1 /* license:BSD-3-Clause
2 * copyright-holders:Aaron Giles
3 ****************************************************************************
4
5 huffman.c
6
7 Static Huffman compression and decompression helpers.
8
9 ****************************************************************************
10
11 Maximum codelength is officially (alphabetsize - 1). This would be 255 bits
12 (since we use 1 byte values). However, it is also dependent upon the number
13 of samples used, as follows:
14
15 2 bits -> 3..4 samples
16 3 bits -> 5..7 samples
17 4 bits -> 8..12 samples
18 5 bits -> 13..20 samples
19 6 bits -> 21..33 samples
20 7 bits -> 34..54 samples
21 8 bits -> 55..88 samples
22 9 bits -> 89..143 samples
23 10 bits -> 144..232 samples
24 11 bits -> 233..376 samples
25 12 bits -> 377..609 samples
26 13 bits -> 610..986 samples
27 14 bits -> 987..1596 samples
28 15 bits -> 1597..2583 samples
29 16 bits -> 2584..4180 samples -> note that a 4k data size guarantees codelength <= 16 bits
30 17 bits -> 4181..6764 samples
31 18 bits -> 6765..10945 samples
32 19 bits -> 10946..17710 samples
33 20 bits -> 17711..28656 samples
34 21 bits -> 28657..46367 samples
35 22 bits -> 46368..75024 samples
36 23 bits -> 75025..121392 samples
37 24 bits -> 121393..196417 samples
38 25 bits -> 196418..317810 samples
39 26 bits -> 317811..514228 samples
40 27 bits -> 514229..832039 samples
41 28 bits -> 832040..1346268 samples
42 29 bits -> 1346269..2178308 samples
43 30 bits -> 2178309..3524577 samples
44 31 bits -> 3524578..5702886 samples
45 32 bits -> 5702887..9227464 samples
46
47 Looking at it differently, here is where powers of 2 fall into these buckets:
48
49 256 samples -> 11 bits max
50 512 samples -> 12 bits max
51 1k samples -> 14 bits max
52 2k samples -> 15 bits max
53 4k samples -> 16 bits max
54 8k samples -> 18 bits max
55 16k samples -> 19 bits max
56 32k samples -> 21 bits max
57 64k samples -> 22 bits max
58 128k samples -> 24 bits max
59 256k samples -> 25 bits max
60 512k samples -> 27 bits max
61 1M samples -> 28 bits max
62 2M samples -> 29 bits max
63 4M samples -> 31 bits max
64 8M samples -> 32 bits max
65
66 ****************************************************************************
67
68 Delta-RLE encoding works as follows:
69
70 Starting value is assumed to be 0. All data is encoded as a delta
71 from the previous value, such that final[i] = final[i - 1] + delta.
72 Long runs of 0s are RLE-encoded as follows:
73
74 0x100 = repeat count of 8
75 0x101 = repeat count of 9
76 0x102 = repeat count of 10
77 0x103 = repeat count of 11
78 0x104 = repeat count of 12
79 0x105 = repeat count of 13
80 0x106 = repeat count of 14
81 0x107 = repeat count of 15
82 0x108 = repeat count of 16
83 0x109 = repeat count of 32
84 0x10a = repeat count of 64
85 0x10b = repeat count of 128
86 0x10c = repeat count of 256
87 0x10d = repeat count of 512
88 0x10e = repeat count of 1024
89 0x10f = repeat count of 2048
90
91 Note that repeat counts are reset at the end of a row, so if a 0 run
92 extends to the end of a row, a large repeat count may be used.
93
94 The reason for starting the run counts at 8 is that 0 is expected to
95 be the most common symbol, and is typically encoded in 1 or 2 bits.
96
97 ***************************************************************************/
98
99 #include <stdlib.h>
100 #include <assert.h>
101 #include <stdio.h>
102 #include <string.h>
103
104 #include "huffman.h"
105
106 #define MAX(x,y) ((x) > (y) ? (x) : (y))
107
108 /***************************************************************************
109 * MACROS
110 ***************************************************************************
111 */
112
113 #define MAKE_LOOKUP(code,bits) (((code) << 5) | ((bits) & 0x1f))
114
115 /***************************************************************************
116 * IMPLEMENTATION
117 ***************************************************************************
118 */
119
120 /*-------------------------------------------------
121 * huffman_context_base - create an encoding/
122 * decoding context
123 *-------------------------------------------------
124 */
125
create_huffman_decoder(int numcodes,int maxbits)126 struct huffman_decoder* create_huffman_decoder(int numcodes, int maxbits)
127 {
128 struct huffman_decoder* decoder = NULL;
129
130 /* limit to 24 bits */
131 if (maxbits > 24)
132 return NULL;
133
134 decoder = (struct huffman_decoder*)malloc(sizeof(struct huffman_decoder));
135 decoder->numcodes = numcodes;
136 decoder->maxbits = maxbits;
137 decoder->lookup = (lookup_value*)malloc(sizeof(lookup_value) * (1 << maxbits));
138 decoder->huffnode = (struct node_t*)malloc(sizeof(struct node_t) * numcodes);
139 decoder->datahisto = NULL;
140 decoder->prevdata = 0;
141 decoder->rleremaining = 0;
142 return decoder;
143 }
144
145 /*-------------------------------------------------
146 * decode_one - decode a single code from the
147 * huffman stream
148 *-------------------------------------------------
149 */
150
huffman_decode_one(struct huffman_decoder * decoder,struct bitstream * bitbuf)151 uint32_t huffman_decode_one(struct huffman_decoder* decoder, struct bitstream* bitbuf)
152 {
153 /* peek ahead to get maxbits worth of data */
154 uint32_t bits = bitstream_peek(bitbuf, decoder->maxbits);
155
156 /* look it up, then remove the actual number of bits for this code */
157 lookup_value lookup = decoder->lookup[bits];
158 bitstream_remove(bitbuf, lookup & 0x1f);
159
160 /* return the value */
161 return lookup >> 5;
162 }
163
164 /*-------------------------------------------------
165 * import_tree_rle - import an RLE-encoded
166 * huffman tree from a source data stream
167 *-------------------------------------------------
168 */
169
huffman_import_tree_rle(struct huffman_decoder * decoder,struct bitstream * bitbuf)170 enum huffman_error huffman_import_tree_rle(struct huffman_decoder* decoder, struct bitstream* bitbuf)
171 {
172 int numbits, curnode;
173 enum huffman_error error;
174
175 /* bits per entry depends on the maxbits */
176 if (decoder->maxbits >= 16)
177 numbits = 5;
178 else if (decoder->maxbits >= 8)
179 numbits = 4;
180 else
181 numbits = 3;
182
183 /* loop until we read all the nodes */
184 for (curnode = 0; curnode < decoder->numcodes; )
185 {
186 /* a non-one value is just raw */
187 int nodebits = bitstream_read(bitbuf, numbits);
188 if (nodebits != 1)
189 decoder->huffnode[curnode++].numbits = nodebits;
190
191 /* a one value is an escape code */
192 else
193 {
194 /* a double 1 is just a single 1 */
195 nodebits = bitstream_read(bitbuf, numbits);
196 if (nodebits == 1)
197 decoder->huffnode[curnode++].numbits = nodebits;
198
199 /* otherwise, we need one for value for the repeat count */
200 else
201 {
202 int repcount = bitstream_read(bitbuf, numbits) + 3;
203 while (repcount--)
204 decoder->huffnode[curnode++].numbits = nodebits;
205 }
206 }
207 }
208
209 /* make sure we ended up with the right number */
210 if (curnode != decoder->numcodes)
211 return HUFFERR_INVALID_DATA;
212
213 /* assign canonical codes for all nodes based on their code lengths */
214 error = huffman_assign_canonical_codes(decoder);
215 if (error != HUFFERR_NONE)
216 return error;
217
218 /* build the lookup table */
219 huffman_build_lookup_table(decoder);
220
221 /* determine final input length and report errors */
222 return bitstream_overflow(bitbuf) ? HUFFERR_INPUT_BUFFER_TOO_SMALL : HUFFERR_NONE;
223 }
224
225
226 /*-------------------------------------------------
227 * import_tree_huffman - import a huffman-encoded
228 * huffman tree from a source data stream
229 *-------------------------------------------------
230 */
231
huffman_import_tree_huffman(struct huffman_decoder * decoder,struct bitstream * bitbuf)232 enum huffman_error huffman_import_tree_huffman(struct huffman_decoder* decoder, struct bitstream* bitbuf)
233 {
234 int start;
235 int last = 0;
236 int count = 0;
237 int index;
238 int curcode;
239 uint8_t rlefullbits = 0;
240 uint32_t temp;
241 enum huffman_error error;
242 /* start by parsing the lengths for the small tree */
243 struct huffman_decoder* smallhuff = create_huffman_decoder(24, 6);
244 smallhuff->huffnode[0].numbits = bitstream_read(bitbuf, 3);
245 start = bitstream_read(bitbuf, 3) + 1;
246 for (index = 1; index < 24; index++)
247 {
248 if (index < start || count == 7)
249 smallhuff->huffnode[index].numbits = 0;
250 else
251 {
252 count = bitstream_read(bitbuf, 3);
253 smallhuff->huffnode[index].numbits = (count == 7) ? 0 : count;
254 }
255 }
256
257 /* then regenerate the tree */
258 error = huffman_assign_canonical_codes(smallhuff);
259 if (error != HUFFERR_NONE)
260 return error;
261 huffman_build_lookup_table(smallhuff);
262
263 /* determine the maximum length of an RLE count */
264 temp = decoder->numcodes - 9;
265 while (temp != 0)
266 temp >>= 1, rlefullbits++;
267
268 /* now process the rest of the data */
269 for (curcode = 0; curcode < decoder->numcodes; )
270 {
271 int value = huffman_decode_one(smallhuff, bitbuf);
272 if (value != 0)
273 decoder->huffnode[curcode++].numbits = last = value - 1;
274 else
275 {
276 int count = bitstream_read(bitbuf, 3) + 2;
277 if (count == 7+2)
278 count += bitstream_read(bitbuf, rlefullbits);
279 for ( ; count != 0 && curcode < decoder->numcodes; count--)
280 decoder->huffnode[curcode++].numbits = last;
281 }
282 }
283
284 /* make sure we ended up with the right number */
285 if (curcode != decoder->numcodes)
286 return HUFFERR_INVALID_DATA;
287
288 /* assign canonical codes for all nodes based on their code lengths */
289 error = huffman_assign_canonical_codes(decoder);
290 if (error != HUFFERR_NONE)
291 return error;
292
293 /* build the lookup table */
294 huffman_build_lookup_table(decoder);
295
296 /* determine final input length and report errors */
297 return bitstream_overflow(bitbuf) ? HUFFERR_INPUT_BUFFER_TOO_SMALL : HUFFERR_NONE;
298 }
299
300 /*-------------------------------------------------
301 * compute_tree_from_histo - common backend for
302 * computing a tree based on the data histogram
303 *-------------------------------------------------
304 */
305
huffman_compute_tree_from_histo(struct huffman_decoder * decoder)306 enum huffman_error huffman_compute_tree_from_histo(struct huffman_decoder* decoder)
307 {
308 int i;
309 uint32_t lowerweight;
310 uint32_t upperweight;
311 /* compute the number of data items in the histogram */
312 uint32_t sdatacount = 0;
313 for (i = 0; i < decoder->numcodes; i++)
314 sdatacount += decoder->datahisto[i];
315
316 /* binary search to achieve the optimum encoding */
317 lowerweight = 0;
318 upperweight = sdatacount * 2;
319 while (1)
320 {
321 /* build a tree using the current weight */
322 uint32_t curweight = (upperweight + lowerweight) / 2;
323 int curmaxbits = huffman_build_tree(decoder, sdatacount, curweight);
324
325 /* apply binary search here */
326 if (curmaxbits <= decoder->maxbits)
327 {
328 lowerweight = curweight;
329
330 /* early out if it worked with the raw weights, or if we're done searching */
331 if (curweight == sdatacount || (upperweight - lowerweight) <= 1)
332 break;
333 }
334 else
335 upperweight = curweight;
336 }
337
338 /* assign canonical codes for all nodes based on their code lengths */
339 return huffman_assign_canonical_codes(decoder);
340 }
341
342 /***************************************************************************
343 * INTERNAL FUNCTIONS
344 ***************************************************************************
345 */
346
347 /*-------------------------------------------------
348 * tree_node_compare - compare two tree nodes
349 * by weight
350 *-------------------------------------------------
351 */
352
huffman_tree_node_compare(const void * item1,const void * item2)353 static int huffman_tree_node_compare(const void *item1, const void *item2)
354 {
355 const struct node_t *node1 = *(const struct node_t **)item1;
356 const struct node_t *node2 = *(const struct node_t **)item2;
357 if (node2->weight != node1->weight)
358 return node2->weight - node1->weight;
359 if (node2->bits - node1->bits == 0)
360 fprintf(stderr, "identical node sort keys, should not happen!\n");
361 return (int)node1->bits - (int)node2->bits;
362 }
363
364 /*-------------------------------------------------
365 * build_tree - build a huffman tree based on the
366 * data distribution
367 *-------------------------------------------------
368 */
369
huffman_build_tree(struct huffman_decoder * decoder,uint32_t totaldata,uint32_t totalweight)370 int huffman_build_tree(struct huffman_decoder* decoder, uint32_t totaldata, uint32_t totalweight)
371 {
372 int curcode;
373 int nextalloc;
374 int listitems = 0;
375 int maxbits = 0;
376 /* make a list of all non-zero nodes */
377 struct node_t** list = (struct node_t**)malloc(sizeof(struct node_t*) * decoder->numcodes * 2);
378 memset(decoder->huffnode, 0, decoder->numcodes * sizeof(decoder->huffnode[0]));
379 for (curcode = 0; curcode < decoder->numcodes; curcode++)
380 if (decoder->datahisto[curcode] != 0)
381 {
382 list[listitems++] = &decoder->huffnode[curcode];
383 decoder->huffnode[curcode].count = decoder->datahisto[curcode];
384 decoder->huffnode[curcode].bits = curcode;
385
386 /* scale the weight by the current effective length, ensuring we don't go to 0 */
387 decoder->huffnode[curcode].weight = ((uint64_t)decoder->datahisto[curcode]) * ((uint64_t)totalweight) / ((uint64_t)totaldata);
388 if (decoder->huffnode[curcode].weight == 0)
389 decoder->huffnode[curcode].weight = 1;
390 }
391
392 #if 0
393 fprintf(stderr, "Pre-sort:\n");
394 for (int i = 0; i < listitems; i++) {
395 fprintf(stderr, "weight: %d code: %d\n", list[i]->m_weight, list[i]->m_bits);
396 }
397 #endif
398
399 /* sort the list by weight, largest weight first */
400 qsort(&list[0], listitems, sizeof(list[0]), huffman_tree_node_compare);
401
402 #if 0
403 fprintf(stderr, "Post-sort:\n");
404 for (int i = 0; i < listitems; i++) {
405 fprintf(stderr, "weight: %d code: %d\n", list[i]->m_weight, list[i]->m_bits);
406 }
407 fprintf(stderr, "===================\n");
408 #endif
409
410 /* now build the tree */
411 nextalloc = decoder->numcodes;
412 while (listitems > 1)
413 {
414 /* remove lowest two items */
415 struct node_t* node1 = &(*list[--listitems]);
416 struct node_t* node0 = &(*list[--listitems]);
417
418 /* create new node */
419 struct node_t* newnode = &decoder->huffnode[nextalloc++];
420 newnode->parent = NULL;
421 node0->parent = node1->parent = newnode;
422 newnode->weight = node0->weight + node1->weight;
423
424 /* insert into list at appropriate location */
425 int curitem;
426 for (curitem = 0; curitem < listitems; curitem++)
427 if (newnode->weight > list[curitem]->weight)
428 {
429 memmove(&list[curitem+1], &list[curitem], (listitems - curitem) * sizeof(list[0]));
430 break;
431 }
432 list[curitem] = newnode;
433 listitems++;
434 }
435
436 /* compute the number of bits in each code, and fill in another histogram */
437 for (curcode = 0; curcode < decoder->numcodes; curcode++)
438 {
439 struct node_t* node = &decoder->huffnode[curcode];
440 node->numbits = 0;
441 node->bits = 0;
442
443 /* if we have a non-zero weight, compute the number of bits */
444 if (node->weight > 0)
445 {
446 /* determine the number of bits for this node */
447 for (struct node_t *curnode = node; curnode->parent != NULL; curnode = curnode->parent)
448 node->numbits++;
449 if (node->numbits == 0)
450 node->numbits = 1;
451
452 /* keep track of the max */
453 maxbits = MAX(maxbits, ((int)node->numbits));
454 }
455 }
456 return maxbits;
457 }
458
459 /*-------------------------------------------------
460 * assign_canonical_codes - assign canonical codes
461 * to all the nodes based on the number of bits
462 * in each
463 *-------------------------------------------------
464 */
465
huffman_assign_canonical_codes(struct huffman_decoder * decoder)466 enum huffman_error huffman_assign_canonical_codes(struct huffman_decoder* decoder)
467 {
468 int curcode, codelen;
469 uint32_t curstart = 0;
470 /* build up a histogram of bit lengths */
471 uint32_t bithisto[33] = { 0 };
472 for (curcode = 0; curcode < decoder->numcodes; curcode++)
473 {
474 struct node_t* node = &decoder->huffnode[curcode];
475 if (node->numbits > decoder->maxbits)
476 return HUFFERR_INTERNAL_INCONSISTENCY;
477 if (node->numbits <= 32)
478 bithisto[node->numbits]++;
479 }
480
481 /* for each code length, determine the starting code number */
482 for (codelen = 32; codelen > 0; codelen--)
483 {
484 uint32_t nextstart = (curstart + bithisto[codelen]) >> 1;
485 if (codelen != 1 && nextstart * 2 != (curstart + bithisto[codelen]))
486 return HUFFERR_INTERNAL_INCONSISTENCY;
487 bithisto[codelen] = curstart;
488 curstart = nextstart;
489 }
490
491 /* now assign canonical codes */
492 for (curcode = 0; curcode < decoder->numcodes; curcode++)
493 {
494 struct node_t* node = &decoder->huffnode[curcode];
495 if (node->numbits > 0)
496 node->bits = bithisto[node->numbits]++;
497 }
498 return HUFFERR_NONE;
499 }
500
501 /*-------------------------------------------------
502 * build_lookup_table - build a lookup table for
503 * fast decoding
504 *-------------------------------------------------
505 */
506
huffman_build_lookup_table(struct huffman_decoder * decoder)507 void huffman_build_lookup_table(struct huffman_decoder* decoder)
508 {
509 int curcode;
510 /* iterate over all codes */
511 for (curcode = 0; curcode < decoder->numcodes; curcode++)
512 {
513 /* process all nodes which have non-zero bits */
514 struct node_t* node = &decoder->huffnode[curcode];
515 if (node->numbits > 0)
516 {
517 /* set up the entry */
518 lookup_value value = MAKE_LOOKUP(curcode, node->numbits);
519
520 /* fill all matching entries */
521 int shift = decoder->maxbits - node->numbits;
522 lookup_value *dest = &decoder->lookup[node->bits << shift];
523 lookup_value *destend = &decoder->lookup[((node->bits + 1) << shift) - 1];
524 while (dest <= destend)
525 *dest++ = value;
526 }
527 }
528 }
529