• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 /* Copyright 2014 Google Inc. All Rights Reserved.
2 
3    Distributed under MIT license.
4    See file LICENSE for detail or copy at https://opensource.org/licenses/MIT
5 */
6 
7 /* Brotli bit stream functions to support the low level format. There are no
8    compression algorithms here, just the right ordering of bits to match the
9    specs. */
10 
11 #include "brotli_bit_stream.h"
12 
13 #include <string.h>  /* memcpy, memset */
14 
15 #include <brotli/types.h>
16 
17 #include "../common/constants.h"
18 #include "../common/context.h"
19 #include "../common/platform.h"
20 #include "entropy_encode.h"
21 #include "entropy_encode_static.h"
22 #include "fast_log.h"
23 #include "histogram.h"
24 #include "memory.h"
25 #include "write_bits.h"
26 
27 #if defined(__cplusplus) || defined(c_plusplus)
28 extern "C" {
29 #endif
30 
31 #define MAX_HUFFMAN_TREE_SIZE (2 * BROTLI_NUM_COMMAND_SYMBOLS + 1)
32 /* The maximum size of Huffman dictionary for distances assuming that
33    NPOSTFIX = 0 and NDIRECT = 0. */
34 #define MAX_SIMPLE_DISTANCE_ALPHABET_SIZE \
35   BROTLI_DISTANCE_ALPHABET_SIZE(0, 0, BROTLI_LARGE_MAX_DISTANCE_BITS)
36 /* MAX_SIMPLE_DISTANCE_ALPHABET_SIZE == 140 */
37 
BlockLengthPrefixCode(uint32_t len)38 static BROTLI_INLINE uint32_t BlockLengthPrefixCode(uint32_t len) {
39   uint32_t code = (len >= 177) ? (len >= 753 ? 20 : 14) : (len >= 41 ? 7 : 0);
40   while (code < (BROTLI_NUM_BLOCK_LEN_SYMBOLS - 1) &&
41       len >= _kBrotliPrefixCodeRanges[code + 1].offset) ++code;
42   return code;
43 }
44 
GetBlockLengthPrefixCode(uint32_t len,size_t * code,uint32_t * n_extra,uint32_t * extra)45 static BROTLI_INLINE void GetBlockLengthPrefixCode(uint32_t len, size_t* code,
46     uint32_t* n_extra, uint32_t* extra) {
47   *code = BlockLengthPrefixCode(len);
48   *n_extra = _kBrotliPrefixCodeRanges[*code].nbits;
49   *extra = len - _kBrotliPrefixCodeRanges[*code].offset;
50 }
51 
52 typedef struct BlockTypeCodeCalculator {
53   size_t last_type;
54   size_t second_last_type;
55 } BlockTypeCodeCalculator;
56 
InitBlockTypeCodeCalculator(BlockTypeCodeCalculator * self)57 static void InitBlockTypeCodeCalculator(BlockTypeCodeCalculator* self) {
58   self->last_type = 1;
59   self->second_last_type = 0;
60 }
61 
NextBlockTypeCode(BlockTypeCodeCalculator * calculator,uint8_t type)62 static BROTLI_INLINE size_t NextBlockTypeCode(
63     BlockTypeCodeCalculator* calculator, uint8_t type) {
64   size_t type_code = (type == calculator->last_type + 1) ? 1u :
65       (type == calculator->second_last_type) ? 0u : type + 2u;
66   calculator->second_last_type = calculator->last_type;
67   calculator->last_type = type;
68   return type_code;
69 }
70 
71 /* |nibblesbits| represents the 2 bits to encode MNIBBLES (0-3)
72    REQUIRES: length > 0
73    REQUIRES: length <= (1 << 24) */
BrotliEncodeMlen(size_t length,uint64_t * bits,size_t * numbits,uint64_t * nibblesbits)74 static void BrotliEncodeMlen(size_t length, uint64_t* bits,
75                              size_t* numbits, uint64_t* nibblesbits) {
76   size_t lg = (length == 1) ? 1 : Log2FloorNonZero((uint32_t)(length - 1)) + 1;
77   size_t mnibbles = (lg < 16 ? 16 : (lg + 3)) / 4;
78   BROTLI_DCHECK(length > 0);
79   BROTLI_DCHECK(length <= (1 << 24));
80   BROTLI_DCHECK(lg <= 24);
81   *nibblesbits = mnibbles - 4;
82   *numbits = mnibbles * 4;
83   *bits = length - 1;
84 }
85 
StoreCommandExtra(const Command * cmd,size_t * storage_ix,uint8_t * storage)86 static BROTLI_INLINE void StoreCommandExtra(
87     const Command* cmd, size_t* storage_ix, uint8_t* storage) {
88   uint32_t copylen_code = CommandCopyLenCode(cmd);
89   uint16_t inscode = GetInsertLengthCode(cmd->insert_len_);
90   uint16_t copycode = GetCopyLengthCode(copylen_code);
91   uint32_t insnumextra = GetInsertExtra(inscode);
92   uint64_t insextraval = cmd->insert_len_ - GetInsertBase(inscode);
93   uint64_t copyextraval = copylen_code - GetCopyBase(copycode);
94   uint64_t bits = (copyextraval << insnumextra) | insextraval;
95   BrotliWriteBits(
96       insnumextra + GetCopyExtra(copycode), bits, storage_ix, storage);
97 }
98 
99 /* Data structure that stores almost everything that is needed to encode each
100    block switch command. */
101 typedef struct BlockSplitCode {
102   BlockTypeCodeCalculator type_code_calculator;
103   uint8_t type_depths[BROTLI_MAX_BLOCK_TYPE_SYMBOLS];
104   uint16_t type_bits[BROTLI_MAX_BLOCK_TYPE_SYMBOLS];
105   uint8_t length_depths[BROTLI_NUM_BLOCK_LEN_SYMBOLS];
106   uint16_t length_bits[BROTLI_NUM_BLOCK_LEN_SYMBOLS];
107 } BlockSplitCode;
108 
109 /* Stores a number between 0 and 255. */
StoreVarLenUint8(size_t n,size_t * storage_ix,uint8_t * storage)110 static void StoreVarLenUint8(size_t n, size_t* storage_ix, uint8_t* storage) {
111   if (n == 0) {
112     BrotliWriteBits(1, 0, storage_ix, storage);
113   } else {
114     size_t nbits = Log2FloorNonZero(n);
115     BrotliWriteBits(1, 1, storage_ix, storage);
116     BrotliWriteBits(3, nbits, storage_ix, storage);
117     BrotliWriteBits(nbits, n - ((size_t)1 << nbits), storage_ix, storage);
118   }
119 }
120 
121 /* Stores the compressed meta-block header.
122    REQUIRES: length > 0
123    REQUIRES: length <= (1 << 24) */
StoreCompressedMetaBlockHeader(BROTLI_BOOL is_final_block,size_t length,size_t * storage_ix,uint8_t * storage)124 static void StoreCompressedMetaBlockHeader(BROTLI_BOOL is_final_block,
125                                            size_t length,
126                                            size_t* storage_ix,
127                                            uint8_t* storage) {
128   uint64_t lenbits;
129   size_t nlenbits;
130   uint64_t nibblesbits;
131 
132   /* Write ISLAST bit. */
133   BrotliWriteBits(1, (uint64_t)is_final_block, storage_ix, storage);
134   /* Write ISEMPTY bit. */
135   if (is_final_block) {
136     BrotliWriteBits(1, 0, storage_ix, storage);
137   }
138 
139   BrotliEncodeMlen(length, &lenbits, &nlenbits, &nibblesbits);
140   BrotliWriteBits(2, nibblesbits, storage_ix, storage);
141   BrotliWriteBits(nlenbits, lenbits, storage_ix, storage);
142 
143   if (!is_final_block) {
144     /* Write ISUNCOMPRESSED bit. */
145     BrotliWriteBits(1, 0, storage_ix, storage);
146   }
147 }
148 
149 /* Stores the uncompressed meta-block header.
150    REQUIRES: length > 0
151    REQUIRES: length <= (1 << 24) */
BrotliStoreUncompressedMetaBlockHeader(size_t length,size_t * storage_ix,uint8_t * storage)152 static void BrotliStoreUncompressedMetaBlockHeader(size_t length,
153                                                    size_t* storage_ix,
154                                                    uint8_t* storage) {
155   uint64_t lenbits;
156   size_t nlenbits;
157   uint64_t nibblesbits;
158 
159   /* Write ISLAST bit.
160      Uncompressed block cannot be the last one, so set to 0. */
161   BrotliWriteBits(1, 0, storage_ix, storage);
162   BrotliEncodeMlen(length, &lenbits, &nlenbits, &nibblesbits);
163   BrotliWriteBits(2, nibblesbits, storage_ix, storage);
164   BrotliWriteBits(nlenbits, lenbits, storage_ix, storage);
165   /* Write ISUNCOMPRESSED bit. */
166   BrotliWriteBits(1, 1, storage_ix, storage);
167 }
168 
BrotliStoreHuffmanTreeOfHuffmanTreeToBitMask(const int num_codes,const uint8_t * code_length_bitdepth,size_t * storage_ix,uint8_t * storage)169 static void BrotliStoreHuffmanTreeOfHuffmanTreeToBitMask(
170     const int num_codes, const uint8_t* code_length_bitdepth,
171     size_t* storage_ix, uint8_t* storage) {
172   static const uint8_t kStorageOrder[BROTLI_CODE_LENGTH_CODES] = {
173     1, 2, 3, 4, 0, 5, 17, 6, 16, 7, 8, 9, 10, 11, 12, 13, 14, 15
174   };
175   /* The bit lengths of the Huffman code over the code length alphabet
176      are compressed with the following static Huffman code:
177        Symbol   Code
178        ------   ----
179        0          00
180        1        1110
181        2         110
182        3          01
183        4          10
184        5        1111 */
185   static const uint8_t kHuffmanBitLengthHuffmanCodeSymbols[6] = {
186      0, 7, 3, 2, 1, 15
187   };
188   static const uint8_t kHuffmanBitLengthHuffmanCodeBitLengths[6] = {
189     2, 4, 3, 2, 2, 4
190   };
191 
192   size_t skip_some = 0;  /* skips none. */
193 
194   /* Throw away trailing zeros: */
195   size_t codes_to_store = BROTLI_CODE_LENGTH_CODES;
196   if (num_codes > 1) {
197     for (; codes_to_store > 0; --codes_to_store) {
198       if (code_length_bitdepth[kStorageOrder[codes_to_store - 1]] != 0) {
199         break;
200       }
201     }
202   }
203   if (code_length_bitdepth[kStorageOrder[0]] == 0 &&
204       code_length_bitdepth[kStorageOrder[1]] == 0) {
205     skip_some = 2;  /* skips two. */
206     if (code_length_bitdepth[kStorageOrder[2]] == 0) {
207       skip_some = 3;  /* skips three. */
208     }
209   }
210   BrotliWriteBits(2, skip_some, storage_ix, storage);
211   {
212     size_t i;
213     for (i = skip_some; i < codes_to_store; ++i) {
214       size_t l = code_length_bitdepth[kStorageOrder[i]];
215       BrotliWriteBits(kHuffmanBitLengthHuffmanCodeBitLengths[l],
216           kHuffmanBitLengthHuffmanCodeSymbols[l], storage_ix, storage);
217     }
218   }
219 }
220 
BrotliStoreHuffmanTreeToBitMask(const size_t huffman_tree_size,const uint8_t * huffman_tree,const uint8_t * huffman_tree_extra_bits,const uint8_t * code_length_bitdepth,const uint16_t * code_length_bitdepth_symbols,size_t * BROTLI_RESTRICT storage_ix,uint8_t * BROTLI_RESTRICT storage)221 static void BrotliStoreHuffmanTreeToBitMask(
222     const size_t huffman_tree_size, const uint8_t* huffman_tree,
223     const uint8_t* huffman_tree_extra_bits, const uint8_t* code_length_bitdepth,
224     const uint16_t* code_length_bitdepth_symbols,
225     size_t* BROTLI_RESTRICT storage_ix, uint8_t* BROTLI_RESTRICT storage) {
226   size_t i;
227   for (i = 0; i < huffman_tree_size; ++i) {
228     size_t ix = huffman_tree[i];
229     BrotliWriteBits(code_length_bitdepth[ix], code_length_bitdepth_symbols[ix],
230                     storage_ix, storage);
231     /* Extra bits */
232     switch (ix) {
233       case BROTLI_REPEAT_PREVIOUS_CODE_LENGTH:
234         BrotliWriteBits(2, huffman_tree_extra_bits[i], storage_ix, storage);
235         break;
236       case BROTLI_REPEAT_ZERO_CODE_LENGTH:
237         BrotliWriteBits(3, huffman_tree_extra_bits[i], storage_ix, storage);
238         break;
239     }
240   }
241 }
242 
StoreSimpleHuffmanTree(const uint8_t * depths,size_t symbols[4],size_t num_symbols,size_t max_bits,size_t * storage_ix,uint8_t * storage)243 static void StoreSimpleHuffmanTree(const uint8_t* depths,
244                                    size_t symbols[4],
245                                    size_t num_symbols,
246                                    size_t max_bits,
247                                    size_t* storage_ix, uint8_t* storage) {
248   /* value of 1 indicates a simple Huffman code */
249   BrotliWriteBits(2, 1, storage_ix, storage);
250   BrotliWriteBits(2, num_symbols - 1, storage_ix, storage);  /* NSYM - 1 */
251 
252   {
253     /* Sort */
254     size_t i;
255     for (i = 0; i < num_symbols; i++) {
256       size_t j;
257       for (j = i + 1; j < num_symbols; j++) {
258         if (depths[symbols[j]] < depths[symbols[i]]) {
259           BROTLI_SWAP(size_t, symbols, j, i);
260         }
261       }
262     }
263   }
264 
265   if (num_symbols == 2) {
266     BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
267     BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
268   } else if (num_symbols == 3) {
269     BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
270     BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
271     BrotliWriteBits(max_bits, symbols[2], storage_ix, storage);
272   } else {
273     BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
274     BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
275     BrotliWriteBits(max_bits, symbols[2], storage_ix, storage);
276     BrotliWriteBits(max_bits, symbols[3], storage_ix, storage);
277     /* tree-select */
278     BrotliWriteBits(1, depths[symbols[0]] == 1 ? 1 : 0, storage_ix, storage);
279   }
280 }
281 
282 /* num = alphabet size
283    depths = symbol depths */
BrotliStoreHuffmanTree(const uint8_t * depths,size_t num,HuffmanTree * tree,size_t * storage_ix,uint8_t * storage)284 void BrotliStoreHuffmanTree(const uint8_t* depths, size_t num,
285                             HuffmanTree* tree,
286                             size_t* storage_ix, uint8_t* storage) {
287   /* Write the Huffman tree into the brotli-representation.
288      The command alphabet is the largest, so this allocation will fit all
289      alphabets. */
290   /* TODO(eustas): fix me */
291   uint8_t huffman_tree[BROTLI_NUM_COMMAND_SYMBOLS];
292   uint8_t huffman_tree_extra_bits[BROTLI_NUM_COMMAND_SYMBOLS];
293   size_t huffman_tree_size = 0;
294   uint8_t code_length_bitdepth[BROTLI_CODE_LENGTH_CODES] = { 0 };
295   uint16_t code_length_bitdepth_symbols[BROTLI_CODE_LENGTH_CODES];
296   uint32_t huffman_tree_histogram[BROTLI_CODE_LENGTH_CODES] = { 0 };
297   size_t i;
298   int num_codes = 0;
299   size_t code = 0;
300 
301   BROTLI_DCHECK(num <= BROTLI_NUM_COMMAND_SYMBOLS);
302 
303   BrotliWriteHuffmanTree(depths, num, &huffman_tree_size, huffman_tree,
304                          huffman_tree_extra_bits);
305 
306   /* Calculate the statistics of the Huffman tree in brotli-representation. */
307   for (i = 0; i < huffman_tree_size; ++i) {
308     ++huffman_tree_histogram[huffman_tree[i]];
309   }
310 
311   for (i = 0; i < BROTLI_CODE_LENGTH_CODES; ++i) {
312     if (huffman_tree_histogram[i]) {
313       if (num_codes == 0) {
314         code = i;
315         num_codes = 1;
316       } else if (num_codes == 1) {
317         num_codes = 2;
318         break;
319       }
320     }
321   }
322 
323   /* Calculate another Huffman tree to use for compressing both the
324      earlier Huffman tree with. */
325   BrotliCreateHuffmanTree(huffman_tree_histogram, BROTLI_CODE_LENGTH_CODES,
326                           5, tree, code_length_bitdepth);
327   BrotliConvertBitDepthsToSymbols(code_length_bitdepth,
328                                   BROTLI_CODE_LENGTH_CODES,
329                                   code_length_bitdepth_symbols);
330 
331   /* Now, we have all the data, let's start storing it */
332   BrotliStoreHuffmanTreeOfHuffmanTreeToBitMask(num_codes, code_length_bitdepth,
333                                                storage_ix, storage);
334 
335   if (num_codes == 1) {
336     code_length_bitdepth[code] = 0;
337   }
338 
339   /* Store the real Huffman tree now. */
340   BrotliStoreHuffmanTreeToBitMask(huffman_tree_size,
341                                   huffman_tree,
342                                   huffman_tree_extra_bits,
343                                   code_length_bitdepth,
344                                   code_length_bitdepth_symbols,
345                                   storage_ix, storage);
346 }
347 
348 /* Builds a Huffman tree from histogram[0:length] into depth[0:length] and
349    bits[0:length] and stores the encoded tree to the bit stream. */
BuildAndStoreHuffmanTree(const uint32_t * histogram,const size_t histogram_length,const size_t alphabet_size,HuffmanTree * tree,uint8_t * depth,uint16_t * bits,size_t * storage_ix,uint8_t * storage)350 static void BuildAndStoreHuffmanTree(const uint32_t* histogram,
351                                      const size_t histogram_length,
352                                      const size_t alphabet_size,
353                                      HuffmanTree* tree,
354                                      uint8_t* depth,
355                                      uint16_t* bits,
356                                      size_t* storage_ix,
357                                      uint8_t* storage) {
358   size_t count = 0;
359   size_t s4[4] = { 0 };
360   size_t i;
361   size_t max_bits = 0;
362   for (i = 0; i < histogram_length; i++) {
363     if (histogram[i]) {
364       if (count < 4) {
365         s4[count] = i;
366       } else if (count > 4) {
367         break;
368       }
369       count++;
370     }
371   }
372 
373   {
374     size_t max_bits_counter = alphabet_size - 1;
375     while (max_bits_counter) {
376       max_bits_counter >>= 1;
377       ++max_bits;
378     }
379   }
380 
381   if (count <= 1) {
382     BrotliWriteBits(4, 1, storage_ix, storage);
383     BrotliWriteBits(max_bits, s4[0], storage_ix, storage);
384     depth[s4[0]] = 0;
385     bits[s4[0]] = 0;
386     return;
387   }
388 
389   memset(depth, 0, histogram_length * sizeof(depth[0]));
390   BrotliCreateHuffmanTree(histogram, histogram_length, 15, tree, depth);
391   BrotliConvertBitDepthsToSymbols(depth, histogram_length, bits);
392 
393   if (count <= 4) {
394     StoreSimpleHuffmanTree(depth, s4, count, max_bits, storage_ix, storage);
395   } else {
396     BrotliStoreHuffmanTree(depth, histogram_length, tree, storage_ix, storage);
397   }
398 }
399 
SortHuffmanTree(const HuffmanTree * v0,const HuffmanTree * v1)400 static BROTLI_INLINE BROTLI_BOOL SortHuffmanTree(
401     const HuffmanTree* v0, const HuffmanTree* v1) {
402   return TO_BROTLI_BOOL(v0->total_count_ < v1->total_count_);
403 }
404 
BrotliBuildAndStoreHuffmanTreeFast(HuffmanTree * tree,const uint32_t * histogram,const size_t histogram_total,const size_t max_bits,uint8_t * depth,uint16_t * bits,size_t * storage_ix,uint8_t * storage)405 void BrotliBuildAndStoreHuffmanTreeFast(HuffmanTree* tree,
406                                         const uint32_t* histogram,
407                                         const size_t histogram_total,
408                                         const size_t max_bits,
409                                         uint8_t* depth, uint16_t* bits,
410                                         size_t* storage_ix,
411                                         uint8_t* storage) {
412   size_t count = 0;
413   size_t symbols[4] = { 0 };
414   size_t length = 0;
415   size_t total = histogram_total;
416   while (total != 0) {
417     if (histogram[length]) {
418       if (count < 4) {
419         symbols[count] = length;
420       }
421       ++count;
422       total -= histogram[length];
423     }
424     ++length;
425   }
426 
427   if (count <= 1) {
428     BrotliWriteBits(4, 1, storage_ix, storage);
429     BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
430     depth[symbols[0]] = 0;
431     bits[symbols[0]] = 0;
432     return;
433   }
434 
435   memset(depth, 0, length * sizeof(depth[0]));
436   {
437     uint32_t count_limit;
438     for (count_limit = 1; ; count_limit *= 2) {
439       HuffmanTree* node = tree;
440       size_t l;
441       for (l = length; l != 0;) {
442         --l;
443         if (histogram[l]) {
444           if (BROTLI_PREDICT_TRUE(histogram[l] >= count_limit)) {
445             InitHuffmanTree(node, histogram[l], -1, (int16_t)l);
446           } else {
447             InitHuffmanTree(node, count_limit, -1, (int16_t)l);
448           }
449           ++node;
450         }
451       }
452       {
453         const int n = (int)(node - tree);
454         HuffmanTree sentinel;
455         int i = 0;      /* Points to the next leaf node. */
456         int j = n + 1;  /* Points to the next non-leaf node. */
457         int k;
458 
459         SortHuffmanTreeItems(tree, (size_t)n, SortHuffmanTree);
460         /* The nodes are:
461            [0, n): the sorted leaf nodes that we start with.
462            [n]: we add a sentinel here.
463            [n + 1, 2n): new parent nodes are added here, starting from
464                         (n+1). These are naturally in ascending order.
465            [2n]: we add a sentinel at the end as well.
466            There will be (2n+1) elements at the end. */
467         InitHuffmanTree(&sentinel, BROTLI_UINT32_MAX, -1, -1);
468         *node++ = sentinel;
469         *node++ = sentinel;
470 
471         for (k = n - 1; k > 0; --k) {
472           int left, right;
473           if (tree[i].total_count_ <= tree[j].total_count_) {
474             left = i;
475             ++i;
476           } else {
477             left = j;
478             ++j;
479           }
480           if (tree[i].total_count_ <= tree[j].total_count_) {
481             right = i;
482             ++i;
483           } else {
484             right = j;
485             ++j;
486           }
487           /* The sentinel node becomes the parent node. */
488           node[-1].total_count_ =
489               tree[left].total_count_ + tree[right].total_count_;
490           node[-1].index_left_ = (int16_t)left;
491           node[-1].index_right_or_value_ = (int16_t)right;
492           /* Add back the last sentinel node. */
493           *node++ = sentinel;
494         }
495         if (BrotliSetDepth(2 * n - 1, tree, depth, 14)) {
496           /* We need to pack the Huffman tree in 14 bits. If this was not
497              successful, add fake entities to the lowest values and retry. */
498           break;
499         }
500       }
501     }
502   }
503   BrotliConvertBitDepthsToSymbols(depth, length, bits);
504   if (count <= 4) {
505     size_t i;
506     /* value of 1 indicates a simple Huffman code */
507     BrotliWriteBits(2, 1, storage_ix, storage);
508     BrotliWriteBits(2, count - 1, storage_ix, storage);  /* NSYM - 1 */
509 
510     /* Sort */
511     for (i = 0; i < count; i++) {
512       size_t j;
513       for (j = i + 1; j < count; j++) {
514         if (depth[symbols[j]] < depth[symbols[i]]) {
515           BROTLI_SWAP(size_t, symbols, j, i);
516         }
517       }
518     }
519 
520     if (count == 2) {
521       BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
522       BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
523     } else if (count == 3) {
524       BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
525       BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
526       BrotliWriteBits(max_bits, symbols[2], storage_ix, storage);
527     } else {
528       BrotliWriteBits(max_bits, symbols[0], storage_ix, storage);
529       BrotliWriteBits(max_bits, symbols[1], storage_ix, storage);
530       BrotliWriteBits(max_bits, symbols[2], storage_ix, storage);
531       BrotliWriteBits(max_bits, symbols[3], storage_ix, storage);
532       /* tree-select */
533       BrotliWriteBits(1, depth[symbols[0]] == 1 ? 1 : 0, storage_ix, storage);
534     }
535   } else {
536     uint8_t previous_value = 8;
537     size_t i;
538     /* Complex Huffman Tree */
539     StoreStaticCodeLengthCode(storage_ix, storage);
540 
541     /* Actual RLE coding. */
542     for (i = 0; i < length;) {
543       const uint8_t value = depth[i];
544       size_t reps = 1;
545       size_t k;
546       for (k = i + 1; k < length && depth[k] == value; ++k) {
547         ++reps;
548       }
549       i += reps;
550       if (value == 0) {
551         BrotliWriteBits(kZeroRepsDepth[reps], kZeroRepsBits[reps],
552                         storage_ix, storage);
553       } else {
554         if (previous_value != value) {
555           BrotliWriteBits(kCodeLengthDepth[value], kCodeLengthBits[value],
556                           storage_ix, storage);
557           --reps;
558         }
559         if (reps < 3) {
560           while (reps != 0) {
561             reps--;
562             BrotliWriteBits(kCodeLengthDepth[value], kCodeLengthBits[value],
563                             storage_ix, storage);
564           }
565         } else {
566           reps -= 3;
567           BrotliWriteBits(kNonZeroRepsDepth[reps], kNonZeroRepsBits[reps],
568                           storage_ix, storage);
569         }
570         previous_value = value;
571       }
572     }
573   }
574 }
575 
IndexOf(const uint8_t * v,size_t v_size,uint8_t value)576 static size_t IndexOf(const uint8_t* v, size_t v_size, uint8_t value) {
577   size_t i = 0;
578   for (; i < v_size; ++i) {
579     if (v[i] == value) return i;
580   }
581   return i;
582 }
583 
MoveToFront(uint8_t * v,size_t index)584 static void MoveToFront(uint8_t* v, size_t index) {
585   uint8_t value = v[index];
586   size_t i;
587   for (i = index; i != 0; --i) {
588     v[i] = v[i - 1];
589   }
590   v[0] = value;
591 }
592 
MoveToFrontTransform(const uint32_t * BROTLI_RESTRICT v_in,const size_t v_size,uint32_t * v_out)593 static void MoveToFrontTransform(const uint32_t* BROTLI_RESTRICT v_in,
594                                  const size_t v_size,
595                                  uint32_t* v_out) {
596   size_t i;
597   uint8_t mtf[256];
598   uint32_t max_value;
599   if (v_size == 0) {
600     return;
601   }
602   max_value = v_in[0];
603   for (i = 1; i < v_size; ++i) {
604     if (v_in[i] > max_value) max_value = v_in[i];
605   }
606   BROTLI_DCHECK(max_value < 256u);
607   for (i = 0; i <= max_value; ++i) {
608     mtf[i] = (uint8_t)i;
609   }
610   {
611     size_t mtf_size = max_value + 1;
612     for (i = 0; i < v_size; ++i) {
613       size_t index = IndexOf(mtf, mtf_size, (uint8_t)v_in[i]);
614       BROTLI_DCHECK(index < mtf_size);
615       v_out[i] = (uint32_t)index;
616       MoveToFront(mtf, index);
617     }
618   }
619 }
620 
621 /* Finds runs of zeros in v[0..in_size) and replaces them with a prefix code of
622    the run length plus extra bits (lower 9 bits is the prefix code and the rest
623    are the extra bits). Non-zero values in v[] are shifted by
624    *max_length_prefix. Will not create prefix codes bigger than the initial
625    value of *max_run_length_prefix. The prefix code of run length L is simply
626    Log2Floor(L) and the number of extra bits is the same as the prefix code. */
RunLengthCodeZeros(const size_t in_size,uint32_t * BROTLI_RESTRICT v,size_t * BROTLI_RESTRICT out_size,uint32_t * BROTLI_RESTRICT max_run_length_prefix)627 static void RunLengthCodeZeros(const size_t in_size,
628     uint32_t* BROTLI_RESTRICT v, size_t* BROTLI_RESTRICT out_size,
629     uint32_t* BROTLI_RESTRICT max_run_length_prefix) {
630   uint32_t max_reps = 0;
631   size_t i;
632   uint32_t max_prefix;
633   for (i = 0; i < in_size;) {
634     uint32_t reps = 0;
635     for (; i < in_size && v[i] != 0; ++i) ;
636     for (; i < in_size && v[i] == 0; ++i) {
637       ++reps;
638     }
639     max_reps = BROTLI_MAX(uint32_t, reps, max_reps);
640   }
641   max_prefix = max_reps > 0 ? Log2FloorNonZero(max_reps) : 0;
642   max_prefix = BROTLI_MIN(uint32_t, max_prefix, *max_run_length_prefix);
643   *max_run_length_prefix = max_prefix;
644   *out_size = 0;
645   for (i = 0; i < in_size;) {
646     BROTLI_DCHECK(*out_size <= i);
647     if (v[i] != 0) {
648       v[*out_size] = v[i] + *max_run_length_prefix;
649       ++i;
650       ++(*out_size);
651     } else {
652       uint32_t reps = 1;
653       size_t k;
654       for (k = i + 1; k < in_size && v[k] == 0; ++k) {
655         ++reps;
656       }
657       i += reps;
658       while (reps != 0) {
659         if (reps < (2u << max_prefix)) {
660           uint32_t run_length_prefix = Log2FloorNonZero(reps);
661           const uint32_t extra_bits = reps - (1u << run_length_prefix);
662           v[*out_size] = run_length_prefix + (extra_bits << 9);
663           ++(*out_size);
664           break;
665         } else {
666           const uint32_t extra_bits = (1u << max_prefix) - 1u;
667           v[*out_size] = max_prefix + (extra_bits << 9);
668           reps -= (2u << max_prefix) - 1u;
669           ++(*out_size);
670         }
671       }
672     }
673   }
674 }
675 
676 #define SYMBOL_BITS 9
677 
678 typedef struct EncodeContextMapArena {
679   uint32_t histogram[BROTLI_MAX_CONTEXT_MAP_SYMBOLS];
680   uint8_t depths[BROTLI_MAX_CONTEXT_MAP_SYMBOLS];
681   uint16_t bits[BROTLI_MAX_CONTEXT_MAP_SYMBOLS];
682 } EncodeContextMapArena;
683 
EncodeContextMap(MemoryManager * m,EncodeContextMapArena * arena,const uint32_t * context_map,size_t context_map_size,size_t num_clusters,HuffmanTree * tree,size_t * storage_ix,uint8_t * storage)684 static void EncodeContextMap(MemoryManager* m,
685                              EncodeContextMapArena* arena,
686                              const uint32_t* context_map,
687                              size_t context_map_size,
688                              size_t num_clusters,
689                              HuffmanTree* tree,
690                              size_t* storage_ix, uint8_t* storage) {
691   size_t i;
692   uint32_t* rle_symbols;
693   uint32_t max_run_length_prefix = 6;
694   size_t num_rle_symbols = 0;
695   uint32_t* BROTLI_RESTRICT const histogram = arena->histogram;
696   static const uint32_t kSymbolMask = (1u << SYMBOL_BITS) - 1u;
697   uint8_t* BROTLI_RESTRICT const depths = arena->depths;
698   uint16_t* BROTLI_RESTRICT const bits = arena->bits;
699 
700   StoreVarLenUint8(num_clusters - 1, storage_ix, storage);
701 
702   if (num_clusters == 1) {
703     return;
704   }
705 
706   rle_symbols = BROTLI_ALLOC(m, uint32_t, context_map_size);
707   if (BROTLI_IS_OOM(m) || BROTLI_IS_NULL(rle_symbols)) return;
708   MoveToFrontTransform(context_map, context_map_size, rle_symbols);
709   RunLengthCodeZeros(context_map_size, rle_symbols,
710                      &num_rle_symbols, &max_run_length_prefix);
711   memset(histogram, 0, sizeof(arena->histogram));
712   for (i = 0; i < num_rle_symbols; ++i) {
713     ++histogram[rle_symbols[i] & kSymbolMask];
714   }
715   {
716     BROTLI_BOOL use_rle = TO_BROTLI_BOOL(max_run_length_prefix > 0);
717     BrotliWriteBits(1, (uint64_t)use_rle, storage_ix, storage);
718     if (use_rle) {
719       BrotliWriteBits(4, max_run_length_prefix - 1, storage_ix, storage);
720     }
721   }
722   BuildAndStoreHuffmanTree(histogram, num_clusters + max_run_length_prefix,
723                            num_clusters + max_run_length_prefix,
724                            tree, depths, bits, storage_ix, storage);
725   for (i = 0; i < num_rle_symbols; ++i) {
726     const uint32_t rle_symbol = rle_symbols[i] & kSymbolMask;
727     const uint32_t extra_bits_val = rle_symbols[i] >> SYMBOL_BITS;
728     BrotliWriteBits(depths[rle_symbol], bits[rle_symbol], storage_ix, storage);
729     if (rle_symbol > 0 && rle_symbol <= max_run_length_prefix) {
730       BrotliWriteBits(rle_symbol, extra_bits_val, storage_ix, storage);
731     }
732   }
733   BrotliWriteBits(1, 1, storage_ix, storage);  /* use move-to-front */
734   BROTLI_FREE(m, rle_symbols);
735 }
736 
737 /* Stores the block switch command with index block_ix to the bit stream. */
StoreBlockSwitch(BlockSplitCode * code,const uint32_t block_len,const uint8_t block_type,BROTLI_BOOL is_first_block,size_t * storage_ix,uint8_t * storage)738 static BROTLI_INLINE void StoreBlockSwitch(BlockSplitCode* code,
739                                            const uint32_t block_len,
740                                            const uint8_t block_type,
741                                            BROTLI_BOOL is_first_block,
742                                            size_t* storage_ix,
743                                            uint8_t* storage) {
744   size_t typecode = NextBlockTypeCode(&code->type_code_calculator, block_type);
745   size_t lencode;
746   uint32_t len_nextra;
747   uint32_t len_extra;
748   if (!is_first_block) {
749     BrotliWriteBits(code->type_depths[typecode], code->type_bits[typecode],
750                     storage_ix, storage);
751   }
752   GetBlockLengthPrefixCode(block_len, &lencode, &len_nextra, &len_extra);
753 
754   BrotliWriteBits(code->length_depths[lencode], code->length_bits[lencode],
755                   storage_ix, storage);
756   BrotliWriteBits(len_nextra, len_extra, storage_ix, storage);
757 }
758 
759 /* Builds a BlockSplitCode data structure from the block split given by the
760    vector of block types and block lengths and stores it to the bit stream. */
BuildAndStoreBlockSplitCode(const uint8_t * types,const uint32_t * lengths,const size_t num_blocks,const size_t num_types,HuffmanTree * tree,BlockSplitCode * code,size_t * storage_ix,uint8_t * storage)761 static void BuildAndStoreBlockSplitCode(const uint8_t* types,
762                                         const uint32_t* lengths,
763                                         const size_t num_blocks,
764                                         const size_t num_types,
765                                         HuffmanTree* tree,
766                                         BlockSplitCode* code,
767                                         size_t* storage_ix,
768                                         uint8_t* storage) {
769   uint32_t type_histo[BROTLI_MAX_BLOCK_TYPE_SYMBOLS];
770   uint32_t length_histo[BROTLI_NUM_BLOCK_LEN_SYMBOLS];
771   size_t i;
772   BlockTypeCodeCalculator type_code_calculator;
773   memset(type_histo, 0, (num_types + 2) * sizeof(type_histo[0]));
774   memset(length_histo, 0, sizeof(length_histo));
775   InitBlockTypeCodeCalculator(&type_code_calculator);
776   for (i = 0; i < num_blocks; ++i) {
777     size_t type_code = NextBlockTypeCode(&type_code_calculator, types[i]);
778     if (i != 0) ++type_histo[type_code];
779     ++length_histo[BlockLengthPrefixCode(lengths[i])];
780   }
781   StoreVarLenUint8(num_types - 1, storage_ix, storage);
782   if (num_types > 1) {  /* TODO(eustas): else? could StoreBlockSwitch occur? */
783     BuildAndStoreHuffmanTree(&type_histo[0], num_types + 2, num_types + 2, tree,
784                              &code->type_depths[0], &code->type_bits[0],
785                              storage_ix, storage);
786     BuildAndStoreHuffmanTree(&length_histo[0], BROTLI_NUM_BLOCK_LEN_SYMBOLS,
787                              BROTLI_NUM_BLOCK_LEN_SYMBOLS,
788                              tree, &code->length_depths[0],
789                              &code->length_bits[0], storage_ix, storage);
790     StoreBlockSwitch(code, lengths[0], types[0], 1, storage_ix, storage);
791   }
792 }
793 
794 /* Stores a context map where the histogram type is always the block type. */
StoreTrivialContextMap(EncodeContextMapArena * arena,size_t num_types,size_t context_bits,HuffmanTree * tree,size_t * storage_ix,uint8_t * storage)795 static void StoreTrivialContextMap(EncodeContextMapArena* arena,
796                                    size_t num_types,
797                                    size_t context_bits,
798                                    HuffmanTree* tree,
799                                    size_t* storage_ix,
800                                    uint8_t* storage) {
801   StoreVarLenUint8(num_types - 1, storage_ix, storage);
802   if (num_types > 1) {
803     size_t repeat_code = context_bits - 1u;
804     size_t repeat_bits = (1u << repeat_code) - 1u;
805     size_t alphabet_size = num_types + repeat_code;
806     uint32_t* BROTLI_RESTRICT const histogram = arena->histogram;
807     uint8_t* BROTLI_RESTRICT const depths = arena->depths;
808     uint16_t* BROTLI_RESTRICT const bits = arena->bits;
809     size_t i;
810     memset(histogram, 0, alphabet_size * sizeof(histogram[0]));
811     /* Write RLEMAX. */
812     BrotliWriteBits(1, 1, storage_ix, storage);
813     BrotliWriteBits(4, repeat_code - 1, storage_ix, storage);
814     histogram[repeat_code] = (uint32_t)num_types;
815     histogram[0] = 1;
816     for (i = context_bits; i < alphabet_size; ++i) {
817       histogram[i] = 1;
818     }
819     BuildAndStoreHuffmanTree(histogram, alphabet_size, alphabet_size,
820                              tree, depths, bits, storage_ix, storage);
821     for (i = 0; i < num_types; ++i) {
822       size_t code = (i == 0 ? 0 : i + context_bits - 1);
823       BrotliWriteBits(depths[code], bits[code], storage_ix, storage);
824       BrotliWriteBits(
825           depths[repeat_code], bits[repeat_code], storage_ix, storage);
826       BrotliWriteBits(repeat_code, repeat_bits, storage_ix, storage);
827     }
828     /* Write IMTF (inverse-move-to-front) bit. */
829     BrotliWriteBits(1, 1, storage_ix, storage);
830   }
831 }
832 
833 /* Manages the encoding of one block category (literal, command or distance). */
834 typedef struct BlockEncoder {
835   size_t histogram_length_;
836   size_t num_block_types_;
837   const uint8_t* block_types_;  /* Not owned. */
838   const uint32_t* block_lengths_;  /* Not owned. */
839   size_t num_blocks_;
840   BlockSplitCode block_split_code_;
841   size_t block_ix_;
842   size_t block_len_;
843   size_t entropy_ix_;
844   uint8_t* depths_;
845   uint16_t* bits_;
846 } BlockEncoder;
847 
InitBlockEncoder(BlockEncoder * self,size_t histogram_length,size_t num_block_types,const uint8_t * block_types,const uint32_t * block_lengths,const size_t num_blocks)848 static void InitBlockEncoder(BlockEncoder* self, size_t histogram_length,
849     size_t num_block_types, const uint8_t* block_types,
850     const uint32_t* block_lengths, const size_t num_blocks) {
851   self->histogram_length_ = histogram_length;
852   self->num_block_types_ = num_block_types;
853   self->block_types_ = block_types;
854   self->block_lengths_ = block_lengths;
855   self->num_blocks_ = num_blocks;
856   InitBlockTypeCodeCalculator(&self->block_split_code_.type_code_calculator);
857   self->block_ix_ = 0;
858   self->block_len_ = num_blocks == 0 ? 0 : block_lengths[0];
859   self->entropy_ix_ = 0;
860   self->depths_ = 0;
861   self->bits_ = 0;
862 }
863 
CleanupBlockEncoder(MemoryManager * m,BlockEncoder * self)864 static void CleanupBlockEncoder(MemoryManager* m, BlockEncoder* self) {
865   BROTLI_FREE(m, self->depths_);
866   BROTLI_FREE(m, self->bits_);
867 }
868 
869 /* Creates entropy codes of block lengths and block types and stores them
870    to the bit stream. */
BuildAndStoreBlockSwitchEntropyCodes(BlockEncoder * self,HuffmanTree * tree,size_t * storage_ix,uint8_t * storage)871 static void BuildAndStoreBlockSwitchEntropyCodes(BlockEncoder* self,
872     HuffmanTree* tree, size_t* storage_ix, uint8_t* storage) {
873   BuildAndStoreBlockSplitCode(self->block_types_, self->block_lengths_,
874       self->num_blocks_, self->num_block_types_, tree, &self->block_split_code_,
875       storage_ix, storage);
876 }
877 
878 /* Stores the next symbol with the entropy code of the current block type.
879    Updates the block type and block length at block boundaries. */
StoreSymbol(BlockEncoder * self,size_t symbol,size_t * storage_ix,uint8_t * storage)880 static void StoreSymbol(BlockEncoder* self, size_t symbol, size_t* storage_ix,
881     uint8_t* storage) {
882   if (self->block_len_ == 0) {
883     size_t block_ix = ++self->block_ix_;
884     uint32_t block_len = self->block_lengths_[block_ix];
885     uint8_t block_type = self->block_types_[block_ix];
886     self->block_len_ = block_len;
887     self->entropy_ix_ = block_type * self->histogram_length_;
888     StoreBlockSwitch(&self->block_split_code_, block_len, block_type, 0,
889         storage_ix, storage);
890   }
891   --self->block_len_;
892   {
893     size_t ix = self->entropy_ix_ + symbol;
894     BrotliWriteBits(self->depths_[ix], self->bits_[ix], storage_ix, storage);
895   }
896 }
897 
898 /* Stores the next symbol with the entropy code of the current block type and
899    context value.
900    Updates the block type and block length at block boundaries. */
StoreSymbolWithContext(BlockEncoder * self,size_t symbol,size_t context,const uint32_t * context_map,size_t * storage_ix,uint8_t * storage,const size_t context_bits)901 static void StoreSymbolWithContext(BlockEncoder* self, size_t symbol,
902     size_t context, const uint32_t* context_map, size_t* storage_ix,
903     uint8_t* storage, const size_t context_bits) {
904   if (self->block_len_ == 0) {
905     size_t block_ix = ++self->block_ix_;
906     uint32_t block_len = self->block_lengths_[block_ix];
907     uint8_t block_type = self->block_types_[block_ix];
908     self->block_len_ = block_len;
909     self->entropy_ix_ = (size_t)block_type << context_bits;
910     StoreBlockSwitch(&self->block_split_code_, block_len, block_type, 0,
911         storage_ix, storage);
912   }
913   --self->block_len_;
914   {
915     size_t histo_ix = context_map[self->entropy_ix_ + context];
916     size_t ix = histo_ix * self->histogram_length_ + symbol;
917     BrotliWriteBits(self->depths_[ix], self->bits_[ix], storage_ix, storage);
918   }
919 }
920 
921 #define FN(X) X ## Literal
922 /* NOLINTNEXTLINE(build/include) */
923 #include "block_encoder_inc.h"
924 #undef FN
925 
926 #define FN(X) X ## Command
927 /* NOLINTNEXTLINE(build/include) */
928 #include "block_encoder_inc.h"
929 #undef FN
930 
931 #define FN(X) X ## Distance
932 /* NOLINTNEXTLINE(build/include) */
933 #include "block_encoder_inc.h"
934 #undef FN
935 
JumpToByteBoundary(size_t * storage_ix,uint8_t * storage)936 static void JumpToByteBoundary(size_t* storage_ix, uint8_t* storage) {
937   *storage_ix = (*storage_ix + 7u) & ~7u;
938   storage[*storage_ix >> 3] = 0;
939 }
940 
941 typedef struct StoreMetablockArena {
942   BlockEncoder literal_enc;
943   BlockEncoder command_enc;
944   BlockEncoder distance_enc;
945   EncodeContextMapArena context_map_arena;
946 } StoreMetablockArena;
947 
BrotliStoreMetaBlock(MemoryManager * m,const uint8_t * input,size_t start_pos,size_t length,size_t mask,uint8_t prev_byte,uint8_t prev_byte2,BROTLI_BOOL is_last,const BrotliEncoderParams * params,ContextType literal_context_mode,const Command * commands,size_t n_commands,const MetaBlockSplit * mb,size_t * storage_ix,uint8_t * storage)948 void BrotliStoreMetaBlock(MemoryManager* m,
949     const uint8_t* input, size_t start_pos, size_t length, size_t mask,
950     uint8_t prev_byte, uint8_t prev_byte2, BROTLI_BOOL is_last,
951     const BrotliEncoderParams* params, ContextType literal_context_mode,
952     const Command* commands, size_t n_commands, const MetaBlockSplit* mb,
953     size_t* storage_ix, uint8_t* storage) {
954 
955   size_t pos = start_pos;
956   size_t i;
957   uint32_t num_distance_symbols = params->dist.alphabet_size_max;
958   uint32_t num_effective_distance_symbols = params->dist.alphabet_size_limit;
959   HuffmanTree* tree;
960   ContextLut literal_context_lut = BROTLI_CONTEXT_LUT(literal_context_mode);
961   StoreMetablockArena* arena = NULL;
962   BlockEncoder* literal_enc = NULL;
963   BlockEncoder* command_enc = NULL;
964   BlockEncoder* distance_enc = NULL;
965   const BrotliDistanceParams* dist = &params->dist;
966   BROTLI_DCHECK(
967       num_effective_distance_symbols <= BROTLI_NUM_HISTOGRAM_DISTANCE_SYMBOLS);
968 
969   StoreCompressedMetaBlockHeader(is_last, length, storage_ix, storage);
970 
971   tree = BROTLI_ALLOC(m, HuffmanTree, MAX_HUFFMAN_TREE_SIZE);
972   arena = BROTLI_ALLOC(m, StoreMetablockArena, 1);
973   if (BROTLI_IS_OOM(m) || BROTLI_IS_NULL(tree) || BROTLI_IS_NULL(arena)) return;
974   literal_enc = &arena->literal_enc;
975   command_enc = &arena->command_enc;
976   distance_enc = &arena->distance_enc;
977   InitBlockEncoder(literal_enc, BROTLI_NUM_LITERAL_SYMBOLS,
978       mb->literal_split.num_types, mb->literal_split.types,
979       mb->literal_split.lengths, mb->literal_split.num_blocks);
980   InitBlockEncoder(command_enc, BROTLI_NUM_COMMAND_SYMBOLS,
981       mb->command_split.num_types, mb->command_split.types,
982       mb->command_split.lengths, mb->command_split.num_blocks);
983   InitBlockEncoder(distance_enc, num_effective_distance_symbols,
984       mb->distance_split.num_types, mb->distance_split.types,
985       mb->distance_split.lengths, mb->distance_split.num_blocks);
986 
987   BuildAndStoreBlockSwitchEntropyCodes(literal_enc, tree, storage_ix, storage);
988   BuildAndStoreBlockSwitchEntropyCodes(command_enc, tree, storage_ix, storage);
989   BuildAndStoreBlockSwitchEntropyCodes(distance_enc, tree, storage_ix, storage);
990 
991   BrotliWriteBits(2, dist->distance_postfix_bits, storage_ix, storage);
992   BrotliWriteBits(
993       4, dist->num_direct_distance_codes >> dist->distance_postfix_bits,
994       storage_ix, storage);
995   for (i = 0; i < mb->literal_split.num_types; ++i) {
996     BrotliWriteBits(2, literal_context_mode, storage_ix, storage);
997   }
998 
999   if (mb->literal_context_map_size == 0) {
1000     StoreTrivialContextMap(
1001         &arena->context_map_arena, mb->literal_histograms_size,
1002         BROTLI_LITERAL_CONTEXT_BITS, tree, storage_ix, storage);
1003   } else {
1004     EncodeContextMap(m, &arena->context_map_arena,
1005         mb->literal_context_map, mb->literal_context_map_size,
1006         mb->literal_histograms_size, tree, storage_ix, storage);
1007     if (BROTLI_IS_OOM(m)) return;
1008   }
1009 
1010   if (mb->distance_context_map_size == 0) {
1011     StoreTrivialContextMap(
1012         &arena->context_map_arena, mb->distance_histograms_size,
1013         BROTLI_DISTANCE_CONTEXT_BITS, tree, storage_ix, storage);
1014   } else {
1015     EncodeContextMap(m, &arena->context_map_arena,
1016         mb->distance_context_map, mb->distance_context_map_size,
1017         mb->distance_histograms_size, tree, storage_ix, storage);
1018     if (BROTLI_IS_OOM(m)) return;
1019   }
1020 
1021   BuildAndStoreEntropyCodesLiteral(m, literal_enc, mb->literal_histograms,
1022       mb->literal_histograms_size, BROTLI_NUM_LITERAL_SYMBOLS, tree,
1023       storage_ix, storage);
1024   if (BROTLI_IS_OOM(m)) return;
1025   BuildAndStoreEntropyCodesCommand(m, command_enc, mb->command_histograms,
1026       mb->command_histograms_size, BROTLI_NUM_COMMAND_SYMBOLS, tree,
1027       storage_ix, storage);
1028   if (BROTLI_IS_OOM(m)) return;
1029   BuildAndStoreEntropyCodesDistance(m, distance_enc, mb->distance_histograms,
1030       mb->distance_histograms_size, num_distance_symbols, tree,
1031       storage_ix, storage);
1032   if (BROTLI_IS_OOM(m)) return;
1033   BROTLI_FREE(m, tree);
1034 
1035   for (i = 0; i < n_commands; ++i) {
1036     const Command cmd = commands[i];
1037     size_t cmd_code = cmd.cmd_prefix_;
1038     StoreSymbol(command_enc, cmd_code, storage_ix, storage);
1039     StoreCommandExtra(&cmd, storage_ix, storage);
1040     if (mb->literal_context_map_size == 0) {
1041       size_t j;
1042       for (j = cmd.insert_len_; j != 0; --j) {
1043         StoreSymbol(literal_enc, input[pos & mask], storage_ix, storage);
1044         ++pos;
1045       }
1046     } else {
1047       size_t j;
1048       for (j = cmd.insert_len_; j != 0; --j) {
1049         size_t context =
1050             BROTLI_CONTEXT(prev_byte, prev_byte2, literal_context_lut);
1051         uint8_t literal = input[pos & mask];
1052         StoreSymbolWithContext(literal_enc, literal, context,
1053             mb->literal_context_map, storage_ix, storage,
1054             BROTLI_LITERAL_CONTEXT_BITS);
1055         prev_byte2 = prev_byte;
1056         prev_byte = literal;
1057         ++pos;
1058       }
1059     }
1060     pos += CommandCopyLen(&cmd);
1061     if (CommandCopyLen(&cmd)) {
1062       prev_byte2 = input[(pos - 2) & mask];
1063       prev_byte = input[(pos - 1) & mask];
1064       if (cmd.cmd_prefix_ >= 128) {
1065         size_t dist_code = cmd.dist_prefix_ & 0x3FF;
1066         uint32_t distnumextra = cmd.dist_prefix_ >> 10;
1067         uint64_t distextra = cmd.dist_extra_;
1068         if (mb->distance_context_map_size == 0) {
1069           StoreSymbol(distance_enc, dist_code, storage_ix, storage);
1070         } else {
1071           size_t context = CommandDistanceContext(&cmd);
1072           StoreSymbolWithContext(distance_enc, dist_code, context,
1073               mb->distance_context_map, storage_ix, storage,
1074               BROTLI_DISTANCE_CONTEXT_BITS);
1075         }
1076         BrotliWriteBits(distnumextra, distextra, storage_ix, storage);
1077       }
1078     }
1079   }
1080   CleanupBlockEncoder(m, distance_enc);
1081   CleanupBlockEncoder(m, command_enc);
1082   CleanupBlockEncoder(m, literal_enc);
1083   BROTLI_FREE(m, arena);
1084   if (is_last) {
1085     JumpToByteBoundary(storage_ix, storage);
1086   }
1087 }
1088 
BuildHistograms(const uint8_t * input,size_t start_pos,size_t mask,const Command * commands,size_t n_commands,HistogramLiteral * lit_histo,HistogramCommand * cmd_histo,HistogramDistance * dist_histo)1089 static void BuildHistograms(const uint8_t* input,
1090                             size_t start_pos,
1091                             size_t mask,
1092                             const Command* commands,
1093                             size_t n_commands,
1094                             HistogramLiteral* lit_histo,
1095                             HistogramCommand* cmd_histo,
1096                             HistogramDistance* dist_histo) {
1097   size_t pos = start_pos;
1098   size_t i;
1099   for (i = 0; i < n_commands; ++i) {
1100     const Command cmd = commands[i];
1101     size_t j;
1102     HistogramAddCommand(cmd_histo, cmd.cmd_prefix_);
1103     for (j = cmd.insert_len_; j != 0; --j) {
1104       HistogramAddLiteral(lit_histo, input[pos & mask]);
1105       ++pos;
1106     }
1107     pos += CommandCopyLen(&cmd);
1108     if (CommandCopyLen(&cmd) && cmd.cmd_prefix_ >= 128) {
1109       HistogramAddDistance(dist_histo, cmd.dist_prefix_ & 0x3FF);
1110     }
1111   }
1112 }
1113 
StoreDataWithHuffmanCodes(const uint8_t * input,size_t start_pos,size_t mask,const Command * commands,size_t n_commands,const uint8_t * lit_depth,const uint16_t * lit_bits,const uint8_t * cmd_depth,const uint16_t * cmd_bits,const uint8_t * dist_depth,const uint16_t * dist_bits,size_t * storage_ix,uint8_t * storage)1114 static void StoreDataWithHuffmanCodes(const uint8_t* input,
1115                                       size_t start_pos,
1116                                       size_t mask,
1117                                       const Command* commands,
1118                                       size_t n_commands,
1119                                       const uint8_t* lit_depth,
1120                                       const uint16_t* lit_bits,
1121                                       const uint8_t* cmd_depth,
1122                                       const uint16_t* cmd_bits,
1123                                       const uint8_t* dist_depth,
1124                                       const uint16_t* dist_bits,
1125                                       size_t* storage_ix,
1126                                       uint8_t* storage) {
1127   size_t pos = start_pos;
1128   size_t i;
1129   for (i = 0; i < n_commands; ++i) {
1130     const Command cmd = commands[i];
1131     const size_t cmd_code = cmd.cmd_prefix_;
1132     size_t j;
1133     BrotliWriteBits(
1134         cmd_depth[cmd_code], cmd_bits[cmd_code], storage_ix, storage);
1135     StoreCommandExtra(&cmd, storage_ix, storage);
1136     for (j = cmd.insert_len_; j != 0; --j) {
1137       const uint8_t literal = input[pos & mask];
1138       BrotliWriteBits(
1139           lit_depth[literal], lit_bits[literal], storage_ix, storage);
1140       ++pos;
1141     }
1142     pos += CommandCopyLen(&cmd);
1143     if (CommandCopyLen(&cmd) && cmd.cmd_prefix_ >= 128) {
1144       const size_t dist_code = cmd.dist_prefix_ & 0x3FF;
1145       const uint32_t distnumextra = cmd.dist_prefix_ >> 10;
1146       const uint32_t distextra = cmd.dist_extra_;
1147       BrotliWriteBits(dist_depth[dist_code], dist_bits[dist_code],
1148                       storage_ix, storage);
1149       BrotliWriteBits(distnumextra, distextra, storage_ix, storage);
1150     }
1151   }
1152 }
1153 
1154 /* TODO(eustas): pull alloc/dealloc to caller? */
1155 typedef struct MetablockArena {
1156   HistogramLiteral lit_histo;
1157   HistogramCommand cmd_histo;
1158   HistogramDistance dist_histo;
1159   /* TODO(eustas): merge bits and depth? */
1160   uint8_t lit_depth[BROTLI_NUM_LITERAL_SYMBOLS];
1161   uint16_t lit_bits[BROTLI_NUM_LITERAL_SYMBOLS];
1162   uint8_t cmd_depth[BROTLI_NUM_COMMAND_SYMBOLS];
1163   uint16_t cmd_bits[BROTLI_NUM_COMMAND_SYMBOLS];
1164   uint8_t dist_depth[MAX_SIMPLE_DISTANCE_ALPHABET_SIZE];
1165   uint16_t dist_bits[MAX_SIMPLE_DISTANCE_ALPHABET_SIZE];
1166   HuffmanTree tree[MAX_HUFFMAN_TREE_SIZE];
1167 } MetablockArena;
1168 
BrotliStoreMetaBlockTrivial(MemoryManager * m,const uint8_t * input,size_t start_pos,size_t length,size_t mask,BROTLI_BOOL is_last,const BrotliEncoderParams * params,const Command * commands,size_t n_commands,size_t * storage_ix,uint8_t * storage)1169 void BrotliStoreMetaBlockTrivial(MemoryManager* m,
1170     const uint8_t* input, size_t start_pos, size_t length, size_t mask,
1171     BROTLI_BOOL is_last, const BrotliEncoderParams* params,
1172     const Command* commands, size_t n_commands,
1173     size_t* storage_ix, uint8_t* storage) {
1174   MetablockArena* arena = BROTLI_ALLOC(m, MetablockArena, 1);
1175   uint32_t num_distance_symbols = params->dist.alphabet_size_max;
1176   if (BROTLI_IS_OOM(m) || BROTLI_IS_NULL(arena)) return;
1177 
1178   StoreCompressedMetaBlockHeader(is_last, length, storage_ix, storage);
1179 
1180   HistogramClearLiteral(&arena->lit_histo);
1181   HistogramClearCommand(&arena->cmd_histo);
1182   HistogramClearDistance(&arena->dist_histo);
1183 
1184   BuildHistograms(input, start_pos, mask, commands, n_commands,
1185                   &arena->lit_histo, &arena->cmd_histo, &arena->dist_histo);
1186 
1187   BrotliWriteBits(13, 0, storage_ix, storage);
1188 
1189   BuildAndStoreHuffmanTree(arena->lit_histo.data_, BROTLI_NUM_LITERAL_SYMBOLS,
1190                            BROTLI_NUM_LITERAL_SYMBOLS, arena->tree,
1191                            arena->lit_depth, arena->lit_bits,
1192                            storage_ix, storage);
1193   BuildAndStoreHuffmanTree(arena->cmd_histo.data_, BROTLI_NUM_COMMAND_SYMBOLS,
1194                            BROTLI_NUM_COMMAND_SYMBOLS, arena->tree,
1195                            arena->cmd_depth, arena->cmd_bits,
1196                            storage_ix, storage);
1197   BuildAndStoreHuffmanTree(arena->dist_histo.data_,
1198                            MAX_SIMPLE_DISTANCE_ALPHABET_SIZE,
1199                            num_distance_symbols, arena->tree,
1200                            arena->dist_depth, arena->dist_bits,
1201                            storage_ix, storage);
1202   StoreDataWithHuffmanCodes(input, start_pos, mask, commands,
1203                             n_commands, arena->lit_depth, arena->lit_bits,
1204                             arena->cmd_depth, arena->cmd_bits,
1205                             arena->dist_depth, arena->dist_bits,
1206                             storage_ix, storage);
1207   BROTLI_FREE(m, arena);
1208   if (is_last) {
1209     JumpToByteBoundary(storage_ix, storage);
1210   }
1211 }
1212 
BrotliStoreMetaBlockFast(MemoryManager * m,const uint8_t * input,size_t start_pos,size_t length,size_t mask,BROTLI_BOOL is_last,const BrotliEncoderParams * params,const Command * commands,size_t n_commands,size_t * storage_ix,uint8_t * storage)1213 void BrotliStoreMetaBlockFast(MemoryManager* m,
1214     const uint8_t* input, size_t start_pos, size_t length, size_t mask,
1215     BROTLI_BOOL is_last, const BrotliEncoderParams* params,
1216     const Command* commands, size_t n_commands,
1217     size_t* storage_ix, uint8_t* storage) {
1218   MetablockArena* arena = BROTLI_ALLOC(m, MetablockArena, 1);
1219   uint32_t num_distance_symbols = params->dist.alphabet_size_max;
1220   uint32_t distance_alphabet_bits =
1221       Log2FloorNonZero(num_distance_symbols - 1) + 1;
1222   if (BROTLI_IS_OOM(m) || BROTLI_IS_NULL(arena)) return;
1223 
1224   StoreCompressedMetaBlockHeader(is_last, length, storage_ix, storage);
1225 
1226   BrotliWriteBits(13, 0, storage_ix, storage);
1227 
1228   if (n_commands <= 128) {
1229     uint32_t histogram[BROTLI_NUM_LITERAL_SYMBOLS] = { 0 };
1230     size_t pos = start_pos;
1231     size_t num_literals = 0;
1232     size_t i;
1233     for (i = 0; i < n_commands; ++i) {
1234       const Command cmd = commands[i];
1235       size_t j;
1236       for (j = cmd.insert_len_; j != 0; --j) {
1237         ++histogram[input[pos & mask]];
1238         ++pos;
1239       }
1240       num_literals += cmd.insert_len_;
1241       pos += CommandCopyLen(&cmd);
1242     }
1243     BrotliBuildAndStoreHuffmanTreeFast(arena->tree, histogram, num_literals,
1244                                        /* max_bits = */ 8,
1245                                        arena->lit_depth, arena->lit_bits,
1246                                        storage_ix, storage);
1247     StoreStaticCommandHuffmanTree(storage_ix, storage);
1248     StoreStaticDistanceHuffmanTree(storage_ix, storage);
1249     StoreDataWithHuffmanCodes(input, start_pos, mask, commands,
1250                               n_commands, arena->lit_depth, arena->lit_bits,
1251                               kStaticCommandCodeDepth,
1252                               kStaticCommandCodeBits,
1253                               kStaticDistanceCodeDepth,
1254                               kStaticDistanceCodeBits,
1255                               storage_ix, storage);
1256   } else {
1257     HistogramClearLiteral(&arena->lit_histo);
1258     HistogramClearCommand(&arena->cmd_histo);
1259     HistogramClearDistance(&arena->dist_histo);
1260     BuildHistograms(input, start_pos, mask, commands, n_commands,
1261                     &arena->lit_histo, &arena->cmd_histo, &arena->dist_histo);
1262     BrotliBuildAndStoreHuffmanTreeFast(arena->tree, arena->lit_histo.data_,
1263                                        arena->lit_histo.total_count_,
1264                                        /* max_bits = */ 8,
1265                                        arena->lit_depth, arena->lit_bits,
1266                                        storage_ix, storage);
1267     BrotliBuildAndStoreHuffmanTreeFast(arena->tree, arena->cmd_histo.data_,
1268                                        arena->cmd_histo.total_count_,
1269                                        /* max_bits = */ 10,
1270                                        arena->cmd_depth, arena->cmd_bits,
1271                                        storage_ix, storage);
1272     BrotliBuildAndStoreHuffmanTreeFast(arena->tree, arena->dist_histo.data_,
1273                                        arena->dist_histo.total_count_,
1274                                        /* max_bits = */
1275                                        distance_alphabet_bits,
1276                                        arena->dist_depth, arena->dist_bits,
1277                                        storage_ix, storage);
1278     StoreDataWithHuffmanCodes(input, start_pos, mask, commands,
1279                               n_commands, arena->lit_depth, arena->lit_bits,
1280                               arena->cmd_depth, arena->cmd_bits,
1281                               arena->dist_depth, arena->dist_bits,
1282                               storage_ix, storage);
1283   }
1284 
1285   BROTLI_FREE(m, arena);
1286 
1287   if (is_last) {
1288     JumpToByteBoundary(storage_ix, storage);
1289   }
1290 }
1291 
1292 /* This is for storing uncompressed blocks (simple raw storage of
1293    bytes-as-bytes). */
BrotliStoreUncompressedMetaBlock(BROTLI_BOOL is_final_block,const uint8_t * BROTLI_RESTRICT input,size_t position,size_t mask,size_t len,size_t * BROTLI_RESTRICT storage_ix,uint8_t * BROTLI_RESTRICT storage)1294 void BrotliStoreUncompressedMetaBlock(BROTLI_BOOL is_final_block,
1295                                       const uint8_t* BROTLI_RESTRICT input,
1296                                       size_t position, size_t mask,
1297                                       size_t len,
1298                                       size_t* BROTLI_RESTRICT storage_ix,
1299                                       uint8_t* BROTLI_RESTRICT storage) {
1300   size_t masked_pos = position & mask;
1301   BrotliStoreUncompressedMetaBlockHeader(len, storage_ix, storage);
1302   JumpToByteBoundary(storage_ix, storage);
1303 
1304   if (masked_pos + len > mask + 1) {
1305     size_t len1 = mask + 1 - masked_pos;
1306     memcpy(&storage[*storage_ix >> 3], &input[masked_pos], len1);
1307     *storage_ix += len1 << 3;
1308     len -= len1;
1309     masked_pos = 0;
1310   }
1311   memcpy(&storage[*storage_ix >> 3], &input[masked_pos], len);
1312   *storage_ix += len << 3;
1313 
1314   /* We need to clear the next 4 bytes to continue to be
1315      compatible with BrotliWriteBits. */
1316   BrotliWriteBitsPrepareStorage(*storage_ix, storage);
1317 
1318   /* Since the uncompressed block itself may not be the final block, add an
1319      empty one after this. */
1320   if (is_final_block) {
1321     BrotliWriteBits(1, 1, storage_ix, storage);  /* islast */
1322     BrotliWriteBits(1, 1, storage_ix, storage);  /* isempty */
1323     JumpToByteBoundary(storage_ix, storage);
1324   }
1325 }
1326 
1327 #if defined(BROTLI_TEST)
GetBlockLengthPrefixCodeForTest(uint32_t len,size_t * code,uint32_t * n_extra,uint32_t * extra)1328 void GetBlockLengthPrefixCodeForTest(uint32_t len, size_t* code,
1329                                      uint32_t* n_extra, uint32_t* extra) {
1330   GetBlockLengthPrefixCode(len, code, n_extra, extra);
1331 }
1332 #endif
1333 
1334 #if defined(__cplusplus) || defined(c_plusplus)
1335 }  /* extern "C" */
1336 #endif
1337