annotate dv.c @ 6471:b3aea89a4f63 libavformat

nutenc: fix unstable floating-point calculations

/*
 * General DV muxer/demuxer
 * Copyright (c) 2003 Roman Shaposhnik
 *
 * Many thanks to Dan Dennedy <[e-mail]> for providing wealth
 * of DV technical info.
 *
 * Raw DV format
 * Copyright (c) 2002 Fabrice Bellard
 *
 * 50 Mbps (DVCPRO50) and 100 Mbps (DVCPRO HD) support
 * Copyright (c) 2006 Daniel Maas <[e-mail]>
 * Funded by BBC Research & Development
 *
 * 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 <time.h>
#include "avformat.h"
#include "libavcodec/dvdata.h"
#include "libavutil/intreadwrite.h"
#include "dv.h"

struct DVDemuxContext {
    const DVprofile*  sys;    /* Current DV profile. E.g.: 525/60, 625/50 */
    AVFormatContext*  fctx;
    AVStream*         vst;
    AVStream*         ast[4];
    AVPacket          audio_pkt[4];
    uint8_t           audio_buf[4][8192];
    int               ach;
    int               frames;
    uint64_t          abytes;
};

static inline uint16_t dv_audio_12to16(uint16_t sample)
{
    uint16_t shift, result;

    sample = (sample < 0x800) ? sample : sample | 0xf000;
    shift  = (sample & 0xf00) >> 8;

    if (shift < 0x2 || shift > 0xd) {
        result = sample;
    } else if (shift < 0x8) {
        shift--;
        result = (sample - (256 * shift)) << shift;
    } else {
        shift = 0xe - shift;
        result = ((sample + ((256 * shift) + 1)) << shift) - 1;
    }

    return result;
}

/*
 * 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;

    switch (t) {
    case dv_audio_source:
        offs = (80*6 + 80*16*3 + 3);
        break;
    case dv_audio_control:
        offs = (80*6 + 80*16*4 + 3);
        break;
    case dv_video_control:
        offs = (80*5 + 48 + 5);
        break;
    default:
        return NULL;
    }

    return frame[offs] == t ? &frame[offs] : NULL;
}

/*
 * 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.
 */
static int dv_extract_audio(uint8_t* frame, uint8_t* ppcm[4],
                            const DVprofile *sys)
{
    int size, chan, i, j, d, of, smpls, freq, quant, half_ch;
    uint16_t lc, rc;
    const uint8_t* as_pack;
    uint8_t *pcm, ipcm;

    as_pack = dv_extract_pack(frame, dv_audio_source);
    if (!as_pack)    /* No audio ? */
        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 - 32kHz */
    quant =  as_pack[4] & 0x07;       /* 0 - 16bit linear, 1 - 12bit nonlinear */

    if (quant > 1)
        return -1; /* unsupported quantization */

    size = (sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */
    half_ch = sys->difseg_size / 2;

    /* We work with 720p frames split in half, thus even frames have
     * channels 0,1 and odd 2,3. */
    ipcm = (sys->height == 720 && !(frame[1] & 0x0C)) ? 2 : 0;
    pcm  = ppcm[ipcm++];

    /* 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) */
                pcm = ppcm[ipcm++];
                if (!pcm)
                    break;
            }

            /* for each AV sequence */
            for (j = 0; j < 9; j++) {
                for (d = 8; d < 80; d += 2) {
                    if (quant == 0) {  /* 16bit quantization */
                        of = sys->audio_shuffle[i][j] + (d - 8) / 2 * sys->audio_stride;
                        if (of*2 >= size)
                            continue;

                        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
                        if (pcm[of*2+1] == 0x80 && pcm[of*2] == 0x00)
                            pcm[of*2+1] = 0;
                    } else {           /* 12bit quantization */
                        lc = ((uint16_t)frame[d]   << 4) |
                             ((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));

                        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: maybe we have to admit
                        pcm[of*2+1] = lc >> 8;   //        that DV is a big-endian PCM
                        of = sys->audio_shuffle[i%half_ch+half_ch][j] +
                            (d - 8) / 3 * sys->audio_stride;
                        pcm[of*2]   = rc & 0xff; // FIXME: maybe we have to admit
                        pcm[of*2+1] = rc >> 8;   //        that DV is a big-endian PCM
                        ++d;
                    }
                }

                frame += 16 * 80; /* 15 Video DIFs + 1 Audio DIF */
            }
        }

        /* next stereo channel (50Mbps and 100Mbps only) */
        pcm = ppcm[ipcm++];
        if (!pcm)
            break;
    }

    return size;
}

static int dv_extract_audio_info(DVDemuxContext* c, uint8_t* frame)
{
    const uint8_t* as_pack;
    int freq, stype, smpls, quant, i, ach;

    as_pack = dv_extract_pack(frame, dv_audio_source);
    if (!as_pack || !c->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 - 32kHz */
    stype = (as_pack[3] & 0x1f);      /* 0 - 2CH, 2 - 4CH, 3 - 8CH */
    quant =  as_pack[4] & 0x07;       /* 0 - 16bit linear, 1 - 12bit nonlinear */

    /* note: ach counts PAIRS of channels (i.e. stereo channels) */
    ach = ((int[4]){  1,  0,  2,  4})[stype];
    if (ach == 1 && quant && freq == 2)
        ach = 2;

    /* 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 = AVMEDIA_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       |= AV_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 (c->sys->audio_min_samples[freq] + smpls) * 4; /* 2ch, 2bytes */;
}

static int dv_extract_video_info(DVDemuxContext *c, uint8_t* frame)
{
    const uint8_t* vsc_pack;
    AVCodecContext* avctx;
    int apt, is16_9;
    int size = 0;

    if (c->sys) {
        avctx = c->vst->codec;

        av_set_pts_info(c->vst, 64, c->sys->time_base.num,
                        c->sys->time_base.den);
        avctx->time_base= c->sys->time_base;
        if (!avctx->width){
            avctx->width = c->sys->width;
            avctx->height = c->sys->height;
        }
        avctx->pix_fmt = c->sys->pix_fmt;

        /* 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)));
        c->vst->sample_aspect_ratio = c->sys->sar[is16_9];
        avctx->bit_rate = av_rescale_q(c->sys->frame_size, (AVRational){8,1},
                                       c->sys->time_base);
        size = c->sys->frame_size;
    }
    return size;
}

/*
 * The following 3 functions constitute our interface to the world
 */

DVDemuxContext* dv_init_demux(AVFormatContext *s)
{
    DVDemuxContext *c;

    c = av_mallocz(sizeof(DVDemuxContext));
    if (!c)
        return NULL;

    c->vst = av_new_stream(s, 0);
    if (!c->vst) {
        av_free(c);
        return NULL;
    }

    c->sys  = NULL;
    c->fctx = s;
    memset(c->ast, 0, sizeof(c->ast));
    c->ach    = 0;
    c->frames = 0;
    c->abytes = 0;

    c->vst->codec->codec_type = AVMEDIA_TYPE_VIDEO;
    c->vst->codec->codec_id   = CODEC_ID_DVVIDEO;
    c->vst->codec->bit_rate   = 25000000;
    c->vst->start_time        = 0;

    return c;
}

int dv_get_packet(DVDemuxContext *c, AVPacket *pkt)
{
    int size = -1;
    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, i;
    uint8_t *ppcm[4] = {0};

    if (buf_size < DV_PROFILE_BYTES ||
        !(c->sys = ff_dv_frame_profile(c->sys, buf, buf_size)) ||
        buf_size < c->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(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;
       ppcm[i] = c->audio_buf[i];
    }
    dv_extract_audio(buf, ppcm, c->sys);

    /* We work with 720p frames split in half, thus even frames have
     * channels 0,1 and odd 2,3. */
    if (c->sys->height == 720) {
        if (buf[1] & 0x0C) {
            c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
        } else {
            c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
            c->abytes += size;
        }
    } else {
        c->abytes += size;
    }

    /* Now it's time to return video packet */
    size = dv_extract_video_info(c, buf);
    av_init_packet(pkt);
    pkt->data         = buf;
    pkt->size         = size;
    pkt->flags       |= AV_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 = ff_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 (size >= 0 && offset > max_offset) offset = max_offset;
    else if (offset < 0) offset = 0;

    return offset;
}

void dv_offset_reset(DVDemuxContext *c, int64_t frame_offset)
{
    c->frames= frame_offset;
    if (c->ach)
        c->abytes= av_rescale_q(c->frames, c->sys->time_base,
                                (AVRational){8, c->ast[0]->codec->bit_rate});
    c->audio_pkt[0].size = c->audio_pkt[1].size = 0;
    c->audio_pkt[2].size = c->audio_pkt[3].size = 0;
}

/************************************************************
 * Implementation of the easiest DV storage of all -- raw DV.
 ************************************************************/

typedef struct RawDVContext {
    DVDemuxContext* dv_demux;
    uint8_t         buf[DV_MAX_FRAME_SIZE];
} RawDVContext;

static int dv_read_header(AVFormatContext *s,
                          AVFormatParameters *ap)
{
    unsigned state, marker_pos = 0;
    RawDVContext *c = s->priv_data;

    c->dv_demux = dv_init_demux(s);
    if (!c->dv_demux)
        return -1;

    state = get_be32(s->pb);
    while ((state & 0xffffff7f) != 0x1f07003f) {
        if (url_feof(s->pb)) {
            av_log(s, AV_LOG_ERROR, "Cannot find DV header.\n");
            return -1;
        }
        if (state == 0x003f0700 || state == 0xff3f0700)
            marker_pos = url_ftell(s->pb);
        if (state == 0xff3f0701 && url_ftell(s->pb) - marker_pos == 80) {
            url_fseek(s->pb, -163, SEEK_CUR);
            state = get_be32(s->pb);
            break;
        }
        state = (state << 8) | get_byte(s->pb);
    }
    AV_WB32(c->buf, state);

    if (get_buffer(s->pb, c->buf + 4, DV_PROFILE_BYTES - 4) <= 0 ||
        url_fseek(s->pb, -DV_PROFILE_BYTES, SEEK_CUR) < 0)
        return AVERROR(EIO);

    c->dv_demux->sys = ff_dv_frame_profile(c->dv_demux->sys, c->buf, DV_PROFILE_BYTES);
    if (!c->dv_demux->sys) {
        av_log(s, AV_LOG_ERROR, "Can't determine profile of DV input stream.\n");
        return -1;
    }

    s->bit_rate = av_rescale_q(c->dv_demux->sys->frame_size, (AVRational){8,1},
                               c->dv_demux->sys->time_base);

    return 0;
}


static int dv_read_packet(AVFormatContext *s, AVPacket *pkt)
{
    int size;
    RawDVContext *c = s->priv_data;

    size = dv_get_packet(c->dv_demux, pkt);

    if (size < 0) {
        if (!c->dv_demux->sys)
            return AVERROR(EIO);
        size = c->dv_demux->sys->frame_size;
        if (get_buffer(s->pb, c->buf, size) <= 0)
            return AVERROR(EIO);

        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);

    dv_offset_reset(c, offset / c->sys->frame_size);

    offset = url_fseek(s->pb, offset, SEEK_SET);
    return (offset < 0) ? offset : 0;
}

static int dv_read_close(AVFormatContext *s)
{
    RawDVContext *c = s->priv_data;
    av_free(c->dv_demux);
    return 0;
}

static int dv_probe(AVProbeData *p)
{
    unsigned state, marker_pos = 0;
    int i;
    int matches = 0;
    int secondary_matches = 0;

    if (p->buf_size < 5)
        return 0;

    state = AV_RB32(p->buf);
    for (i = 4; i < p->buf_size; i++) {
        if ((state & 0xffffff7f) == 0x1f07003f)
            matches++;
        // 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++;
        if (state == 0x003f0700 || state == 0xff3f0700)
            marker_pos = i;
        if (state == 0xff3f0701 && i - marker_pos == 80)
            matches++;
        state = (state << 8) | p->buf[i];
    }

    if (matches && p->buf_size / matches < 1024*1024) {
        if (matches > 4 || (secondary_matches >= 10 && p->buf_size / secondary_matches < 24000))
            return AVPROBE_SCORE_MAX*3/4; // not max to avoid dv in mov to match
        return AVPROBE_SCORE_MAX/4;
    }
    return 0;
}

#if CONFIG_DV_DEMUXER
AVInputFormat dv_demuxer = {
    "dv",
    NULL_IF_CONFIG_SMALL("DV video format"),
    sizeof(RawDVContext),
    dv_probe,
    dv_read_header,
    dv_read_packet,
    dv_read_close,
    dv_read_seek,
    .extensions = "dv,dif",
};
#endif