|
1
|
/* inftrees.c -- generate Huffman trees for efficient decoding |
|
2
|
* Copyright (C) 1995-2026 Mark Adler |
|
3
|
* For conditions of distribution and use, see copyright notice in zlib.h |
|
4
|
*/ |
|
5
|
|
|
6
|
#ifdef MAKEFIXED |
|
7
|
# ifndef BUILDFIXED |
|
8
|
# define BUILDFIXED |
|
9
|
# endif |
|
10
|
#endif |
|
11
|
#ifdef BUILDFIXED |
|
12
|
# define Z_ONCE |
|
13
|
#endif |
|
14
|
|
|
15
|
#include "zutil.h" |
|
16
|
#include "inftrees.h" |
|
17
|
#include "inflate.h" |
|
18
|
|
|
19
|
#ifndef NULL |
|
20
|
# define NULL 0 |
|
21
|
#endif |
|
22
|
|
|
23
|
#define MAXBITS 15 |
|
24
|
|
|
25
|
const char inflate_copyright[] = |
|
26
|
" inflate 1.3.2 Copyright 1995-2026 Mark Adler "; |
|
27
|
/* |
|
28
|
If you use the zlib library in a product, an acknowledgment is welcome |
|
29
|
in the documentation of your product. If for some reason you cannot |
|
30
|
include such an acknowledgment, I would appreciate that you keep this |
|
31
|
copyright string in the executable of your product. |
|
32
|
*/ |
|
33
|
|
|
34
|
/* |
|
35
|
Build a set of tables to decode the provided canonical Huffman code. |
|
36
|
The code lengths are lens[0..codes-1]. The result starts at *table, |
|
37
|
whose indices are 0..2^bits-1. work is a writable array of at least |
|
38
|
lens shorts, which is used as a work area. type is the type of code |
|
39
|
to be generated, CODES, LENS, or DISTS. On return, zero is success, |
|
40
|
-1 is an invalid code, and +1 means that ENOUGH isn't enough. table |
|
41
|
on return points to the next available entry's address. bits is the |
|
42
|
requested root table index bits, and on return it is the actual root |
|
43
|
table index bits. It will differ if the request is greater than the |
|
44
|
longest code or if it is less than the shortest code. |
|
45
|
*/ |
|
46
|
int ZLIB_INTERNAL inflate_table(codetype type, unsigned short FAR *lens, |
|
47
|
unsigned codes, code FAR * FAR *table, |
|
48
|
unsigned FAR *bits, unsigned short FAR *work) { |
|
49
|
unsigned len; /* a code's length in bits */ |
|
50
|
unsigned sym; /* index of code symbols */ |
|
51
|
unsigned min, max; /* minimum and maximum code lengths */ |
|
52
|
unsigned root; /* number of index bits for root table */ |
|
53
|
unsigned curr; /* number of index bits for current table */ |
|
54
|
unsigned drop; /* code bits to drop for sub-table */ |
|
55
|
int left; /* number of prefix codes available */ |
|
56
|
unsigned used; /* code entries in table used */ |
|
57
|
unsigned huff; /* Huffman code */ |
|
58
|
unsigned incr; /* for incrementing code, index */ |
|
59
|
unsigned fill; /* index for replicating entries */ |
|
60
|
unsigned low; /* low bits for current root entry */ |
|
61
|
unsigned mask; /* mask for low root bits */ |
|
62
|
code here; /* table entry for duplication */ |
|
63
|
code FAR *next; /* next available space in table */ |
|
64
|
const unsigned short FAR *base = NULL; /* base value table to use */ |
|
65
|
const unsigned short FAR *extra = NULL; /* extra bits table to use */ |
|
66
|
unsigned match = 0; /* use base and extra for symbol >= match */ |
|
67
|
unsigned short count[MAXBITS+1]; /* number of codes of each length */ |
|
68
|
unsigned short offs[MAXBITS+1]; /* offsets in table for each length */ |
|
69
|
static const unsigned short lbase[31] = { /* Length codes 257..285 base */ |
|
70
|
3, 4, 5, 6, 7, 8, 9, 10, 11, 13, 15, 17, 19, 23, 27, 31, |
|
71
|
35, 43, 51, 59, 67, 83, 99, 115, 131, 163, 195, 227, 258, 0, 0}; |
|
72
|
static const unsigned short lext[31] = { /* Length codes 257..285 extra */ |
|
73
|
16, 16, 16, 16, 16, 16, 16, 16, 17, 17, 17, 17, 18, 18, 18, 18, |
|
74
|
19, 19, 19, 19, 20, 20, 20, 20, 21, 21, 21, 21, 16, 199, 75}; |
|
75
|
static const unsigned short dbase[32] = { /* Distance codes 0..29 base */ |
|
76
|
1, 2, 3, 4, 5, 7, 9, 13, 17, 25, 33, 49, 65, 97, 129, 193, |
|
77
|
257, 385, 513, 769, 1025, 1537, 2049, 3073, 4097, 6145, |
|
78
|
8193, 12289, 16385, 24577, 0, 0}; |
|
79
|
static const unsigned short dext[32] = { /* Distance codes 0..29 extra */ |
|
80
|
16, 16, 16, 16, 17, 17, 18, 18, 19, 19, 20, 20, 21, 21, 22, 22, |
|
81
|
23, 23, 24, 24, 25, 25, 26, 26, 27, 27, |
|
82
|
28, 28, 29, 29, 64, 64}; |
|
83
|
|
|
84
|
/* |
|
85
|
Process a set of code lengths to create a canonical Huffman code. The |
|
86
|
code lengths are lens[0..codes-1]. Each length corresponds to the |
|
87
|
symbols 0..codes-1. The Huffman code is generated by first sorting the |
|
88
|
symbols by length from short to long, and retaining the symbol order |
|
89
|
for codes with equal lengths. Then the code starts with all zero bits |
|
90
|
for the first code of the shortest length, and the codes are integer |
|
91
|
increments for the same length, and zeros are appended as the length |
|
92
|
increases. For the deflate format, these bits are stored backwards |
|
93
|
from their more natural integer increment ordering, and so when the |
|
94
|
decoding tables are built in the large loop below, the integer codes |
|
95
|
are incremented backwards. |
|
96
|
|
|
97
|
This routine assumes, but does not check, that all of the entries in |
|
98
|
lens[] are in the range 0..MAXBITS. The caller must assure this. |
|
99
|
1..MAXBITS is interpreted as that code length. zero means that that |
|
100
|
symbol does not occur in this code. |
|
101
|
|
|
102
|
The codes are sorted by computing a count of codes for each length, |
|
103
|
creating from that a table of starting indices for each length in the |
|
104
|
sorted table, and then entering the symbols in order in the sorted |
|
105
|
table. The sorted table is work[], with that space being provided by |
|
106
|
the caller. |
|
107
|
|
|
108
|
The length counts are used for other purposes as well, i.e. finding |
|
109
|
the minimum and maximum length codes, determining if there are any |
|
110
|
codes at all, checking for a valid set of lengths, and looking ahead |
|
111
|
at length counts to determine sub-table sizes when building the |
|
112
|
decoding tables. |
|
113
|
*/ |
|
114
|
|
|
115
|
/* accumulate lengths for codes (assumes lens[] all in 0..MAXBITS) */ |
|
116
|
for (len = 0; len <= MAXBITS; len++) |
|
117
|
count[len] = 0; |
|
118
|
for (sym = 0; sym < codes; sym++) |
|
119
|
count[lens[sym]]++; |
|
120
|
|
|
121
|
/* bound code lengths, force root to be within code lengths */ |
|
122
|
root = *bits; |
|
123
|
for (max = MAXBITS; max >= 1; max--) |
|
124
|
if (count[max] != 0) break; |
|
125
|
if (root > max) root = max; |
|
126
|
if (max == 0) { /* no symbols to code at all */ |
|
127
|
here.op = (unsigned char)64; /* invalid code marker */ |
|
128
|
here.bits = (unsigned char)1; |
|
129
|
here.val = (unsigned short)0; |
|
130
|
*(*table)++ = here; /* make a table to force an error */ |
|
131
|
*(*table)++ = here; |
|
132
|
*bits = 1; |
|
133
|
return 0; /* no symbols, but wait for decoding to report error */ |
|
134
|
} |
|
135
|
for (min = 1; min < max; min++) |
|
136
|
if (count[min] != 0) break; |
|
137
|
if (root < min) root = min; |
|
138
|
|
|
139
|
/* check for an over-subscribed or incomplete set of lengths */ |
|
140
|
left = 1; |
|
141
|
for (len = 1; len <= MAXBITS; len++) { |
|
142
|
left <<= 1; |
|
143
|
left -= count[len]; |
|
144
|
if (left < 0) return -1; /* over-subscribed */ |
|
145
|
} |
|
146
|
if (left > 0 && (type == CODES || max != 1)) |
|
147
|
return -1; /* incomplete set */ |
|
148
|
|
|
149
|
/* generate offsets into symbol table for each length for sorting */ |
|
150
|
offs[1] = 0; |
|
151
|
for (len = 1; len < MAXBITS; len++) |
|
152
|
offs[len + 1] = offs[len] + count[len]; |
|
153
|
|
|
154
|
/* sort symbols by length, by symbol order within each length */ |
|
155
|
for (sym = 0; sym < codes; sym++) |
|
156
|
if (lens[sym] != 0) work[offs[lens[sym]]++] = (unsigned short)sym; |
|
157
|
|
|
158
|
/* |
|
159
|
Create and fill in decoding tables. In this loop, the table being |
|
160
|
filled is at next and has curr index bits. The code being used is huff |
|
161
|
with length len. That code is converted to an index by dropping drop |
|
162
|
bits off of the bottom. For codes where len is less than drop + curr, |
|
163
|
those top drop + curr - len bits are incremented through all values to |
|
164
|
fill the table with replicated entries. |
|
165
|
|
|
166
|
root is the number of index bits for the root table. When len exceeds |
|
167
|
root, sub-tables are created pointed to by the root entry with an index |
|
168
|
of the low root bits of huff. This is saved in low to check for when a |
|
169
|
new sub-table should be started. drop is zero when the root table is |
|
170
|
being filled, and drop is root when sub-tables are being filled. |
|
171
|
|
|
172
|
When a new sub-table is needed, it is necessary to look ahead in the |
|
173
|
code lengths to determine what size sub-table is needed. The length |
|
174
|
counts are used for this, and so count[] is decremented as codes are |
|
175
|
entered in the tables. |
|
176
|
|
|
177
|
used keeps track of how many table entries have been allocated from the |
|
178
|
provided *table space. It is checked for LENS and DIST tables against |
|
179
|
the constants ENOUGH_LENS and ENOUGH_DISTS to guard against changes in |
|
180
|
the initial root table size constants. See the comments in inftrees.h |
|
181
|
for more information. |
|
182
|
|
|
183
|
sym increments through all symbols, and the loop terminates when |
|
184
|
all codes of length max, i.e. all codes, have been processed. This |
|
185
|
routine permits incomplete codes, so another loop after this one fills |
|
186
|
in the rest of the decoding tables with invalid code markers. |
|
187
|
*/ |
|
188
|
|
|
189
|
/* set up for code type */ |
|
190
|
switch (type) { |
|
191
|
case CODES: |
|
192
|
match = 20; |
|
193
|
break; |
|
194
|
case LENS: |
|
195
|
base = lbase; |
|
196
|
extra = lext; |
|
197
|
match = 257; |
|
198
|
break; |
|
199
|
case DISTS: |
|
200
|
base = dbase; |
|
201
|
extra = dext; |
|
202
|
} |
|
203
|
|
|
204
|
/* initialize state for loop */ |
|
205
|
huff = 0; /* starting code */ |
|
206
|
sym = 0; /* starting code symbol */ |
|
207
|
len = min; /* starting code length */ |
|
208
|
next = *table; /* current table to fill in */ |
|
209
|
curr = root; /* current table index bits */ |
|
210
|
drop = 0; /* current bits to drop from code for index */ |
|
211
|
low = (unsigned)(-1); /* trigger new sub-table when len > root */ |
|
212
|
used = 1U << root; /* use root table entries */ |
|
213
|
mask = used - 1; /* mask for comparing low */ |
|
214
|
|
|
215
|
/* check available table space */ |
|
216
|
if ((type == LENS && used > ENOUGH_LENS) || |
|
217
|
(type == DISTS && used > ENOUGH_DISTS)) |
|
218
|
return 1; |
|
219
|
|
|
220
|
/* process all codes and make table entries */ |
|
221
|
for (;;) { |
|
222
|
/* create table entry */ |
|
223
|
here.bits = (unsigned char)(len - drop); |
|
224
|
if (work[sym] + 1U < match) { |
|
225
|
here.op = (unsigned char)0; |
|
226
|
here.val = work[sym]; |
|
227
|
} |
|
228
|
else if (work[sym] >= match) { |
|
229
|
here.op = (unsigned char)(extra[work[sym] - match]); |
|
230
|
here.val = base[work[sym] - match]; |
|
231
|
} |
|
232
|
else { |
|
233
|
here.op = (unsigned char)(32 + 64); /* end of block */ |
|
234
|
here.val = 0; |
|
235
|
} |
|
236
|
|
|
237
|
/* replicate for those indices with low len bits equal to huff */ |
|
238
|
incr = 1U << (len - drop); |
|
239
|
fill = 1U << curr; |
|
240
|
min = fill; /* save offset to next table */ |
|
241
|
do { |
|
242
|
fill -= incr; |
|
243
|
next[(huff >> drop) + fill] = here; |
|
244
|
} while (fill != 0); |
|
245
|
|
|
246
|
/* backwards increment the len-bit code huff */ |
|
247
|
incr = 1U << (len - 1); |
|
248
|
while (huff & incr) |
|
249
|
incr >>= 1; |
|
250
|
if (incr != 0) { |
|
251
|
huff &= incr - 1; |
|
252
|
huff += incr; |
|
253
|
} |
|
254
|
else |
|
255
|
huff = 0; |
|
256
|
|
|
257
|
/* go to next symbol, update count, len */ |
|
258
|
sym++; |
|
259
|
if (--(count[len]) == 0) { |
|
260
|
if (len == max) break; |
|
261
|
len = lens[work[sym]]; |
|
262
|
} |
|
263
|
|
|
264
|
/* create new sub-table if needed */ |
|
265
|
if (len > root && (huff & mask) != low) { |
|
266
|
/* if first time, transition to sub-tables */ |
|
267
|
if (drop == 0) |
|
268
|
drop = root; |
|
269
|
|
|
270
|
/* increment past last table */ |
|
271
|
next += min; /* here min is 1 << curr */ |
|
272
|
|
|
273
|
/* determine length of next table */ |
|
274
|
curr = len - drop; |
|
275
|
left = (int)(1 << curr); |
|
276
|
while (curr + drop < max) { |
|
277
|
left -= count[curr + drop]; |
|
278
|
if (left <= 0) break; |
|
279
|
curr++; |
|
280
|
left <<= 1; |
|
281
|
} |
|
282
|
|
|
283
|
/* check for enough space */ |
|
284
|
used += 1U << curr; |
|
285
|
if ((type == LENS && used > ENOUGH_LENS) || |
|
286
|
(type == DISTS && used > ENOUGH_DISTS)) |
|
287
|
return 1; |
|
288
|
|
|
289
|
/* point entry in root table to sub-table */ |
|
290
|
low = huff & mask; |
|
291
|
(*table)[low].op = (unsigned char)curr; |
|
292
|
(*table)[low].bits = (unsigned char)root; |
|
293
|
(*table)[low].val = (unsigned short)(next - *table); |
|
294
|
} |
|
295
|
} |
|
296
|
|
|
297
|
/* fill in remaining table entry if code is incomplete (guaranteed to have |
|
298
|
at most one remaining entry, since if the code is incomplete, the |
|
299
|
maximum code length that was allowed to get this far is one bit) */ |
|
300
|
if (huff != 0) { |
|
301
|
here.op = (unsigned char)64; /* invalid code marker */ |
|
302
|
here.bits = (unsigned char)(len - drop); |
|
303
|
here.val = (unsigned short)0; |
|
304
|
next[huff] = here; |
|
305
|
} |
|
306
|
|
|
307
|
/* set return parameters */ |
|
308
|
*table += used; |
|
309
|
*bits = root; |
|
310
|
return 0; |
|
311
|
} |
|
312
|
|
|
313
|
#ifdef BUILDFIXED |
|
314
|
/* |
|
315
|
If this is compiled with BUILDFIXED defined, and if inflate will be used in |
|
316
|
multiple threads, and if atomics are not available, then inflate() must be |
|
317
|
called with a fixed block (e.g. 0x03 0x00) to initialize the tables and must |
|
318
|
return before any other threads are allowed to call inflate. |
|
319
|
*/ |
|
320
|
|
|
321
|
static code *lenfix, *distfix; |
|
322
|
static code fixed[544]; |
|
323
|
|
|
324
|
/* State for z_once(). */ |
|
325
|
local z_once_t built = Z_ONCE_INIT; |
|
326
|
|
|
327
|
local void buildtables(void) { |
|
328
|
unsigned sym, bits; |
|
329
|
static code *next; |
|
330
|
unsigned short lens[288], work[288]; |
|
331
|
|
|
332
|
/* literal/length table */ |
|
333
|
sym = 0; |
|
334
|
while (sym < 144) lens[sym++] = 8; |
|
335
|
while (sym < 256) lens[sym++] = 9; |
|
336
|
while (sym < 280) lens[sym++] = 7; |
|
337
|
while (sym < 288) lens[sym++] = 8; |
|
338
|
next = fixed; |
|
339
|
lenfix = next; |
|
340
|
bits = 9; |
|
341
|
inflate_table(LENS, lens, 288, &(next), &(bits), work); |
|
342
|
|
|
343
|
/* distance table */ |
|
344
|
sym = 0; |
|
345
|
while (sym < 32) lens[sym++] = 5; |
|
346
|
distfix = next; |
|
347
|
bits = 5; |
|
348
|
inflate_table(DISTS, lens, 32, &(next), &(bits), work); |
|
349
|
} |
|
350
|
#else /* !BUILDFIXED */ |
|
351
|
# include "inffixed.h" |
|
352
|
#endif /* BUILDFIXED */ |
|
353
|
|
|
354
|
/* |
|
355
|
Return state with length and distance decoding tables and index sizes set to |
|
356
|
fixed code decoding. Normally this returns fixed tables from inffixed.h. |
|
357
|
If BUILDFIXED is defined, then instead this routine builds the tables the |
|
358
|
first time it's called, and returns those tables the first time and |
|
359
|
thereafter. This reduces the size of the code by about 2K bytes, in |
|
360
|
exchange for a little execution time. However, BUILDFIXED should not be |
|
361
|
used for threaded applications if atomics are not available, as it will |
|
362
|
not be thread-safe. |
|
363
|
*/ |
|
364
|
void inflate_fixed(struct inflate_state FAR *state) { |
|
365
|
#ifdef BUILDFIXED |
|
366
|
z_once(&built, buildtables); |
|
367
|
#endif /* BUILDFIXED */ |
|
368
|
state->lencode = lenfix; |
|
369
|
state->lenbits = 9; |
|
370
|
state->distcode = distfix; |
|
371
|
state->distbits = 5; |
|
372
|
} |
|
373
|
|
|
374
|
#ifdef MAKEFIXED |
|
375
|
#include <stdio.h> |
|
376
|
|
|
377
|
/* |
|
378
|
Write out the inffixed.h that will be #include'd above. Defining MAKEFIXED |
|
379
|
also defines BUILDFIXED, so the tables are built on the fly. main() writes |
|
380
|
those tables to stdout, which would directed to inffixed.h. Compile this |
|
381
|
along with zutil.c: |
|
382
|
|
|
383
|
cc -DMAKEFIXED -o fix inftrees.c zutil.c |
|
384
|
./fix > inffixed.h |
|
385
|
*/ |
|
386
|
int main(void) { |
|
387
|
unsigned low, size; |
|
388
|
struct inflate_state state; |
|
389
|
|
|
390
|
inflate_fixed(&state); |
|
391
|
puts("/* inffixed.h -- table for decoding fixed codes"); |
|
392
|
puts(" * Generated automatically by makefixed()."); |
|
393
|
puts(" */"); |
|
394
|
puts(""); |
|
395
|
puts("/* WARNING: this file should *not* be used by applications."); |
|
396
|
puts(" It is part of the implementation of this library and is"); |
|
397
|
puts(" subject to change. Applications should only use zlib.h."); |
|
398
|
puts(" */"); |
|
399
|
puts(""); |
|
400
|
size = 1U << 9; |
|
401
|
printf("static const code lenfix[%u] = {", size); |
|
402
|
low = 0; |
|
403
|
for (;;) { |
|
404
|
if ((low % 7) == 0) printf("\n "); |
|
405
|
printf("{%u,%u,%d}", (low & 127) == 99 ? 64 : state.lencode[low].op, |
|
406
|
state.lencode[low].bits, state.lencode[low].val); |
|
407
|
if (++low == size) break; |
|
408
|
putchar(','); |
|
409
|
} |
|
410
|
puts("\n};"); |
|
411
|
size = 1U << 5; |
|
412
|
printf("\nstatic const code distfix[%u] = {", size); |
|
413
|
low = 0; |
|
414
|
for (;;) { |
|
415
|
if ((low % 6) == 0) printf("\n "); |
|
416
|
printf("{%u,%u,%d}", state.distcode[low].op, state.distcode[low].bits, |
|
417
|
state.distcode[low].val); |
|
418
|
if (++low == size) break; |
|
419
|
putchar(','); |
|
420
|
} |
|
421
|
puts("\n};"); |
|
422
|
return 0; |
|
423
|
} |
|
424
|
#endif /* MAKEFIXED */ |
|
425
|
|