1 |
root |
1.1 |
/* |
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 |
|
|
#if __GNUC__ >= 3 |
42 |
|
|
# define expect(expr,value) __builtin_expect ((expr),(value)) |
43 |
|
|
# define inline inline |
44 |
|
|
#else |
45 |
|
|
# define expect(expr,value) (expr) |
46 |
|
|
# define inline static |
47 |
|
|
#endif |
48 |
|
|
|
49 |
|
|
#define expect_false(expr) expect ((expr) != 0, 0) |
50 |
|
|
#define expect_true(expr) expect ((expr) != 0, 1) |
51 |
|
|
|
52 |
|
|
/* |
53 |
|
|
* compressed format |
54 |
|
|
* |
55 |
|
|
* 000LLLLL <L+1> ; literal, L+1=1..33 octets |
56 |
|
|
* LLLooooo oooooooo ; backref L+1=1..7 octets, o+1=1..4096 offset |
57 |
|
|
* 111ooooo LLLLLLLL oooooooo ; backref L+8 octets, o+1=1..4096 offset |
58 |
|
|
* |
59 |
|
|
*/ |
60 |
|
|
|
61 |
|
|
unsigned int |
62 |
|
|
lzf_compress_best (const void *const in_data, unsigned int in_len, |
63 |
root |
1.3 |
void *out_data, unsigned int out_len |
64 |
|
|
#if LZF_STATE_ARG |
65 |
|
|
, LZF_STATE_BEST state |
66 |
|
|
#endif |
67 |
|
|
) |
68 |
root |
1.1 |
{ |
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 |
root |
1.3 |
#if !LZF_STATE_ARG |
75 |
|
|
LZF_STATE_BEST state; |
76 |
|
|
#endif |
77 |
|
|
#define STATE state[0] |
78 |
root |
1.1 |
|
79 |
|
|
int lit; |
80 |
|
|
|
81 |
|
|
if (!in_len || !out_len) |
82 |
|
|
return 0; |
83 |
|
|
|
84 |
|
|
lit = 0; op++; /* start run */ |
85 |
|
|
|
86 |
|
|
lit++; *op++ = *ip++; |
87 |
|
|
|
88 |
|
|
while (ip < in_end - 2) |
89 |
|
|
{ |
90 |
|
|
int best_l = 0; |
91 |
|
|
const u8 *best_p; |
92 |
root |
1.3 |
int e = (in_end - ip < LZF_MAX_REF ? in_end - ip : LZF_MAX_REF) - 1; |
93 |
|
|
unsigned int res = ((unsigned long)ip) & (LZF_MAX_OFF - 1); |
94 |
root |
1.1 |
u16 hash = HASH (ip); |
95 |
|
|
u16 diff; |
96 |
root |
1.3 |
const u8 *b = ip < (u8 *)in_data + LZF_MAX_OFF ? in_data : ip - LZF_MAX_OFF; |
97 |
|
|
const u8 *p = STATE.first [hash]; |
98 |
|
|
STATE.prev [res] = ip - p; /* update ptr to previous hash match */ |
99 |
|
|
STATE.first [hash] = ip; /* first hash match is here */ |
100 |
root |
1.1 |
|
101 |
|
|
if (p < ip) |
102 |
root |
1.3 |
while (p >= b) |
103 |
|
|
{ |
104 |
|
|
if (p[2] == ip[2]) /* first two bytes almost always match */ |
105 |
|
|
if (p[best_l] == ip[best_l]) /* new match must be longer than the old one to qualify */ |
106 |
|
|
if (p[1] == ip[1] && p[0] == ip[0]) /* just to be sure */ |
107 |
|
|
{ |
108 |
|
|
int l = 3; |
109 |
|
|
|
110 |
|
|
while (p[l] == ip[l] && l < e) |
111 |
|
|
++l; |
112 |
|
|
|
113 |
|
|
if (l >= best_l) |
114 |
|
|
{ |
115 |
|
|
best_p = p; |
116 |
|
|
best_l = l; |
117 |
|
|
} |
118 |
|
|
} |
119 |
|
|
|
120 |
|
|
diff = STATE.prev [((unsigned long)p) & (LZF_MAX_OFF - 1)]; |
121 |
|
|
p = diff ? p - diff : 0; |
122 |
|
|
} |
123 |
root |
1.1 |
|
124 |
|
|
if (best_l) |
125 |
|
|
{ |
126 |
|
|
int len = best_l; |
127 |
|
|
int off = ip - best_p - 1; |
128 |
|
|
|
129 |
|
|
if (expect_false (op + 3 + 1 >= out_end)) /* first a faster conservative test */ |
130 |
|
|
if (op - !lit + 3 + 1 >= out_end) /* second the exact but rare test */ |
131 |
|
|
return 0; |
132 |
|
|
|
133 |
|
|
op [- lit - 1] = lit - 1; /* stop run */ |
134 |
|
|
op -= !lit; /* undo run if length is zero */ |
135 |
|
|
|
136 |
|
|
len -= 2; /* len is now #octets - 1 */ |
137 |
|
|
ip++; |
138 |
|
|
|
139 |
|
|
if (len < 7) |
140 |
|
|
{ |
141 |
|
|
*op++ = (off >> 8) + (len << 5); |
142 |
|
|
} |
143 |
|
|
else |
144 |
|
|
{ |
145 |
|
|
*op++ = (off >> 8) + ( 7 << 5); |
146 |
|
|
*op++ = len - 7; |
147 |
|
|
} |
148 |
|
|
|
149 |
|
|
*op++ = off; |
150 |
|
|
|
151 |
|
|
lit = 0; op++; /* start run */ |
152 |
|
|
|
153 |
|
|
ip += len + 1; |
154 |
|
|
|
155 |
|
|
if (expect_false (ip >= in_end - 2)) |
156 |
|
|
break; |
157 |
|
|
|
158 |
|
|
ip -= len + 1; |
159 |
|
|
|
160 |
|
|
//printf (" fill %p for %d\n", ip, len);//D |
161 |
|
|
do |
162 |
|
|
{ |
163 |
|
|
u16 hash = HASH (ip); |
164 |
root |
1.3 |
res = ((unsigned long)ip) & (LZF_MAX_OFF - 1); |
165 |
root |
1.1 |
|
166 |
root |
1.3 |
p = STATE.first [hash]; |
167 |
|
|
STATE.prev [res] = ip - p; /* update ptr to previous hash match */ |
168 |
|
|
STATE.first [hash] = ip; /* first hash match is here */ |
169 |
root |
1.1 |
|
170 |
|
|
ip++; |
171 |
|
|
} |
172 |
|
|
while (len--); |
173 |
|
|
} |
174 |
|
|
else |
175 |
|
|
{ |
176 |
|
|
/* one more literal byte we must copy */ |
177 |
|
|
if (expect_false (op >= out_end)) |
178 |
|
|
return 0; |
179 |
|
|
|
180 |
|
|
lit++; *op++ = *ip++; |
181 |
|
|
|
182 |
root |
1.3 |
if (expect_false (lit == LZF_MAX_LIT)) |
183 |
root |
1.1 |
{ |
184 |
|
|
op [- lit - 1] = lit - 1; /* stop run */ |
185 |
|
|
lit = 0; op++; /* start run */ |
186 |
|
|
} |
187 |
|
|
} |
188 |
|
|
} |
189 |
|
|
|
190 |
|
|
if (op + 3 > out_end) /* at most 3 bytes can be missing here */ |
191 |
|
|
return 0; |
192 |
|
|
|
193 |
|
|
while (ip < in_end) |
194 |
|
|
{ |
195 |
|
|
lit++; *op++ = *ip++; |
196 |
|
|
|
197 |
root |
1.3 |
if (expect_false (lit == LZF_MAX_LIT)) |
198 |
root |
1.1 |
{ |
199 |
|
|
op [- lit - 1] = lit - 1; /* stop run */ |
200 |
|
|
lit = 0; op++; /* start run */ |
201 |
|
|
} |
202 |
|
|
} |
203 |
|
|
|
204 |
|
|
op [- lit - 1] = lit - 1; /* end run */ |
205 |
|
|
op -= !lit; /* undo run if length is zero */ |
206 |
|
|
|
207 |
|
|
return op - (u8 *)out_data; |
208 |
root |
1.3 |
|
209 |
|
|
#undef STATE |
210 |
root |
1.1 |
} |
211 |
|
|
|
212 |
root |
1.3 |
|