1 /*
2 * Audible AA demuxer
3 * Copyright (c) 2015 Vesselin Bontchev
4 *
5 * Header parsing is borrowed from https://github.com/jteeuwen/audible project.
6 * Copyright (c) 2001-2014, Jim Teeuwen
7 *
8 * Redistribution and use in source and binary forms, with or without modification,
9 * are permitted provided that the following conditions are met:
10 *
11 * 1. Redistributions of source code must retain the above copyright notice, this
12 * list of conditions and the following disclaimer.
13 *
14 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
15 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
16 * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
17 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR
18 * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
19 * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
20 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
21 * ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
22 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
23 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
24 */
25
26 #include "avformat.h"
27 #include "demux.h"
28 #include "internal.h"
29 #include "libavutil/avstring.h"
30 #include "libavutil/dict.h"
31 #include "libavutil/intreadwrite.h"
32 #include "libavutil/tea.h"
33 #include "libavutil/opt.h"
34
35 #define AA_MAGIC 1469084982 /* this identifies an audible .aa file */
36 #define MAX_TOC_ENTRIES 16
37 #define MAX_DICTIONARY_ENTRIES 128
38 #define TEA_BLOCK_SIZE 8
39 #define CHAPTER_HEADER_SIZE 8
40 #define TIMEPREC 1000
41 #define MP3_FRAME_SIZE 104
42
43 typedef struct AADemuxContext {
44 AVClass *class;
45 uint8_t *aa_fixed_key;
46 int aa_fixed_key_len;
47 int codec_second_size;
48 int current_codec_second_size;
49 int chapter_idx;
50 struct AVTEA *tea_ctx;
51 uint8_t file_key[16];
52 int64_t current_chapter_size;
53 int64_t content_start;
54 int64_t content_end;
55 int seek_offset;
56 } AADemuxContext;
57
get_second_size(char * codec_name)58 static int get_second_size(char *codec_name)
59 {
60 int result = -1;
61
62 if (!strcmp(codec_name, "mp332")) {
63 result = 3982;
64 } else if (!strcmp(codec_name, "acelp16")) {
65 result = 2000;
66 } else if (!strcmp(codec_name, "acelp85")) {
67 result = 1045;
68 }
69
70 return result;
71 }
72
aa_read_header(AVFormatContext * s)73 static int aa_read_header(AVFormatContext *s)
74 {
75 int largest_idx = -1;
76 uint32_t toc_size, npairs, header_seed = 0, start;
77 char codec_name[64] = {0};
78 uint8_t buf[24];
79 int64_t largest_size = -1, current_size = -1, chapter_pos;
80 struct toc_entry {
81 uint32_t offset;
82 uint32_t size;
83 } TOC[MAX_TOC_ENTRIES];
84 uint8_t header_key[16] = {0};
85 AADemuxContext *c = s->priv_data;
86 char file_key[2 * sizeof(c->file_key) + 1];
87 AVIOContext *pb = s->pb;
88 AVStream *st;
89 FFStream *sti;
90 int ret;
91
92 /* parse .aa header */
93 avio_skip(pb, 4); // file size
94 avio_skip(pb, 4); // magic string
95 toc_size = avio_rb32(pb); // TOC size
96 avio_skip(pb, 4); // unidentified integer
97 if (toc_size > MAX_TOC_ENTRIES || toc_size < 2)
98 return AVERROR_INVALIDDATA;
99 for (uint32_t i = 0; i < toc_size; i++) { // read TOC
100 avio_skip(pb, 4); // TOC entry index
101 TOC[i].offset = avio_rb32(pb); // block offset
102 TOC[i].size = avio_rb32(pb); // block size
103 }
104 avio_skip(pb, 24); // header termination block (ignored)
105 npairs = avio_rb32(pb); // read dictionary entries
106 if (npairs > MAX_DICTIONARY_ENTRIES)
107 return AVERROR_INVALIDDATA;
108 for (uint32_t i = 0; i < npairs; i++) {
109 char key[128], val[128];
110 uint32_t nkey, nval;
111
112 avio_skip(pb, 1); // unidentified integer
113 nkey = avio_rb32(pb); // key string length
114 nval = avio_rb32(pb); // value string length
115 avio_get_str(pb, nkey, key, sizeof(key));
116 avio_get_str(pb, nval, val, sizeof(val));
117 if (!strcmp(key, "codec")) {
118 av_log(s, AV_LOG_DEBUG, "Codec is <%s>\n", val);
119 av_strlcpy(codec_name, val, sizeof(codec_name));
120 } else if (!strcmp(key, "HeaderSeed")) {
121 av_log(s, AV_LOG_DEBUG, "HeaderSeed is <%s>\n", val);
122 header_seed = atoi(val);
123 } else if (!strcmp(key, "HeaderKey")) { // this looks like "1234567890 1234567890 1234567890 1234567890"
124 uint32_t header_key_part[4];
125 av_log(s, AV_LOG_DEBUG, "HeaderKey is <%s>\n", val);
126
127 ret = sscanf(val, "%"SCNu32"%"SCNu32"%"SCNu32"%"SCNu32,
128 &header_key_part[0], &header_key_part[1], &header_key_part[2], &header_key_part[3]);
129 if (ret != 4)
130 return AVERROR_INVALIDDATA;
131
132 for (int idx = 0; idx < 4; idx++)
133 AV_WB32(&header_key[idx * 4], header_key_part[idx]); // convert each part to BE!
134 ff_data_to_hex(key, header_key, sizeof(header_key), 1);
135 av_log(s, AV_LOG_DEBUG, "Processed HeaderKey is %s\n", key);
136 } else {
137 av_dict_set(&s->metadata, key, val, 0);
138 }
139 }
140
141 /* verify fixed key */
142 if (c->aa_fixed_key_len != 16) {
143 av_log(s, AV_LOG_ERROR, "aa_fixed_key value needs to be 16 bytes!\n");
144 return AVERROR(EINVAL);
145 }
146
147 /* verify codec */
148 if ((c->codec_second_size = get_second_size(codec_name)) == -1) {
149 av_log(s, AV_LOG_ERROR, "unknown codec <%s>!\n", codec_name);
150 return AVERROR(EINVAL);
151 }
152
153 /* decryption key derivation */
154 c->tea_ctx = av_tea_alloc();
155 if (!c->tea_ctx)
156 return AVERROR(ENOMEM);
157 av_tea_init(c->tea_ctx, c->aa_fixed_key, 16);
158 for (int i = 0; i < 6; i++)
159 AV_WB32(buf + 4 * i, header_seed + i);
160 av_tea_crypt(c->tea_ctx, buf, buf, 3, NULL, 0);
161 AV_WN64(c->file_key, AV_RN64(buf + 2) ^ AV_RN64(header_key));
162 AV_WN64(c->file_key + 8, AV_RN64(buf + 10) ^ AV_RN64(header_key + 8));
163 ff_data_to_hex(file_key, c->file_key, sizeof(c->file_key), 1);
164 av_log(s, AV_LOG_DEBUG, "File key is %s\n", file_key);
165 av_tea_init(c->tea_ctx, c->file_key, 16);
166
167 /* decoder setup */
168 st = avformat_new_stream(s, NULL);
169 if (!st)
170 return AVERROR(ENOMEM);
171 sti = ffstream(st);
172 st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
173 if (!strcmp(codec_name, "mp332")) {
174 st->codecpar->codec_id = AV_CODEC_ID_MP3;
175 st->codecpar->sample_rate = 22050;
176 sti->need_parsing = AVSTREAM_PARSE_FULL_RAW;
177 avpriv_set_pts_info(st, 64, 8, 32000 * TIMEPREC);
178 // encoded audio frame is MP3_FRAME_SIZE bytes (+1 with padding, unlikely)
179 } else if (!strcmp(codec_name, "acelp85")) {
180 st->codecpar->codec_id = AV_CODEC_ID_SIPR;
181 st->codecpar->block_align = 19;
182 st->codecpar->ch_layout.nb_channels = 1;
183 st->codecpar->sample_rate = 8500;
184 st->codecpar->bit_rate = 8500;
185 sti->need_parsing = AVSTREAM_PARSE_FULL_RAW;
186 avpriv_set_pts_info(st, 64, 8, 8500 * TIMEPREC);
187 } else if (!strcmp(codec_name, "acelp16")) {
188 st->codecpar->codec_id = AV_CODEC_ID_SIPR;
189 st->codecpar->block_align = 20;
190 st->codecpar->ch_layout.nb_channels = 1;
191 st->codecpar->sample_rate = 16000;
192 st->codecpar->bit_rate = 16000;
193 sti->need_parsing = AVSTREAM_PARSE_FULL_RAW;
194 avpriv_set_pts_info(st, 64, 8, 16000 * TIMEPREC);
195 }
196
197 /* determine, and jump to audio start offset */
198 for (uint32_t i = 1; i < toc_size; i++) { // skip the first entry!
199 current_size = TOC[i].size;
200 if (current_size > largest_size) {
201 largest_idx = i;
202 largest_size = current_size;
203 }
204 }
205 start = TOC[largest_idx].offset;
206 avio_seek(pb, start, SEEK_SET);
207
208 // extract chapter positions. since all formats have constant bit rate, use it
209 // as time base in bytes/s, for easy stream position <-> timestamp conversion
210 st->start_time = 0;
211 c->content_start = start;
212 c->content_end = start + largest_size;
213
214 while ((chapter_pos = avio_tell(pb)) >= 0 && chapter_pos < c->content_end) {
215 unsigned chapter_idx = s->nb_chapters;
216 uint32_t chapter_size = avio_rb32(pb);
217 if (chapter_size == 0 || avio_feof(pb))
218 break;
219 chapter_pos -= start + CHAPTER_HEADER_SIZE * chapter_idx;
220 avio_skip(pb, 4 + chapter_size);
221 if (!avpriv_new_chapter(s, chapter_idx, st->time_base,
222 chapter_pos * TIMEPREC,
223 (chapter_pos + chapter_size) * TIMEPREC, NULL))
224 return AVERROR(ENOMEM);
225 }
226
227 st->duration = (largest_size - CHAPTER_HEADER_SIZE * s->nb_chapters) * TIMEPREC;
228
229 avpriv_update_cur_dts(s, st, 0);
230 avio_seek(pb, start, SEEK_SET);
231 c->current_chapter_size = 0;
232 c->seek_offset = 0;
233
234 return 0;
235 }
236
aa_read_packet(AVFormatContext * s,AVPacket * pkt)237 static int aa_read_packet(AVFormatContext *s, AVPacket *pkt)
238 {
239 int ret;
240 AADemuxContext *c = s->priv_data;
241 uint64_t pos = avio_tell(s->pb);
242
243 // are we at the end of the audio content?
244 if (pos >= c->content_end) {
245 return AVERROR_EOF;
246 }
247
248 // are we at the start of a chapter?
249 if (c->current_chapter_size == 0) {
250 c->current_chapter_size = avio_rb32(s->pb);
251 if (c->current_chapter_size == 0) {
252 return AVERROR_EOF;
253 }
254 av_log(s, AV_LOG_DEBUG, "Chapter %d (%" PRId64 " bytes)\n", c->chapter_idx, c->current_chapter_size);
255 c->chapter_idx = c->chapter_idx + 1;
256 avio_skip(s->pb, 4); // data start offset
257 c->current_codec_second_size = c->codec_second_size;
258 }
259
260 // is this the last block in this chapter?
261 if (c->current_chapter_size / c->current_codec_second_size == 0) {
262 c->current_codec_second_size = c->current_chapter_size % c->current_codec_second_size;
263 }
264
265 ret = av_get_packet(s->pb, pkt, c->current_codec_second_size);
266 if (ret != c->current_codec_second_size)
267 return AVERROR_EOF;
268
269 // decrypt c->current_codec_second_size bytes in blocks of TEA_BLOCK_SIZE
270 // trailing bytes are left unencrypted!
271 av_tea_crypt(c->tea_ctx, pkt->data, pkt->data,
272 c->current_codec_second_size / TEA_BLOCK_SIZE, NULL, 1);
273
274 // update state
275 c->current_chapter_size = c->current_chapter_size - c->current_codec_second_size;
276 if (c->current_chapter_size <= 0)
277 c->current_chapter_size = 0;
278
279 if (c->seek_offset > c->current_codec_second_size)
280 c->seek_offset = 0; // ignore wrong estimate
281 pkt->data += c->seek_offset;
282 pkt->size -= c->seek_offset;
283 c->seek_offset = 0;
284
285 return 0;
286 }
287
aa_read_seek(AVFormatContext * s,int stream_index,int64_t timestamp,int flags)288 static int aa_read_seek(AVFormatContext *s,
289 int stream_index, int64_t timestamp, int flags)
290 {
291 AADemuxContext *c = s->priv_data;
292 AVChapter *ch;
293 int64_t chapter_pos, chapter_start, chapter_size;
294 int chapter_idx = 0;
295
296 // find chapter containing seek timestamp
297 if (timestamp < 0)
298 timestamp = 0;
299
300 while (chapter_idx < s->nb_chapters && timestamp >= s->chapters[chapter_idx]->end) {
301 ++chapter_idx;
302 }
303
304 if (chapter_idx >= s->nb_chapters) {
305 chapter_idx = s->nb_chapters - 1;
306 if (chapter_idx < 0) return -1; // there is no chapter.
307 timestamp = s->chapters[chapter_idx]->end;
308 }
309
310 ch = s->chapters[chapter_idx];
311
312 // sync by clamping timestamp to nearest valid block position in its chapter
313 chapter_size = ch->end / TIMEPREC - ch->start / TIMEPREC;
314 chapter_pos = av_rescale_rnd((timestamp - ch->start) / TIMEPREC,
315 1, c->codec_second_size,
316 (flags & AVSEEK_FLAG_BACKWARD) ? AV_ROUND_DOWN : AV_ROUND_UP)
317 * c->codec_second_size;
318 if (chapter_pos >= chapter_size)
319 chapter_pos = chapter_size;
320 chapter_start = c->content_start + (ch->start / TIMEPREC) + CHAPTER_HEADER_SIZE * (1 + chapter_idx);
321
322 // reinit read state
323 avio_seek(s->pb, chapter_start + chapter_pos, SEEK_SET);
324 c->current_codec_second_size = c->codec_second_size;
325 c->current_chapter_size = chapter_size - chapter_pos;
326 c->chapter_idx = 1 + chapter_idx;
327
328 // for unaligned frames, estimate offset of first frame in block (assume no padding)
329 if (s->streams[0]->codecpar->codec_id == AV_CODEC_ID_MP3) {
330 c->seek_offset = (MP3_FRAME_SIZE - chapter_pos % MP3_FRAME_SIZE) % MP3_FRAME_SIZE;
331 }
332
333 avpriv_update_cur_dts(s, s->streams[0], ch->start + (chapter_pos + c->seek_offset) * TIMEPREC);
334
335 return 1;
336 }
337
aa_probe(const AVProbeData * p)338 static int aa_probe(const AVProbeData *p)
339 {
340 uint8_t *buf = p->buf;
341
342 // first 4 bytes are file size, next 4 bytes are the magic
343 if (AV_RB32(buf+4) != AA_MAGIC)
344 return 0;
345
346 return AVPROBE_SCORE_MAX / 2;
347 }
348
aa_read_close(AVFormatContext * s)349 static int aa_read_close(AVFormatContext *s)
350 {
351 AADemuxContext *c = s->priv_data;
352
353 av_freep(&c->tea_ctx);
354
355 return 0;
356 }
357
358 #define OFFSET(x) offsetof(AADemuxContext, x)
359 static const AVOption aa_options[] = {
360 { "aa_fixed_key", // extracted from libAAX_SDK.so and AAXSDKWin.dll files!
361 "Fixed key used for handling Audible AA files", OFFSET(aa_fixed_key),
362 AV_OPT_TYPE_BINARY, {.str="77214d4b196a87cd520045fd2a51d673"},
363 .flags = AV_OPT_FLAG_DECODING_PARAM },
364 { NULL },
365 };
366
367 static const AVClass aa_class = {
368 .class_name = "aa",
369 .item_name = av_default_item_name,
370 .option = aa_options,
371 .version = LIBAVUTIL_VERSION_INT,
372 };
373
374 const AVInputFormat ff_aa_demuxer = {
375 .name = "aa",
376 .long_name = NULL_IF_CONFIG_SMALL("Audible AA format files"),
377 .priv_class = &aa_class,
378 .priv_data_size = sizeof(AADemuxContext),
379 .extensions = "aa",
380 .read_probe = aa_probe,
381 .read_header = aa_read_header,
382 .read_packet = aa_read_packet,
383 .read_seek = aa_read_seek,
384 .read_close = aa_read_close,
385 .flags = AVFMT_NO_BYTE_SEEK | AVFMT_NOGENSEARCH,
386 .flags_internal = FF_FMT_INIT_CLEANUP,
387 };
388