* encrypted file */
302 eid = AV_RB16(&buf[6]);
303 if (eid != -1 && eid != -128 && decrypt_init(s, extra_meta, buf) < 0) {
304 ff_id3v2_free_extra_meta(&extra_meta);
305 return -1;
306 }
307
308 ff_id3v2_free_extra_meta(&extra_meta);
309
310 codec_params = AV_RB24(&buf[33]);
311
312 st = avformat_new_stream(s, NULL);
313 if (!st)
314 return AVERROR(ENOMEM);
315
316 st->start_time = 0;
317 st->codec->codec_type = AVMEDIA_TYPE_AUDIO;
318 st->codec->codec_tag = buf[32];
319 st->codec->codec_id = ff_codec_get_id(codec_oma_tags, st->codec->codec_tag);
320
321 switch (buf[32]) {
322 case OMA_CODECID_ATRAC3:
323 samplerate = srate_tab[(codec_params >> 13) & 7]*100;
324 if (samplerate != 44100)
325 av_log_ask_for_sample(s, "Unsupported sample rate: %d\n",
326 samplerate);
327
328 framesize = (codec_params & 0x3FF) * 8;
329 jsflag = (codec_params >> 17) & 1; /* get stereo coding mode, 1 for joint-stereo */
330 st->codec->channels = 2;
331 st->codec->sample_rate = samplerate;
332 st->codec->bit_rate = st->codec->sample_rate * framesize * 8 / 1024;
333
334 /* fake the atrac3 extradata (wav format, makes stream copy to wav work) */
335 st->codec->extradata_size = 14;
336 edata = av_mallocz(14 + FF_INPUT_BUFFER_PADDING_SIZE);
337 if (!edata)
338 return AVERROR(ENOMEM);
339
340 st->codec->extradata = edata;
341 AV_WL16(&edata[0], 1); // always 1
342 AV_WL32(&edata[2], samplerate); // samples rate
343 AV_WL16(&edata[6], jsflag); // coding mode
344 AV_WL16(&edata[8], jsflag); // coding mode
345 AV_WL16(&edata[10], 1); // always 1
346 // AV_WL16(&edata[12], 0); // always 0
347
348 avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
349 break;
350 case OMA_CODECID_ATRAC3P:
351 st->codec->channels = (codec_params >> 10) & 7;
352 framesize = ((codec_params & 0x3FF) * 8) + 8;
353 st->codec->sample_rate = srate_tab[(codec_params >> 13) & 7]*100;
354 st->codec->bit_rate = st->codec->sample_rate * framesize * 8 / 1024;
355 avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
356 av_log(s, AV_LOG_ERROR, "Unsupported codec ATRAC3+!\n");
357 break;
358 case OMA_CODECID_MP3:
359 st->need_parsing = AVSTREAM_PARSE_FULL;
360 framesize = 1024;
361 break;
362 case OMA_CODECID_LPCM:
363 /* PCM 44.1 kHz 16 bit stereo big-endian */
364 st->codec->channels = 2;
365 st->codec->sample_rate = 44100;
366 framesize = 1024;
367 /* bit rate = sample rate x PCM block align (= 4) x 8 */
368 st->codec->bit_rate = st->codec->sample_rate * 32;
369 st->codec->bits_per_coded_sample = av_get_bits_per_sample(st->codec->codec_id);
370 avpriv_set_pts_info(st, 64, 1, st->codec->sample_rate);
371 break;
372 default:
373 av_log(s, AV_LOG_ERROR, "Unsupported codec %d!\n",buf[32]);
374 return -1;
375 }
376
377 st->codec->block_align = framesize;
378
379 return 0;
380 }
381
382
383 static int oma_read_packet(AVFormatContext *s, AVPacket *pkt)
384 {
385 OMAContext *oc = s->priv_data;
386 int packet_size = s->streams[0]->codec->block_align;
387 int ret = av_get_packet(s->pb, pkt, packet_size);
388
389 if (ret <= 0)
390 return AVERROR(EIO);
391
392 pkt->stream_index = 0;
393
394 if (oc->encrypted) {
395 /* previous unencrypted block saved in IV for the next packet (CBC mode) */
396 av_des_crypt(&oc->av_des, pkt->data, pkt->data, (packet_size >> 3), oc->iv, 1);
397 }
398
399 return ret;
400 }
401
402 static int oma_read_probe(AVProbeData *p)
403 {
404 const uint8_t *buf;
405 unsigned tag_len = 0;
406
407 buf = p->buf;
408
409 if (p->buf_size < ID3v2_HEADER_SIZE ||
410 !ff_id3v2_match(buf, ID3v2_EA3_MAGIC) ||
411 buf[3] != 3 || // version must be 3
412 buf[4]) // flags byte zero
413 return 0;
414
415 tag_len = ff_id3v2_tag_len(buf);
416
417 /* This check cannot overflow as tag_len has at most 28 bits */
418 if (p->buf_size < tag_len + 5)
419 /* EA3 header comes late, might be outside of the probe buffer */
420 return AVPROBE_SCORE_MAX / 2;
421
422 buf += tag_len;
423
424 if (!memcmp(buf, "EA3", 3) && !buf[4] && buf[5] == EA3_HEADER_SIZE)
425 return AVPROBE_SCORE_MAX;
426 else
427 return 0;
428 }
429
430 static int oma_read_seek(struct AVFormatContext *s, int stream_index, int64_t timestamp, int flags)
431 {
432 OMAContext *oc = s->priv_data;
433
434 pcm_read_seek(s, stream_index, timestamp, flags);
435
436 if (oc->encrypted) {
437 /* readjust IV for CBC */
438 int64_t pos = avio_tell(s->pb);
439 if (pos < oc->content_start)
440 memset(oc->iv, 0, 8);
441 else {
442 if (avio_seek(s->pb, -8, SEEK_CUR) < 0 || avio_read(s->pb, oc->iv, 8) < 8) {
443 memset(oc->iv, 0, 8);
444 return -1;
445 }
446 }
447 }
448
449 return 0;
450 }
451
452 AVInputFormat ff_oma_demuxer = {
453 .name = "oma",
454 .long_name = NULL_IF_CONFIG_SMALL("Sony OpenMG audio"),
455 .priv_data_size = sizeof(OMAContext),
456 .read_probe = oma_read_probe,
457 .read_header = oma_read_header,
458 .read_packet = oma_read_packet,
459 .read_seek = oma_read_seek,
460 .flags = AVFMT_GENERIC_INDEX,
461 .extensions = "oma,omg,aa3",
462 .codec_tag = (const AVCodecTag* const []){codec_oma_tags, 0},
463 };
464