You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.

1047 lines
34 KiB

/*
* "NUT" Container Format demuxer
* Copyright (c) 2004-2006 Michael Niedermayer
* Copyright (c) 2003 Alex Beregszaszi
*
* This file is part of Libav.
*
* Libav 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.
*
* Libav 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 Libav; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
#include "libavutil/avstring.h"
#include "libavutil/bswap.h"
#include "libavutil/dict.h"
#include "libavutil/mathematics.h"
#include "libavutil/tree.h"
#include "avio_internal.h"
#include "nut.h"
#include "riff.h"
#undef NDEBUG
#include <assert.h>
#define NUT_MAX_STREAMS 256 /* arbitrary sanity check value */
static int get_str(AVIOContext *bc, char *string, unsigned int maxlen)
{
unsigned int len = ffio_read_varlen(bc);
if (len && maxlen)
avio_read(bc, string, FFMIN(len, maxlen));
while (len > maxlen) {
avio_r8(bc);
len--;
}
if (maxlen)
string[FFMIN(len, maxlen - 1)] = 0;
if (maxlen == len)
return -1;
else
return 0;
}
static int64_t get_s(AVIOContext *bc)
{
int64_t v = ffio_read_varlen(bc) + 1;
if (v & 1)
return -(v >> 1);
else
return (v >> 1);
}
static uint64_t get_fourcc(AVIOContext *bc)
{
unsigned int len = ffio_read_varlen(bc);
if (len == 2)
return avio_rl16(bc);
else if (len == 4)
return avio_rl32(bc);
else
return -1;
}
static int get_packetheader(NUTContext *nut, AVIOContext *bc,
int calculate_checksum, uint64_t startcode)
{
int64_t size;
startcode = av_be2ne64(startcode);
startcode = ff_crc04C11DB7_update(0, (uint8_t*) &startcode, 8);
ffio_init_checksum(bc, ff_crc04C11DB7_update, startcode);
size = ffio_read_varlen(bc);
if (size > 4096)
avio_rb32(bc);
if (ffio_get_checksum(bc) && size > 4096)
return -1;
ffio_init_checksum(bc, calculate_checksum ? ff_crc04C11DB7_update : NULL, 0);
return size;
}
static uint64_t find_any_startcode(AVIOContext *bc, int64_t pos)
{
uint64_t state = 0;
if (pos >= 0)
/* Note, this may fail if the stream is not seekable, but that should
* not matter, as in this case we simply start where we currently are */
avio_seek(bc, pos, SEEK_SET);
while (!bc->eof_reached) {
state = (state << 8) | avio_r8(bc);
if ((state >> 56) != 'N')
continue;
switch (state) {
case MAIN_STARTCODE:
case STREAM_STARTCODE:
case SYNCPOINT_STARTCODE:
case INFO_STARTCODE:
case INDEX_STARTCODE:
return state;
}
}
return 0;
}
/**
* Find the given startcode.
* @param code the startcode
* @param pos the start position of the search, or -1 if the current position
* @return the position of the startcode or -1 if not found
*/
static int64_t find_startcode(AVIOContext *bc, uint64_t code, int64_t pos)
{
for (;;) {
uint64_t startcode = find_any_startcode(bc, pos);
if (startcode == code)
return avio_tell(bc) - 8;
else if (startcode == 0)
return -1;
pos = -1;
}
}
static int nut_probe(AVProbeData *p)
{
int i;
uint64_t code = 0;
for (i = 0; i < p->buf_size; i++) {
code = (code << 8) | p->buf[i];
if (code == MAIN_STARTCODE)
return AVPROBE_SCORE_MAX;
}
return 0;
}
#define GET_V(dst, check) \
do { \
tmp = ffio_read_varlen(bc); \
if (!(check)) { \
av_log(s, AV_LOG_ERROR, "Error " #dst " is (%"PRId64")\n", tmp); \
return AVERROR_INVALIDDATA; \
} \
dst = tmp; \
} while (0)
static int skip_reserved(AVIOContext *bc, int64_t pos)
{
pos -= avio_tell(bc);
if (pos < 0) {
avio_seek(bc, pos, SEEK_CUR);
return AVERROR_INVALIDDATA;
} else {
while (pos--)
avio_r8(bc);
return 0;
}
}
static int decode_main_header(NUTContext *nut)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
uint64_t tmp, end;
unsigned int stream_count;
int i, j, count;
int tmp_stream, tmp_mul, tmp_pts, tmp_size, tmp_res, tmp_head_idx;
end = get_packetheader(nut, bc, 1, MAIN_STARTCODE);
end += avio_tell(bc);
nut->version = ffio_read_varlen(bc);
if (nut->version < NUT_MIN_VERSION &&
nut->version > NUT_MAX_VERSION) {
av_log(s, AV_LOG_ERROR, "Version %d not supported.\n",
nut->version);
return AVERROR(ENOSYS);
}
GET_V(stream_count, tmp > 0 && tmp <= NUT_MAX_STREAMS);
nut->max_distance = ffio_read_varlen(bc);
if (nut->max_distance > 65536) {
av_log(s, AV_LOG_DEBUG, "max_distance %d\n", nut->max_distance);
nut->max_distance = 65536;
}
GET_V(nut->time_base_count, tmp > 0 && tmp < INT_MAX / sizeof(AVRational));
nut->time_base = av_malloc(nut->time_base_count * sizeof(AVRational));
if (!nut->time_base)
return AVERROR(ENOMEM);
for (i = 0; i < nut->time_base_count; i++) {
GET_V(nut->time_base[i].num, tmp > 0 && tmp < (1ULL << 31));
GET_V(nut->time_base[i].den, tmp > 0 && tmp < (1ULL << 31));
if (av_gcd(nut->time_base[i].num, nut->time_base[i].den) != 1) {
av_log(s, AV_LOG_ERROR, "invalid time base %d/%d\n",
nut->time_base[i].num,
nut->time_base[i].den);
return AVERROR_INVALIDDATA;
}
}
tmp_pts = 0;
tmp_mul = 1;
tmp_stream = 0;
tmp_head_idx = 0;
for (i = 0; i < 256;) {
int tmp_flags = ffio_read_varlen(bc);
int tmp_fields = ffio_read_varlen(bc);
if (tmp_fields > 0)
tmp_pts = get_s(bc);
if (tmp_fields > 1)
tmp_mul = ffio_read_varlen(bc);
if (tmp_fields > 2)
tmp_stream = ffio_read_varlen(bc);
if (tmp_fields > 3)
tmp_size = ffio_read_varlen(bc);
else
tmp_size = 0;
if (tmp_fields > 4)
tmp_res = ffio_read_varlen(bc);
else
tmp_res = 0;
if (tmp_fields > 5)
count = ffio_read_varlen(bc);
else
count = tmp_mul - tmp_size;
if (tmp_fields > 6)
get_s(bc);
if (tmp_fields > 7)
tmp_head_idx = ffio_read_varlen(bc);
while (tmp_fields-- > 8)
ffio_read_varlen(bc);
if (count == 0 || i + count > 256) {
av_log(s, AV_LOG_ERROR, "illegal count %d at %d\n", count, i);
return AVERROR_INVALIDDATA;
}
if (tmp_stream >= stream_count) {
av_log(s, AV_LOG_ERROR, "illegal stream number %d >= %d\n",
tmp_stream, stream_count);
return AVERROR_INVALIDDATA;
}
for (j = 0; j < count; j++, i++) {
if (i == 'N') {
nut->frame_code[i].flags = FLAG_INVALID;
j--;
continue;
}
nut->frame_code[i].flags = tmp_flags;
nut->frame_code[i].pts_delta = tmp_pts;
nut->frame_code[i].stream_id = tmp_stream;
nut->frame_code[i].size_mul = tmp_mul;
nut->frame_code[i].size_lsb = tmp_size + j;
nut->frame_code[i].reserved_count = tmp_res;
nut->frame_code[i].header_idx = tmp_head_idx;
}
}
assert(nut->frame_code['N'].flags == FLAG_INVALID);
if (end > avio_tell(bc) + 4) {
int rem = 1024;
GET_V(nut->header_count, tmp < 128U);
nut->header_count++;
for (i = 1; i < nut->header_count; i++) {
uint8_t *hdr;
GET_V(nut->header_len[i], tmp > 0 && tmp < 256);
if (rem < nut->header_len[i]) {
av_log(s, AV_LOG_ERROR,
"invalid elision header %d : %d > %d\n",
i, nut->header_len[i], rem);
return AVERROR_INVALIDDATA;
}
rem -= nut->header_len[i];
hdr = av_malloc(nut->header_len[i]);
if (!hdr)
return AVERROR(ENOMEM);
avio_read(bc, hdr, nut->header_len[i]);
nut->header[i] = hdr;
}
assert(nut->header_len[0] == 0);
}
// flags had been effectively introduced in version 4
if (nut->version > NUT_STABLE_VERSION) {
nut->flags = ffio_read_varlen(bc);
}
if (skip_reserved(bc, end) || ffio_get_checksum(bc)) {
av_log(s, AV_LOG_ERROR, "main header checksum mismatch\n");
return AVERROR_INVALIDDATA;
}
nut->stream = av_mallocz(sizeof(StreamContext) * stream_count);
if (!nut->stream)
return AVERROR(ENOMEM);
for (i = 0; i < stream_count; i++)
avformat_new_stream(s, NULL);
return 0;
}
static int decode_stream_header(NUTContext *nut)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
StreamContext *stc;
int class, stream_id;
uint64_t tmp, end;
AVStream *st;
end = get_packetheader(nut, bc, 1, STREAM_STARTCODE);
end += avio_tell(bc);
GET_V(stream_id, tmp < s->nb_streams && !nut->stream[tmp].time_base);
stc = &nut->stream[stream_id];
st = s->streams[stream_id];
if (!st)
return AVERROR(ENOMEM);
class = ffio_read_varlen(bc);
tmp = get_fourcc(bc);
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
st->codecpar->codec_tag = tmp;
switch (class) {
case 0:
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
st->codecpar->codec_type = AVMEDIA_TYPE_VIDEO;
st->codecpar->codec_id = av_codec_get_id((const AVCodecTag * const []) {
ff_nut_video_tags,
ff_codec_bmp_tags,
0
},
tmp);
break;
case 1:
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
st->codecpar->codec_type = AVMEDIA_TYPE_AUDIO;
st->codecpar->codec_id = av_codec_get_id((const AVCodecTag * const []) {
ff_nut_audio_tags,
ff_codec_wav_tags,
0
},
tmp);
break;
case 2:
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
st->codecpar->codec_type = AVMEDIA_TYPE_SUBTITLE;
st->codecpar->codec_id = ff_codec_get_id(ff_nut_subtitle_tags, tmp);
break;
case 3:
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
st->codecpar->codec_type = AVMEDIA_TYPE_DATA;
st->codecpar->codec_id = ff_codec_get_id(ff_nut_data_tags, tmp);
break;
default:
av_log(s, AV_LOG_ERROR, "unknown stream class (%d)\n", class);
return AVERROR(ENOSYS);
}
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
if (class < 3 && st->codecpar->codec_id == AV_CODEC_ID_NONE)
av_log(s, AV_LOG_ERROR,
"Unknown codec tag '0x%04x' for stream number %d\n",
(unsigned int) tmp, stream_id);
GET_V(stc->time_base_id, tmp < nut->time_base_count);
GET_V(stc->msb_pts_shift, tmp < 16);
stc->max_pts_distance = ffio_read_varlen(bc);
GET_V(stc->decode_delay, tmp < 1000); // sanity limit, raise this if Moore's law is true
ffio_read_varlen(bc); // stream flags
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
GET_V(st->codecpar->extradata_size, tmp < (1 << 30));
if (st->codecpar->extradata_size) {
st->codecpar->extradata = av_mallocz(st->codecpar->extradata_size +
AV_INPUT_BUFFER_PADDING_SIZE);
if (!st->codecpar->extradata)
return AVERROR(ENOMEM);
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
avio_read(bc, st->codecpar->extradata, st->codecpar->extradata_size);
}
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
if (st->codecpar->codec_type == AVMEDIA_TYPE_VIDEO) {
GET_V(st->codecpar->width, tmp > 0);
GET_V(st->codecpar->height, tmp > 0);
st->sample_aspect_ratio.num = ffio_read_varlen(bc);
st->sample_aspect_ratio.den = ffio_read_varlen(bc);
if ((!st->sample_aspect_ratio.num) != (!st->sample_aspect_ratio.den)) {
av_log(s, AV_LOG_ERROR, "invalid aspect ratio %d/%d\n",
st->sample_aspect_ratio.num, st->sample_aspect_ratio.den);
return AVERROR_INVALIDDATA;
}
ffio_read_varlen(bc); /* csp type */
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
} else if (st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
GET_V(st->codecpar->sample_rate, tmp > 0);
ffio_read_varlen(bc); // samplerate_den
lavf: replace AVStream.codec with AVStream.codecpar Currently, AVStream contains an embedded AVCodecContext instance, which is used by demuxers to export stream parameters to the caller and by muxers to receive stream parameters from the caller. It is also used internally as the codec context that is passed to parsers. In addition, it is also widely used by the callers as the decoding (when demuxer) or encoding (when muxing) context, though this has been officially discouraged since Libav 11. There are multiple important problems with this approach: - the fields in AVCodecContext are in general one of * stream parameters * codec options * codec state However, it's not clear which ones are which. It is consequently unclear which fields are a demuxer allowed to set or a muxer allowed to read. This leads to erratic behaviour depending on whether decoding or encoding is being performed or not (and whether it uses the AVStream embedded codec context). - various synchronization issues arising from the fact that the same context is used by several different APIs (muxers/demuxers, parsers, bitstream filters and encoders/decoders) simultaneously, with there being no clear rules for who can modify what and the different processes being typically delayed with respect to each other. - avformat_find_stream_info() making it necessary to support opening and closing a single codec context multiple times, thus complicating the semantics of freeing various allocated objects in the codec context. Those problems are resolved by replacing the AVStream embedded codec context with a newly added AVCodecParameters instance, which stores only the stream parameters exported by the demuxers or read by the muxers.
11 years ago
GET_V(st->codecpar->channels, tmp > 0);
}
if (skip_reserved(bc, end) || ffio_get_checksum(bc)) {
av_log(s, AV_LOG_ERROR,
"stream header %d checksum mismatch\n", stream_id);
return AVERROR_INVALIDDATA;
}
stc->time_base = &nut->time_base[stc->time_base_id];
avpriv_set_pts_info(s->streams[stream_id], 63, stc->time_base->num,
stc->time_base->den);
return 0;
}
static void set_disposition_bits(AVFormatContext *avf, char *value,
int stream_id)
{
int flag = 0, i;
for (i = 0; ff_nut_dispositions[i].flag; ++i)
if (!strcmp(ff_nut_dispositions[i].str, value))
flag = ff_nut_dispositions[i].flag;
if (!flag)
av_log(avf, AV_LOG_INFO, "unknown disposition type '%s'\n", value);
for (i = 0; i < avf->nb_streams; ++i)
if (stream_id == i || stream_id == -1)
avf->streams[i]->disposition |= flag;
}
static int decode_info_header(NUTContext *nut)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
uint64_t tmp, chapter_start, chapter_len;
unsigned int stream_id_plus1, count;
int chapter_id, i;
int64_t value, end;
char name[256], str_value[1024], type_str[256];
const char *type;
int *event_flags = NULL;
AVChapter *chapter = NULL;
AVStream *st = NULL;
AVDictionary **metadata = NULL;
int metadata_flag = 0;
end = get_packetheader(nut, bc, 1, INFO_STARTCODE);
end += avio_tell(bc);
GET_V(stream_id_plus1, tmp <= s->nb_streams);
chapter_id = get_s(bc);
chapter_start = ffio_read_varlen(bc);
chapter_len = ffio_read_varlen(bc);
count = ffio_read_varlen(bc);
if (chapter_id && !stream_id_plus1) {
int64_t start = chapter_start / nut->time_base_count;
chapter = avpriv_new_chapter(s, chapter_id,
nut->time_base[chapter_start %
nut->time_base_count],
start, start + chapter_len, NULL);
if (!chapter) {
av_log(s, AV_LOG_ERROR, "Could not create chapter.\n");
return AVERROR(ENOMEM);
}
metadata = &chapter->metadata;
} else if (stream_id_plus1) {
st = s->streams[stream_id_plus1 - 1];
metadata = &st->metadata;
event_flags = &st->event_flags;
metadata_flag = AVSTREAM_EVENT_FLAG_METADATA_UPDATED;
} else {
metadata = &s->metadata;
event_flags = &s->event_flags;
metadata_flag = AVFMT_EVENT_FLAG_METADATA_UPDATED;
}
for (i = 0; i < count; i++) {
get_str(bc, name, sizeof(name));
value = get_s(bc);
if (value == -1) {
type = "UTF-8";
get_str(bc, str_value, sizeof(str_value));
} else if (value == -2) {
get_str(bc, type_str, sizeof(type_str));
type = type_str;
get_str(bc, str_value, sizeof(str_value));
} else if (value == -3) {
type = "s";
value = get_s(bc);
} else if (value == -4) {
type = "t";
value = ffio_read_varlen(bc);
} else if (value < -4) {
type = "r";
get_s(bc);
} else {
type = "v";
}
if (stream_id_plus1 > s->nb_streams) {
av_log(s, AV_LOG_WARNING,
"invalid stream id %d for info packet\n",
stream_id_plus1);
continue;
}
if (!strcmp(type, "UTF-8")) {
if (chapter_id == 0 && !strcmp(name, "Disposition")) {
set_disposition_bits(s, str_value, stream_id_plus1 - 1);
continue;
}
if (metadata && av_strcasecmp(name, "Uses") &&
av_strcasecmp(name, "Depends") && av_strcasecmp(name, "Replaces")) {
if (event_flags)
*event_flags |= metadata_flag;
av_dict_set(metadata, name, str_value, 0);
}
}
}
if (skip_reserved(bc, end) || ffio_get_checksum(bc)) {
av_log(s, AV_LOG_ERROR, "info header checksum mismatch\n");
return AVERROR_INVALIDDATA;
}
return 0;
}
static int decode_syncpoint(NUTContext *nut, int64_t *ts, int64_t *back_ptr)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
int64_t end, tmp;
int ret;
nut->last_syncpoint_pos = avio_tell(bc) - 8;
end = get_packetheader(nut, bc, 1, SYNCPOINT_STARTCODE);
end += avio_tell(bc);
tmp = ffio_read_varlen(bc);
*back_ptr = nut->last_syncpoint_pos - 16 * ffio_read_varlen(bc);
if (*back_ptr < 0)
return -1;
ff_nut_reset_ts(nut, nut->time_base[tmp % nut->time_base_count],
tmp / nut->time_base_count);
if (nut->flags & NUT_BROADCAST) {
tmp = ffio_read_varlen(bc);
av_log(s, AV_LOG_VERBOSE, "Syncpoint wallclock %"PRId64"\n",
av_rescale_q(tmp / nut->time_base_count,
nut->time_base[tmp % nut->time_base_count],
AV_TIME_BASE_Q));
}
if (skip_reserved(bc, end) || ffio_get_checksum(bc)) {
av_log(s, AV_LOG_ERROR, "sync point checksum mismatch\n");
return AVERROR_INVALIDDATA;
}
*ts = tmp / s->nb_streams *
av_q2d(nut->time_base[tmp % s->nb_streams]) * AV_TIME_BASE;
if ((ret = ff_nut_add_sp(nut, nut->last_syncpoint_pos, *back_ptr, *ts)) < 0)
return ret;
return 0;
}
static int find_and_decode_index(NUTContext *nut)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
uint64_t tmp, end;
int i, j, syncpoint_count;
int64_t filesize = avio_size(bc);
int64_t *syncpoints;
int8_t *has_keyframe;
int ret = AVERROR_INVALIDDATA;
avio_seek(bc, filesize - 12, SEEK_SET);
avio_seek(bc, filesize - avio_rb64(bc), SEEK_SET);
if (avio_rb64(bc) != INDEX_STARTCODE) {
av_log(s, AV_LOG_WARNING, "no index at the end\n");
return ret;
}
end = get_packetheader(nut, bc, 1, INDEX_STARTCODE);
end += avio_tell(bc);
ffio_read_varlen(bc); // max_pts
GET_V(syncpoint_count, tmp < INT_MAX / 8 && tmp > 0);
syncpoints = av_malloc(sizeof(int64_t) * syncpoint_count);
has_keyframe = av_malloc(sizeof(int8_t) * (syncpoint_count + 1));
if (!syncpoints || !has_keyframe) {
ret = AVERROR(ENOMEM);
goto fail;
}
for (i = 0; i < syncpoint_count; i++) {
syncpoints[i] = ffio_read_varlen(bc);
if (syncpoints[i] <= 0)
goto fail;
if (i)
syncpoints[i] += syncpoints[i - 1];
}
for (i = 0; i < s->nb_streams; i++) {
int64_t last_pts = -1;
for (j = 0; j < syncpoint_count;) {
uint64_t x = ffio_read_varlen(bc);
int type = x & 1;
int n = j;
x >>= 1;
if (type) {
int flag = x & 1;
x >>= 1;
if (n + x >= syncpoint_count + 1) {
av_log(s, AV_LOG_ERROR, "index overflow A\n");
goto fail;
}
while (x--)
has_keyframe[n++] = flag;
has_keyframe[n++] = !flag;
} else {
while (x != 1) {
if (n >= syncpoint_count + 1) {
av_log(s, AV_LOG_ERROR, "index overflow B\n");
goto fail;
}
has_keyframe[n++] = x & 1;
x >>= 1;
}
}
if (has_keyframe[0]) {
av_log(s, AV_LOG_ERROR, "keyframe before first syncpoint in index\n");
goto fail;
}
assert(n <= syncpoint_count + 1);
for (; j < n && j < syncpoint_count; j++) {
if (has_keyframe[j]) {
uint64_t B, A = ffio_read_varlen(bc);
if (!A) {
A = ffio_read_varlen(bc);
B = ffio_read_varlen(bc);
// eor_pts[j][i] = last_pts + A + B
} else
B = 0;
av_add_index_entry(s->streams[i], 16 * syncpoints[j - 1],
last_pts + A, 0, 0, AVINDEX_KEYFRAME);
last_pts += A + B;
}
}
}
}
if (skip_reserved(bc, end) || ffio_get_checksum(bc)) {
av_log(s, AV_LOG_ERROR, "index checksum mismatch\n");
goto fail;
}
ret = 0;
fail:
av_free(syncpoints);
av_free(has_keyframe);
return ret;
}
static int nut_read_close(AVFormatContext *s)
{
NUTContext *nut = s->priv_data;
int i;
av_freep(&nut->time_base);
av_freep(&nut->stream);
ff_nut_free_sp(nut);
for (i = 1; i < nut->header_count; i++)
av_freep(&nut->header[i]);
return 0;
}
static int nut_read_header(AVFormatContext *s)
{
NUTContext *nut = s->priv_data;
AVIOContext *bc = s->pb;
int64_t pos;
int initialized_stream_count;
nut->avf = s;
/* main header */
pos = 0;
do {
pos = find_startcode(bc, MAIN_STARTCODE, pos) + 1;
if (pos < 0 + 1) {
av_log(s, AV_LOG_ERROR, "No main startcode found.\n");
goto fail;
}
} while (decode_main_header(nut) < 0);
/* stream headers */
pos = 0;
for (initialized_stream_count = 0; initialized_stream_count < s->nb_streams;) {
pos = find_startcode(bc, STREAM_STARTCODE, pos) + 1;
if (pos < 0 + 1) {
av_log(s, AV_LOG_ERROR, "Not all stream headers found.\n");
goto fail;
}
if (decode_stream_header(nut) >= 0)
initialized_stream_count++;
}
/* info headers */
pos = 0;
for (;;) {
uint64_t startcode = find_any_startcode(bc, pos);
pos = avio_tell(bc);
if (startcode == 0) {
av_log(s, AV_LOG_ERROR, "EOF before video frames\n");
goto fail;
} else if (startcode == SYNCPOINT_STARTCODE) {
nut->next_startcode = startcode;
break;
} else if (startcode != INFO_STARTCODE) {
continue;
}
decode_info_header(nut);
}
s->internal->data_offset = pos - 8;
if (bc->seekable & AVIO_SEEKABLE_NORMAL) {
int64_t orig_pos = avio_tell(bc);
find_and_decode_index(nut);
avio_seek(bc, orig_pos, SEEK_SET);
}
assert(nut->next_startcode == SYNCPOINT_STARTCODE);
ff_metadata_conv_ctx(s, NULL, ff_nut_metadata_conv);
return 0;
fail:
nut_read_close(s);
return AVERROR_INVALIDDATA;
}
static int decode_frame_header(NUTContext *nut, int64_t *pts, int *stream_id,
uint8_t *header_idx, int frame_code)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
StreamContext *stc;
int size, flags, size_mul, pts_delta, i, reserved_count;
uint64_t tmp;
if (!(nut->flags & NUT_PIPE) &&
avio_tell(bc) > nut->last_syncpoint_pos + nut->max_distance) {
av_log(s, AV_LOG_ERROR,
"Last frame must have been damaged %"PRId64" > %"PRId64" + %d\n",
avio_tell(bc), nut->last_syncpoint_pos, nut->max_distance);
return AVERROR_INVALIDDATA;
}
flags = nut->frame_code[frame_code].flags;
size_mul = nut->frame_code[frame_code].size_mul;
size = nut->frame_code[frame_code].size_lsb;
*stream_id = nut->frame_code[frame_code].stream_id;
pts_delta = nut->frame_code[frame_code].pts_delta;
reserved_count = nut->frame_code[frame_code].reserved_count;
*header_idx = nut->frame_code[frame_code].header_idx;
if (flags & FLAG_INVALID)
return AVERROR_INVALIDDATA;
if (flags & FLAG_CODED)
flags ^= ffio_read_varlen(bc);
if (flags & FLAG_STREAM_ID) {
GET_V(*stream_id, tmp < s->nb_streams);
}
stc = &nut->stream[*stream_id];
if (flags & FLAG_CODED_PTS) {
int coded_pts = ffio_read_varlen(bc);
// FIXME check last_pts validity?
if (coded_pts < (1 << stc->msb_pts_shift)) {
*pts = ff_lsb2full(stc, coded_pts);
} else
*pts = coded_pts - (1 << stc->msb_pts_shift);
} else
*pts = stc->last_pts + pts_delta;
if (flags & FLAG_SIZE_MSB)
size += size_mul * ffio_read_varlen(bc);
if (flags & FLAG_MATCH_TIME)
get_s(bc);
if (flags & FLAG_HEADER_IDX)
*header_idx = ffio_read_varlen(bc);
if (flags & FLAG_RESERVED)
reserved_count = ffio_read_varlen(bc);
for (i = 0; i < reserved_count; i++)
ffio_read_varlen(bc);
if (*header_idx >= (unsigned)nut->header_count) {
av_log(s, AV_LOG_ERROR, "header_idx invalid\n");
return AVERROR_INVALIDDATA;
}
if (size > 4096)
*header_idx = 0;
size -= nut->header_len[*header_idx];
if (flags & FLAG_CHECKSUM) {
avio_rb32(bc); // FIXME check this
} else if (!(nut->flags & NUT_PIPE) &&
size > 2 * nut->max_distance ||
FFABS(stc->last_pts - *pts) > stc->max_pts_distance) {
av_log(s, AV_LOG_ERROR, "frame size > 2max_distance and no checksum\n");
return AVERROR_INVALIDDATA;
}
stc->last_pts = *pts;
stc->last_flags = flags;
return size;
}
static int decode_frame(NUTContext *nut, AVPacket *pkt, int frame_code)
{
AVFormatContext *s = nut->avf;
AVIOContext *bc = s->pb;
int size, stream_id, discard, ret;
int64_t pts, last_IP_pts;
StreamContext *stc;
uint8_t header_idx;
size = decode_frame_header(nut, &pts, &stream_id, &header_idx, frame_code);
if (size < 0)
return size;
stc = &nut->stream[stream_id];
if (stc->last_flags & FLAG_KEY)
stc->skip_until_key_frame = 0;
discard = s->streams[stream_id]->discard;
last_IP_pts = s->streams[stream_id]->last_IP_pts;
if ((discard >= AVDISCARD_NONKEY && !(stc->last_flags & FLAG_KEY)) ||
(discard >= AVDISCARD_BIDIR && last_IP_pts != AV_NOPTS_VALUE &&
last_IP_pts > pts) ||
discard >= AVDISCARD_ALL ||
stc->skip_until_key_frame) {
avio_skip(bc, size);
return 1;
}
ret = av_new_packet(pkt, size + nut->header_len[header_idx]);
if (ret < 0)
return ret;
if (nut->header[header_idx])
memcpy(pkt->data, nut->header[header_idx], nut->header_len[header_idx]);
pkt->pos = avio_tell(bc); // FIXME
avio_read(bc, pkt->data + nut->header_len[header_idx], size);
pkt->stream_index = stream_id;
if (stc->last_flags & FLAG_KEY)
pkt->flags |= AV_PKT_FLAG_KEY;
pkt->pts = pts;
return 0;
}
static int nut_read_packet(AVFormatContext *s, AVPacket *pkt)
{
NUTContext *nut = s->priv_data;
AVIOContext *bc = s->pb;
int i, frame_code = 0, ret, skip;
int64_t ts, back_ptr;
for (;;) {
int64_t pos = avio_tell(bc);
uint64_t tmp = nut->next_startcode;
nut->next_startcode = 0;
if (tmp) {
pos -= 8;
} else {
frame_code = avio_r8(bc);
if (bc->eof_reached)
return AVERROR_EOF;
if (frame_code == 'N') {
tmp = frame_code;
for (i = 1; i < 8; i++)
tmp = (tmp << 8) + avio_r8(bc);
}
}
switch (tmp) {
case MAIN_STARTCODE:
case STREAM_STARTCODE:
case INDEX_STARTCODE:
skip = get_packetheader(nut, bc, 0, tmp);
avio_skip(bc, skip);
break;
case INFO_STARTCODE:
if (decode_info_header(nut) < 0)
goto resync;
break;
case SYNCPOINT_STARTCODE:
if (decode_syncpoint(nut, &ts, &back_ptr) < 0)
goto resync;
frame_code = avio_r8(bc);
case 0:
ret = decode_frame(nut, pkt, frame_code);
if (ret == 0)
return 0;
else if (ret == 1) // OK but discard packet
break;
default:
resync:
av_log(s, AV_LOG_DEBUG, "syncing from %"PRId64"\n", pos);
tmp = find_any_startcode(bc, nut->last_syncpoint_pos + 1);
if (tmp == 0)
return AVERROR_INVALIDDATA;
av_log(s, AV_LOG_DEBUG, "sync\n");
nut->next_startcode = tmp;
}
}
}
static int64_t nut_read_timestamp(AVFormatContext *s, int stream_index,
int64_t *pos_arg, int64_t pos_limit)
{
NUTContext *nut = s->priv_data;
AVIOContext *bc = s->pb;
int64_t pos, pts, back_ptr;
av_log(s, AV_LOG_DEBUG, "read_timestamp(X,%d,%"PRId64",%"PRId64")\n",
stream_index, *pos_arg, pos_limit);
pos = *pos_arg;
do {
pos = find_startcode(bc, SYNCPOINT_STARTCODE, pos) + 1;
if (pos < 1) {
assert(nut->next_startcode == 0);
av_log(s, AV_LOG_ERROR, "read_timestamp failed.\n");
return AV_NOPTS_VALUE;
}
} while (decode_syncpoint(nut, &pts, &back_ptr) < 0);
*pos_arg = pos - 1;
assert(nut->last_syncpoint_pos == *pos_arg);
av_log(s, AV_LOG_DEBUG, "return %"PRId64" %"PRId64"\n", pts, back_ptr);
if (stream_index == -1)
return pts;
else if (stream_index == -2)
return back_ptr;
return AV_NOPTS_VALUE;
}
static int read_seek(AVFormatContext *s, int stream_index,
int64_t pts, int flags)
{
NUTContext *nut = s->priv_data;
AVStream *st = s->streams[stream_index];
Syncpoint dummy = { .ts = pts * av_q2d(st->time_base) * AV_TIME_BASE };
Syncpoint nopts_sp = { .ts = AV_NOPTS_VALUE, .back_ptr = AV_NOPTS_VALUE };
Syncpoint *sp, *next_node[2] = { &nopts_sp, &nopts_sp };
int64_t pos, pos2, ts;
int i;
if (nut->flags & NUT_PIPE) {
return AVERROR(ENOSYS);
}
if (st->index_entries) {
int index = av_index_search_timestamp(st, pts, flags);
if (index < 0)
return -1;
pos2 = st->index_entries[index].pos;
ts = st->index_entries[index].timestamp;
} else {
av_tree_find(nut->syncpoints, &dummy,
(int (*)(void *, const void *)) ff_nut_sp_pts_cmp,
(void **) next_node);
av_log(s, AV_LOG_DEBUG, "%"PRIu64"-%"PRIu64" %"PRId64"-%"PRId64"\n",
next_node[0]->pos, next_node[1]->pos, next_node[0]->ts,
next_node[1]->ts);
pos = ff_gen_search(s, -1, dummy.ts, next_node[0]->pos,
next_node[1]->pos, next_node[1]->pos,
next_node[0]->ts, next_node[1]->ts,
AVSEEK_FLAG_BACKWARD, &ts, nut_read_timestamp);
if (!(flags & AVSEEK_FLAG_BACKWARD)) {
dummy.pos = pos + 16;
next_node[1] = &nopts_sp;
av_tree_find(nut->syncpoints, &dummy,
(int (*)(void *, const void *)) ff_nut_sp_pos_cmp,
(void **) next_node);
pos2 = ff_gen_search(s, -2, dummy.pos, next_node[0]->pos,
next_node[1]->pos, next_node[1]->pos,
next_node[0]->back_ptr, next_node[1]->back_ptr,
flags, &ts, nut_read_timestamp);
if (pos2 >= 0)
pos = pos2;
// FIXME dir but I think it does not matter
}
dummy.pos = pos;
sp = av_tree_find(nut->syncpoints, &dummy,
(int (*)(void *, const void *)) ff_nut_sp_pos_cmp,
NULL);
assert(sp);
pos2 = sp->back_ptr - 15;
}
av_log(NULL, AV_LOG_DEBUG, "SEEKTO: %"PRId64"\n", pos2);
pos = find_startcode(s->pb, SYNCPOINT_STARTCODE, pos2);
avio_seek(s->pb, pos, SEEK_SET);
av_log(NULL, AV_LOG_DEBUG, "SP: %"PRId64"\n", pos);
if (pos2 > pos || pos2 + 15 < pos)
av_log(NULL, AV_LOG_ERROR, "no syncpoint at backptr pos\n");
for (i = 0; i < s->nb_streams; i++)
nut->stream[i].skip_until_key_frame = 1;
return 0;
}
AVInputFormat ff_nut_demuxer = {
.name = "nut",
.long_name = NULL_IF_CONFIG_SMALL("NUT"),
.priv_data_size = sizeof(NUTContext),
.read_probe = nut_probe,
.read_header = nut_read_header,
.read_packet = nut_read_packet,
.read_close = nut_read_close,
.read_seek = read_seek,
.extensions = "nut",
.codec_tag = ff_nut_codec_tags,
};