#include <time.h>
#include "avformat.h"
#include "dvdata.h"
+#include "dv.h"
+
+struct DVDemuxContext {
+ AVFormatContext* fctx;
+ AVStream* vst;
+ AVStream* ast[2];
+ AVPacket audio_pkt[2];
+ uint8_t audio_buf[2][8192];
+ int ach;
+ int frames;
+ uint64_t abytes;
+};
-typedef struct DVDemuxContext {
- AVPacket audio_pkt;
- AVStream *vst;
- AVStream *ast;
-} DVDemuxContext;
-
-typedef struct DVMuxContext {
+struct DVMuxContext {
const DVprofile* sys; /* Current DV profile. E.g.: 525/60, 625/50 */
uint8_t frame_buf[144000]; /* frame under contruction */
FifoBuffer audio_data; /* Fifo for storing excessive amounts of PCM */
uint8_t aspect; /* Aspect ID 0 - 4:3, 7 - 16:9 */
int has_audio; /* frame under contruction has audio */
int has_video; /* frame under contruction has video */
-} DVMuxContext;
+};
enum dv_section_type {
dv_sect_header = 0x1f,
case dv_timecode:
ct = (time_t)(c->frames / ((float)c->sys->frame_rate /
(float)c->sys->frame_rate_base));
- localtime_r(&ct, &tc);
+ brktimegm(ct, &tc);
/*
* LTC drop-frame frame counter drops two frames (0 and 1) every
* minute, unless it is exactly divisible by 10
case dv_viedo_recdate: /* VAUX recording date */
ct = c->start_time + (time_t)(c->frames /
((float)c->sys->frame_rate / (float)c->sys->frame_rate_base));
- localtime_r(&ct, &tc);
+ brktimegm(ct, &tc);
buf[1] = 0xff; /* ds, tm, tens of time zone, units of time zone */
/* 0xff is very likely to be "unknown" */
buf[2] = (3 << 6) | /* reserved -- always 1 */
((tc.tm_mday / 10) << 4) | /* Tens of day */
(tc.tm_mday % 10); /* Units of day */
buf[3] = /* we set high 4 bits to 0, shouldn't we set them to week? */
- (tc.tm_mon % 10); /* Units of month */
+ ((tc.tm_mon / 10) << 4) | /* Tens of month */
+ (tc.tm_mon % 10); /* Units of month */
buf[4] = (((tc.tm_year % 100) / 10) << 4) | /* Tens of year */
(tc.tm_year % 10); /* Units of year */
break;
case dv_video_rectime: /* VAUX recording time */
ct = c->start_time + (time_t)(c->frames /
((float)c->sys->frame_rate / (float)c->sys->frame_rate_base));
- localtime_r(&ct, &tc);
+ brktimegm(ct, &tc);
buf[1] = (3 << 6) | /* reserved -- always 1 */
0x3f; /* tens of frame, units of frame: 0x3f - "unknown" ? */
buf[2] = (1 << 7) | /* reserved -- always 1 */
/* DV Audio/Video: 135 Video DIFs + 9 Audio DIFs */
for (j = 0; j < 135; j++) {
if (j%15 == 0) {
+ memset(buf, 0xff, 80);
buf += dv_write_dif_id(dv_sect_audio, i, j/15, buf);
- buf += dv_write_pack(dv_aaux_packs_dist[i][j/15], c, buf);
- buf += 72; /* shuffled PCM audio */
+ buf += 77; /* audio control & shuffled PCM audio */
}
buf += dv_write_dif_id(dv_sect_video, i, j, buf);
buf += 77; /* 1 video macro block: 1 bytes control
static void dv_inject_audio(DVMuxContext *c, const uint8_t* pcm, uint8_t* frame_ptr)
{
- int i, j, d, of;
+ int i, j, d, of, size;
+ size = 4 * dv_audio_frame_size(c->sys, c->frames);
for (i = 0; i < c->sys->difseg_size; i++) {
frame_ptr += 6 * 80; /* skip DIF segment header */
for (j = 0; j < 9; j++) {
+ dv_write_pack(dv_aaux_packs_dist[i][j], c, &frame_ptr[3]);
for (d = 8; d < 80; d+=2) {
of = c->sys->audio_shuffle[i][j] + (d - 8)/2 * c->sys->audio_stride;
+ if (of*2 >= size)
+ continue;
+
frame_ptr[d] = pcm[of*2+1]; // FIXME: may be we have to admit
frame_ptr[d+1] = pcm[of*2]; // that DV is a big endian PCM
}
* 3. Audio is always returned as 16bit linear samples: 12bit nonlinear samples
* are converted into 16bit linear ones.
*/
-static int dv_extract_audio(uint8_t* frame, uint8_t* pcm)
+static int dv_extract_audio(uint8_t* frame, uint8_t* pcm, uint8_t* pcm2)
{
- int size, i, j, d, of, smpls, freq, quant;
+ int size, i, j, d, of, smpls, freq, quant, half_ch;
uint16_t lc, rc;
const DVprofile* sys;
const uint8_t* as_pack;
return -1; /* Unsupported quantization */
size = (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */
+ half_ch = sys->difseg_size/2;
/* for each DIF segment */
for (i = 0; i < sys->difseg_size; i++) {
frame += 6 * 80; /* skip DIF segment header */
+ if (quant == 1 && i == half_ch) {
+ if (!pcm2)
+ break;
+ else
+ pcm = pcm2;
+ }
+
for (j = 0; j < 9; j++) {
for (d = 8; d < 80; d += 2) {
if (quant == 0) { /* 16bit quantization */
if (pcm[of*2+1] == 0x80 && pcm[of*2] == 0x00)
pcm[of*2+1] = 0;
} else { /* 12bit quantization */
- if (i >= sys->difseg_size/2)
- goto out; /* We're not doing 4ch at this time */
-
lc = ((uint16_t)frame[d] << 4) |
((uint16_t)frame[d+2] >> 4);
rc = ((uint16_t)frame[d+1] << 4) |
lc = (lc == 0x800 ? 0 : dv_audio_12to16(lc));
rc = (rc == 0x800 ? 0 : dv_audio_12to16(rc));
- of = sys->audio_shuffle[i][j] + (d - 8)/3 * sys->audio_stride;
+ of = sys->audio_shuffle[i%half_ch][j] + (d - 8)/3 * sys->audio_stride;
if (of*2 >= size)
continue;
pcm[of*2] = lc & 0xff; // FIXME: may be we have to admit
pcm[of*2+1] = lc >> 8; // that DV is a big endian PCM
- of = sys->audio_shuffle[i+sys->difseg_size/2][j] +
+ of = sys->audio_shuffle[i%half_ch+half_ch][j] +
(d - 8)/3 * sys->audio_stride;
pcm[of*2] = rc & 0xff; // FIXME: may be we have to admit
pcm[of*2+1] = rc >> 8; // that DV is a big endian PCM
}
}
-out:
return size;
}
-static int dv_extract_audio_info(uint8_t* frame, AVCodecContext* avctx)
+static int dv_extract_audio_info(DVDemuxContext* c, uint8_t* frame)
{
const uint8_t* as_pack;
const DVprofile* sys;
- int freq, smpls;
+ int freq, smpls, quant, i, ach;
sys = dv_frame_profile(frame);
as_pack = dv_extract_pack(frame, dv_audio_source);
- if (!as_pack || !sys) /* No audio ? */
- return -1;
+ if (!as_pack || !sys) { /* No audio ? */
+ c->ach = 0;
+ return 0;
+ }
smpls = as_pack[1] & 0x3f; /* samples in this frame - min. samples */
freq = (as_pack[4] >> 3) & 0x07; /* 0 - 48KHz, 1 - 44,1kHz, 2 - 32 kHz */
-
- avctx->sample_rate = dv_audio_frequency[freq];
- avctx->channels = 2;
- avctx->bit_rate = avctx->channels * avctx->sample_rate * 16;
-
+ quant = as_pack[4] & 0x07; /* 0 - 16bit linear, 1 - 12bit nonlinear */
+ ach = (quant && freq == 2) ? 2 : 1;
+
+ /* Dynamic handling of the audio streams in DV */
+ for (i=0; i<ach; i++) {
+ if (!c->ast[i]) {
+ c->ast[i] = av_new_stream(c->fctx, 0);
+ if (!c->ast[i])
+ break;
+ av_set_pts_info(c->ast[i], 64, 1, 30000);
+ c->ast[i]->codec->codec_type = CODEC_TYPE_AUDIO;
+ c->ast[i]->codec->codec_id = CODEC_ID_PCM_S16LE;
+
+ av_init_packet(&c->audio_pkt[i]);
+ c->audio_pkt[i].size = 0;
+ c->audio_pkt[i].data = c->audio_buf[i];
+ c->audio_pkt[i].stream_index = c->ast[i]->index;
+ c->audio_pkt[i].flags |= PKT_FLAG_KEY;
+ }
+ c->ast[i]->codec->sample_rate = dv_audio_frequency[freq];
+ c->ast[i]->codec->channels = 2;
+ c->ast[i]->codec->bit_rate = 2 * dv_audio_frequency[freq] * 16;
+ c->ast[i]->start_time = 0;
+ }
+ c->ach = i;
+
return (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */;
}
-static int dv_extract_video_info(uint8_t* frame, AVCodecContext* avctx)
+static int dv_extract_video_info(DVDemuxContext *c, uint8_t* frame)
{
const DVprofile* sys;
const uint8_t* vsc_pack;
- int apt;
+ AVCodecContext* avctx;
+ int apt, is16_9;
int size = 0;
sys = dv_frame_profile(frame);
if (sys) {
- apt = frame[4] & 0x07;
- avctx->frame_rate = sys->frame_rate;
- avctx->frame_rate_base = sys->frame_rate_base;
+ avctx = c->vst->codec;
+
+ av_set_pts_info(c->vst, 64, sys->frame_rate_base, sys->frame_rate);
+ avctx->time_base= (AVRational){sys->frame_rate_base, sys->frame_rate};
avctx->width = sys->width;
avctx->height = sys->height;
avctx->pix_fmt = sys->pix_fmt;
- vsc_pack = dv_extract_pack(frame, dv_video_control);
- if (vsc_pack && (vsc_pack[2] & 0x07) == (apt?0x02:0x07))
- avctx->aspect_ratio = 16.0 / 9.0;
- else
- avctx->aspect_ratio = 4.0 / 3.0;
+
+ /* finding out SAR is a little bit messy */
+ vsc_pack = dv_extract_pack(frame, dv_video_control);
+ apt = frame[4] & 0x07;
+ is16_9 = (vsc_pack && ((vsc_pack[2] & 0x07) == 0x02 ||
+ (!apt && (vsc_pack[2] & 0x07) == 0x07)));
+ avctx->sample_aspect_ratio = sys->sar[is16_9];
+
size = sys->frame_size;
}
return size;
if (c->has_audio && c->has_video) { /* must be a stale frame */
dv_format_frame(c, *frame);
c->frames++;
- c->has_audio = c->has_video = 0;
+ if (c->has_audio > 0)
+ c->has_audio = 0;
+ c->has_video = 0;
}
- if (st->codec.codec_type == CODEC_TYPE_VIDEO) {
+ if (st->codec->codec_type == CODEC_TYPE_VIDEO) {
/* FIXME: we have to have more sensible approach than this one */
if (c->has_video)
- fprintf(stderr, "Can't process DV frame #%d. Insufficient audio data or severe sync problem.\n", c->frames);
+ av_log(st->codec, AV_LOG_ERROR, "Can't process DV frame #%d. Insufficient audio data or severe sync problem.\n", c->frames);
dv_inject_video(c, data, *frame);
c->has_video = 1;
data_size = 0;
+ if (c->has_audio < 0)
+ goto out;
}
reqasize = 4 * dv_audio_frame_size(c->sys, c->frames);
fsize = fifo_size(&c->audio_data, c->audio_data.rptr);
- if (st->codec.codec_type == CODEC_TYPE_AUDIO || (c->has_video && fsize >= reqasize)) {
+ if (st->codec->codec_type == CODEC_TYPE_AUDIO || (c->has_video && fsize >= reqasize)) {
if (fsize + data_size >= reqasize && !c->has_audio) {
if (fsize >= reqasize) {
fifo_read(&c->audio_data, &pcm[0], reqasize, &c->audio_data.rptr);
/* FIXME: we have to have more sensible approach than this one */
if (fifo_size(&c->audio_data, c->audio_data.rptr) + data_size >= 100*AVCODEC_MAX_AUDIO_FRAME_SIZE)
- fprintf(stderr, "Can't process DV frame #%d. Insufficient video data or severe sync problem.\n", c->frames);
+ av_log(st->codec, AV_LOG_ERROR, "Can't process DV frame #%d. Insufficient video data or severe sync problem.\n", c->frames);
fifo_write(&c->audio_data, (uint8_t *)data, data_size, &c->audio_data.wptr);
}
+out:
return (c->has_audio && c->has_video) ? c->sys->frame_size : 0;
}
DVMuxContext* dv_init_mux(AVFormatContext* s)
{
DVMuxContext *c;
- AVStream *vst;
- AVStream *ast;
+ AVStream *vst = NULL;
+ AVStream *ast = NULL;
+ int i;
+
+ if (s->nb_streams > 2)
+ return NULL;
c = av_mallocz(sizeof(DVMuxContext));
if (!c)
return NULL;
- if (s->nb_streams != 2)
- goto bail_out;
-
/* We have to sort out where audio and where video stream is */
- if (s->streams[0]->codec.codec_type == CODEC_TYPE_VIDEO &&
- s->streams[1]->codec.codec_type == CODEC_TYPE_AUDIO) {
- vst = s->streams[0];
- ast = s->streams[1];
+ for (i=0; i<s->nb_streams; i++) {
+ switch (s->streams[i]->codec->codec_type) {
+ case CODEC_TYPE_VIDEO:
+ vst = s->streams[i];
+ break;
+ case CODEC_TYPE_AUDIO:
+ ast = s->streams[i];
+ break;
+ default:
+ goto bail_out;
+ }
}
- else if (s->streams[1]->codec.codec_type == CODEC_TYPE_VIDEO &&
- s->streams[0]->codec.codec_type == CODEC_TYPE_AUDIO) {
- vst = s->streams[1];
- ast = s->streams[0];
- } else
- goto bail_out;
-
+
/* Some checks -- DV format is very picky about its incoming streams */
- if (vst->codec.codec_id != CODEC_ID_DVVIDEO ||
- ast->codec.codec_id != CODEC_ID_PCM_S16LE)
- goto bail_out;
- if (ast->codec.sample_rate != 48000 ||
- ast->codec.channels != 2)
- goto bail_out;
-
- c->sys = dv_codec_profile(&vst->codec);
+ if (!vst || vst->codec->codec_id != CODEC_ID_DVVIDEO)
+ goto bail_out;
+ if (ast && (ast->codec->codec_id != CODEC_ID_PCM_S16LE ||
+ ast->codec->sample_rate != 48000 ||
+ ast->codec->channels != 2))
+ goto bail_out;
+ c->sys = dv_codec_profile(vst->codec);
if (!c->sys)
goto bail_out;
/* Ok, everything seems to be in working order */
c->frames = 0;
- c->has_audio = c->has_video = 0;
- c->start_time = time(NULL);
+ c->has_audio = ast ? 0 : -1;
+ c->has_video = 0;
+ c->start_time = (time_t)s->timestamp;
c->aspect = 0; /* 4:3 is the default */
- if (vst->codec.aspect_ratio == 16.0 / 9.0)
+ if ((int)(av_q2d(vst->codec->sample_aspect_ratio) * vst->codec->width / vst->codec->height * 10) == 17) /* 16:9 */
c->aspect = 0x07;
- if (fifo_init(&c->audio_data, 100*AVCODEC_MAX_AUDIO_FRAME_SIZE) < 0)
+ if (ast && fifo_init(&c->audio_data, 100*AVCODEC_MAX_AUDIO_FRAME_SIZE) < 0)
goto bail_out;
dv_format_frame(c, &c->frame_buf[0]);
void dv_delete_mux(DVMuxContext *c)
{
fifo_free(&c->audio_data);
- av_free(c);
}
-DVDemuxContext* dv_init_demux(AVFormatContext *s, int vid, int aid)
+DVDemuxContext* dv_init_demux(AVFormatContext *s)
{
DVDemuxContext *c;
if (!c)
return NULL;
- c->vst = (vid < s->nb_streams) ? s->streams[vid] : av_new_stream(s, vid);
- c->ast = (aid < s->nb_streams) ? s->streams[aid] : av_new_stream(s, aid);
- if (!c->vst || !c->ast)
- goto fail;
-
- c->vst->codec.codec_type = CODEC_TYPE_VIDEO;
- c->vst->codec.codec_id = CODEC_ID_DVVIDEO;
- c->vst->codec.bit_rate = 25000000;
-
- c->ast->codec.codec_type = CODEC_TYPE_AUDIO;
- c->ast->codec.codec_id = CODEC_ID_PCM_S16LE;
-
- c->audio_pkt.size = 0;
+ c->vst = av_new_stream(s, 0);
+ if (!c->vst) {
+ av_free(c);
+ return NULL;
+ }
+
+ c->fctx = s;
+ c->ast[0] = c->ast[1] = NULL;
+ c->ach = 0;
+ c->frames = 0;
+ c->abytes = 0;
+
+ c->vst->codec->codec_type = CODEC_TYPE_VIDEO;
+ c->vst->codec->codec_id = CODEC_ID_DVVIDEO;
+ c->vst->codec->bit_rate = 25000000;
+ c->vst->start_time = 0;
return c;
-
-fail:
- if (c->vst)
- av_free(c->vst);
- if (c->ast)
- av_free(c->ast);
- av_free(c);
- return NULL;
-}
-
-static void __destruct_pkt(struct AVPacket *pkt)
-{
- pkt->data = NULL; pkt->size = 0;
- return;
}
int dv_get_packet(DVDemuxContext *c, AVPacket *pkt)
{
int size = -1;
-
- if (c->audio_pkt.size) {
- *pkt = c->audio_pkt;
- c->audio_pkt.size = 0;
- size = pkt->size;
+ int i;
+
+ for (i=0; i<c->ach; i++) {
+ if (c->ast[i] && c->audio_pkt[i].size) {
+ *pkt = c->audio_pkt[i];
+ c->audio_pkt[i].size = 0;
+ size = pkt->size;
+ break;
+ }
}
return size;
int dv_produce_packet(DVDemuxContext *c, AVPacket *pkt,
uint8_t* buf, int buf_size)
{
- int size;
+ int size, i;
+ const DVprofile* sys = dv_frame_profile(buf);
- if (buf_size < 4 || buf_size < dv_frame_profile(buf)->frame_size)
+ if (buf_size < 4 || buf_size < sys->frame_size)
return -1; /* Broken frame, or not enough data */
/* Queueing audio packet */
/* FIXME: in case of no audio/bad audio we have to do something */
- size = dv_extract_audio_info(buf, &c->ast->codec);
- if (av_new_packet(&c->audio_pkt, size) < 0)
- return AVERROR_NOMEM;
- dv_extract_audio(buf, c->audio_pkt.data);
- c->audio_pkt.size = size;
- c->audio_pkt.stream_index = c->ast->id;
+ size = dv_extract_audio_info(c, buf);
+ for (i=0; i<c->ach; i++) {
+ c->audio_pkt[i].size = size;
+ c->audio_pkt[i].pts = c->abytes * 30000*8 / c->ast[i]->codec->bit_rate;
+ }
+ dv_extract_audio(buf, c->audio_buf[0], c->audio_buf[1]);
+ c->abytes += size;
/* Now it's time to return video packet */
- size = dv_extract_video_info(buf, &c->vst->codec);
+ size = dv_extract_video_info(c, buf);
av_init_packet(pkt);
- pkt->destruct = __destruct_pkt;
pkt->data = buf;
pkt->size = size;
pkt->flags |= PKT_FLAG_KEY;
pkt->stream_index = c->vst->id;
+ pkt->pts = c->frames;
+ c->frames++;
+
return size;
}
+static int64_t dv_frame_offset(AVFormatContext *s, DVDemuxContext *c,
+ int64_t timestamp, int flags)
+{
+ // FIXME: sys may be wrong if last dv_read_packet() failed (buffer is junk)
+ const DVprofile* sys = dv_codec_profile(c->vst->codec);
+ int64_t offset;
+ int64_t size = url_fsize(&s->pb);
+ int64_t max_offset = ((size-1) / sys->frame_size) * sys->frame_size;
+
+ offset = sys->frame_size * timestamp;
+
+ if (offset > max_offset) offset = max_offset;
+ else if (offset < 0) offset = 0;
+
+ return offset;
+}
+
+void dv_flush_audio_packets(DVDemuxContext *c)
+{
+ c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
+}
+
/************************************************************
* Implementation of the easiest DV storage of all -- raw DV.
************************************************************/
typedef struct RawDVContext {
- void* dv_demux;
- uint8_t buf[144000];
+ uint8_t buf[144000];
+ DVDemuxContext* dv_demux;
} RawDVContext;
static int dv_read_header(AVFormatContext *s,
AVFormatParameters *ap)
{
RawDVContext *c = s->priv_data;
- c->dv_demux = dv_init_demux(s, 0, 1);
+ const DVprofile* sys;
+ c->dv_demux = dv_init_demux(s);
+ if (!c->dv_demux)
+ return -1;
+
+ if (get_buffer(&s->pb, c->buf, 4) <= 0 || url_fseek(&s->pb, -4, SEEK_CUR) < 0)
+ return AVERROR_IO;
+
+ sys = dv_frame_profile(c->buf);
+ s->bit_rate = av_rescale(sys->frame_size * 8, sys->frame_rate, sys->frame_rate_base);
- return c->dv_demux ? 0 : -1;
+ return 0;
}
if (size < 0) {
if (get_buffer(&s->pb, c->buf, 4) <= 0)
- return -EIO;
+ return AVERROR_IO;
size = dv_frame_profile(c->buf)->frame_size;
if (get_buffer(&s->pb, c->buf + 4, size - 4) <= 0)
- return -EIO;
+ return AVERROR_IO;
size = dv_produce_packet(c->dv_demux, pkt, c->buf, size);
}
return size;
}
+static int dv_read_seek(AVFormatContext *s, int stream_index,
+ int64_t timestamp, int flags)
+{
+ RawDVContext *r = s->priv_data;
+ DVDemuxContext *c = r->dv_demux;
+ int64_t offset= dv_frame_offset(s, c, timestamp, flags);
+ const DVprofile* sys = dv_codec_profile(c->vst->codec);
+
+ c->frames= offset / sys->frame_size;
+ if (c->ach)
+ c->abytes= av_rescale(c->frames,
+ c->ast[0]->codec->bit_rate * (int64_t)sys->frame_rate_base,
+ 8*sys->frame_rate);
+
+ dv_flush_audio_packets(c);
+ return url_fseek(&s->pb, offset, SEEK_SET);
+}
+
static int dv_read_close(AVFormatContext *s)
{
RawDVContext *c = s->priv_data;
{
s->priv_data = dv_init_mux(s);
if (!s->priv_data) {
- fprintf(stderr, "Can't initialize DV format!\n"
+ av_log(s, AV_LOG_ERROR, "Can't initialize DV format!\n"
"Make sure that you supply exactly two streams:\n"
" video: 25fps or 29.97fps, audio: 2ch/48Khz/PCM\n");
return -1;
return 0;
}
-static int dv_write_packet(struct AVFormatContext *s,
- int stream_index,
- const uint8_t *buf, int size, int64_t pts)
+static int dv_write_packet(struct AVFormatContext *s, AVPacket *pkt)
{
uint8_t* frame;
int fsize;
- fsize = dv_assemble_frame((DVMuxContext *)s->priv_data, s->streams[stream_index],
- buf, size, &frame);
+ fsize = dv_assemble_frame((DVMuxContext *)s->priv_data, s->streams[pkt->stream_index],
+ pkt->data, pkt->size, &frame);
if (fsize > 0) {
put_buffer(&s->pb, frame, fsize);
put_flush_packet(&s->pb);
}
- return fsize;
+ return 0;
}
/*
dv_read_header,
dv_read_packet,
dv_read_close,
- .extensions = "dv",
+ dv_read_seek,
+ .extensions = "dv,dif",
};
static AVOutputFormat dv_oformat = {
dv_write_trailer,
};
-int dv_init(void)
+int ff_dv_init(void)
{
av_register_input_format(&dv_iformat);
av_register_output_format(&dv_oformat);