libavformat/dv.c
115329f1
 /*
  * General DV muxer/demuxer
064cf251
  * Copyright (c) 2003 Roman Shaposhnik
7458ccbb
  *
  * Many thanks to Dan Dennedy <dan@dennedy.org> for providing wealth
  * of DV technical info.
  *
f20dca40
  * Raw DV format
406792e7
  * Copyright (c) 2002 Fabrice Bellard
f20dca40
  *
f8007a14
  * 50 Mbps (DVCPRO50) and 100 Mbps (DVCPRO HD) support
0b297700
  * Copyright (c) 2006 Daniel Maas <dmaas@maasdigital.com>
f8007a14
  * Funded by BBC Research & Development
0b297700
  *
b78e7197
  * This file is part of FFmpeg.
  *
  * FFmpeg is free software; you can redistribute it and/or
f20dca40
  * modify it under the terms of the GNU Lesser General Public
  * License as published by the Free Software Foundation; either
b78e7197
  * version 2.1 of the License, or (at your option) any later version.
f20dca40
  *
b78e7197
  * FFmpeg is distributed in the hope that it will be useful,
f20dca40
  * 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
b78e7197
  * License along with FFmpeg; if not, write to the Free Software
5509bffa
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
f20dca40
  */
7458ccbb
 #include <time.h>
f20dca40
 #include "avformat.h"
245976da
 #include "libavcodec/dvdata.h"
67d44b84
 #include "libavutil/intreadwrite.h"
ddaae6a9
 #include "dv.h"
 
 struct DVDemuxContext {
0b297700
     const DVprofile*  sys;    /* Current DV profile. E.g.: 525/60, 625/50 */
f94036f1
     AVFormatContext*  fctx;
     AVStream*         vst;
     AVStream*         ast[4];
     AVPacket          audio_pkt[4];
     uint8_t           audio_buf[4][8192];
     int               ach;
     int               frames;
     uint64_t          abytes;
ddaae6a9
 };
f20dca40
 
7458ccbb
 static inline uint16_t dv_audio_12to16(uint16_t sample)
 {
     uint16_t shift, result;
115329f1
 
7458ccbb
     sample = (sample < 0x800) ? sample : sample | 0xf000;
f94036f1
     shift  = (sample & 0xf00) >> 8;
7458ccbb
 
     if (shift < 0x2 || shift > 0xd) {
bb270c08
         result = sample;
7458ccbb
     } else if (shift < 0x8) {
         shift--;
bb270c08
         result = (sample - (256 * shift)) << shift;
7458ccbb
     } else {
bb270c08
         shift = 0xe - shift;
         result = ((sample + ((256 * shift) + 1)) << shift) - 1;
7458ccbb
     }
 
     return result;
 }
 
115329f1
 /*
7458ccbb
  * This is the dumbest implementation of all -- it simply looks at
  * a fixed offset and if pack isn't there -- fails. We might want
  * to have a fallback mechanism for complete search of missing packs.
  */
 static const uint8_t* dv_extract_pack(uint8_t* frame, enum dv_pack_type t)
 {
     int offs;
115329f1
 
7458ccbb
     switch (t) {
     case dv_audio_source:
f94036f1
         offs = (80*6 + 80*16*3 + 3);
         break;
7458ccbb
     case dv_audio_control:
f94036f1
         offs = (80*6 + 80*16*4 + 3);
         break;
7458ccbb
     case dv_video_control:
f94036f1
         offs = (80*5 + 48 + 5);
         break;
7458ccbb
     default:
f94036f1
         return NULL;
115329f1
     }
7458ccbb
 
ccd425e7
     return frame[offs] == t ? &frame[offs] : NULL;
7458ccbb
 }
 
115329f1
 /*
7458ccbb
  * There's a couple of assumptions being made here:
  * 1. By default we silence erroneous (0x8000/16bit 0x800/12bit) audio samples.
  *    We can pass them upwards when ffmpeg will be ready to deal with them.
  * 2. We don't do software emphasis.
  * 3. Audio is always returned as 16bit linear samples: 12bit nonlinear samples
  *    are converted into 16bit linear ones.
  */
f8007a14
 static int dv_extract_audio(uint8_t* frame, uint8_t* ppcm[4],
0b297700
                             const DVprofile *sys)
7458ccbb
 {
0b297700
     int size, chan, i, j, d, of, smpls, freq, quant, half_ch;
7458ccbb
     uint16_t lc, rc;
     const uint8_t* as_pack;
f8007a14
     uint8_t *pcm, ipcm;
115329f1
 
7458ccbb
     as_pack = dv_extract_pack(frame, dv_audio_source);
     if (!as_pack)    /* No audio ? */
bb270c08
         return 0;
115329f1
 
f94036f1
     smpls =  as_pack[1] & 0x3f;       /* samples in this frame - min. samples */
     freq  = (as_pack[4] >> 3) & 0x07; /* 0 - 48kHz, 1 - 44,1kHz, 2 - 32kHz */
     quant =  as_pack[4] & 0x07;       /* 0 - 16bit linear, 1 - 12bit nonlinear */
115329f1
 
7458ccbb
     if (quant > 1)
75f2c209
         return -1; /* unsupported quantization */
7458ccbb
 
     size = (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */
f94036f1
     half_ch = sys->difseg_size / 2;
7458ccbb
 
75f2c209
     /* We work with 720p frames split in half, thus even frames have
f94036f1
      * channels 0,1 and odd 2,3. */
     ipcm = (sys->height == 720 && !(frame[1] & 0x0C)) ? 2 : 0;
     pcm  = ppcm[ipcm++];
f8007a14
 
0b297700
     /* for each DIF channel */
     for (chan = 0; chan < sys->n_difchan; chan++) {
         /* 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) {
                 /* next stereo channel (12bit mode only) */
f8007a14
                 pcm = ppcm[ipcm++];
                 if (!pcm)
0b297700
                     break;
             }
bb270c08
 
0b297700
             /* for each AV sequence */
             for (j = 0; j < 9; j++) {
                 for (d = 8; d < 80; d += 2) {
                     if (quant == 0) {  /* 16bit quantization */
f94036f1
                         of = sys->audio_shuffle[i][j] + (d - 8) / 2 * sys->audio_stride;
0b297700
                         if (of*2 >= size)
                             continue;
 
910f02a0
                         pcm[of*2]   = frame[d+1]; // FIXME: maybe we have to admit
                         pcm[of*2+1] = frame[d];   //        that DV is a big-endian PCM
0b297700
                         if (pcm[of*2+1] == 0x80 && pcm[of*2] == 0x00)
                             pcm[of*2+1] = 0;
                     } else {           /* 12bit quantization */
f94036f1
                         lc = ((uint16_t)frame[d]   << 4) |
0b297700
                              ((uint16_t)frame[d+2] >> 4);
                         rc = ((uint16_t)frame[d+1] << 4) |
                              ((uint16_t)frame[d+2] & 0x0f);
                         lc = (lc == 0x800 ? 0 : dv_audio_12to16(lc));
                         rc = (rc == 0x800 ? 0 : dv_audio_12to16(rc));
 
f94036f1
                         of = sys->audio_shuffle[i%half_ch][j] + (d - 8) / 3 * sys->audio_stride;
0b297700
                         if (of*2 >= size)
                             continue;
 
910f02a0
                         pcm[of*2]   = lc & 0xff; // FIXME: maybe we have to admit
                         pcm[of*2+1] = lc >> 8;   //        that DV is a big-endian PCM
0b297700
                         of = sys->audio_shuffle[i%half_ch+half_ch][j] +
f94036f1
                             (d - 8) / 3 * sys->audio_stride;
910f02a0
                         pcm[of*2]   = rc & 0xff; // FIXME: maybe we have to admit
                         pcm[of*2+1] = rc >> 8;   //        that DV is a big-endian PCM
0b297700
                         ++d;
                     }
                 }
 
                 frame += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */
             }
7458ccbb
         }
0b297700
 
f8007a14
         /* next stereo channel (50Mbps and 100Mbps only) */
         pcm = ppcm[ipcm++];
         if (!pcm)
0b297700
             break;
7458ccbb
     }
 
     return size;
 }
 
ddaae6a9
 static int dv_extract_audio_info(DVDemuxContext* c, uint8_t* frame)
7458ccbb
 {
     const uint8_t* as_pack;
0b297700
     int freq, stype, smpls, quant, i, ach;
7458ccbb
 
     as_pack = dv_extract_pack(frame, dv_audio_source);
0b297700
     if (!as_pack || !c->sys) {    /* No audio ? */
ddaae6a9
         c->ach = 0;
bb270c08
         return 0;
ddaae6a9
     }
115329f1
 
f94036f1
     smpls =  as_pack[1] & 0x3f;       /* samples in this frame - min. samples */
     freq  = (as_pack[4] >> 3) & 0x07; /* 0 - 48kHz, 1 - 44,1kHz, 2 - 32kHz */
     stype = (as_pack[3] & 0x1f);      /* 0 - 2CH, 2 - 4CH, 3 - 8CH */
     quant =  as_pack[4] & 0x07;       /* 0 - 16bit linear, 1 - 12bit nonlinear */
0b297700
 
     /* note: ach counts PAIRS of channels (i.e. stereo channels) */
51f4155f
     ach = ((int[4]){  1,  0,  2,  4})[stype];
a53621fc
     if (ach == 1 && quant && freq == 2)
         ach = 2;
50883632
 
     /* Dynamic handling of the audio streams in DV */
f94036f1
     for (i = 0; i < ach; i++) {
50883632
        if (!c->ast[i]) {
            c->ast[i] = av_new_stream(c->fctx, 0);
bb270c08
            if (!c->ast[i])
                break;
            av_set_pts_info(c->ast[i], 64, 1, 30000);
72415b2a
            c->ast[i]->codec->codec_type = AVMEDIA_TYPE_AUDIO;
f94036f1
            c->ast[i]->codec->codec_id   = CODEC_ID_PCM_S16LE;
664f5836
 
            av_init_packet(&c->audio_pkt[i]);
f94036f1
            c->audio_pkt[i].size         = 0;
            c->audio_pkt[i].data         = c->audio_buf[i];
664f5836
            c->audio_pkt[i].stream_index = c->ast[i]->index;
cc947f04
            c->audio_pkt[i].flags       |= AV_PKT_FLAG_KEY;
50883632
        }
01f4895c
        c->ast[i]->codec->sample_rate = dv_audio_frequency[freq];
f94036f1
        c->ast[i]->codec->channels    = 2;
        c->ast[i]->codec->bit_rate    = 2 * dv_audio_frequency[freq] * 16;
        c->ast[i]->start_time         = 0;
ddaae6a9
     }
50883632
     c->ach = i;
115329f1
 
0b297700
     return (c->sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */;
7458ccbb
 }
 
ddaae6a9
 static int dv_extract_video_info(DVDemuxContext *c, uint8_t* frame)
7458ccbb
 {
     const uint8_t* vsc_pack;
ddaae6a9
     AVCodecContext* avctx;
     int apt, is16_9;
7458ccbb
     int size = 0;
115329f1
 
0b297700
     if (c->sys) {
01f4895c
         avctx = c->vst->codec;
115329f1
 
f94036f1
         av_set_pts_info(c->vst, 64, c->sys->time_base.num,
                         c->sys->time_base.den);
4e92dabf
         avctx->time_base= c->sys->time_base;
f94036f1
         if (!avctx->width){
0b297700
             avctx->width = c->sys->width;
             avctx->height = c->sys->height;
4325ffd0
         }
0b297700
         avctx->pix_fmt = c->sys->pix_fmt;
115329f1
 
bb270c08
         /* finding out SAR is a little bit messy */
         vsc_pack = dv_extract_pack(frame, dv_video_control);
f94036f1
         apt      = frame[4] & 0x07;
         is16_9   = (vsc_pack && ((vsc_pack[2] & 0x07) == 0x02 ||
                                 (!apt && (vsc_pack[2] & 0x07) == 0x07)));
59729451
         c->vst->sample_aspect_ratio = c->sys->sar[is16_9];
4e92dabf
         avctx->bit_rate = av_rescale_q(c->sys->frame_size, (AVRational){8,1},
                                        c->sys->time_base);
0b297700
         size = c->sys->frame_size;
7458ccbb
     }
     return size;
 }
 
115329f1
 /*
d4fc91b4
  * The following 3 functions constitute our interface to the world
7458ccbb
  */
 
ddaae6a9
 DVDemuxContext* dv_init_demux(AVFormatContext *s)
7458ccbb
 {
     DVDemuxContext *c;
 
     c = av_mallocz(sizeof(DVDemuxContext));
     if (!c)
         return NULL;
 
ddaae6a9
     c->vst = av_new_stream(s, 0);
50883632
     if (!c->vst) {
115329f1
         av_free(c);
50883632
         return NULL;
     }
9ee91c2f
 
f94036f1
     c->sys  = NULL;
ddaae6a9
     c->fctx = s;
34fab90f
     memset(c->ast, 0, sizeof(c->ast));
f94036f1
     c->ach    = 0;
8066e59f
     c->frames = 0;
     c->abytes = 0;
664f5836
 
72415b2a
     c->vst->codec->codec_type = AVMEDIA_TYPE_VIDEO;
f94036f1
     c->vst->codec->codec_id   = CODEC_ID_DVVIDEO;
     c->vst->codec->bit_rate   = 25000000;
     c->vst->start_time        = 0;
115329f1
 
7458ccbb
     return c;
f20dca40
 }
 
7458ccbb
 int dv_get_packet(DVDemuxContext *c, AVPacket *pkt)
f20dca40
 {
7458ccbb
     int size = -1;
ddaae6a9
     int i;
 
f94036f1
     for (i = 0; i < c->ach; i++) {
ddaae6a9
        if (c->ast[i] && c->audio_pkt[i].size) {
            *pkt = c->audio_pkt[i];
bb270c08
            c->audio_pkt[i].size = 0;
            size = pkt->size;
            break;
ddaae6a9
        }
425ed6e2
     }
7458ccbb
 
     return size;
 }
 
115329f1
 int dv_produce_packet(DVDemuxContext *c, AVPacket *pkt,
148ffcd2
                       uint8_t* buf, int buf_size, int64_t pos)
7458ccbb
 {
ddaae6a9
     int size, i;
f8007a14
     uint8_t *ppcm[4] = {0};
115329f1
 
c4e7baa8
     if (buf_size < DV_PROFILE_BYTES ||
c76911bd
         !(c->sys = ff_dv_frame_profile(c->sys, buf, buf_size)) ||
c4e7baa8
         buf_size < c->sys->frame_size) {
           return -1;   /* Broken frame, or not enough data */
     }
7458ccbb
 
     /* Queueing audio packet */
     /* FIXME: in case of no audio/bad audio we have to do something */
ddaae6a9
     size = dv_extract_audio_info(c, buf);
f94036f1
     for (i = 0; i < c->ach; i++) {
148ffcd2
        c->audio_pkt[i].pos  = pos;
664f5836
        c->audio_pkt[i].size = size;
01f4895c
        c->audio_pkt[i].pts  = c->abytes * 30000*8 / c->ast[i]->codec->bit_rate;
f8007a14
        ppcm[i] = c->audio_buf[i];
ddaae6a9
     }
f8007a14
     dv_extract_audio(buf, ppcm, c->sys);
115329f1
 
75f2c209
     /* We work with 720p frames split in half, thus even frames have
      * channels 0,1 and odd 2,3. */
f8007a14
     if (c->sys->height == 720) {
8a4ea200
         if (buf[1] & 0x0C) {
f8007a14
             c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
8a4ea200
         } else {
f8007a14
             c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
8a4ea200
             c->abytes += size;
         }
     } else {
         c->abytes += size;
f8007a14
     }
 
7458ccbb
     /* Now it's time to return video packet */
ddaae6a9
     size = dv_extract_video_info(c, buf);
ec820113
     av_init_packet(pkt);
f94036f1
     pkt->data         = buf;
148ffcd2
     pkt->pos          = pos;
f94036f1
     pkt->size         = size;
cc947f04
     pkt->flags       |= AV_PKT_FLAG_KEY;
7458ccbb
     pkt->stream_index = c->vst->id;
f94036f1
     pkt->pts          = c->frames;
115329f1
 
8066e59f
     c->frames++;
ddaae6a9
 
7458ccbb
     return size;
 }
115329f1
 
83582de4
 static int64_t dv_frame_offset(AVFormatContext *s, DVDemuxContext *c,
                               int64_t timestamp, int flags)
8066e59f
 {
0bf4e2b8
     // FIXME: sys may be wrong if last dv_read_packet() failed (buffer is junk)
c76911bd
     const DVprofile* sys = ff_dv_codec_profile(c->vst->codec);
eb1bde3e
     int64_t offset;
db44ea96
     int64_t size = avio_size(s->pb) - s->data_offset;
83582de4
     int64_t max_offset = ((size-1) / sys->frame_size) * sys->frame_size;
0bf4e2b8
 
eb1bde3e
     offset = sys->frame_size * timestamp;
115329f1
 
d40a999a
     if (size >= 0 && offset > max_offset) offset = max_offset;
83582de4
     else if (offset < 0) offset = 0;
8066e59f
 
fe9d2edb
     return offset + s->data_offset;
8066e59f
 }
 
6eb2de74
 void dv_offset_reset(DVDemuxContext *c, int64_t frame_offset)
664f5836
 {
6eb2de74
     c->frames= frame_offset;
     if (c->ach)
4e92dabf
         c->abytes= av_rescale_q(c->frames, c->sys->time_base,
                                 (AVRational){8, c->ast[0]->codec->bit_rate});
664f5836
     c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
f8007a14
     c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
664f5836
 }
 
7458ccbb
 /************************************************************
  * Implementation of the easiest DV storage of all -- raw DV.
  ************************************************************/
115329f1
 
7458ccbb
 typedef struct RawDVContext {
ddaae6a9
     DVDemuxContext* dv_demux;
0b297700
     uint8_t         buf[DV_MAX_FRAME_SIZE];
7458ccbb
 } RawDVContext;
 
 static int dv_read_header(AVFormatContext *s,
                           AVFormatParameters *ap)
 {
70606675
     unsigned state, marker_pos = 0;
7458ccbb
     RawDVContext *c = s->priv_data;
0b297700
 
ddaae6a9
     c->dv_demux = dv_init_demux(s);
c4e55470
     if (!c->dv_demux)
         return -1;
115329f1
 
e63a3628
     state = avio_rb32(s->pb);
67d44b84
     while ((state & 0xffffff7f) != 0x1f07003f) {
         if (url_feof(s->pb)) {
             av_log(s, AV_LOG_ERROR, "Cannot find DV header.\n");
             return -1;
         }
70606675
         if (state == 0x003f0700 || state == 0xff3f0700)
384c9c2f
             marker_pos = avio_tell(s->pb);
         if (state == 0xff3f0701 && avio_tell(s->pb) - marker_pos == 80) {
f59d8ff8
             avio_seek(s->pb, -163, SEEK_CUR);
e63a3628
             state = avio_rb32(s->pb);
70606675
             break;
         }
e63a3628
         state = (state << 8) | avio_r8(s->pb);
67d44b84
     }
     AV_WB32(c->buf, state);
 
e63a3628
     if (avio_read(s->pb, c->buf + 4, DV_PROFILE_BYTES - 4) <= 0 ||
f59d8ff8
         avio_seek(s->pb, -DV_PROFILE_BYTES, SEEK_CUR) < 0)
6f3e0b21
         return AVERROR(EIO);
c4e55470
 
c76911bd
     c->dv_demux->sys = ff_dv_frame_profile(c->dv_demux->sys, c->buf, DV_PROFILE_BYTES);
f8007a14
     if (!c->dv_demux->sys) {
         av_log(s, AV_LOG_ERROR, "Can't determine profile of DV input stream.\n");
         return -1;
     }
 
4e92dabf
     s->bit_rate = av_rescale_q(c->dv_demux->sys->frame_size, (AVRational){8,1},
                                c->dv_demux->sys->time_base);
115329f1
 
c4e55470
     return 0;
7458ccbb
 }
 
 
 static int dv_read_packet(AVFormatContext *s, AVPacket *pkt)
 {
     int size;
115329f1
     RawDVContext *c = s->priv_data;
 
7458ccbb
     size = dv_get_packet(c->dv_demux, pkt);
115329f1
 
7458ccbb
     if (size < 0) {
148ffcd2
         int64_t pos = avio_tell(s->pb);
6d8b21f3
         if (!c->dv_demux->sys)
             return AVERROR(EIO);
0b297700
         size = c->dv_demux->sys->frame_size;
e63a3628
         if (avio_read(s->pb, c->buf, size) <= 0)
6f3e0b21
             return AVERROR(EIO);
7458ccbb
 
148ffcd2
         size = dv_produce_packet(c->dv_demux, pkt, c->buf, size, pos);
115329f1
     }
 
7458ccbb
     return size;
f20dca40
 }
 
115329f1
 static int dv_read_seek(AVFormatContext *s, int stream_index,
83582de4
                        int64_t timestamp, int flags)
8066e59f
 {
f94036f1
     RawDVContext *r   = s->priv_data;
83582de4
     DVDemuxContext *c = r->dv_demux;
f94036f1
     int64_t offset    = dv_frame_offset(s, c, timestamp, flags);
83582de4
 
6eb2de74
     dv_offset_reset(c, offset / c->sys->frame_size);
115329f1
 
f59d8ff8
     offset = avio_seek(s->pb, offset, SEEK_SET);
f94036f1
     return (offset < 0) ? offset : 0;
8066e59f
 }
 
c18a2692
 static int dv_read_close(AVFormatContext *s)
f20dca40
 {
7458ccbb
     RawDVContext *c = s->priv_data;
     av_free(c->dv_demux);
f20dca40
     return 0;
 }
 
e34a5996
 static int dv_probe(AVProbeData *p)
 {
70606675
     unsigned state, marker_pos = 0;
e34a5996
     int i;
6377de61
     int matches = 0;
f6765f3f
     int secondary_matches = 0;
e34a5996
 
     if (p->buf_size < 5)
         return 0;
 
     state = AV_RB32(p->buf);
     for (i = 4; i < p->buf_size; i++) {
         if ((state & 0xffffff7f) == 0x1f07003f)
6377de61
             matches++;
f6765f3f
         // any section header, also with seq/chan num != 0,
         // should appear around every 12000 bytes, at least 10 per frame
         if ((state & 0xff07ff7f) == 0x1f07003f)
             secondary_matches++;
70606675
         if (state == 0x003f0700 || state == 0xff3f0700)
             marker_pos = i;
         if (state == 0xff3f0701 && i - marker_pos == 80)
6377de61
             matches++;
e34a5996
         state = (state << 8) | p->buf[i];
     }
 
6377de61
     if (matches && p->buf_size / matches < 1024*1024) {
f6765f3f
         if (matches > 4 || (secondary_matches >= 10 && p->buf_size / secondary_matches < 24000))
6377de61
             return AVPROBE_SCORE_MAX*3/4; // not max to avoid dv in mov to match
         return AVPROBE_SCORE_MAX/4;
     }
e34a5996
     return 0;
 }
 
b250f9c6
 #if CONFIG_DV_DEMUXER
66355be3
 AVInputFormat ff_dv_demuxer = {
118e91f3
     "dv",
bde15e74
     NULL_IF_CONFIG_SMALL("DV video format"),
7458ccbb
     sizeof(RawDVContext),
e34a5996
     dv_probe,
118e91f3
     dv_read_header,
     dv_read_packet,
     dv_read_close,
8066e59f
     dv_read_seek,
490c735d
     .extensions = "dv,dif",
118e91f3
 };
ff70e601
 #endif