brotli/enc/block_splitter.cc

// Copyright 2013 Google Inc. All Rights Reserved.
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.
//
// Block split point selection utilities.

#include "./block_splitter.h"

#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include <algorithm>
#include <map>

#include "./cluster.h"
#include "./command.h"
#include "./fast_log.h"
#include "./histogram.h"

namespace brotli {

static const int kMaxLiteralHistograms = 48;
static const int kMaxCommandHistograms = 50;
static const double kLiteralBlockSwitchCost = 26;
static const double kCommandBlockSwitchCost = 13.5;
static const double kDistanceBlockSwitchCost = 14.6;
static const int kLiteralStrideLength = 70;
static const int kCommandStrideLength = 40;
static const int kSymbolsPerLiteralHistogram = 550;
static const int kSymbolsPerCommandHistogram = 530;
static const int kSymbolsPerDistanceHistogram = 550;
static const int kMinLengthForBlockSplitting = 128;
static const int kIterMulForRefining = 2;
static const int kMinItersForRefining = 100;

void CopyLiteralsToByteArray(const std::vector<Command>& cmds,
                             const uint8_t* data,
                             std::vector<uint8_t>* literals) {
  // Count how many we have.
  size_t total_length = 0;
  for (int i = 0; i < cmds.size(); ++i) {
    total_length += cmds[i].insert_length_;
  }
  if (total_length == 0) {
    return;
  }

  // Allocate.
  literals->resize(total_length);

  // Loop again, and copy this time.
  size_t pos = 0;
  size_t from_pos = 0;
  for (int i = 0; i < cmds.size() && pos < total_length; ++i) {
    memcpy(&(*literals)[pos], data + from_pos, cmds[i].insert_length_);
    pos += cmds[i].insert_length_;
    from_pos += cmds[i].insert_length_ + cmds[i].copy_length_;
  }
}

void CopyCommandsToByteArray(const std::vector<Command>& cmds,
                             std::vector<uint16_t>* insert_and_copy_codes,
                             std::vector<uint8_t>* distance_prefixes) {
  for (int i = 0; i < cmds.size(); ++i) {
    const Command& cmd = cmds[i];
    insert_and_copy_codes->push_back(cmd.command_prefix_);
    if (cmd.copy_length_ > 0 && cmd.distance_prefix_ != 0xffff) {
      distance_prefixes->push_back(cmd.distance_prefix_);
    }
  }
}

template<int kSize>
double HistogramAddEval(const Histogram<kSize>& a,
                        const Histogram<kSize>& b) {
  int total = a.total_count_ + b.total_count_;
  double retval = total * FastLog2(total);
  for (int i = 0; i < kSize; ++i) {
    int count = a.data_[i] + b.data_[i];
    retval -= count * FastLog2(count);
  }
  return retval;
}

template<typename HistogramType, typename DataType>
void InitialEntropyCodes(const DataType* data, size_t length,
                         int literals_per_histogram,
                         int max_histograms,
                         size_t stride,
                         std::vector<HistogramType>* vec) {
  int total_histograms = length / literals_per_histogram + 1;
  if (total_histograms > max_histograms) {
    total_histograms = max_histograms;
  }
  unsigned int seed = 7;
  int block_length = length / total_histograms;
  for (int i = 0; i < total_histograms; ++i) {
    int pos = length * i / total_histograms;
    if (i != 0) {
      pos += rand_r(&seed) % block_length;
    }
    if (pos + stride >= length) {
      pos = length - stride - 1;
    }
    HistogramType histo;
    histo.Add(data + pos, stride);
    vec->push_back(histo);
  }
}

template<typename HistogramType>
int FindClosest(const HistogramType& sample,
                const std::vector<HistogramType>& vec) {
  double best_distance = 1e99;
  int best_ix = 0;
  for (int i = 0; i < vec.size(); ++i) {
    double distance = HistogramAddEval(sample, vec[i]);
    if (distance < best_distance) {
      best_ix = i;
      best_distance = distance;
    }
  }
  return best_ix;
}

template<typename HistogramType, typename DataType>
void RandomSample(unsigned int* seed,
                  const DataType* data,
                  size_t length,
                  size_t stride,
                  HistogramType* sample) {
  size_t pos = rand_r(seed) % (length - stride);
  sample->Add(data + pos, stride);
}

template<typename HistogramType, typename DataType>
void RefineEntropyCodes(const DataType* data, size_t length,
                        size_t stride,
                        std::vector<HistogramType>* vec) {
  const int iters =
      kIterMulForRefining * length / stride + kMinItersForRefining;
  unsigned int seed = 7;
  for (int iter = 0; iter < iters; ++iter) {
    HistogramType sample;
    RandomSample(&seed, data, length, stride, &sample);
    int ix = FindClosest(sample, *vec);
    (*vec)[ix].AddHistogram(sample);
  }
}

inline static float BitCost(int total, int count) {
  return count == 0 ? FastLog2(total) + 2 : FastLog2(total) - FastLog2(count);
}

template<typename DataType, int kSize>
void FindBlocks(const DataType* data, const size_t length,
                const double block_switch_bitcost,
                const std::vector<Histogram<kSize> > &vec,
                uint8_t *block_id) {
  if (vec.size() <= 1) {
    for (int i = 0; i < length; ++i) {
      block_id[i] = 0;
    }
    return;
  }
  int vecsize = vec.size();
  double* insert_cost = new double[kSize * vecsize];
  memset(insert_cost, 0, sizeof(insert_cost[0]) * kSize * vecsize);
  for (int i = 0; i < kSize; ++i) {
    for (int j = 0; j < vecsize; ++j) {
      insert_cost[i * vecsize + j] =
          BitCost(vec[j].total_count_, vec[j].data_[i]);
    }
  }
  double *cost = new double[vecsize];
  memset(cost, 0, sizeof(cost[0]) * vecsize);
  bool* switch_signal = new bool[length * vecsize];
  memset(switch_signal, 0, sizeof(switch_signal[0]) * length * vecsize);
  // After each iteration of this loop, cost[k] will contain the difference
  // between the minimum cost of arriving at the current byte position using
  // entropy code k, and the minimum cost of arriving at the current byte
  // position. This difference is capped at the block switch cost, and if it
  // reaches block switch cost, it means that when we trace back from the last
  // position, we need to switch here.
  for (size_t byte_ix = 0; byte_ix < length; ++byte_ix) {
    int ix = byte_ix * vecsize;
    int insert_cost_ix = data[byte_ix] * vecsize;
    double min_cost = 1e99;
    for (int k = 0; k < vecsize; ++k) {
      // We are coding the symbol in data[byte_ix] with entropy code k.
      cost[k] += insert_cost[insert_cost_ix + k];
      if (cost[k] < min_cost) {
        min_cost = cost[k];
        block_id[byte_ix] = k;
      }
    }
    double block_switch_cost = block_switch_bitcost;
    // More blocks for the beginning.
    if (byte_ix < 2000) {
      block_switch_cost *= 0.77 + 0.07 * byte_ix / 2000;
    }
    for (int k = 0; k < vecsize; ++k) {
      cost[k] -= min_cost;
      if (cost[k] >= block_switch_cost) {
        cost[k] = block_switch_cost;
        switch_signal[ix + k] = true;
      }
    }
  }
  // Now trace back from the last position and switch at the marked places.
  int byte_ix = length - 1;
  int ix = byte_ix * vecsize;
  int cur_id = block_id[byte_ix];
  while (byte_ix > 0) {
    --byte_ix;
    ix -= vecsize;
    if (switch_signal[ix + cur_id]) {
      cur_id = block_id[byte_ix];
    }
    block_id[byte_ix] = cur_id;
  }
  delete[] insert_cost;
  delete[] cost;
  delete[] switch_signal;
}

int RemapBlockIds(uint8_t* block_ids, const size_t length) {
  std::map<uint8_t, uint8_t> new_id;
  int next_id = 0;
  for (int i = 0; i < length; ++i) {
    if (new_id.find(block_ids[i]) == new_id.end()) {
      new_id[block_ids[i]] = next_id;
      ++next_id;
    }
  }
  for (int i = 0; i < length; ++i) {
    block_ids[i] = new_id[block_ids[i]];
  }
  return next_id;
}

template<typename HistogramType, typename DataType>
void BuildBlockHistograms(const DataType* data, const size_t length,
                          uint8_t* block_ids,
                          std::vector<HistogramType>* histograms) {
  int num_types = RemapBlockIds(block_ids, length);
  histograms->clear();
  histograms->resize(num_types);
  for (int i = 0; i < length; ++i) {
    (*histograms)[block_ids[i]].Add(data[i]);
  }
}

template<typename HistogramType, typename DataType>
void ClusterBlocks(const DataType* data, const size_t length,
                   uint8_t* block_ids) {
  std::vector<HistogramType> histograms;
  std::vector<int> block_index(length);
  int cur_idx = 0;
  HistogramType cur_histogram;
  for (int i = 0; i < length; ++i) {
    bool block_boundary = (i + 1 == length || block_ids[i] != block_ids[i + 1]);
    block_index[i] = cur_idx;
    cur_histogram.Add(data[i]);
    if (block_boundary) {
      histograms.push_back(cur_histogram);
      cur_histogram.Clear();
      ++cur_idx;
    }
  }
  std::vector<HistogramType> clustered_histograms;
  std::vector<int> histogram_symbols;
  // Block ids need to fit in one byte and there are two ids reserved for
  // indicating 'same as last' and 'last plus one'.
  static const int kMaxNumberOfBlockTypes = 254;
  ClusterHistograms(histograms, 1, histograms.size(),
                    kMaxNumberOfBlockTypes,
                    &clustered_histograms,
                    &histogram_symbols);
  for (int i = 0; i < length; ++i) {
    block_ids[i] = histogram_symbols[block_index[i]];
  }
}

void BuildBlockSplit(const std::vector<uint8_t>& block_ids, BlockSplit* split) {
  int cur_id = block_ids[0];
  int cur_length = 1;
  split->num_types_ = -1;
  for (int i = 1; i < block_ids.size(); ++i) {
    if (block_ids[i] != cur_id) {
      split->types_.push_back(cur_id);
      split->lengths_.push_back(cur_length);
      split->num_types_ = std::max(split->num_types_, cur_id);
      cur_id = block_ids[i];
      cur_length = 0;
    }
    ++cur_length;
  }
  split->types_.push_back(cur_id);
  split->lengths_.push_back(cur_length);
  split->num_types_ = std::max(split->num_types_, cur_id);
  ++split->num_types_;
}

template<typename HistogramType, typename DataType>
void SplitByteVector(const std::vector<DataType>& data,
                     const int literals_per_histogram,
                     const int max_histograms,
                     const int sampling_stride_length,
                     const double block_switch_cost,
                     BlockSplit* split) {
  if (data.empty()) {
    split->num_types_ = 0;
    return;
  } else if (data.size() < kMinLengthForBlockSplitting) {
    split->num_types_ = 1;
    split->types_.push_back(0);
    split->lengths_.push_back(data.size());
    return;
  }
  std::vector<HistogramType> histograms;
  // Find good entropy codes.
  InitialEntropyCodes(data.data(), data.size(),
                      literals_per_histogram,
                      max_histograms,
                      sampling_stride_length,
                      &histograms);
  RefineEntropyCodes(data.data(), data.size(),
                     sampling_stride_length,
                     &histograms);
  // Find a good path through literals with the good entropy codes.
  std::vector<uint8_t> block_ids(data.size());
  for (int i = 0; i < 10; ++i) {
    FindBlocks(data.data(), data.size(),
               block_switch_cost,
               histograms,
               &block_ids[0]);
    BuildBlockHistograms(data.data(), data.size(), &block_ids[0], &histograms);
  }
  ClusterBlocks<HistogramType>(data.data(), data.size(), &block_ids[0]);
  BuildBlockSplit(block_ids, split);
}

void SplitBlock(const std::vector<Command>& cmds,
                const uint8_t* data,
                BlockSplit* literal_split,
                BlockSplit* insert_and_copy_split,
                BlockSplit* dist_split) {
  // Create a continuous array of literals.
  std::vector<uint8_t> literals;
  CopyLiteralsToByteArray(cmds, data, &literals);

  // Compute prefix codes for commands.
  std::vector<uint16_t> insert_and_copy_codes;
  std::vector<uint8_t> distance_prefixes;
  CopyCommandsToByteArray(cmds,
                          &insert_and_copy_codes,
                          &distance_prefixes);

  SplitByteVector<HistogramLiteral>(
      literals,
      kSymbolsPerLiteralHistogram, kMaxLiteralHistograms,
      kLiteralStrideLength, kLiteralBlockSwitchCost,
      literal_split);
  SplitByteVector<HistogramCommand>(
      insert_and_copy_codes,
      kSymbolsPerCommandHistogram, kMaxCommandHistograms,
      kCommandStrideLength, kCommandBlockSwitchCost,
      insert_and_copy_split);
  SplitByteVector<HistogramDistance>(
      distance_prefixes,
      kSymbolsPerDistanceHistogram, kMaxCommandHistograms,
      kCommandStrideLength, kDistanceBlockSwitchCost,
      dist_split);
}

void SplitBlockByTotalLength(const std::vector<Command>& all_commands,
                             int input_size,
                             int target_length,
                             std::vector<std::vector<Command> >* blocks) {
  int num_blocks = input_size / target_length + 1;
  int length_limit = input_size / num_blocks + 1;
  int total_length = 0;
  std::vector<Command> cur_block;
  for (int i = 0; i < all_commands.size(); ++i) {
    const Command& cmd = all_commands[i];
    int cmd_length = cmd.insert_length_ + cmd.copy_length_;
    if (total_length > length_limit) {
      blocks->push_back(cur_block);
      cur_block.clear();
      total_length = 0;
    }
    cur_block.push_back(cmd);
    total_length += cmd_length;
  }
  blocks->push_back(cur_block);
}

}  // namespace brotli