CABACWriter.cpp

/* The copyright in this software is being made available under the BSD
* License, included below. This software may be subject to other third party
* and contributor rights, including patent rights, and no such rights are
* granted under this license.
*
* Copyright (c) 2010-2019, ITU/ISO/IEC
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions are met:
*
*  * Redistributions of source code must retain the above copyright notice,
*    this list of conditions and the following disclaimer.
*  * Redistributions in binary form must reproduce the above copyright notice,
*    this list of conditions and the following disclaimer in the documentation
*    and/or other materials provided with the distribution.
*  * Neither the name of the ITU/ISO/IEC nor the names of its contributors may
*    be used to endorse or promote products derived from this software without
*    specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
* AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS
* BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
* CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
* SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
* INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
* CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
* ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF
* THE POSSIBILITY OF SUCH DAMAGE.
*/

/** \file     CABACWriter.cpp
 *  \brief    Writer for low level syntax
 */

#include "CommonLib/Contexts.h"
#include "CABACWriter.h"

#include "EncLib.h"

#include "CommonLib/UnitTools.h"
#include "CommonLib/dtrace_buffer.h"
#include "CommonLib/BinaryDecisionTree.h"

#include <map>
#include <algorithm>
#include <limits>


//! \ingroup EncoderLib
//! \{

void CABACWriter::initCtxModels( const Slice& slice )
{
  int       qp                = slice.getSliceQp();
  SliceType sliceType         = slice.getSliceType();
  SliceType encCABACTableIdx  = slice.getEncCABACTableIdx();
  if( !slice.isIntra() && (encCABACTableIdx==B_SLICE || encCABACTableIdx==P_SLICE) && slice.getPPS()->getCabacInitPresentFlag() )
  {
    sliceType = encCABACTableIdx;
  }
  m_BinEncoder.reset( qp, (int)sliceType );
}



template <class BinProbModel>
SliceType xGetCtxInitId( const Slice& slice, const BinEncIf& binEncoder, Ctx& ctxTest )
{
  const CtxStore<BinProbModel>& ctxStoreTest = static_cast<const CtxStore<BinProbModel>&>( ctxTest );
  const CtxStore<BinProbModel>& ctxStoreRef  = static_cast<const CtxStore<BinProbModel>&>( binEncoder.getCtx() );
  int qp = slice.getSliceQp();
  if( !slice.isIntra() )
  {
    SliceType aSliceTypeChoices[] = { B_SLICE, P_SLICE };
    uint64_t  bestCost            = std::numeric_limits<uint64_t>::max();
    SliceType bestSliceType       = aSliceTypeChoices[0];
    for (uint32_t idx=0; idx<2; idx++)
    {
      uint64_t  curCost           = 0;
      SliceType curSliceType      = aSliceTypeChoices[idx];
      ctxTest.init( qp, (int)curSliceType );
      for( int k = 0; k < Ctx::NumberOfContexts; k++ )
      {
        if( binEncoder.getNumBins(k) > 0 )
        {
          curCost += uint64_t( binEncoder.getNumBins(k) ) * ctxStoreRef[k].estFracExcessBits( ctxStoreTest[k] );
        }
      }
      if (curCost < bestCost)
      {
        bestSliceType = curSliceType;
        bestCost      = curCost;
      }
    }
    return bestSliceType;
  }
  else
  {
    return I_SLICE;
  }
}


SliceType CABACWriter::getCtxInitId( const Slice& slice )
{
  switch( m_TestCtx.getBPMType() )
  {
  case BPM_Std:   return  xGetCtxInitId<BinProbModel_Std>   ( slice, m_BinEncoder, m_TestCtx );
  default:        return  NUMBER_OF_SLICE_TYPES;
  }
}



unsigned estBits( BinEncIf& binEnc, const std::vector<bool>& bins, const Ctx& ctx, const int ctxId, const uint8_t winSize )
{
  binEnc.initCtxAndWinSize( ctxId, ctx, winSize );
  binEnc.start();
  const std::size_t numBins   = bins.size();
  unsigned          startBits = binEnc.getNumWrittenBits();
  for( std::size_t binId = 0; binId < numBins; binId++ )
  {
    unsigned  bin = ( bins[binId] ? 1 : 0 );
    binEnc.encodeBin( bin, ctxId );
  }
  unsigned endBits    = binEnc.getNumWrittenBits();
  unsigned codedBits  = endBits - startBits;
  return   codedBits;
}





//================================================================================
//  clause 7.3.8.1
//--------------------------------------------------------------------------------
//    void  end_of_slice()
//================================================================================

void CABACWriter::end_of_slice()
{
  m_BinEncoder.encodeBinTrm ( 1 );
  m_BinEncoder.finish       ();
}




//================================================================================
//  clause 7.3.8.2
//--------------------------------------------------------------------------------
//    bool  coding_tree_unit( cs, area, qp, ctuRsAddr, skipSao )
//================================================================================

void CABACWriter::coding_tree_unit( CodingStructure& cs, const UnitArea& area, int (&qps)[2], unsigned ctuRsAddr, bool skipSao /* = false */ )
{
  CUCtx cuCtx( qps[CH_L] );
  Partitioner *partitioner = PartitionerFactory::get( *cs.slice );

  partitioner->initCtu( area, CH_L, *cs.slice );

  if( !skipSao )
  {
    sao( *cs.slice, ctuRsAddr );
  }

  for( int compIdx = 0; compIdx < MAX_NUM_COMPONENT; compIdx++ )
  {
    codeAlfCtuEnableFlag( cs, ctuRsAddr, compIdx );
  }

  if ( CS::isDualITree(cs) && cs.pcv->chrFormat != CHROMA_400 && cs.pcv->maxCUWidth > 64 )
  {
    CUCtx chromaCuCtx(qps[CH_C]);
    Partitioner *chromaPartitioner = PartitionerFactory::get(*cs.slice);
    chromaPartitioner->initCtu(area, CH_C, *cs.slice);
    coding_tree(cs, *partitioner, cuCtx, chromaPartitioner, &chromaCuCtx);
    qps[CH_L] = cuCtx.qp;
    qps[CH_C] = chromaCuCtx.qp;

    delete chromaPartitioner;
  }
  else
  {
    coding_tree( cs, *partitioner, cuCtx );
    qps[CH_L] = cuCtx.qp;
    if( CS::isDualITree( cs ) && cs.pcv->chrFormat != CHROMA_400 )
    {
      CUCtx cuCtxChroma( qps[CH_C] );
      partitioner->initCtu( area, CH_C, *cs.slice );
      coding_tree( cs, *partitioner, cuCtxChroma );
      qps[CH_C] = cuCtxChroma.qp;
    }
  }

  delete partitioner;
}





//================================================================================
//  clause 7.3.8.3
//--------------------------------------------------------------------------------
//    void  sao             ( slice, ctuRsAddr )
//    void  sao_block_pars  ( saoPars, bitDepths, sliceEnabled, leftMergeAvail, aboveMergeAvail, onlyEstMergeInfo )
//    void  sao_offset_pars ( ctbPars, compID, sliceEnabled, bitDepth )
//================================================================================

void CABACWriter::sao( const Slice& slice, unsigned ctuRsAddr )
{
  const SPS& sps = *slice.getSPS();
  if( !sps.getUseSAO() )
  {
    return;
  }

  CodingStructure&     cs                     = *slice.getPic()->cs;
  const PreCalcValues& pcv                    = *cs.pcv;
  const SAOBlkParam&  sao_ctu_pars            = cs.picture->getSAO()[ctuRsAddr];
  bool                slice_sao_luma_flag     = ( slice.getSaoEnabledFlag( CHANNEL_TYPE_LUMA ) );
  bool                slice_sao_chroma_flag   = ( slice.getSaoEnabledFlag( CHANNEL_TYPE_CHROMA ) && sps.getChromaFormatIdc() != CHROMA_400 );
  if( !slice_sao_luma_flag && !slice_sao_chroma_flag )
  {
    return;
  }

  bool                sliceEnabled[3]         = { slice_sao_luma_flag, slice_sao_chroma_flag, slice_sao_chroma_flag };
  int                 frame_width_in_ctus     = pcv.widthInCtus;
  int                 ry                      = ctuRsAddr      / frame_width_in_ctus;
  int                 rx                      = ctuRsAddr - ry * frame_width_in_ctus;
  const Position      pos                     ( rx * cs.pcv->maxCUWidth, ry * cs.pcv->maxCUHeight );
  const unsigned      curSliceIdx             = slice.getIndependentSliceIdx();
#if HEVC_TILES_WPP
  const unsigned      curTileIdx              = cs.picture->tileMap->getTileIdxMap( pos );
  bool                leftMergeAvail          = cs.getCURestricted( pos.offset( -(int)pcv.maxCUWidth, 0  ), curSliceIdx, curTileIdx, CH_L ) ? true : false;
  bool                aboveMergeAvail         = cs.getCURestricted( pos.offset( 0, -(int)pcv.maxCUHeight ), curSliceIdx, curTileIdx, CH_L ) ? true : false;
#else
  bool                leftMergeAvail          = cs.getCURestricted( pos.offset( -(int)pcv.maxCUWidth, 0  ), curSliceIdx, CH_L ) ? true : false;
  bool                aboveMergeAvail         = cs.getCURestricted( pos.offset( 0, -(int)pcv.maxCUHeight ), curSliceIdx, CH_L ) ? true : false;
#endif
  sao_block_pars( sao_ctu_pars, sps.getBitDepths(), sliceEnabled, leftMergeAvail, aboveMergeAvail, false );
}


void CABACWriter::sao_block_pars( const SAOBlkParam& saoPars, const BitDepths& bitDepths, bool* sliceEnabled, bool leftMergeAvail, bool aboveMergeAvail, bool onlyEstMergeInfo )
{
  bool isLeftMerge  = false;
  bool isAboveMerge = false;
  if( leftMergeAvail )
  {
    // sao_merge_left_flag
    isLeftMerge   = ( saoPars[COMPONENT_Y].modeIdc == SAO_MODE_MERGE && saoPars[COMPONENT_Y].typeIdc == SAO_MERGE_LEFT );
    m_BinEncoder.encodeBin( (isLeftMerge), Ctx::SaoMergeFlag() );
  }
  if( aboveMergeAvail && !isLeftMerge )
  {
    // sao_merge_above_flag
    isAboveMerge  = ( saoPars[COMPONENT_Y].modeIdc == SAO_MODE_MERGE && saoPars[COMPONENT_Y].typeIdc == SAO_MERGE_ABOVE );
    m_BinEncoder.encodeBin( (isAboveMerge), Ctx::SaoMergeFlag() );
  }
  if( onlyEstMergeInfo )
  {
    return; //only for RDO
  }
  if( !isLeftMerge && !isAboveMerge )
  {
    // explicit parameters
    for( int compIdx=0; compIdx < MAX_NUM_COMPONENT; compIdx++ )
    {
      sao_offset_pars( saoPars[compIdx], ComponentID(compIdx), sliceEnabled[compIdx], bitDepths.recon[ toChannelType(ComponentID(compIdx)) ] );
    }
  }
}


void CABACWriter::sao_offset_pars( const SAOOffset& ctbPars, ComponentID compID, bool sliceEnabled, int bitDepth )
{
  if( !sliceEnabled )
  {
    CHECK( ctbPars.modeIdc != SAO_MODE_OFF, "Sao must be off, if it is disabled on slice level" );
    return;
  }
  const bool isFirstCompOfChType = ( getFirstComponentOfChannel( toChannelType(compID) ) == compID );

  if( isFirstCompOfChType )
  {
    // sao_type_idx_luma / sao_type_idx_chroma
    if( ctbPars.modeIdc == SAO_MODE_OFF )
    {
      m_BinEncoder.encodeBin  ( 0, Ctx::SaoTypeIdx() );
    }
    else if( ctbPars.typeIdc == SAO_TYPE_BO )
    {
      m_BinEncoder.encodeBin  ( 1, Ctx::SaoTypeIdx() );
      m_BinEncoder.encodeBinEP( 0 );
    }
    else
    {
      CHECK(!( ctbPars.typeIdc < SAO_TYPE_START_BO ), "Unspecified error");
      m_BinEncoder.encodeBin  ( 1, Ctx::SaoTypeIdx() );
      m_BinEncoder.encodeBinEP( 1 );
    }
  }

  if( ctbPars.modeIdc == SAO_MODE_NEW )
  {
    const int maxOffsetQVal = SampleAdaptiveOffset::getMaxOffsetQVal( bitDepth );
    int       numClasses    = ( ctbPars.typeIdc == SAO_TYPE_BO ? 4 : NUM_SAO_EO_CLASSES );
    int       k             = 0;
    int       offset[4];
    for( int i = 0; i < numClasses; i++ )
    {
      if( ctbPars.typeIdc != SAO_TYPE_BO && i == SAO_CLASS_EO_PLAIN )
      {
        continue;
      }
      int classIdx = ( ctbPars.typeIdc == SAO_TYPE_BO ? ( ctbPars.typeAuxInfo + i ) % NUM_SAO_BO_CLASSES : i );
      offset[k++]  = ctbPars.offset[classIdx];
    }

    // sao_offset_abs
    for( int i = 0; i < 4; i++ )
    {
      unsigned absOffset = ( offset[i] < 0 ? -offset[i] : offset[i] );
      unary_max_eqprob( absOffset, maxOffsetQVal );
    }

    // band offset mode
    if( ctbPars.typeIdc == SAO_TYPE_BO )
    {
      // sao_offset_sign
      for( int i = 0; i < 4; i++ )
      {
        if( offset[i] )
        {
          m_BinEncoder.encodeBinEP( (offset[i] < 0) );
        }
      }
      // sao_band_position
      m_BinEncoder.encodeBinsEP( ctbPars.typeAuxInfo, NUM_SAO_BO_CLASSES_LOG2 );
    }
    // edge offset mode
    else
    {
      if( isFirstCompOfChType )
      {
        // sao_eo_class_luma / sao_eo_class_chroma
        CHECK( ctbPars.typeIdc - SAO_TYPE_START_EO < 0, "sao edge offset class is outside valid range" );
        m_BinEncoder.encodeBinsEP( ctbPars.typeIdc - SAO_TYPE_START_EO, NUM_SAO_EO_TYPES_LOG2 );
      }
    }
  }
}



//================================================================================
//  clause 7.3.8.4
//--------------------------------------------------------------------------------
//    void  coding_tree       ( cs, partitioner, cuCtx )
//    void  split_cu_flag     ( split, cs, partitioner )
//    void  split_cu_mode_mt  ( split, cs, partitioner )
//================================================================================

void CABACWriter::coding_tree(const CodingStructure& cs, Partitioner& partitioner, CUCtx& cuCtx, Partitioner* pPartitionerChroma, CUCtx* pCuCtxChroma)
{
  const PPS      &pps         = *cs.pps;
  const UnitArea &currArea    = partitioner.currArea();
  const CodingUnit &cu        = *cs.getCU( currArea.blocks[partitioner.chType], partitioner.chType );

  // Reset delta QP coding flag and ChromaQPAdjustemt coding flag
  if( pps.getUseDQP() && partitioner.currDepth <= pps.getMaxCuDQPDepth() )
  {
    cuCtx.isDQPCoded          = false;
  }
  if( cs.slice->getUseChromaQpAdj() && partitioner.currDepth <= pps.getPpsRangeExtension().getDiffCuChromaQpOffsetDepth() )
  {
    cuCtx.isChromaQpAdjCoded  = false;
  }
  // Reset delta QP coding flag and ChromaQPAdjustemt coding flag
  if (CS::isDualITree(cs) && pPartitionerChroma != nullptr)
  {
    if (pps.getUseDQP() && pPartitionerChroma->currDepth <= pps.getMaxCuDQPDepth())
    {
      pCuCtxChroma->isDQPCoded = false;
    }
    if (cs.slice->getUseChromaQpAdj() && pPartitionerChroma->currDepth <= pps.getPpsRangeExtension().getDiffCuChromaQpOffsetDepth())
    {
      pCuCtxChroma->isChromaQpAdjCoded = false;
    }
  }

  const PartSplit implicitSplit = partitioner.getImplicitSplit( cs );

  // QT
  bool canQtSplit = partitioner.canSplit( CU_QUAD_SPLIT, cs );

  if( canQtSplit )
  {
    // split_cu_flag
    bool qtSplit = implicitSplit == CU_QUAD_SPLIT;

    if( !qtSplit && implicitSplit != CU_QUAD_SPLIT )
    {
      qtSplit = ( cu.qtDepth > partitioner.currQtDepth );
      split_cu_flag( qtSplit, cs, partitioner );
    }

    // quad-tree split
    if( qtSplit )
    {
      if (CS::isDualITree(cs) && pPartitionerChroma != nullptr && (partitioner.currArea().lwidth() >= 64 || partitioner.currArea().lheight() >= 64))
      {
        partitioner.splitCurrArea(CU_QUAD_SPLIT, cs);
        pPartitionerChroma->splitCurrArea(CU_QUAD_SPLIT, cs);
        bool beContinue = true;
        bool lumaContinue = true;
        bool chromaContinue = true;

        while (beContinue)
        {
          if (partitioner.currArea().lwidth() > 64 || partitioner.currArea().lheight() > 64)
          {
            if (cs.picture->blocks[partitioner.chType].contains(partitioner.currArea().blocks[partitioner.chType].pos()))
            {
              coding_tree(cs, partitioner, cuCtx, pPartitionerChroma, pCuCtxChroma);
            }
            lumaContinue = partitioner.nextPart(cs);
            chromaContinue = pPartitionerChroma->nextPart(cs);
            CHECK(lumaContinue != chromaContinue, "luma chroma partition should be matched");
            beContinue = lumaContinue;
          }
          else
          {
            //dual tree coding under 64x64 block
            if (cs.picture->blocks[partitioner.chType].contains(partitioner.currArea().blocks[partitioner.chType].pos()))
            {
              coding_tree(cs, partitioner, cuCtx);
            }
            lumaContinue = partitioner.nextPart(cs);
            if (cs.picture->blocks[pPartitionerChroma->chType].contains(pPartitionerChroma->currArea().blocks[pPartitionerChroma->chType].pos()))
            {
              coding_tree(cs, *pPartitionerChroma, *pCuCtxChroma);
            }
            chromaContinue = pPartitionerChroma->nextPart(cs);
            CHECK(lumaContinue != chromaContinue, "luma chroma partition should be matched");
            beContinue = lumaContinue;
          }
        }
        partitioner.exitCurrSplit();
        pPartitionerChroma->exitCurrSplit();

      }
      else
      {
      partitioner.splitCurrArea( CU_QUAD_SPLIT, cs );

      do
      {
        if( cs.picture->blocks[partitioner.chType].contains( partitioner.currArea().blocks[partitioner.chType].pos() ) )
        {
          coding_tree( cs, partitioner, cuCtx );
        }
      } while( partitioner.nextPart( cs ) );

      partitioner.exitCurrSplit();
      }
      return;
    }
  }

  {
    bool mtSplit = partitioner.canSplit( CU_MT_SPLIT, cs );

    if( mtSplit )
    {
      const PartSplit splitMode = CU::getSplitAtDepth( cu, partitioner.currDepth );

      split_cu_mode_mt( splitMode, cs, partitioner );

      if( splitMode != CU_DONT_SPLIT )
      {
        partitioner.splitCurrArea( splitMode, cs );
        do
        {
          if( cs.picture->blocks[partitioner.chType].contains( partitioner.currArea().blocks[partitioner.chType].pos() ) )
          {
            coding_tree( cs, partitioner, cuCtx );
          }
        } while( partitioner.nextPart( cs ) );

        partitioner.exitCurrSplit();
        return;
      }
    }
  }

  // Predict QP on start of quantization group
  if( pps.getUseDQP() && !cuCtx.isDQPCoded && CU::isQGStart( cu, partitioner ) )
  {
    cuCtx.qp = CU::predictQP( cu, cuCtx.qp );
  }


  // coding unit
  coding_unit( cu, partitioner, cuCtx );

  DTRACE_COND( ( isEncoding() ), g_trace_ctx, D_QP, "x=%d, y=%d, w=%d, h=%d, qp=%d\n", cu.Y().x, cu.Y().y, cu.Y().width, cu.Y().height, cu.qp );
  DTRACE_BLOCK_REC_COND( ( !isEncoding() ), cs.picture->getRecoBuf( cu ), cu, cu.predMode );
}

void CABACWriter::split_cu_flag( bool split, const CodingStructure& cs, Partitioner& partitioner )
{
  unsigned maxQTDepth = ( g_aucLog2[cs.sps->getCTUSize()] - g_aucLog2[cs.sps->getMinQTSize(cs.slice->getSliceType(), partitioner.chType)] );
  if( partitioner.currDepth == maxQTDepth )
  {
    return;
  }
  unsigned  ctxId = DeriveCtx::CtxCUsplit( cs, partitioner );
  m_BinEncoder.encodeBin( (split), Ctx::SplitFlag(ctxId) );

  DTRACE( g_trace_ctx, D_SYNTAX, "split_cu_flag() ctx=%d split=%d\n", ctxId, split ? 1 : 0 );
}

void CABACWriter::split_cu_mode_mt(const PartSplit split, const CodingStructure& cs, Partitioner& partitioner)
{
  unsigned ctxIdBT = DeriveCtx::CtxBTsplit( cs, partitioner );

  unsigned width   = partitioner.currArea().lumaSize().width;
  unsigned height  = partitioner.currArea().lumaSize().height;

  DecisionTree dt( g_mtSplitDTT );

  dt.setAvail( DTT_SPLIT_BT_HORZ,  partitioner.canSplit( CU_HORZ_SPLIT, cs ) );
  dt.setAvail( DTT_SPLIT_BT_VERT,  partitioner.canSplit( CU_VERT_SPLIT, cs ) );

  dt.setAvail( DTT_SPLIT_TT_HORZ,  partitioner.canSplit( CU_TRIH_SPLIT, cs ) );
  dt.setAvail( DTT_SPLIT_TT_VERT,  partitioner.canSplit( CU_TRIV_SPLIT, cs ) );
  dt.setAvail( DTT_SPLIT_NO_SPLIT, partitioner.canSplit( CU_DONT_SPLIT, cs ) );

  unsigned btSCtxId = width == height ? 0 : ( width > height ? 1 : 2 );
  dt.setCtxId( DTT_SPLIT_DO_SPLIT_DECISION,   Ctx::BTSplitFlag( ctxIdBT ) );
  dt.setCtxId( DTT_SPLIT_HV_DECISION,         Ctx::BTSplitFlag( 12 + btSCtxId ) );

  dt.setCtxId( DTT_SPLIT_H_IS_BT_12_DECISION, Ctx::BTSplitFlag( 15 ) );
  dt.setCtxId( DTT_SPLIT_V_IS_BT_12_DECISION, Ctx::BTSplitFlag( 15 ) );


  encode_sparse_dt( dt, split == CU_DONT_SPLIT ? ( unsigned ) DTT_SPLIT_NO_SPLIT : ( unsigned ) split );

  DTRACE(g_trace_ctx, D_SYNTAX, "split_cu_mode_mt() ctx=%d split=%d\n", ctxIdBT, split);
}


//================================================================================
//  clause 7.3.8.5
//--------------------------------------------------------------------------------
//    void  coding_unit               ( cu, partitioner, cuCtx )
//    void  cu_transquant_bypass_flag ( cu )
//    void  cu_skip_flag              ( cu )
//    void  pred_mode                 ( cu )
//    void  part_mode                 ( cu )
//    void  pcm_flag                  ( cu )
//    void  pcm_samples               ( tu )
//    void  cu_pred_data              ( pus )
//    void  cu_lic_flag               ( cu )
//    void  intra_luma_pred_modes     ( pus )
//    void  intra_chroma_pred_mode    ( pu )
//    void  cu_residual               ( cu, partitioner, cuCtx )
//    void  rqt_root_cbf              ( cu )
//    void  end_of_ctu                ( cu, cuCtx )
//================================================================================

void CABACWriter::coding_unit( const CodingUnit& cu, Partitioner& partitioner, CUCtx& cuCtx )
{
  CodingStructure& cs = *cu.cs;
  cs.chType = partitioner.chType;
  // transquant bypass flag
  if( cs.pps->getTransquantBypassEnabledFlag() )
  {
    cu_transquant_bypass_flag( cu );
  }

  // skip flag
  if (!cs.slice->isIntra() && cu.Y().valid())
  {
    cu_skip_flag( cu );
  }


  // skip data
  if( cu.skip )
  {
    CHECK( !cu.firstPU->mergeFlag, "Merge flag has to be on!" );
    PredictionUnit&   pu = *cu.firstPU;
    prediction_unit ( pu );
    end_of_ctu      ( cu, cuCtx );
    return;
  }

  // pcm samples
  if( CU::isIntra(cu) )
  {
    pcm_data( cu, partitioner );
    if( cu.ipcm )
    {
      end_of_ctu( cu, cuCtx );
      return;
    }
  }

  // prediction mode and partitioning data
  pred_mode ( cu );

  extend_ref_line(cu);


  // prediction data ( intra prediction modes / reference indexes + motion vectors )
  cu_pred_data( cu );

  // residual data ( coded block flags + transform coefficient levels )
  cu_residual( cu, partitioner, cuCtx );

  // end of cu
  end_of_ctu( cu, cuCtx );
}


void CABACWriter::cu_transquant_bypass_flag( const CodingUnit& cu )
{
  m_BinEncoder.encodeBin( (cu.transQuantBypass), Ctx::TransquantBypassFlag() );
}


void CABACWriter::cu_skip_flag( const CodingUnit& cu )
{
  unsigned ctxId = DeriveCtx::CtxSkipFlag( cu );
  m_BinEncoder.encodeBin( ( cu.skip ), Ctx::SkipFlag( ctxId ) );

  DTRACE( g_trace_ctx, D_SYNTAX, "cu_skip_flag() ctx=%d skip=%d\n", ctxId, cu.skip ? 1 : 0 );
  if (cu.skip)
  {
    m_BinEncoder.encodeBin(cu.mmvdSkip, Ctx::MmvdFlag(0));
    DTRACE(g_trace_ctx, D_SYNTAX, "mmvd_cu_skip_flag() ctx=%d mmvd_skip=%d\n", 0, cu.mmvdSkip ? 1 : 0);
  }
}


void CABACWriter::pred_mode( const CodingUnit& cu )
{
  if( cu.cs->slice->isIntra() )
  {
    return;
  }
  m_BinEncoder.encodeBin( ( CU::isIntra( cu ) ), Ctx::PredMode() );
}

void CABACWriter::pcm_data( const CodingUnit& cu, Partitioner& partitioner  )
{
  pcm_flag( cu, partitioner );
  if( cu.ipcm )
  {
    m_BinEncoder.pcmAlignBits();
    pcm_samples( *cu.firstTU );
  }
}

void CABACWriter::pcm_flag( const CodingUnit& cu, Partitioner& partitioner )
{
  const SPS& sps = *cu.cs->sps;
  if( !sps.getUsePCM() || partitioner.currArea().lwidth() > (1 << sps.getPCMLog2MaxSize()) || partitioner.currArea().lwidth() < (1 << sps.getPCMLog2MinSize()) 
      || partitioner.currArea().lheight() > (1 << sps.getPCMLog2MaxSize()) || partitioner.currArea().lheight() < (1 << sps.getPCMLog2MinSize()) )
  {
    return;
  }
  m_BinEncoder.encodeBinTrm( cu.ipcm );
}


void CABACWriter::cu_pred_data( const CodingUnit& cu )
{
  if( CU::isIntra( cu ) )
  {
    intra_luma_pred_modes  ( cu );
    intra_chroma_pred_modes( cu );
    return;
  }
  if (!cu.Y().valid()) // dual tree chroma CU
  {
    return;
  }
  for( auto &pu : CU::traversePUs( cu ) )
  {
    prediction_unit( pu );
  }

  imv_mode   ( cu );

  cu_gbi_flag( cu );

}

void CABACWriter::cu_gbi_flag(const CodingUnit& cu)
{
  if(!CU::isGBiIdxCoded(cu))
  {
    return;
  }

  CHECK(!(GBI_NUM > 1 && (GBI_NUM == 2 || (GBI_NUM & 0x01) == 1)), " !( GBI_NUM > 1 && ( GBI_NUM == 2 || ( GBI_NUM & 0x01 ) == 1 ) ) ");
  const uint8_t gbiCodingIdx = (uint8_t)g_GbiCodingOrder[CU::getValidGbiIdx(cu)];

  int ctxId = 0;

  int32_t numGBi = (cu.slice->getCheckLDC()) ? 5 : 3;

  m_BinEncoder.encodeBin((gbiCodingIdx == 0 ? 1 : 0), Ctx::GBiIdx(ctxId));

  if(numGBi > 2 && gbiCodingIdx != 0)
  {
    uint32_t prefixNumBits = numGBi - 2;
    uint32_t step = 1;
    uint8_t prefixSymbol = gbiCodingIdx;

    int ctxIdGBi = 4;
    uint8_t idx = 1;
    for(int ui = 0; ui < prefixNumBits; ++ui)
    {
      if (prefixSymbol == idx)
      {
        m_BinEncoder.encodeBin(1, Ctx::GBiIdx(ctxIdGBi));
        break;
      }
      else
      {
        m_BinEncoder.encodeBin(0, Ctx::GBiIdx(ctxIdGBi));
        ctxIdGBi += step;
        idx += step;
      }
    }
  }

  DTRACE(g_trace_ctx, D_SYNTAX, "cu_gbi_flag() gbi_idx=%d\n", cu.GBiIdx ? 1 : 0);
}

void CABACWriter::xWriteTruncBinCode(uint32_t symbol, uint32_t maxSymbol)
{
  int thresh;
  if (maxSymbol > 256)
  {
    int threshVal = 1 << 8;
    thresh = 8;
    while (threshVal <= maxSymbol)
    {
      thresh++;
      threshVal <<= 1;
    }
    thresh--;
  }
  else
  {
    thresh = g_tbMax[maxSymbol];
  }

  int val = 1 << thresh;
  assert(val <= maxSymbol);
  assert((val << 1) > maxSymbol);
  assert(symbol < maxSymbol);
  int b = maxSymbol - val;
  assert(b < val);
  if (symbol < val - b)
  {
    m_BinEncoder.encodeBinsEP(symbol, thresh);
  }
  else
  {
    symbol += val - b;
    assert(symbol < (val << 1));
    assert((symbol >> 1) >= val - b);
    m_BinEncoder.encodeBinsEP(symbol, thresh + 1);
  }
}

void CABACWriter::extend_ref_line(const PredictionUnit& pu)
{
  const CodingUnit& cu = *pu.cu;
  if (!cu.Y().valid() || cu.predMode != MODE_INTRA || !isLuma(cu.chType))
  {
    return;
  }
  bool isFirstLineOfCtu = (((cu.block(COMPONENT_Y).y)&((cu.cs->sps)->getMaxCUWidth() - 1)) == 0);
  if (isFirstLineOfCtu)
  {
    return;
  }
  int multiRefIdx = pu.multiRefIdx;
  if (MRL_NUM_REF_LINES > 1)
  {
    m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[0], Ctx::MultiRefLineIdx(0));
    if (MRL_NUM_REF_LINES > 2 && multiRefIdx != MULTI_REF_LINE_IDX[0])
    {
      m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[1], Ctx::MultiRefLineIdx(1));
      if (MRL_NUM_REF_LINES > 3 && multiRefIdx != MULTI_REF_LINE_IDX[1])
      {
        m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[2], Ctx::MultiRefLineIdx(2));
      }
    }
  }
}

void CABACWriter::extend_ref_line(const CodingUnit& cu)
{
  if (!cu.Y().valid() || cu.predMode != MODE_INTRA || !isLuma(cu.chType))
  {
    return;
  }

  const int numBlocks = CU::getNumPUs(cu);
  const PredictionUnit* pu = cu.firstPU;

  for (int k = 0; k < numBlocks; k++)
  {
    bool isFirstLineOfCtu = (((cu.block(COMPONENT_Y).y)&((cu.cs->sps)->getMaxCUWidth() - 1)) == 0);
    if (isFirstLineOfCtu)
    {
      return;
    }
    int multiRefIdx = pu->multiRefIdx;
    if (MRL_NUM_REF_LINES > 1)
    {
      m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[0], Ctx::MultiRefLineIdx(0));
      if (MRL_NUM_REF_LINES > 2 && multiRefIdx != MULTI_REF_LINE_IDX[0])
      {
        m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[1], Ctx::MultiRefLineIdx(1));
        if (MRL_NUM_REF_LINES > 3 && multiRefIdx != MULTI_REF_LINE_IDX[1])
        {
          m_BinEncoder.encodeBin(multiRefIdx != MULTI_REF_LINE_IDX[2], Ctx::MultiRefLineIdx(2));
        }
      }

    }
    pu = pu->next;
  }
}

void CABACWriter::intra_luma_pred_modes( const CodingUnit& cu )
{
  if( !cu.Y().valid() )
  {
    return;
  }

  const int numMPMs   = NUM_MOST_PROBABLE_MODES;
  const int numBlocks = CU::getNumPUs( cu );
  unsigned  mpm_preds   [4][numMPMs];
  unsigned  mpm_idxs    [4];
  unsigned  ipred_modes [4];

  const PredictionUnit* pu = cu.firstPU;

  // prev_intra_luma_pred_flag
  for( int k = 0; k < numBlocks; k++ )
  {
    unsigned*  mpm_pred   = mpm_preds[k];
    unsigned&  mpm_idx    = mpm_idxs[k];
    unsigned&  ipred_mode = ipred_modes[k];

    PU::getIntraMPMs( *pu, mpm_pred );

    ipred_mode = pu->intraDir[0];
    mpm_idx    = numMPMs;
    for( unsigned idx = 0; idx < numMPMs; idx++ )
    {
      if( ipred_mode == mpm_pred[idx] )
      {
        mpm_idx = idx;
        break;
      }
    }
    if (pu->multiRefIdx)
    {
      CHECK(mpm_idx >= numMPMs, "use of non-MPM");
    }
    else
    m_BinEncoder.encodeBin( mpm_idx < numMPMs, Ctx::IPredMode[0]() );

    pu = pu->next;
  }

  pu = cu.firstPU;

  // mpm_idx / rem_intra_luma_pred_mode
  for( int k = 0; k < numBlocks; k++ )
  {
    const unsigned& mpm_idx = mpm_idxs[k];
    if( mpm_idx < numMPMs )
    {
      {
        m_BinEncoder.encodeBinEP( mpm_idx > 0 );
        if( mpm_idx )
        {
          m_BinEncoder.encodeBinEP( mpm_idx > 1 );
        }
        if (mpm_idx > 1)
        {
          m_BinEncoder.encodeBinEP(mpm_idx > 2);
        }
        if (mpm_idx > 2)
        {
          m_BinEncoder.encodeBinEP(mpm_idx > 3);
        }
        if (mpm_idx > 3)
        {
          m_BinEncoder.encodeBinEP(mpm_idx > 4);
        }
      }
    }
    else
    {
      unsigned* mpm_pred   = mpm_preds[k];
      unsigned  ipred_mode = ipred_modes[k];

      // sorting of MPMs
      std::sort( mpm_pred, mpm_pred + numMPMs );

      {
        for (int idx = numMPMs - 1; idx >= 0; idx--)
        {
          if (ipred_mode > mpm_pred[idx])
          {
            ipred_mode--;
          }
        }
        CHECK(ipred_mode >= 64, "Incorrect mode");
        xWriteTruncBinCode(ipred_mode, NUM_LUMA_MODE - NUM_MOST_PROBABLE_MODES);  // Remaining mode is truncated binary coded
      }
    }

    DTRACE( g_trace_ctx, D_SYNTAX, "intra_luma_pred_modes() idx=%d pos=(%d,%d) mode=%d\n", k, pu->lumaPos().x, pu->lumaPos().y, pu->intraDir[0] );
    pu = pu->next;
  }
}


void CABACWriter::intra_luma_pred_mode( const PredictionUnit& pu )
{

  // prev_intra_luma_pred_flag
  const int numMPMs  = NUM_MOST_PROBABLE_MODES;
  unsigned  mpm_pred[numMPMs];
  
  PU::getIntraMPMs( pu, mpm_pred );

  unsigned ipred_mode = pu.intraDir[0];
  unsigned mpm_idx = numMPMs;

  for( int idx = 0; idx < numMPMs; idx++ )
  {
    if( ipred_mode == mpm_pred[idx] )
    {
      mpm_idx = idx;
      break;
    }
  }
  if (pu.multiRefIdx)
  {
    CHECK(mpm_idx >= numMPMs, "use of non-MPM");
  }
  else
  m_BinEncoder.encodeBin( mpm_idx < numMPMs, Ctx::IPredMode[0]() );

  // mpm_idx / rem_intra_luma_pred_mode
  if( mpm_idx < numMPMs )
  {
    {
      m_BinEncoder.encodeBinEP( mpm_idx > 0 );
      if( mpm_idx )
      {
        m_BinEncoder.encodeBinEP( mpm_idx > 1 );
      }
      if (mpm_idx > 1)
      {
        m_BinEncoder.encodeBinEP(mpm_idx > 2);
      }
      if (mpm_idx > 2)
      {
        m_BinEncoder.encodeBinEP(mpm_idx > 3);
      }
      if (mpm_idx > 3)
      {
        m_BinEncoder.encodeBinEP(mpm_idx > 4);
      }
    }
  }