1 |
/* |
2 |
* Copyright (c) 2000-2012 Marc Alexander Lehmann <schmorp@schmorp.de> |
3 |
* |
4 |
* Redistribution and use in source and binary forms, with or without modifica- |
5 |
* tion, are permitted provided that the following conditions are met: |
6 |
* |
7 |
* 1. Redistributions of source code must retain the above copyright notice, |
8 |
* this list of conditions and the following disclaimer. |
9 |
* |
10 |
* 2. Redistributions in binary form must reproduce the above copyright |
11 |
* notice, this list of conditions and the following disclaimer in the |
12 |
* documentation and/or other materials provided with the distribution. |
13 |
* |
14 |
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR IMPLIED |
15 |
* WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MER- |
16 |
* CHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO |
17 |
* EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPE- |
18 |
* CIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
19 |
* PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; |
20 |
* OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, |
21 |
* WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTH- |
22 |
* ERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED |
23 |
* OF THE POSSIBILITY OF SUCH DAMAGE. |
24 |
* |
25 |
* Alternatively, the contents of this file may be used under the terms of |
26 |
* the GNU General Public License ("GPL") version 2 or any later version, |
27 |
* in which case the provisions of the GPL are applicable instead of |
28 |
* the above. If you wish to allow the use of your version of this file |
29 |
* only under the terms of the GPL and not to allow others to use your |
30 |
* version of this file under the BSD license, indicate your decision |
31 |
* by deleting the provisions above and replace them with the notice |
32 |
* and other provisions required by the GPL. If you do not delete the |
33 |
* provisions above, a recipient may use your version of this file under |
34 |
* either the BSD or the GPL. |
35 |
*/ |
36 |
|
37 |
#include "lzfP.h" |
38 |
|
39 |
#define HASH(p) (p[0] << 6) ^ (p[1] << 3) ^ p[2] |
40 |
|
41 |
#define MAX_LIT (1 << 5) |
42 |
#define MAX_OFF (1 << 13) |
43 |
#define MAX_REF ((1 << 8) + (1 << 3)) |
44 |
|
45 |
#if __GNUC__ >= 3 |
46 |
# define expect(expr,value) __builtin_expect ((expr),(value)) |
47 |
# define inline inline |
48 |
#else |
49 |
# define expect(expr,value) (expr) |
50 |
# define inline static |
51 |
#endif |
52 |
|
53 |
#define expect_false(expr) expect ((expr) != 0, 0) |
54 |
#define expect_true(expr) expect ((expr) != 0, 1) |
55 |
|
56 |
/* |
57 |
* compressed format |
58 |
* |
59 |
* 000LLLLL <L+1> ; literal, L+1=1..33 octets |
60 |
* LLLooooo oooooooo ; backref L+1=1..7 octets, o+1=1..4096 offset |
61 |
* 111ooooo LLLLLLLL oooooooo ; backref L+8 octets, o+1=1..4096 offset |
62 |
* |
63 |
*/ |
64 |
|
65 |
unsigned int |
66 |
lzf_compress_best (const void *const in_data, unsigned int in_len, |
67 |
void *out_data, unsigned int out_len) |
68 |
{ |
69 |
const u8 *ip = (const u8 *)in_data; |
70 |
u8 *op = (u8 *)out_data; |
71 |
const u8 *in_end = ip + in_len; |
72 |
u8 *out_end = op + out_len; |
73 |
|
74 |
const u8 *first [1 << (6+8)]; /* most recent occurance of a match */ |
75 |
u16 prev [MAX_OFF]; /* how many bytes to go backwards for te next match */ |
76 |
|
77 |
int lit; |
78 |
|
79 |
if (!in_len || !out_len) |
80 |
return 0; |
81 |
|
82 |
lit = 0; op++; /* start run */ |
83 |
|
84 |
lit++; *op++ = *ip++; |
85 |
|
86 |
while (ip < in_end - 2) |
87 |
{ |
88 |
int best_l = 0; |
89 |
const u8 *best_p; |
90 |
int e = (in_end - ip < MAX_REF ? in_end - ip : MAX_REF) - 1; |
91 |
unsigned int res = ((unsigned int)ip) & (MAX_OFF - 1); |
92 |
u16 hash = HASH (ip); |
93 |
u16 diff; |
94 |
const u8 *b = ip < (u8 *)in_data + MAX_OFF ? in_data : ip - MAX_OFF; |
95 |
const u8 *p = first [hash]; |
96 |
prev [res] = ip - p; /* update ptr to previous hash match */ |
97 |
first [hash] = ip; /* first hash match is here */ |
98 |
|
99 |
if (p < ip) |
100 |
while (p >= b) |
101 |
{ |
102 |
if (p[2] == ip[2]) /* first two bytes almost always match */ |
103 |
if (p[best_l] == ip[best_l]) /* new match must be longer than the old one to qualify */ |
104 |
if (p[1] == ip[1] && p[0] == ip[0]) /* just to be sure */ |
105 |
{ |
106 |
int l = 3; |
107 |
|
108 |
while (p[l] == ip[l] && l < e) |
109 |
++l; |
110 |
|
111 |
if (l >= best_l) |
112 |
{ |
113 |
best_p = p; |
114 |
best_l = l; |
115 |
} |
116 |
} |
117 |
|
118 |
diff = prev [((unsigned int)p) & (MAX_OFF - 1)]; |
119 |
p = diff ? p - diff : 0; |
120 |
} |
121 |
|
122 |
if (best_l) |
123 |
{ |
124 |
int len = best_l; |
125 |
int off = ip - best_p - 1; |
126 |
|
127 |
if (expect_false (op + 3 + 1 >= out_end)) /* first a faster conservative test */ |
128 |
if (op - !lit + 3 + 1 >= out_end) /* second the exact but rare test */ |
129 |
return 0; |
130 |
|
131 |
op [- lit - 1] = lit - 1; /* stop run */ |
132 |
op -= !lit; /* undo run if length is zero */ |
133 |
|
134 |
len -= 2; /* len is now #octets - 1 */ |
135 |
ip++; |
136 |
|
137 |
if (len < 7) |
138 |
{ |
139 |
*op++ = (off >> 8) + (len << 5); |
140 |
} |
141 |
else |
142 |
{ |
143 |
*op++ = (off >> 8) + ( 7 << 5); |
144 |
*op++ = len - 7; |
145 |
} |
146 |
|
147 |
*op++ = off; |
148 |
|
149 |
lit = 0; op++; /* start run */ |
150 |
|
151 |
ip += len + 1; |
152 |
|
153 |
if (expect_false (ip >= in_end - 2)) |
154 |
break; |
155 |
|
156 |
ip -= len + 1; |
157 |
|
158 |
//printf (" fill %p for %d\n", ip, len);//D |
159 |
do |
160 |
{ |
161 |
u16 hash = HASH (ip); |
162 |
res = ((unsigned int)ip) & (MAX_OFF - 1); |
163 |
|
164 |
p = first [hash]; |
165 |
prev [res] = ip - p; /* update ptr to previous hash match */ |
166 |
first [hash] = ip; /* first hash match is here */ |
167 |
|
168 |
ip++; |
169 |
} |
170 |
while (len--); |
171 |
} |
172 |
else |
173 |
{ |
174 |
/* one more literal byte we must copy */ |
175 |
if (expect_false (op >= out_end)) |
176 |
return 0; |
177 |
|
178 |
lit++; *op++ = *ip++; |
179 |
|
180 |
if (expect_false (lit == MAX_LIT)) |
181 |
{ |
182 |
op [- lit - 1] = lit - 1; /* stop run */ |
183 |
lit = 0; op++; /* start run */ |
184 |
} |
185 |
} |
186 |
} |
187 |
|
188 |
if (op + 3 > out_end) /* at most 3 bytes can be missing here */ |
189 |
return 0; |
190 |
|
191 |
while (ip < in_end) |
192 |
{ |
193 |
lit++; *op++ = *ip++; |
194 |
|
195 |
if (expect_false (lit == MAX_LIT)) |
196 |
{ |
197 |
op [- lit - 1] = lit - 1; /* stop run */ |
198 |
lit = 0; op++; /* start run */ |
199 |
} |
200 |
} |
201 |
|
202 |
op [- lit - 1] = lit - 1; /* end run */ |
203 |
op -= !lit; /* undo run if length is zero */ |
204 |
|
205 |
return op - (u8 *)out_data; |
206 |
} |
207 |
|