1 | /* Copyright 2010 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 | /* Entropy encoding (Huffman) utilities. */ |
8 | |
9 | #include "./entropy_encode.h" |
10 | |
11 | #include <string.h> /* memset */ |
12 | |
13 | #include "../common/constants.h" |
14 | #include "../common/platform.h" |
15 | #include <brotli/types.h> |
16 | |
17 | #if defined(__cplusplus) || defined(c_plusplus) |
18 | extern "C" { |
19 | #endif |
20 | |
21 | BROTLI_BOOL BrotliSetDepth( |
22 | int p0, HuffmanTree* pool, uint8_t* depth, int max_depth) { |
23 | int stack[16]; |
24 | int level = 0; |
25 | int p = p0; |
26 | BROTLI_DCHECK(max_depth <= 15); |
27 | stack[0] = -1; |
28 | while (BROTLI_TRUE) { |
29 | if (pool[p].index_left_ >= 0) { |
30 | level++; |
31 | if (level > max_depth) return BROTLI_FALSE; |
32 | stack[level] = pool[p].index_right_or_value_; |
33 | p = pool[p].index_left_; |
34 | continue; |
35 | } else { |
36 | depth[pool[p].index_right_or_value_] = (uint8_t)level; |
37 | } |
38 | while (level >= 0 && stack[level] == -1) level--; |
39 | if (level < 0) return BROTLI_TRUE; |
40 | p = stack[level]; |
41 | stack[level] = -1; |
42 | } |
43 | } |
44 | |
45 | /* Sort the root nodes, least popular first. */ |
46 | static BROTLI_INLINE BROTLI_BOOL SortHuffmanTree( |
47 | const HuffmanTree* v0, const HuffmanTree* v1) { |
48 | if (v0->total_count_ != v1->total_count_) { |
49 | return TO_BROTLI_BOOL(v0->total_count_ < v1->total_count_); |
50 | } |
51 | return TO_BROTLI_BOOL(v0->index_right_or_value_ > v1->index_right_or_value_); |
52 | } |
53 | |
54 | /* This function will create a Huffman tree. |
55 | |
56 | The catch here is that the tree cannot be arbitrarily deep. |
57 | Brotli specifies a maximum depth of 15 bits for "code trees" |
58 | and 7 bits for "code length code trees." |
59 | |
60 | count_limit is the value that is to be faked as the minimum value |
61 | and this minimum value is raised until the tree matches the |
62 | maximum length requirement. |
63 | |
64 | This algorithm is not of excellent performance for very long data blocks, |
65 | especially when population counts are longer than 2**tree_limit, but |
66 | we are not planning to use this with extremely long blocks. |
67 | |
68 | See http://en.wikipedia.org/wiki/Huffman_coding */ |
69 | void BrotliCreateHuffmanTree(const uint32_t* data, |
70 | const size_t length, |
71 | const int tree_limit, |
72 | HuffmanTree* tree, |
73 | uint8_t* depth) { |
74 | uint32_t count_limit; |
75 | HuffmanTree sentinel; |
76 | InitHuffmanTree(&sentinel, BROTLI_UINT32_MAX, -1, -1); |
77 | /* For block sizes below 64 kB, we never need to do a second iteration |
78 | of this loop. Probably all of our block sizes will be smaller than |
79 | that, so this loop is mostly of academic interest. If we actually |
80 | would need this, we would be better off with the Katajainen algorithm. */ |
81 | for (count_limit = 1; ; count_limit *= 2) { |
82 | size_t n = 0; |
83 | size_t i; |
84 | size_t j; |
85 | size_t k; |
86 | for (i = length; i != 0;) { |
87 | --i; |
88 | if (data[i]) { |
89 | const uint32_t count = BROTLI_MAX(uint32_t, data[i], count_limit); |
90 | InitHuffmanTree(&tree[n++], count, -1, (int16_t)i); |
91 | } |
92 | } |
93 | |
94 | if (n == 1) { |
95 | depth[tree[0].index_right_or_value_] = 1; /* Only one element. */ |
96 | break; |
97 | } |
98 | |
99 | SortHuffmanTreeItems(tree, n, SortHuffmanTree); |
100 | |
101 | /* The nodes are: |
102 | [0, n): the sorted leaf nodes that we start with. |
103 | [n]: we add a sentinel here. |
104 | [n + 1, 2n): new parent nodes are added here, starting from |
105 | (n+1). These are naturally in ascending order. |
106 | [2n]: we add a sentinel at the end as well. |
107 | There will be (2n+1) elements at the end. */ |
108 | tree[n] = sentinel; |
109 | tree[n + 1] = sentinel; |
110 | |
111 | i = 0; /* Points to the next leaf node. */ |
112 | j = n + 1; /* Points to the next non-leaf node. */ |
113 | for (k = n - 1; k != 0; --k) { |
114 | size_t left, right; |
115 | if (tree[i].total_count_ <= tree[j].total_count_) { |
116 | left = i; |
117 | ++i; |
118 | } else { |
119 | left = j; |
120 | ++j; |
121 | } |
122 | if (tree[i].total_count_ <= tree[j].total_count_) { |
123 | right = i; |
124 | ++i; |
125 | } else { |
126 | right = j; |
127 | ++j; |
128 | } |
129 | |
130 | { |
131 | /* The sentinel node becomes the parent node. */ |
132 | size_t j_end = 2 * n - k; |
133 | tree[j_end].total_count_ = |
134 | tree[left].total_count_ + tree[right].total_count_; |
135 | tree[j_end].index_left_ = (int16_t)left; |
136 | tree[j_end].index_right_or_value_ = (int16_t)right; |
137 | |
138 | /* Add back the last sentinel node. */ |
139 | tree[j_end + 1] = sentinel; |
140 | } |
141 | } |
142 | if (BrotliSetDepth((int)(2 * n - 1), &tree[0], depth, tree_limit)) { |
143 | /* We need to pack the Huffman tree in tree_limit bits. If this was not |
144 | successful, add fake entities to the lowest values and retry. */ |
145 | break; |
146 | } |
147 | } |
148 | } |
149 | |
150 | static void Reverse(uint8_t* v, size_t start, size_t end) { |
151 | --end; |
152 | while (start < end) { |
153 | uint8_t tmp = v[start]; |
154 | v[start] = v[end]; |
155 | v[end] = tmp; |
156 | ++start; |
157 | --end; |
158 | } |
159 | } |
160 | |
161 | static void BrotliWriteHuffmanTreeRepetitions( |
162 | const uint8_t previous_value, |
163 | const uint8_t value, |
164 | size_t repetitions, |
165 | size_t* tree_size, |
166 | uint8_t* tree, |
167 | uint8_t* ) { |
168 | BROTLI_DCHECK(repetitions > 0); |
169 | if (previous_value != value) { |
170 | tree[*tree_size] = value; |
171 | extra_bits_data[*tree_size] = 0; |
172 | ++(*tree_size); |
173 | --repetitions; |
174 | } |
175 | if (repetitions == 7) { |
176 | tree[*tree_size] = value; |
177 | extra_bits_data[*tree_size] = 0; |
178 | ++(*tree_size); |
179 | --repetitions; |
180 | } |
181 | if (repetitions < 3) { |
182 | size_t i; |
183 | for (i = 0; i < repetitions; ++i) { |
184 | tree[*tree_size] = value; |
185 | extra_bits_data[*tree_size] = 0; |
186 | ++(*tree_size); |
187 | } |
188 | } else { |
189 | size_t start = *tree_size; |
190 | repetitions -= 3; |
191 | while (BROTLI_TRUE) { |
192 | tree[*tree_size] = BROTLI_REPEAT_PREVIOUS_CODE_LENGTH; |
193 | extra_bits_data[*tree_size] = repetitions & 0x3; |
194 | ++(*tree_size); |
195 | repetitions >>= 2; |
196 | if (repetitions == 0) { |
197 | break; |
198 | } |
199 | --repetitions; |
200 | } |
201 | Reverse(tree, start, *tree_size); |
202 | Reverse(extra_bits_data, start, *tree_size); |
203 | } |
204 | } |
205 | |
206 | static void BrotliWriteHuffmanTreeRepetitionsZeros( |
207 | size_t repetitions, |
208 | size_t* tree_size, |
209 | uint8_t* tree, |
210 | uint8_t* ) { |
211 | if (repetitions == 11) { |
212 | tree[*tree_size] = 0; |
213 | extra_bits_data[*tree_size] = 0; |
214 | ++(*tree_size); |
215 | --repetitions; |
216 | } |
217 | if (repetitions < 3) { |
218 | size_t i; |
219 | for (i = 0; i < repetitions; ++i) { |
220 | tree[*tree_size] = 0; |
221 | extra_bits_data[*tree_size] = 0; |
222 | ++(*tree_size); |
223 | } |
224 | } else { |
225 | size_t start = *tree_size; |
226 | repetitions -= 3; |
227 | while (BROTLI_TRUE) { |
228 | tree[*tree_size] = BROTLI_REPEAT_ZERO_CODE_LENGTH; |
229 | extra_bits_data[*tree_size] = repetitions & 0x7; |
230 | ++(*tree_size); |
231 | repetitions >>= 3; |
232 | if (repetitions == 0) { |
233 | break; |
234 | } |
235 | --repetitions; |
236 | } |
237 | Reverse(tree, start, *tree_size); |
238 | Reverse(extra_bits_data, start, *tree_size); |
239 | } |
240 | } |
241 | |
242 | void BrotliOptimizeHuffmanCountsForRle(size_t length, uint32_t* counts, |
243 | uint8_t* good_for_rle) { |
244 | size_t nonzero_count = 0; |
245 | size_t stride; |
246 | size_t limit; |
247 | size_t sum; |
248 | const size_t streak_limit = 1240; |
249 | /* Let's make the Huffman code more compatible with RLE encoding. */ |
250 | size_t i; |
251 | for (i = 0; i < length; i++) { |
252 | if (counts[i]) { |
253 | ++nonzero_count; |
254 | } |
255 | } |
256 | if (nonzero_count < 16) { |
257 | return; |
258 | } |
259 | while (length != 0 && counts[length - 1] == 0) { |
260 | --length; |
261 | } |
262 | if (length == 0) { |
263 | return; /* All zeros. */ |
264 | } |
265 | /* Now counts[0..length - 1] does not have trailing zeros. */ |
266 | { |
267 | size_t nonzeros = 0; |
268 | uint32_t smallest_nonzero = 1 << 30; |
269 | for (i = 0; i < length; ++i) { |
270 | if (counts[i] != 0) { |
271 | ++nonzeros; |
272 | if (smallest_nonzero > counts[i]) { |
273 | smallest_nonzero = counts[i]; |
274 | } |
275 | } |
276 | } |
277 | if (nonzeros < 5) { |
278 | /* Small histogram will model it well. */ |
279 | return; |
280 | } |
281 | if (smallest_nonzero < 4) { |
282 | size_t zeros = length - nonzeros; |
283 | if (zeros < 6) { |
284 | for (i = 1; i < length - 1; ++i) { |
285 | if (counts[i - 1] != 0 && counts[i] == 0 && counts[i + 1] != 0) { |
286 | counts[i] = 1; |
287 | } |
288 | } |
289 | } |
290 | } |
291 | if (nonzeros < 28) { |
292 | return; |
293 | } |
294 | } |
295 | /* 2) Let's mark all population counts that already can be encoded |
296 | with an RLE code. */ |
297 | memset(good_for_rle, 0, length); |
298 | { |
299 | /* Let's not spoil any of the existing good RLE codes. |
300 | Mark any seq of 0's that is longer as 5 as a good_for_rle. |
301 | Mark any seq of non-0's that is longer as 7 as a good_for_rle. */ |
302 | uint32_t symbol = counts[0]; |
303 | size_t step = 0; |
304 | for (i = 0; i <= length; ++i) { |
305 | if (i == length || counts[i] != symbol) { |
306 | if ((symbol == 0 && step >= 5) || |
307 | (symbol != 0 && step >= 7)) { |
308 | size_t k; |
309 | for (k = 0; k < step; ++k) { |
310 | good_for_rle[i - k - 1] = 1; |
311 | } |
312 | } |
313 | step = 1; |
314 | if (i != length) { |
315 | symbol = counts[i]; |
316 | } |
317 | } else { |
318 | ++step; |
319 | } |
320 | } |
321 | } |
322 | /* 3) Let's replace those population counts that lead to more RLE codes. |
323 | Math here is in 24.8 fixed point representation. */ |
324 | stride = 0; |
325 | limit = 256 * (counts[0] + counts[1] + counts[2]) / 3 + 420; |
326 | sum = 0; |
327 | for (i = 0; i <= length; ++i) { |
328 | if (i == length || good_for_rle[i] || |
329 | (i != 0 && good_for_rle[i - 1]) || |
330 | (256 * counts[i] - limit + streak_limit) >= 2 * streak_limit) { |
331 | if (stride >= 4 || (stride >= 3 && sum == 0)) { |
332 | size_t k; |
333 | /* The stride must end, collapse what we have, if we have enough (4). */ |
334 | size_t count = (sum + stride / 2) / stride; |
335 | if (count == 0) { |
336 | count = 1; |
337 | } |
338 | if (sum == 0) { |
339 | /* Don't make an all zeros stride to be upgraded to ones. */ |
340 | count = 0; |
341 | } |
342 | for (k = 0; k < stride; ++k) { |
343 | /* We don't want to change value at counts[i], |
344 | that is already belonging to the next stride. Thus - 1. */ |
345 | counts[i - k - 1] = (uint32_t)count; |
346 | } |
347 | } |
348 | stride = 0; |
349 | sum = 0; |
350 | if (i < length - 2) { |
351 | /* All interesting strides have a count of at least 4, */ |
352 | /* at least when non-zeros. */ |
353 | limit = 256 * (counts[i] + counts[i + 1] + counts[i + 2]) / 3 + 420; |
354 | } else if (i < length) { |
355 | limit = 256 * counts[i]; |
356 | } else { |
357 | limit = 0; |
358 | } |
359 | } |
360 | ++stride; |
361 | if (i != length) { |
362 | sum += counts[i]; |
363 | if (stride >= 4) { |
364 | limit = (256 * sum + stride / 2) / stride; |
365 | } |
366 | if (stride == 4) { |
367 | limit += 120; |
368 | } |
369 | } |
370 | } |
371 | } |
372 | |
373 | static void DecideOverRleUse(const uint8_t* depth, const size_t length, |
374 | BROTLI_BOOL* use_rle_for_non_zero, |
375 | BROTLI_BOOL* use_rle_for_zero) { |
376 | size_t total_reps_zero = 0; |
377 | size_t total_reps_non_zero = 0; |
378 | size_t count_reps_zero = 1; |
379 | size_t count_reps_non_zero = 1; |
380 | size_t i; |
381 | for (i = 0; i < length;) { |
382 | const uint8_t value = depth[i]; |
383 | size_t reps = 1; |
384 | size_t k; |
385 | for (k = i + 1; k < length && depth[k] == value; ++k) { |
386 | ++reps; |
387 | } |
388 | if (reps >= 3 && value == 0) { |
389 | total_reps_zero += reps; |
390 | ++count_reps_zero; |
391 | } |
392 | if (reps >= 4 && value != 0) { |
393 | total_reps_non_zero += reps; |
394 | ++count_reps_non_zero; |
395 | } |
396 | i += reps; |
397 | } |
398 | *use_rle_for_non_zero = |
399 | TO_BROTLI_BOOL(total_reps_non_zero > count_reps_non_zero * 2); |
400 | *use_rle_for_zero = TO_BROTLI_BOOL(total_reps_zero > count_reps_zero * 2); |
401 | } |
402 | |
403 | void BrotliWriteHuffmanTree(const uint8_t* depth, |
404 | size_t length, |
405 | size_t* tree_size, |
406 | uint8_t* tree, |
407 | uint8_t* ) { |
408 | uint8_t previous_value = BROTLI_INITIAL_REPEATED_CODE_LENGTH; |
409 | size_t i; |
410 | BROTLI_BOOL use_rle_for_non_zero = BROTLI_FALSE; |
411 | BROTLI_BOOL use_rle_for_zero = BROTLI_FALSE; |
412 | |
413 | /* Throw away trailing zeros. */ |
414 | size_t new_length = length; |
415 | for (i = 0; i < length; ++i) { |
416 | if (depth[length - i - 1] == 0) { |
417 | --new_length; |
418 | } else { |
419 | break; |
420 | } |
421 | } |
422 | |
423 | /* First gather statistics on if it is a good idea to do RLE. */ |
424 | if (length > 50) { |
425 | /* Find RLE coding for longer codes. |
426 | Shorter codes seem not to benefit from RLE. */ |
427 | DecideOverRleUse(depth, new_length, |
428 | &use_rle_for_non_zero, &use_rle_for_zero); |
429 | } |
430 | |
431 | /* Actual RLE coding. */ |
432 | for (i = 0; i < new_length;) { |
433 | const uint8_t value = depth[i]; |
434 | size_t reps = 1; |
435 | if ((value != 0 && use_rle_for_non_zero) || |
436 | (value == 0 && use_rle_for_zero)) { |
437 | size_t k; |
438 | for (k = i + 1; k < new_length && depth[k] == value; ++k) { |
439 | ++reps; |
440 | } |
441 | } |
442 | if (value == 0) { |
443 | BrotliWriteHuffmanTreeRepetitionsZeros( |
444 | reps, tree_size, tree, extra_bits_data); |
445 | } else { |
446 | BrotliWriteHuffmanTreeRepetitions(previous_value, |
447 | value, reps, tree_size, |
448 | tree, extra_bits_data); |
449 | previous_value = value; |
450 | } |
451 | i += reps; |
452 | } |
453 | } |
454 | |
455 | static uint16_t BrotliReverseBits(size_t num_bits, uint16_t bits) { |
456 | static const size_t kLut[16] = { /* Pre-reversed 4-bit values. */ |
457 | 0x00, 0x08, 0x04, 0x0C, 0x02, 0x0A, 0x06, 0x0E, |
458 | 0x01, 0x09, 0x05, 0x0D, 0x03, 0x0B, 0x07, 0x0F |
459 | }; |
460 | size_t retval = kLut[bits & 0x0F]; |
461 | size_t i; |
462 | for (i = 4; i < num_bits; i += 4) { |
463 | retval <<= 4; |
464 | bits = (uint16_t)(bits >> 4); |
465 | retval |= kLut[bits & 0x0F]; |
466 | } |
467 | retval >>= ((0 - num_bits) & 0x03); |
468 | return (uint16_t)retval; |
469 | } |
470 | |
471 | /* 0..15 are values for bits */ |
472 | #define MAX_HUFFMAN_BITS 16 |
473 | |
474 | void BrotliConvertBitDepthsToSymbols(const uint8_t* depth, |
475 | size_t len, |
476 | uint16_t* bits) { |
477 | /* In Brotli, all bit depths are [1..15] |
478 | 0 bit depth means that the symbol does not exist. */ |
479 | uint16_t bl_count[MAX_HUFFMAN_BITS] = { 0 }; |
480 | uint16_t next_code[MAX_HUFFMAN_BITS]; |
481 | size_t i; |
482 | int code = 0; |
483 | for (i = 0; i < len; ++i) { |
484 | ++bl_count[depth[i]]; |
485 | } |
486 | bl_count[0] = 0; |
487 | next_code[0] = 0; |
488 | for (i = 1; i < MAX_HUFFMAN_BITS; ++i) { |
489 | code = (code + bl_count[i - 1]) << 1; |
490 | next_code[i] = (uint16_t)code; |
491 | } |
492 | for (i = 0; i < len; ++i) { |
493 | if (depth[i]) { |
494 | bits[i] = BrotliReverseBits(depth[i], next_code[depth[i]]++); |
495 | } |
496 | } |
497 | } |
498 | |
499 | #if defined(__cplusplus) || defined(c_plusplus) |
500 | } /* extern "C" */ |
501 | #endif |
502 | |