10#define HistogramType FN(Histogram)
14typedef struct FN(BlockSplitter) {
16 size_t alphabet_size_;
18 size_t min_block_size_;
23 double split_threshold_;
28 size_t* histograms_size_;
35 size_t target_block_size_;
39 size_t curr_histogram_ix_;
41 size_t last_histogram_ix_[2];
43 double last_entropy_[2];
45 size_t merge_last_count_;
48static void FN(InitBlockSplitter)(
50 size_t min_block_size,
double split_threshold,
size_t num_symbols,
52 size_t max_num_blocks = num_symbols / min_block_size + 1;
55 size_t max_num_types =
57 self->alphabet_size_ = alphabet_size;
58 self->min_block_size_ = min_block_size;
59 self->split_threshold_ = split_threshold;
60 self->num_blocks_ = 0;
62 self->histograms_size_ = histograms_size;
63 self->target_block_size_ = min_block_size;
64 self->block_size_ = 0;
65 self->curr_histogram_ix_ = 0;
66 self->merge_last_count_ = 0;
68 split->types,
split->types_alloc_size, max_num_blocks);
70 split->lengths,
split->lengths_alloc_size, max_num_blocks);
72 self->split_->num_blocks = max_num_blocks;
74 *histograms_size = max_num_types;
76 self->histograms_ = *histograms;
79 FN(HistogramClear)(&self->histograms_[0]);
80 self->last_histogram_ix_[0] = self->last_histogram_ix_[1] = 0;
87static void FN(BlockSplitterFinishBlock)(
90 double* last_entropy = self->last_entropy_;
93 BROTLI_MAX(
size_t, self->block_size_, self->min_block_size_);
94 if (self->num_blocks_ == 0) {
96 split->lengths[0] = (uint32_t)self->block_size_;
99 BitsEntropy(histograms[0].data_, self->alphabet_size_);
100 last_entropy[1] = last_entropy[0];
103 ++self->curr_histogram_ix_;
104 if (self->curr_histogram_ix_ < *self->histograms_size_)
105 FN(HistogramClear)(&histograms[self->curr_histogram_ix_]);
106 self->block_size_ = 0;
107 }
else if (self->block_size_ > 0) {
108 double entropy = BitsEntropy(histograms[self->curr_histogram_ix_].data_,
109 self->alphabet_size_);
110 double combined_entropy[2];
113 for (j = 0; j < 2; ++j) {
114 size_t last_histogram_ix = self->last_histogram_ix_[j];
115 self->combined_histo[j] = histograms[self->curr_histogram_ix_];
116 FN(HistogramAddHistogram)(&self->combined_histo[j],
117 &histograms[last_histogram_ix]);
118 combined_entropy[j] = BitsEntropy(
119 &self->combined_histo[j].data_[0], self->alphabet_size_);
120 diff[j] = combined_entropy[j] - entropy - last_entropy[j];
124 diff[0] > self->split_threshold_ &&
125 diff[1] > self->split_threshold_) {
127 split->lengths[self->num_blocks_] = (uint32_t)self->block_size_;
128 split->types[self->num_blocks_] = (uint8_t)
split->num_types;
129 self->last_histogram_ix_[1] = self->last_histogram_ix_[0];
130 self->last_histogram_ix_[0] = (uint8_t)
split->num_types;
131 last_entropy[1] = last_entropy[0];
132 last_entropy[0] = entropy;
135 ++self->curr_histogram_ix_;
136 if (self->curr_histogram_ix_ < *self->histograms_size_)
137 FN(HistogramClear)(&histograms[self->curr_histogram_ix_]);
138 self->block_size_ = 0;
139 self->merge_last_count_ = 0;
140 self->target_block_size_ = self->min_block_size_;
141 }
else if (diff[1] < diff[0] - 20.0) {
143 split->lengths[self->num_blocks_] = (uint32_t)self->block_size_;
144 split->types[self->num_blocks_] =
split->types[self->num_blocks_ - 2];
145 BROTLI_SWAP(
size_t, self->last_histogram_ix_, 0, 1);
146 histograms[self->last_histogram_ix_[0]] = self->combined_histo[1];
147 last_entropy[1] = last_entropy[0];
148 last_entropy[0] = combined_entropy[1];
150 self->block_size_ = 0;
151 FN(HistogramClear)(&histograms[self->curr_histogram_ix_]);
152 self->merge_last_count_ = 0;
153 self->target_block_size_ = self->min_block_size_;
156 split->lengths[self->num_blocks_ - 1] += (uint32_t)self->block_size_;
157 histograms[self->last_histogram_ix_[0]] = self->combined_histo[0];
158 last_entropy[0] = combined_entropy[0];
159 if (
split->num_types == 1) {
160 last_entropy[1] = last_entropy[0];
162 self->block_size_ = 0;
163 FN(HistogramClear)(&histograms[self->curr_histogram_ix_]);
164 if (++self->merge_last_count_ > 1) {
165 self->target_block_size_ += self->min_block_size_;
170 *self->histograms_size_ =
split->num_types;
171 split->num_blocks = self->num_blocks_;
177static void FN(BlockSplitterAddSymbol)(
FN(BlockSplitter)* self,
size_t symbol) {
178 FN(HistogramAdd)(&self->histograms_[self->curr_histogram_ix_], symbol);
180 if (self->block_size_ == self->target_block_size_) {
#define BROTLI_MAX_NUMBER_OF_BLOCK_TYPES
Definition constants.h:23
#define FN(X)
Definition backward_references.c:51
#define HistogramType
Definition bit_cost_inc.h:10
#define BROTLI_ALLOC(M, T, N)
Definition memory.h:44
#define BROTLI_ENSURE_CAPACITY(M, T, A, C, R)
Definition memory.h:81
#define BROTLI_IS_OOM(M)
Definition memory.h:54
#define BROTLI_IS_NULL(A)
Definition memory.h:68
Buffer split(Buffer &buffer, ZSTD_outBuffer &outBuffer)
Definition Pzstd.cpp:248
str diff(bytes a, bytes b)
Definition run.py:91
Definition block_splitter.h:22
#define BROTLI_FALSE
Definition types.h:53
#define BROTLI_BOOL
Definition types.h:49