mirror of
https://github.com/FFmpeg/FFmpeg.git
synced 2024-12-23 12:43:46 +02:00
dd25ca02ae
Signed-off-by: James Almer <jamrial@gmail.com>
394 lines
11 KiB
C
394 lines
11 KiB
C
/*
|
|
* Apple HTTP Live Streaming Sample Encryption/Decryption
|
|
*
|
|
* Copyright (c) 2021 Nachiket Tarate
|
|
*
|
|
* This file is part of FFmpeg.
|
|
*
|
|
* FFmpeg is free software; you can redistribute it and/or
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
* License as published by the Free Software Foundation; either
|
|
* version 2.1 of the License, or (at your option) any later version.
|
|
*
|
|
* FFmpeg is distributed in the hope that it will be useful,
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
* Lesser General Public License for more details.
|
|
*
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
* License along with FFmpeg; if not, write to the Free Software
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
|
*/
|
|
|
|
/**
|
|
* @file
|
|
* Apple HTTP Live Streaming Sample Encryption
|
|
* https://developer.apple.com/library/ios/documentation/AudioVideo/Conceptual/HLS_Sample_Encryption
|
|
*/
|
|
|
|
#include "libavutil/channel_layout.h"
|
|
|
|
#include "hls_sample_encryption.h"
|
|
|
|
#include "libavcodec/adts_header.h"
|
|
#include "libavcodec/adts_parser.h"
|
|
#include "libavcodec/ac3_parser_internal.h"
|
|
|
|
|
|
typedef struct NALUnit {
|
|
uint8_t *data;
|
|
int type;
|
|
int length;
|
|
int start_code_length;
|
|
} NALUnit;
|
|
|
|
typedef struct AudioFrame {
|
|
uint8_t *data;
|
|
int length;
|
|
int header_length;
|
|
} AudioFrame;
|
|
|
|
typedef struct CodecParserContext {
|
|
const uint8_t *buf_ptr;
|
|
const uint8_t *buf_end;
|
|
} CodecParserContext;
|
|
|
|
static const int eac3_sample_rate_tab[] = { 48000, 44100, 32000, 0 };
|
|
|
|
void ff_hls_senc_read_audio_setup_info(HLSAudioSetupInfo *info, const uint8_t *buf, size_t size)
|
|
{
|
|
if (size < 8)
|
|
return;
|
|
|
|
info->codec_tag = AV_RL32(buf);
|
|
|
|
if (info->codec_tag == MKTAG('z','a','a','c'))
|
|
info->codec_id = AV_CODEC_ID_AAC;
|
|
else if (info->codec_tag == MKTAG('z','a','c','3'))
|
|
info->codec_id = AV_CODEC_ID_AC3;
|
|
else if (info->codec_tag == MKTAG('z','e','c','3'))
|
|
info->codec_id = AV_CODEC_ID_EAC3;
|
|
else
|
|
info->codec_id = AV_CODEC_ID_NONE;
|
|
|
|
buf += 4;
|
|
info->priming = AV_RL16(buf);
|
|
buf += 2;
|
|
info->version = *buf++;
|
|
info->setup_data_length = *buf++;
|
|
|
|
if (info->setup_data_length > size - 8)
|
|
info->setup_data_length = size - 8;
|
|
|
|
if (info->setup_data_length > HLS_MAX_AUDIO_SETUP_DATA_LEN)
|
|
return;
|
|
|
|
memcpy(info->setup_data, buf, info->setup_data_length);
|
|
}
|
|
|
|
int ff_hls_senc_parse_audio_setup_info(AVStream *st, HLSAudioSetupInfo *info)
|
|
{
|
|
int ret = 0;
|
|
|
|
st->codecpar->codec_tag = info->codec_tag;
|
|
|
|
if (st->codecpar->codec_id == AV_CODEC_ID_AAC)
|
|
return 0;
|
|
|
|
if (st->codecpar->codec_id != AV_CODEC_ID_AC3 && st->codecpar->codec_id != AV_CODEC_ID_EAC3)
|
|
return AVERROR_INVALIDDATA;
|
|
|
|
if (st->codecpar->codec_id == AV_CODEC_ID_AC3) {
|
|
AC3HeaderInfo *ac3hdr = NULL;
|
|
|
|
ret = avpriv_ac3_parse_header(&ac3hdr, info->setup_data, info->setup_data_length);
|
|
if (ret < 0) {
|
|
if (ret != AVERROR(ENOMEM))
|
|
av_free(ac3hdr);
|
|
return ret;
|
|
}
|
|
|
|
st->codecpar->sample_rate = ac3hdr->sample_rate;
|
|
av_channel_layout_uninit(&st->codecpar->ch_layout);
|
|
av_channel_layout_from_mask(&st->codecpar->ch_layout, ac3hdr->channel_layout);
|
|
st->codecpar->bit_rate = ac3hdr->bit_rate;
|
|
|
|
av_free(ac3hdr);
|
|
} else { /* Parse 'dec3' EC3SpecificBox */
|
|
GetBitContext gb;
|
|
uint64_t mask;
|
|
int data_rate, fscod, acmod, lfeon;
|
|
|
|
ret = init_get_bits8(&gb, info->setup_data, info->setup_data_length);
|
|
if (ret < 0)
|
|
return AVERROR_INVALIDDATA;
|
|
|
|
data_rate = get_bits(&gb, 13);
|
|
skip_bits(&gb, 3);
|
|
fscod = get_bits(&gb, 2);
|
|
skip_bits(&gb, 10);
|
|
acmod = get_bits(&gb, 3);
|
|
lfeon = get_bits(&gb, 1);
|
|
|
|
st->codecpar->sample_rate = eac3_sample_rate_tab[fscod];
|
|
|
|
mask = ff_ac3_channel_layout_tab[acmod];
|
|
if (lfeon)
|
|
mask |= AV_CH_LOW_FREQUENCY;
|
|
|
|
av_channel_layout_uninit(&st->codecpar->ch_layout);
|
|
av_channel_layout_from_mask(&st->codecpar->ch_layout, mask);
|
|
|
|
st->codecpar->bit_rate = data_rate*1000;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
/*
|
|
* Remove start code emulation prevention 0x03 bytes
|
|
*/
|
|
static void remove_scep_3_bytes(NALUnit *nalu)
|
|
{
|
|
int i = 0;
|
|
int j = 0;
|
|
|
|
uint8_t *data = nalu->data;
|
|
|
|
while (i < nalu->length) {
|
|
if (nalu->length - i > 3 && AV_RB24(&data[i]) == 0x000003) {
|
|
data[j++] = data[i++];
|
|
data[j++] = data[i++];
|
|
i++;
|
|
} else {
|
|
data[j++] = data[i++];
|
|
}
|
|
}
|
|
|
|
nalu->length = j;
|
|
}
|
|
|
|
static int get_next_nal_unit(CodecParserContext *ctx, NALUnit *nalu)
|
|
{
|
|
const uint8_t *nalu_start = ctx->buf_ptr;
|
|
|
|
if (ctx->buf_end - ctx->buf_ptr >= 4 && AV_RB32(ctx->buf_ptr) == 0x00000001)
|
|
nalu->start_code_length = 4;
|
|
else if (ctx->buf_end - ctx->buf_ptr >= 3 && AV_RB24(ctx->buf_ptr) == 0x000001)
|
|
nalu->start_code_length = 3;
|
|
else /* No start code at the beginning of the NAL unit */
|
|
return -1;
|
|
|
|
ctx->buf_ptr += nalu->start_code_length;
|
|
|
|
while (ctx->buf_ptr < ctx->buf_end) {
|
|
if (ctx->buf_end - ctx->buf_ptr >= 4 && AV_RB32(ctx->buf_ptr) == 0x00000001)
|
|
break;
|
|
else if (ctx->buf_end - ctx->buf_ptr >= 3 && AV_RB24(ctx->buf_ptr) == 0x000001)
|
|
break;
|
|
ctx->buf_ptr++;
|
|
}
|
|
|
|
nalu->data = (uint8_t *)nalu_start + nalu->start_code_length;
|
|
nalu->length = ctx->buf_ptr - nalu->data;
|
|
nalu->type = *nalu->data & 0x1F;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int decrypt_nal_unit(HLSCryptoContext *crypto_ctx, NALUnit *nalu)
|
|
{
|
|
int ret = 0;
|
|
int rem_bytes;
|
|
uint8_t *data;
|
|
uint8_t iv[16];
|
|
|
|
ret = av_aes_init(crypto_ctx->aes_ctx, crypto_ctx->key, 16 * 8, 1);
|
|
if (ret < 0)
|
|
return ret;
|
|
|
|
/* Remove start code emulation prevention 0x03 bytes */
|
|
remove_scep_3_bytes(nalu);
|
|
|
|
data = nalu->data + 32;
|
|
rem_bytes = nalu->length - 32;
|
|
|
|
memcpy(iv, crypto_ctx->iv, 16);
|
|
|
|
while (rem_bytes > 0) {
|
|
if (rem_bytes > 16) {
|
|
av_aes_crypt(crypto_ctx->aes_ctx, data, data, 1, iv, 1);
|
|
data += 16;
|
|
rem_bytes -= 16;
|
|
}
|
|
data += FFMIN(144, rem_bytes);
|
|
rem_bytes -= FFMIN(144, rem_bytes);
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int decrypt_video_frame(HLSCryptoContext *crypto_ctx, AVPacket *pkt)
|
|
{
|
|
int ret = 0;
|
|
CodecParserContext ctx;
|
|
NALUnit nalu;
|
|
uint8_t *data_ptr;
|
|
int move_nalu = 0;
|
|
|
|
memset(&ctx, 0, sizeof(ctx));
|
|
ctx.buf_ptr = pkt->data;
|
|
ctx.buf_end = pkt->data + pkt->size;
|
|
|
|
data_ptr = pkt->data;
|
|
|
|
while (ctx.buf_ptr < ctx.buf_end) {
|
|
memset(&nalu, 0, sizeof(nalu));
|
|
ret = get_next_nal_unit(&ctx, &nalu);
|
|
if (ret < 0)
|
|
return ret;
|
|
if ((nalu.type == 0x01 || nalu.type == 0x05) && nalu.length > 48) {
|
|
int encrypted_nalu_length = nalu.length;
|
|
ret = decrypt_nal_unit(crypto_ctx, &nalu);
|
|
if (ret < 0)
|
|
return ret;
|
|
move_nalu = nalu.length != encrypted_nalu_length;
|
|
}
|
|
if (move_nalu)
|
|
memmove(data_ptr, nalu.data - nalu.start_code_length, nalu.start_code_length + nalu.length);
|
|
data_ptr += nalu.start_code_length + nalu.length;
|
|
}
|
|
|
|
av_shrink_packet(pkt, data_ptr - pkt->data);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int get_next_adts_frame(CodecParserContext *ctx, AudioFrame *frame)
|
|
{
|
|
int ret = 0;
|
|
|
|
AACADTSHeaderInfo *adts_hdr = NULL;
|
|
|
|
/* Find next sync word 0xFFF */
|
|
while (ctx->buf_ptr < ctx->buf_end - 1) {
|
|
if (*ctx->buf_ptr == 0xFF && (*(ctx->buf_ptr + 1) & 0xF0) == 0xF0)
|
|
break;
|
|
ctx->buf_ptr++;
|
|
}
|
|
|
|
if (ctx->buf_ptr >= ctx->buf_end - 1)
|
|
return -1;
|
|
|
|
frame->data = (uint8_t*)ctx->buf_ptr;
|
|
|
|
ret = avpriv_adts_header_parse (&adts_hdr, frame->data, ctx->buf_end - frame->data);
|
|
if (ret < 0)
|
|
return ret;
|
|
|
|
frame->header_length = adts_hdr->crc_absent ? AV_AAC_ADTS_HEADER_SIZE : AV_AAC_ADTS_HEADER_SIZE + 2;
|
|
frame->length = adts_hdr->frame_length;
|
|
|
|
av_free(adts_hdr);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int get_next_ac3_eac3_sync_frame(CodecParserContext *ctx, AudioFrame *frame)
|
|
{
|
|
int ret = 0;
|
|
|
|
AC3HeaderInfo *hdr = NULL;
|
|
|
|
/* Find next sync word 0x0B77 */
|
|
while (ctx->buf_ptr < ctx->buf_end - 1) {
|
|
if (*ctx->buf_ptr == 0x0B && *(ctx->buf_ptr + 1) == 0x77)
|
|
break;
|
|
ctx->buf_ptr++;
|
|
}
|
|
|
|
if (ctx->buf_ptr >= ctx->buf_end - 1)
|
|
return -1;
|
|
|
|
frame->data = (uint8_t*)ctx->buf_ptr;
|
|
frame->header_length = 0;
|
|
|
|
ret = avpriv_ac3_parse_header(&hdr, frame->data, ctx->buf_end - frame->data);
|
|
if (ret < 0) {
|
|
if (ret != AVERROR(ENOMEM))
|
|
av_free(hdr);
|
|
return ret;
|
|
}
|
|
|
|
frame->length = hdr->frame_size;
|
|
|
|
av_free(hdr);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int get_next_sync_frame(enum AVCodecID codec_id, CodecParserContext *ctx, AudioFrame *frame)
|
|
{
|
|
if (codec_id == AV_CODEC_ID_AAC)
|
|
return get_next_adts_frame(ctx, frame);
|
|
else if (codec_id == AV_CODEC_ID_AC3 || codec_id == AV_CODEC_ID_EAC3)
|
|
return get_next_ac3_eac3_sync_frame(ctx, frame);
|
|
else
|
|
return AVERROR_INVALIDDATA;
|
|
}
|
|
|
|
static int decrypt_sync_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AudioFrame *frame)
|
|
{
|
|
int ret = 0;
|
|
uint8_t *data;
|
|
int num_of_encrypted_blocks;
|
|
|
|
ret = av_aes_init(crypto_ctx->aes_ctx, crypto_ctx->key, 16 * 8, 1);
|
|
if (ret < 0)
|
|
return ret;
|
|
|
|
data = frame->data + frame->header_length + 16;
|
|
|
|
num_of_encrypted_blocks = (frame->length - frame->header_length - 16)/16;
|
|
|
|
av_aes_crypt(crypto_ctx->aes_ctx, data, data, num_of_encrypted_blocks, crypto_ctx->iv, 1);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int decrypt_audio_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AVPacket *pkt)
|
|
{
|
|
int ret = 0;
|
|
CodecParserContext ctx;
|
|
AudioFrame frame;
|
|
|
|
memset(&ctx, 0, sizeof(ctx));
|
|
ctx.buf_ptr = pkt->data;
|
|
ctx.buf_end = pkt->data + pkt->size;
|
|
|
|
while (ctx.buf_ptr < ctx.buf_end) {
|
|
memset(&frame, 0, sizeof(frame));
|
|
ret = get_next_sync_frame(codec_id, &ctx, &frame);
|
|
if (ret < 0)
|
|
return ret;
|
|
if (frame.length - frame.header_length > 31) {
|
|
ret = decrypt_sync_frame(codec_id, crypto_ctx, &frame);
|
|
if (ret < 0)
|
|
return ret;
|
|
}
|
|
ctx.buf_ptr += frame.length;
|
|
}
|
|
|
|
return 0;
|
|
}
|
|
|
|
int ff_hls_senc_decrypt_frame(enum AVCodecID codec_id, HLSCryptoContext *crypto_ctx, AVPacket *pkt)
|
|
{
|
|
if (codec_id == AV_CODEC_ID_H264)
|
|
return decrypt_video_frame(crypto_ctx, pkt);
|
|
else if (codec_id == AV_CODEC_ID_AAC || codec_id == AV_CODEC_ID_AC3 || codec_id == AV_CODEC_ID_EAC3)
|
|
return decrypt_audio_frame(codec_id, crypto_ctx, pkt);
|
|
|
|
return AVERROR_INVALIDDATA;
|
|
}
|