/src/ffmpeg/libavformat/hls_sample_encryption.c
Line | Count | Source (jump to first uncovered line) |
1 | | /* |
2 | | * Apple HTTP Live Streaming Sample Encryption/Decryption |
3 | | * |
4 | | * Copyright (c) 2021 Nachiket Tarate |
5 | | * |
6 | | * This file is part of FFmpeg. |
7 | | * |
8 | | * FFmpeg is free software; you can redistribute it and/or |
9 | | * modify it under the terms of the GNU Lesser General Public |
10 | | * License as published by the Free Software Foundation; either |
11 | | * version 2.1 of the License, or (at your option) any later version. |
12 | | * |
13 | | * FFmpeg is distributed in the hope that it will be useful, |
14 | | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
15 | | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
16 | | * Lesser General Public License for more details. |
17 | | * |
18 | | * You should have received a copy of the GNU Lesser General Public |
19 | | * License along with FFmpeg; if not, write to the Free Software |
20 | | * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
21 | | */ |
22 | | |
23 | | /** |
24 | | * @file |
25 | | * Apple HTTP Live Streaming Sample Encryption |
26 | | * https://developer.apple.com/library/ios/documentation/AudioVideo/Conceptual/HLS_Sample_Encryption |
27 | | */ |
28 | | |
29 | | #include "libavutil/aes.h" |
30 | | #include "libavutil/channel_layout.h" |
31 | | #include "libavutil/mem.h" |
32 | | |
33 | | #include "hls_sample_encryption.h" |
34 | | |
35 | | #include "libavcodec/adts_header.h" |
36 | | #include "libavcodec/adts_parser.h" |
37 | | #include "libavcodec/ac3tab.h" |
38 | | #include "libavcodec/ac3_parser_internal.h" |
39 | | |
40 | | |
41 | | typedef struct NALUnit { |
42 | | uint8_t *data; |
43 | | int type; |
44 | | int length; |
45 | | int start_code_length; |
46 | | } NALUnit; |
47 | | |
48 | | typedef struct AudioFrame { |
49 | | uint8_t *data; |
50 | | int length; |
51 | | int header_length; |
52 | | } AudioFrame; |
53 | | |
54 | | typedef struct CodecParserContext { |
55 | | const uint8_t *buf_ptr; |
56 | | const uint8_t *buf_end; |
57 | | } CodecParserContext; |
58 | | |
59 | | static const int eac3_sample_rate_tab[] = { 48000, 44100, 32000, 0 }; |
60 | | |
61 | | void ff_hls_senc_read_audio_setup_info(HLSAudioSetupInfo *info, const uint8_t *buf, size_t size) |
62 | 0 | { |
63 | 0 | if (size < 8) |
64 | 0 | return; |
65 | | |
66 | 0 | info->codec_tag = AV_RL32(buf); |
67 | | |
68 | | /* Always keep this list in sync with the one from hls_read_header() */ |
69 | 0 | if (info->codec_tag == MKTAG('z','a','a','c')) |
70 | 0 | info->codec_id = AV_CODEC_ID_AAC; |
71 | 0 | else if (info->codec_tag == MKTAG('z','a','c','3')) |
72 | 0 | info->codec_id = AV_CODEC_ID_AC3; |
73 | 0 | else if (info->codec_tag == MKTAG('z','e','c','3')) |
74 | 0 | info->codec_id = AV_CODEC_ID_EAC3; |
75 | 0 | else |
76 | 0 | info->codec_id = AV_CODEC_ID_NONE; |
77 | |
|
78 | 0 | buf += 4; |
79 | 0 | info->priming = AV_RL16(buf); |
80 | 0 | buf += 2; |
81 | 0 | info->version = *buf++; |
82 | 0 | info->setup_data_length = *buf++; |
83 | |
|
84 | 0 | if (info->setup_data_length > size - 8) |
85 | 0 | info->setup_data_length = size - 8; |
86 | |
|
87 | 0 | if (info->setup_data_length > HLS_MAX_AUDIO_SETUP_DATA_LEN) |
88 | 0 | return; |
89 | | |
90 | 0 | memcpy(info->setup_data, buf, info->setup_data_length); |
91 | 0 | } |
92 | | |
93 | | int ff_hls_senc_parse_audio_setup_info(AVStream *st, HLSAudioSetupInfo *info) |
94 | 0 | { |
95 | 0 | int ret = 0; |
96 | |
|
97 | 0 | st->codecpar->codec_tag = info->codec_tag; |
98 | |
|
99 | 0 | if (st->codecpar->codec_id == AV_CODEC_ID_AAC) |
100 | 0 | return 0; |
101 | | |
102 | 0 | if (st->codecpar->codec_id != AV_CODEC_ID_AC3 && st->codecpar->codec_id != AV_CODEC_ID_EAC3) |
103 | 0 | return AVERROR_INVALIDDATA; |
104 | | |
105 | 0 | if (st->codecpar->codec_id == AV_CODEC_ID_AC3) { |
106 | 0 | AC3HeaderInfo *ac3hdr = NULL; |
107 | |
|
108 | 0 | ret = avpriv_ac3_parse_header(&ac3hdr, info->setup_data, info->setup_data_length); |
109 | 0 | if (ret < 0) { |
110 | 0 | av_free(ac3hdr); |
111 | 0 | return ret; |
112 | 0 | } |
113 | | |
114 | 0 | st->codecpar->sample_rate = ac3hdr->sample_rate; |
115 | 0 | av_channel_layout_uninit(&st->codecpar->ch_layout); |
116 | 0 | av_channel_layout_from_mask(&st->codecpar->ch_layout, ac3hdr->channel_layout); |
117 | 0 | st->codecpar->bit_rate = ac3hdr->bit_rate; |
118 | |
|
119 | 0 | av_free(ac3hdr); |
120 | 0 | } else { /* Parse 'dec3' EC3SpecificBox */ |
121 | 0 | GetBitContext gb; |
122 | 0 | uint64_t mask; |
123 | 0 | int data_rate, fscod, acmod, lfeon; |
124 | |
|
125 | 0 | ret = init_get_bits8(&gb, info->setup_data, info->setup_data_length); |
126 | 0 | if (ret < 0) |
127 | 0 | return AVERROR_INVALIDDATA; |
128 | | |
129 | 0 | data_rate = get_bits(&gb, 13); |
130 | 0 | skip_bits(&gb, 3); |
131 | 0 | fscod = get_bits(&gb, 2); |
132 | 0 | skip_bits(&gb, 10); |
133 | 0 | acmod = get_bits(&gb, 3); |
134 | 0 | lfeon = get_bits(&gb, 1); |
135 | |
|
136 | 0 | st->codecpar->sample_rate = eac3_sample_rate_tab[fscod]; |
137 | |
|
138 | 0 | mask = ff_ac3_channel_layout_tab[acmod]; |
139 | 0 | if (lfeon) |
140 | 0 | mask |= AV_CH_LOW_FREQUENCY; |
141 | |
|
142 | 0 | av_channel_layout_uninit(&st->codecpar->ch_layout); |
143 | 0 | av_channel_layout_from_mask(&st->codecpar->ch_layout, mask); |
144 | |
|
145 | 0 | st->codecpar->bit_rate = data_rate*1000; |
146 | 0 | } |
147 | | |
148 | 0 | return 0; |
149 | 0 | } |
150 | | |
151 | | /* |
152 | | * Remove start code emulation prevention 0x03 bytes |
153 | | */ |
154 | | static void remove_scep_3_bytes(NALUnit *nalu) |
155 | 0 | { |
156 | 0 | int i = 0; |
157 | 0 | int j = 0; |
158 | |
|
159 | 0 | uint8_t *data = nalu->data; |
160 | |
|
161 | 0 | while (i < nalu->length) { |
162 | 0 | if (nalu->length - i > 3 && AV_RB24(&data[i]) == 0x000003) { |
163 | 0 | data[j++] = data[i++]; |
164 | 0 | data[j++] = data[i++]; |
165 | 0 | i++; |
166 | 0 | } else { |
167 | 0 | data[j++] = data[i++]; |
168 | 0 | } |
169 | 0 | } |
170 | |
|
171 | 0 | nalu->length = j; |
172 | 0 | } |
173 | | |
174 | | static int get_next_nal_unit(CodecParserContext *ctx, NALUnit *nalu) |
175 | 0 | { |
176 | 0 | const uint8_t *nalu_start = ctx->buf_ptr; |
177 | |
|
178 | 0 | if (ctx->buf_end - ctx->buf_ptr >= 4 && AV_RB32(ctx->buf_ptr) == 0x00000001) |
179 | 0 | nalu->start_code_length = 4; |
180 | 0 | else if (ctx->buf_end - ctx->buf_ptr >= 3 && AV_RB24(ctx->buf_ptr) == 0x000001) |
181 | 0 | nalu->start_code_length = 3; |
182 | 0 | else /* No start code at the beginning of the NAL unit */ |
183 | 0 | return -1; |
184 | | |
185 | 0 | ctx->buf_ptr += nalu->start_code_length; |
186 | |
|
187 | 0 | while (ctx->buf_ptr < ctx->buf_end) { |
188 | 0 | if (ctx->buf_end - ctx->buf_ptr >= 4 && AV_RB32(ctx->buf_ptr) == 0x00000001) |
189 | 0 | break; |
190 | 0 | else if (ctx->buf_end - ctx->buf_ptr >= 3 && AV_RB24(ctx->buf_ptr) == 0x000001) |
191 | 0 | break; |
192 | 0 | ctx->buf_ptr++; |
193 | 0 | } |
194 | |
|
195 | 0 | nalu->data = (uint8_t *)nalu_start + nalu->start_code_length; |
196 | 0 | nalu->length = ctx->buf_ptr - nalu->data; |
197 | 0 | nalu->type = *nalu->data & 0x1F; |
198 | |
|
199 | 0 | return 0; |
200 | 0 | } |
201 | | |
202 | | static int decrypt_nal_unit(HLSCryptoContext *crypto_ctx, NALUnit *nalu) |
203 | 0 | { |
204 | 0 | int ret = 0; |
205 | 0 | int rem_bytes; |
206 | 0 | uint8_t *data; |
207 | 0 | uint8_t iv[16]; |
208 | |
|
209 | 0 | ret = av_aes_init(crypto_ctx->aes_ctx, crypto_ctx->key, 16 * 8, 1); |
210 | 0 | if (ret < 0) |
211 | 0 | return ret; |
212 | | |
213 | | /* Remove start code emulation prevention 0x03 bytes */ |
214 | 0 | remove_scep_3_bytes(nalu); |
215 | |
|
216 | 0 | data = nalu->data + 32; |
217 | 0 | rem_bytes = nalu->length - 32; |
218 | |
|
219 | 0 | memcpy(iv, crypto_ctx->iv, 16); |
220 | |
|
221 | 0 | while (rem_bytes > 0) { |
222 | 0 | if (rem_bytes > 16) { |
223 | 0 | av_aes_crypt(crypto_ctx->aes_ctx, data, data, 1, iv, 1); |
224 | 0 | data += 16; |
225 | 0 | rem_bytes -= 16; |
226 | 0 | } |
227 | 0 | data += FFMIN(144, rem_bytes); |
228 | 0 | rem_bytes -= FFMIN(144, rem_bytes); |
229 | 0 | } |
230 | |
|
231 | 0 | return 0; |
232 | 0 | } |
233 | | |
234 | | static int decrypt_video_frame(HLSCryptoContext *crypto_ctx, AVPacket *pkt) |
235 | 0 | { |
236 | 0 | int ret = 0; |
237 | 0 | CodecParserContext ctx; |
238 | 0 | NALUnit nalu; |
239 | 0 | uint8_t *data_ptr; |
240 | 0 | int move_nalu = 0; |
241 | |
|
242 | 0 | memset(&ctx, 0, sizeof(ctx)); |
243 | 0 | ctx.buf_ptr = pkt->data; |
244 | 0 | ctx.buf_end = pkt->data + pkt->size; |
245 | |
|
246 | 0 | data_ptr = pkt->data; |
247 | |
|
248 | 0 | while (ctx.buf_ptr < ctx.buf_end) { |
249 | 0 | memset(&nalu, 0, sizeof(nalu)); |
250 | 0 | ret = get_next_nal_unit(&ctx, &nalu); |
251 | 0 | if (ret < 0) |
252 | 0 | return ret; |
253 | 0 | if ((nalu.type == 0x01 || nalu.type == 0x05) && nalu.length > 48) { |
254 | 0 | int encrypted_nalu_length = nalu.length; |
255 | 0 | ret = decrypt_nal_unit(crypto_ctx, &nalu); |
256 | 0 | if (ret < 0) |
257 | 0 | return ret; |
258 | 0 | move_nalu = nalu.length != encrypted_nalu_length; |
259 | 0 | } |
260 | 0 | if (move_nalu) |
261 | 0 | memmove(data_ptr, nalu.data - nalu.start_code_length, nalu.start_code_length + nalu.length); |
262 | 0 | data_ptr += nalu.start_code_length + nalu.length; |
263 | 0 | } |
264 | | |
265 | 0 | av_shrink_packet(pkt, data_ptr - pkt->data); |
266 | |
|
267 | 0 | return 0; |
268 | 0 | } |
269 | | |
270 | | static int get_next_adts_frame(CodecParserContext *ctx, AudioFrame *frame) |
271 | 0 | { |
272 | 0 | int ret = 0; |
273 | |
|
274 | 0 | AACADTSHeaderInfo *adts_hdr = NULL; |
275 | | |
276 | | /* Find next sync word 0xFFF */ |
277 | 0 | while (ctx->buf_ptr < ctx->buf_end - 1) { |
278 | 0 | if (*ctx->buf_ptr == 0xFF && (*(ctx->buf_ptr + 1) & 0xF0) == 0xF0) |
279 | 0 | break; |
280 | 0 | ctx->buf_ptr++; |
281 | 0 | } |
282 | |
|
283 | 0 | if (ctx->buf_ptr >= ctx->buf_end - 1) |
284 | 0 | return -1; |
285 | | |
286 | 0 | frame->data = (uint8_t*)ctx->buf_ptr; |
287 | |
|
288 | 0 | ret = avpriv_adts_header_parse (&adts_hdr, frame->data, ctx->buf_end - frame->data); |
289 | 0 | if (ret < 0) |
290 | 0 | return ret; |
291 | | |
292 | 0 | frame->header_length = adts_hdr->crc_absent ? AV_AAC_ADTS_HEADER_SIZE : AV_AAC_ADTS_HEADER_SIZE + 2; |
293 | 0 | frame->length = adts_hdr->frame_length; |
294 | |
|
295 | 0 | av_free(adts_hdr); |
296 | |
|
297 | 0 | return 0; |
298 | 0 | } |
299 | | |
300 | | static int get_next_ac3_eac3_sync_frame(CodecParserContext *ctx, AudioFrame *frame) |
301 | 0 | { |
302 | 0 | int ret = 0; |
303 | |
|
304 | 0 | AC3HeaderInfo *hdr = NULL; |
305 | | |
306 | | /* Find next sync word 0x0B77 */ |
307 | 0 | while (ctx->buf_ptr < ctx->buf_end - 1) { |
308 | 0 | if (*ctx->buf_ptr == 0x0B && *(ctx->buf_ptr + 1) == 0x77) |
309 | 0 | break; |
310 | 0 | ctx->buf_ptr++; |
311 | 0 | } |
312 | |
|
313 | 0 | if (ctx->buf_ptr >= ctx->buf_end - 1) |
314 | 0 | return -1; |
315 | | |
316 | 0 | frame->data = (uint8_t*)ctx->buf_ptr; |
317 | 0 | frame->header_length = 0; |
318 | |
|
319 | 0 | ret = avpriv_ac3_parse_header(&hdr, frame->data, ctx->buf_end - frame->data); |
320 | 0 | if (ret < 0) { |
321 | 0 | av_free(hdr); |
322 | 0 | return ret; |
323 | 0 | } |
324 | | |
325 | 0 | frame->length = hdr->frame_size; |
326 | |
|
327 | 0 | av_free(hdr); |
328 | |
|
329 | 0 | return 0; |
330 | 0 | } |
331 | | |
332 | | static int get_next_sync_frame(enum AVCodecID codec_id, CodecParserContext *ctx, AudioFrame *frame) |
333 | 0 | { |
334 | 0 | if (codec_id == AV_CODEC_ID_AAC) |
335 | 0 | return get_next_adts_frame(ctx, frame); |
336 | 0 | else if (codec_id == AV_CODEC_ID_AC3 || codec_id == AV_CODEC_ID_EAC3) |
337 | 0 | return get_next_ac3_eac3_sync_frame(ctx, frame); |
338 | 0 | else |
339 | 0 | return AVERROR_INVALIDDATA; |
340 | 0 | } |
341 | | |
342 | | static int decrypt_sync_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AudioFrame *frame) |
343 | 0 | { |
344 | 0 | int ret = 0; |
345 | 0 | uint8_t *data; |
346 | 0 | int num_of_encrypted_blocks; |
347 | |
|
348 | 0 | ret = av_aes_init(crypto_ctx->aes_ctx, crypto_ctx->key, 16 * 8, 1); |
349 | 0 | if (ret < 0) |
350 | 0 | return ret; |
351 | | |
352 | 0 | data = frame->data + frame->header_length + 16; |
353 | |
|
354 | 0 | num_of_encrypted_blocks = (frame->length - frame->header_length - 16)/16; |
355 | |
|
356 | 0 | av_aes_crypt(crypto_ctx->aes_ctx, data, data, num_of_encrypted_blocks, crypto_ctx->iv, 1); |
357 | |
|
358 | 0 | return 0; |
359 | 0 | } |
360 | | |
361 | | static int decrypt_audio_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AVPacket *pkt) |
362 | 0 | { |
363 | 0 | int ret = 0; |
364 | 0 | CodecParserContext ctx; |
365 | 0 | AudioFrame frame; |
366 | |
|
367 | 0 | memset(&ctx, 0, sizeof(ctx)); |
368 | 0 | ctx.buf_ptr = pkt->data; |
369 | 0 | ctx.buf_end = pkt->data + pkt->size; |
370 | |
|
371 | 0 | while (ctx.buf_ptr < ctx.buf_end) { |
372 | 0 | memset(&frame, 0, sizeof(frame)); |
373 | 0 | ret = get_next_sync_frame(codec_id, &ctx, &frame); |
374 | 0 | if (ret < 0) |
375 | 0 | return ret; |
376 | 0 | if (frame.length - frame.header_length > 31) { |
377 | 0 | ret = decrypt_sync_frame(codec_id, crypto_ctx, &frame); |
378 | 0 | if (ret < 0) |
379 | 0 | return ret; |
380 | 0 | } |
381 | 0 | ctx.buf_ptr += frame.length; |
382 | 0 | } |
383 | | |
384 | 0 | return 0; |
385 | 0 | } |
386 | | |
387 | | int ff_hls_senc_decrypt_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AVPacket *pkt) |
388 | 0 | { |
389 | 0 | if (codec_id == AV_CODEC_ID_H264) |
390 | 0 | return decrypt_video_frame(crypto_ctx, pkt); |
391 | 0 | else if (codec_id == AV_CODEC_ID_AAC || codec_id == AV_CODEC_ID_AC3 || codec_id == AV_CODEC_ID_EAC3) |
392 | 0 | return decrypt_audio_frame(codec_id, crypto_ctx, pkt); |
393 | | |
394 | 0 | return AVERROR_INVALIDDATA; |
395 | 0 | } |