av/codec/h264/h264dec/decode.go

/*
DESCRIPTION
  decode.go provides slice decoding functionality.

AUTHORS
  Saxon A. Nelson-Milton <saxon@ausocean.org>

LICENSE
  Copyright (C) 2019 the Australian Ocean Lab (AusOcean).

  It is free software: you can redistribute it and/or modify them
  under the terms of the GNU General Public License as published by the
  Free Software Foundation, either version 3 of the License, or (at your
  option) any later version.

  It 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 General Public License
  for more details.

  You should have received a copy of the GNU General Public License
  in gpl.txt.  If not, see http://www.gnu.org/licenses.
*/

package h264dec

import (
	"errors"
	"fmt"
)

// NB: this is a placeholder.
func decode(vid *VideoStream, ctx *SliceContext) error {
	var err error
	vid.topFieldOrderCnt, vid.bottomFieldOrderCnt, err = decodePicOrderCnt(vid, ctx)
	if err != nil {
		return fmt.Errorf("could not derive topFieldOrderCnt and bottomFieldOrderCnt, failed with error: %v", err)
	}

	// According to 8.2.1 after decoding picture.
	if ctx.elements[0].MemoryManagementControlOperation == 5 {
		tempPicOrderCnt := picOrderCnt(ctx)
		vid.topFieldOrderCnt = vid.topFieldOrderCnt - tempPicOrderCnt
		vid.bottomFieldOrderCnt = vid.bottomFieldOrderCnt - tempPicOrderCnt
	}
	return nil
}

// TODO: complete this. Unsure how to determine if pic is frame or a
// complementary field pair.
// picOrderCnt as defined by section 8.2.1.
func picOrderCnt(ctx *SliceContext) int {
	panic("not implemented")
}

// decodePicOrderCnt derives topFieldOrderCnt and bottomFieldOrderCnt based
// on the PicOrderCntType using the process defined in section 8.2.1 of the
// specifications.
func decodePicOrderCnt(vid *VideoStream, ctx *SliceContext) (topFieldOrderCnt, bottomFieldOrderCnt int, err error) {
	// There are some steps listed in 8.2.1 regarding cases when not an IDR frame,
	// but we're not yet dealing with that.
	// TODO: write steps dealing with not IDR frame.
	if !vid.idrPicFlag {
		panic("not implemented")
	}

	switch ctx.PicOrderCountType {
	case 0:
		topFieldOrderCnt, bottomFieldOrderCnt = decodePicOrderCntType0(vid, ctx)
	case 1:
		topFieldOrderCnt, bottomFieldOrderCnt = decodePicOrderCntType1(vid, ctx)
	case 2:
		topFieldOrderCnt, bottomFieldOrderCnt = decodePicOrderCntType2(vid, ctx)
	default:
		err = errors.New("invalid PicOrderCountType")
	}

	// TODO: check DiffPicOrderCnt( picA, picB ) once picOrderCnt( picX ) is
	// worked out.
	return
}

// picOrderCntType0 is used to return topFieldOrderCnt and bottomFieldOrderCnt
// when pic_order_cnt_type i.e vid.PicOrderCntType == 0, using the process
// defined in section 8.2.1.1 of the specifications. If topFieldOrderCnt or
// bottomFieldOrderCnt are -1 they are unset.
func decodePicOrderCntType0(vid *VideoStream, ctx *SliceContext) (topFieldOrderCnt, bottomFieldOrderCnt int) {
	prevPicOrderCntMsb, prevPicOrderCntLsb := 0, 0
	topFieldOrderCnt, bottomFieldOrderCnt = -1, -1

	// NB: We're currently only handling IDRs so panic.
	if !vid.idrPicFlag {
		panic("not implemented")
	}

	switch {
	case (ctx.PicOrderCntLsb < prevPicOrderCntLsb) && ((prevPicOrderCntLsb - ctx.PicOrderCntLsb) >= (vid.maxPicOrderCntLsb / 2)):
		vid.picOrderCntMsb = prevPicOrderCntMsb + vid.maxPicOrderCntLsb
	case (ctx.PicOrderCntLsb > prevPicOrderCntLsb) && ((ctx.PicOrderCntLsb - prevPicOrderCntLsb) > (vid.maxPicOrderCntLsb / 2)):
		vid.picOrderCntMsb = prevPicOrderCntMsb - vid.maxPicOrderCntLsb
	default:
		vid.picOrderCntMsb = prevPicOrderCntMsb
	}

	if !ctx.BottomField {
		topFieldOrderCnt = vid.picOrderCntMsb + ctx.PicOrderCntLsb
	} else {
		if !ctx.FieldPic {
			bottomFieldOrderCnt = topFieldOrderCnt + ctx.DeltaPicOrderCntBottom
		} else {
			bottomFieldOrderCnt = vid.picOrderCntMsb + ctx.PicOrderCntLsb
		}
	}
	return
}

// picOrderCntType1 is used to return topFieldOrderCnt and bottomFieldOrderCnt
// when vic.PicOrderCntType == 1 according to logic defined in section 8.2.1.2
// of the specifications. If topFieldOrderCnt or bottomFieldOrderCnt are -1,
// then they are considered unset.
func decodePicOrderCntType1(vid *VideoStream, ctx *SliceContext) (topFieldOrderCnt, bottomFieldOrderCnt int) {
	topFieldOrderCnt, bottomFieldOrderCnt = -1, -1

	// TODO: this will be prevFrameNum when we do frames other than IDR.
	_ = vid.priorPic.FrameNum

	if vid.idrPicFlag {
		vid.frameNumOffset = 0
	} else {
		panic("not implemented")
	}

	absFrameNum := 0
	if ctx.NumRefFramesInPicOrderCntCycle != 0 {
		absFrameNum = vid.frameNumOffset + ctx.FrameNum
	}

	if ctx.RefIdc == 0 && absFrameNum > 0 {
		absFrameNum = absFrameNum - 1
	}

	var expectedPicOrderCnt int
	if absFrameNum > 0 {
		picOrderCntCycleCnt := (absFrameNum - 1) / int(ctx.NumRefFramesInPicOrderCntCycle)
		frameNumInPicOrderCntCycle := (absFrameNum - 1) % int(ctx.NumRefFramesInPicOrderCntCycle)
		expectedPicOrderCnt = picOrderCntCycleCnt * vid.expectedDeltaPerPicOrderCntCycle
		for i := 0; i <= frameNumInPicOrderCntCycle; i++ {
			expectedPicOrderCnt = expectedPicOrderCnt + ctx.OffsetForRefFrameList[i]
		}
	}

	if ctx.RefIdc == 0 {
		expectedPicOrderCnt = expectedPicOrderCnt + int(ctx.OffsetForNonRefPic)
	}

	switch {
	case !ctx.FieldPic:
		topFieldOrderCnt = expectedPicOrderCnt + ctx.DeltaPicOrderCnt[0]
		bottomFieldOrderCnt = topFieldOrderCnt + int(ctx.OffsetForTopToBottomField) + ctx.DeltaPicOrderCnt[1]
	case !ctx.BottomField:
		topFieldOrderCnt = expectedPicOrderCnt + ctx.DeltaPicOrderCnt[0]
	default:
		bottomFieldOrderCnt = expectedPicOrderCnt + int(ctx.OffsetForTopToBottomField) + ctx.DeltaPicOrderCnt[0]
	}
	return
}

// picOrderCntType2 is used to return topFieldOrderCnt and bottomFieldOrderCnt
// when vic.PicOrderCntType == 1 according to logic defined in section 8.2.1.3
// of the specifications. If topFieldOrderCnt or bottomFieldOrderCnt are -1,
// then they are considered unset.
func decodePicOrderCntType2(vid *VideoStream, ctx *SliceContext) (topFieldOrderCnt, bottomFieldOrderCnt int) {
	topFieldOrderCnt, bottomFieldOrderCnt = -1, -1

	// TODO: this will be prevFrameNum when we do frames other than IDR.
	_ = vid.priorPic.FrameNum

	if vid.idrPicFlag {
		vid.frameNumOffset = 0
	} else {
		panic("not implemented")
	}

	var tempPicOrderCnt int
	switch {
	case vid.idrPicFlag:
		tempPicOrderCnt = 0
	case ctx.RefIdc == 0:
		tempPicOrderCnt = 2*(vid.frameNumOffset+ctx.FrameNum) - 1
	default:
		tempPicOrderCnt = 2 * (vid.frameNumOffset + ctx.FrameNum)
	}

	switch {
	case ctx.FieldPic:
		topFieldOrderCnt = tempPicOrderCnt
		bottomFieldOrderCnt = tempPicOrderCnt
	case ctx.BottomField:
		bottomFieldOrderCnt = tempPicOrderCnt
	default:
		topFieldOrderCnt = tempPicOrderCnt
	}
	return
}