mirror of https://git.ffmpeg.org/ffmpeg.git
375 lines
12 KiB
C
375 lines
12 KiB
C
/*
|
|
* FFM (ffserver live feed) muxer
|
|
* Copyright (c) 2001 Fabrice Bellard
|
|
*
|
|
* 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
|
|
*/
|
|
|
|
#include "libavutil/intreadwrite.h"
|
|
#include "libavutil/intfloat.h"
|
|
#include "libavutil/avassert.h"
|
|
#include "libavutil/parseutils.h"
|
|
#include "libavutil/opt.h"
|
|
#include "avformat.h"
|
|
#include "avio_internal.h"
|
|
#include "internal.h"
|
|
#include "ffm.h"
|
|
|
|
static void flush_packet(AVFormatContext *s)
|
|
{
|
|
FFMContext *ffm = s->priv_data;
|
|
int fill_size, h;
|
|
AVIOContext *pb = s->pb;
|
|
|
|
fill_size = ffm->packet_end - ffm->packet_ptr;
|
|
memset(ffm->packet_ptr, 0, fill_size);
|
|
|
|
av_assert1(avio_tell(pb) % ffm->packet_size == 0);
|
|
|
|
/* put header */
|
|
avio_wb16(pb, PACKET_ID);
|
|
avio_wb16(pb, fill_size);
|
|
avio_wb64(pb, ffm->dts);
|
|
h = ffm->frame_offset;
|
|
if (ffm->first_packet)
|
|
h |= 0x8000;
|
|
avio_wb16(pb, h);
|
|
avio_write(pb, ffm->packet, ffm->packet_end - ffm->packet);
|
|
avio_flush(pb);
|
|
|
|
/* prepare next packet */
|
|
ffm->frame_offset = 0; /* no key frame */
|
|
ffm->packet_ptr = ffm->packet;
|
|
ffm->first_packet = 0;
|
|
}
|
|
|
|
/* 'first' is true if first data of a frame */
|
|
static void ffm_write_data(AVFormatContext *s,
|
|
const uint8_t *buf, int size,
|
|
int64_t dts, int header)
|
|
{
|
|
FFMContext *ffm = s->priv_data;
|
|
int len;
|
|
|
|
if (header && ffm->frame_offset == 0) {
|
|
ffm->frame_offset = ffm->packet_ptr - ffm->packet + FFM_HEADER_SIZE;
|
|
ffm->dts = dts;
|
|
}
|
|
|
|
/* write as many packets as needed */
|
|
while (size > 0) {
|
|
len = ffm->packet_end - ffm->packet_ptr;
|
|
if (len > size)
|
|
len = size;
|
|
memcpy(ffm->packet_ptr, buf, len);
|
|
|
|
ffm->packet_ptr += len;
|
|
buf += len;
|
|
size -= len;
|
|
if (ffm->packet_ptr >= ffm->packet_end)
|
|
flush_packet(s);
|
|
}
|
|
}
|
|
|
|
static void write_header_chunk(AVIOContext *pb, AVIOContext *dpb, unsigned id)
|
|
{
|
|
uint8_t *dyn_buf;
|
|
int dyn_size= avio_close_dyn_buf(dpb, &dyn_buf);
|
|
avio_wb32(pb, id);
|
|
avio_wb32(pb, dyn_size);
|
|
avio_write(pb, dyn_buf, dyn_size);
|
|
av_free(dyn_buf);
|
|
}
|
|
|
|
static int ffm_write_header_codec_private_ctx(AVFormatContext *s, AVCodecContext *ctx, int type)
|
|
{
|
|
AVIOContext *pb = s->pb;
|
|
AVIOContext *tmp;
|
|
char *buf = NULL;
|
|
int ret;
|
|
const AVCodec *enc = ctx->codec ? ctx->codec : avcodec_find_encoder(ctx->codec_id);
|
|
|
|
if (!enc) {
|
|
av_log(s, AV_LOG_WARNING, "Stream codec is not found. Codec private options are not stored.\n");
|
|
return 0;
|
|
}
|
|
if (ctx->priv_data && enc->priv_class && enc->priv_data_size) {
|
|
if ((ret = av_opt_serialize(ctx->priv_data, AV_OPT_FLAG_ENCODING_PARAM | type,
|
|
AV_OPT_SERIALIZE_SKIP_DEFAULTS, &buf, '=', ',')) < 0)
|
|
return ret;
|
|
if (buf && strlen(buf)) {
|
|
if (avio_open_dyn_buf(&tmp) < 0) {
|
|
av_free(buf);
|
|
return AVERROR(ENOMEM);
|
|
}
|
|
avio_put_str(tmp, buf);
|
|
write_header_chunk(pb, tmp, MKBETAG('C', 'P', 'R', 'V'));
|
|
}
|
|
av_free(buf);
|
|
}
|
|
return 0;
|
|
}
|
|
|
|
static int ffm_write_header_codec_ctx(AVIOContext *pb, AVCodecContext *ctx, unsigned tag, int type)
|
|
{
|
|
AVIOContext *tmp;
|
|
char *buf = NULL;
|
|
int ret, need_coma = 0;
|
|
|
|
#define SKIP_DEFAULTS AV_OPT_SERIALIZE_SKIP_DEFAULTS
|
|
#define OPT_FLAGS_EXACT AV_OPT_SERIALIZE_OPT_FLAGS_EXACT
|
|
#define ENC AV_OPT_FLAG_ENCODING_PARAM
|
|
|
|
if (avio_open_dyn_buf(&tmp) < 0)
|
|
return AVERROR(ENOMEM);
|
|
if ((ret = av_opt_serialize(ctx, ENC | type, SKIP_DEFAULTS, &buf, '=', ',')) < 0)
|
|
goto fail;
|
|
if (buf && strlen(buf)) {
|
|
avio_write(tmp, buf, strlen(buf));
|
|
av_freep(&buf);
|
|
need_coma = 1;
|
|
}
|
|
if ((ret = av_opt_serialize(ctx, 0, SKIP_DEFAULTS | OPT_FLAGS_EXACT, &buf, '=', ',')) < 0)
|
|
goto fail;
|
|
if (buf && strlen(buf)) {
|
|
if (need_coma)
|
|
avio_w8(tmp, ',');
|
|
avio_write(tmp, buf, strlen(buf));
|
|
}
|
|
av_freep(&buf);
|
|
avio_w8(tmp, 0);
|
|
write_header_chunk(pb, tmp, tag);
|
|
return 0;
|
|
fail:
|
|
av_free(buf);
|
|
ffio_free_dyn_buf(&tmp);
|
|
return ret;
|
|
|
|
#undef SKIP_DEFAULTS
|
|
#undef OPT_FLAGS_EXACT
|
|
#undef ENC
|
|
}
|
|
|
|
static int ffm_write_recommended_config(AVIOContext *pb, AVCodecContext *ctx, unsigned tag,
|
|
const char *configuration)
|
|
{
|
|
int ret;
|
|
const AVCodec *enc = ctx->codec ? ctx->codec : avcodec_find_encoder(ctx->codec_id);
|
|
AVIOContext *tmp;
|
|
AVDictionaryEntry *t = NULL;
|
|
AVDictionary *all = NULL, *comm = NULL, *prv = NULL;
|
|
char *buf = NULL;
|
|
|
|
if (!enc || !enc->priv_class || !enc->priv_data_size) {
|
|
/* codec is not known/has no private options, so save everything as common options */
|
|
if (avio_open_dyn_buf(&tmp) < 0)
|
|
return AVERROR(ENOMEM);
|
|
avio_put_str(tmp, configuration);
|
|
write_header_chunk(pb, tmp, tag);
|
|
return 0;
|
|
}
|
|
|
|
if ((ret = av_dict_parse_string(&all, configuration, "=", ",", 0)) < 0)
|
|
return ret;
|
|
|
|
while ((t = av_dict_get(all, "", t, AV_DICT_IGNORE_SUFFIX))) {
|
|
if (av_opt_find((void *)&enc->priv_class, t->key, NULL, 0, AV_OPT_SEARCH_FAKE_OBJ)) {
|
|
if ((ret = av_dict_set(&prv, t->key, t->value, 0)) < 0)
|
|
goto fail;
|
|
} else if ((ret = av_dict_set(&comm, t->key, t->value, 0)) < 0)
|
|
goto fail;
|
|
}
|
|
|
|
if (comm) {
|
|
if ((ret = av_dict_get_string(comm, &buf, '=', ',')) < 0 ||
|
|
(ret = avio_open_dyn_buf(&tmp)) < 0)
|
|
goto fail;
|
|
avio_put_str(tmp, buf);
|
|
av_freep(&buf);
|
|
write_header_chunk(pb, tmp, tag);
|
|
}
|
|
if (prv) {
|
|
if ((ret = av_dict_get_string(prv, &buf, '=', ',')) < 0 ||
|
|
(ret = avio_open_dyn_buf(&tmp)) < 0)
|
|
goto fail;
|
|
avio_put_str(tmp, buf);
|
|
write_header_chunk(pb, tmp, MKBETAG('C', 'P', 'R', 'V'));
|
|
}
|
|
|
|
fail:
|
|
av_free(buf);
|
|
av_dict_free(&all);
|
|
av_dict_free(&comm);
|
|
av_dict_free(&prv);
|
|
return ret;
|
|
}
|
|
|
|
static int ffm_write_header(AVFormatContext *s)
|
|
{
|
|
FFMContext *ffm = s->priv_data;
|
|
AVDictionaryEntry *t;
|
|
AVStream *st;
|
|
AVIOContext *pb = s->pb;
|
|
AVCodecContext *codec;
|
|
int bit_rate, i, ret;
|
|
|
|
if (t = av_dict_get(s->metadata, "creation_time", NULL, 0)) {
|
|
ret = av_parse_time(&ffm->start_time, t->value, 0);
|
|
if (ret < 0)
|
|
return ret;
|
|
}
|
|
|
|
ffm->packet_size = FFM_PACKET_SIZE;
|
|
|
|
/* header */
|
|
avio_wl32(pb, MKTAG('F', 'F', 'M', '2'));
|
|
avio_wb32(pb, ffm->packet_size);
|
|
avio_wb64(pb, 0); /* current write position */
|
|
|
|
if(avio_open_dyn_buf(&pb) < 0)
|
|
return AVERROR(ENOMEM);
|
|
|
|
avio_wb32(pb, s->nb_streams);
|
|
bit_rate = 0;
|
|
for(i=0;i<s->nb_streams;i++) {
|
|
st = s->streams[i];
|
|
bit_rate += st->codec->bit_rate;
|
|
}
|
|
avio_wb32(pb, bit_rate);
|
|
|
|
write_header_chunk(s->pb, pb, MKBETAG('M', 'A', 'I', 'N'));
|
|
|
|
/* list of streams */
|
|
for(i=0;i<s->nb_streams;i++) {
|
|
st = s->streams[i];
|
|
avpriv_set_pts_info(st, 64, 1, 1000000);
|
|
if(avio_open_dyn_buf(&pb) < 0)
|
|
return AVERROR(ENOMEM);
|
|
|
|
codec = st->codec;
|
|
/* generic info */
|
|
avio_wb32(pb, codec->codec_id);
|
|
avio_w8(pb, codec->codec_type);
|
|
avio_wb32(pb, codec->bit_rate);
|
|
avio_wb32(pb, codec->flags);
|
|
avio_wb32(pb, codec->flags2);
|
|
avio_wb32(pb, codec->debug);
|
|
if (codec->flags & AV_CODEC_FLAG_GLOBAL_HEADER) {
|
|
avio_wb32(pb, codec->extradata_size);
|
|
avio_write(pb, codec->extradata, codec->extradata_size);
|
|
}
|
|
write_header_chunk(s->pb, pb, MKBETAG('C', 'O', 'M', 'M'));
|
|
/* specific info */
|
|
switch(codec->codec_type) {
|
|
case AVMEDIA_TYPE_VIDEO:
|
|
if (st->recommended_encoder_configuration) {
|
|
av_log(NULL, AV_LOG_DEBUG, "writing recommended configuration: %s\n",
|
|
st->recommended_encoder_configuration);
|
|
if ((ret = ffm_write_recommended_config(s->pb, codec, MKBETAG('S', '2', 'V', 'I'),
|
|
st->recommended_encoder_configuration)) < 0)
|
|
return ret;
|
|
} else if ((ret = ffm_write_header_codec_ctx(s->pb, codec, MKBETAG('S', '2', 'V', 'I'), AV_OPT_FLAG_VIDEO_PARAM)) < 0 ||
|
|
(ret = ffm_write_header_codec_private_ctx(s, codec, AV_OPT_FLAG_VIDEO_PARAM)) < 0)
|
|
return ret;
|
|
break;
|
|
case AVMEDIA_TYPE_AUDIO:
|
|
if (st->recommended_encoder_configuration) {
|
|
av_log(NULL, AV_LOG_DEBUG, "writing recommended configuration: %s\n",
|
|
st->recommended_encoder_configuration);
|
|
if ((ret = ffm_write_recommended_config(s->pb, codec, MKBETAG('S', '2', 'A', 'U'),
|
|
st->recommended_encoder_configuration)) < 0)
|
|
return ret;
|
|
} else if ((ret = ffm_write_header_codec_ctx(s->pb, codec, MKBETAG('S', '2', 'A', 'U'), AV_OPT_FLAG_AUDIO_PARAM)) < 0 ||
|
|
(ret = ffm_write_header_codec_private_ctx(s, codec, AV_OPT_FLAG_AUDIO_PARAM)) < 0)
|
|
return ret;
|
|
break;
|
|
default:
|
|
return -1;
|
|
}
|
|
}
|
|
pb = s->pb;
|
|
|
|
avio_wb64(pb, 0); // end of header
|
|
|
|
/* flush until end of block reached */
|
|
while ((avio_tell(pb) % ffm->packet_size) != 0)
|
|
avio_w8(pb, 0);
|
|
|
|
avio_flush(pb);
|
|
|
|
/* init packet mux */
|
|
ffm->packet_ptr = ffm->packet;
|
|
ffm->packet_end = ffm->packet + ffm->packet_size - FFM_HEADER_SIZE;
|
|
av_assert0(ffm->packet_end >= ffm->packet);
|
|
ffm->frame_offset = 0;
|
|
ffm->dts = 0;
|
|
ffm->first_packet = 1;
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int ffm_write_packet(AVFormatContext *s, AVPacket *pkt)
|
|
{
|
|
FFMContext *ffm = s->priv_data;
|
|
int64_t dts;
|
|
uint8_t header[FRAME_HEADER_SIZE+4];
|
|
int header_size = FRAME_HEADER_SIZE;
|
|
|
|
dts = ffm->start_time + pkt->dts;
|
|
/* packet size & key_frame */
|
|
header[0] = pkt->stream_index;
|
|
header[1] = 0;
|
|
if (pkt->flags & AV_PKT_FLAG_KEY)
|
|
header[1] |= FLAG_KEY_FRAME;
|
|
AV_WB24(header+2, pkt->size);
|
|
AV_WB24(header+5, pkt->duration);
|
|
AV_WB64(header+8, ffm->start_time + pkt->pts);
|
|
if (pkt->pts != pkt->dts) {
|
|
header[1] |= FLAG_DTS;
|
|
AV_WB32(header+16, pkt->pts - pkt->dts);
|
|
header_size += 4;
|
|
}
|
|
ffm_write_data(s, header, header_size, dts, 1);
|
|
ffm_write_data(s, pkt->data, pkt->size, dts, 0);
|
|
|
|
return 0;
|
|
}
|
|
|
|
static int ffm_write_trailer(AVFormatContext *s)
|
|
{
|
|
FFMContext *ffm = s->priv_data;
|
|
|
|
/* flush packets */
|
|
if (ffm->packet_ptr > ffm->packet)
|
|
flush_packet(s);
|
|
|
|
return 0;
|
|
}
|
|
|
|
AVOutputFormat ff_ffm_muxer = {
|
|
.name = "ffm",
|
|
.long_name = NULL_IF_CONFIG_SMALL("FFM (FFserver live feed)"),
|
|
.extensions = "ffm",
|
|
.priv_data_size = sizeof(FFMContext),
|
|
.audio_codec = AV_CODEC_ID_MP2,
|
|
.video_codec = AV_CODEC_ID_MPEG1VIDEO,
|
|
.write_header = ffm_write_header,
|
|
.write_packet = ffm_write_packet,
|
|
.write_trailer = ffm_write_trailer,
|
|
.flags = AVFMT_TS_NEGATIVE,
|
|
};
|