1 | /* |
1 | /* |
2 | * Copyright (c) 2000-2007 Marc Alexander Lehmann <schmorp@schmorp.de> |
2 | * Copyright (c) 2000-2008 Marc Alexander Lehmann <schmorp@schmorp.de> |
3 | * |
3 | * |
4 | * Redistribution and use in source and binary forms, with or without modifica- |
4 | * Redistribution and use in source and binary forms, with or without modifica- |
5 | * tion, are permitted provided that the following conditions are met: |
5 | * tion, are permitted provided that the following conditions are met: |
6 | * |
6 | * |
7 | * 1. Redistributions of source code must retain the above copyright notice, |
7 | * 1. Redistributions of source code must retain the above copyright notice, |
… | |
… | |
111 | u8 *op = (u8 *)out_data; |
111 | u8 *op = (u8 *)out_data; |
112 | const u8 *in_end = ip + in_len; |
112 | const u8 *in_end = ip + in_len; |
113 | u8 *out_end = op + out_len; |
113 | u8 *out_end = op + out_len; |
114 | const u8 *ref; |
114 | const u8 *ref; |
115 | |
115 | |
|
|
116 | /* off requires a type wide enough to hold a general pointer difference. |
|
|
117 | * ISO C doesn't have that (size_t might not be enough and ptrdiff_t only |
|
|
118 | * works for differences within a single object). We also assume that no |
|
|
119 | * no bit pattern traps. Since the only platform that is both non-POSIX |
|
|
120 | * and fails to support both assumptions is windows 64 bit, we make a |
|
|
121 | * special workaround for it. |
|
|
122 | */ |
|
|
123 | #if defined (WIN32) && defined (_M_X64) |
|
|
124 | unsigned _int64 off; /* workaround for missing POSIX compliance */ |
|
|
125 | #else |
|
|
126 | unsigned long off; |
|
|
127 | #endif |
116 | unsigned int hval; |
128 | unsigned int hval; |
117 | unsigned long off; |
|
|
118 | int lit; |
129 | int lit; |
119 | |
130 | |
120 | if (!in_len || !out_len) |
131 | if (!in_len || !out_len) |
121 | return 0; |
132 | return 0; |
122 | |
133 | |
… | |
… | |
136 | hval = NEXT (hval, ip); |
147 | hval = NEXT (hval, ip); |
137 | hslot = htab + IDX (hval); |
148 | hslot = htab + IDX (hval); |
138 | ref = *hslot; *hslot = ip; |
149 | ref = *hslot; *hslot = ip; |
139 | |
150 | |
140 | if (1 |
151 | if (1 |
141 | #if INIT_HTAB && !USE_MEMCPY |
152 | #if INIT_HTAB |
142 | && ref < ip /* the next test will actually take care of this, but this is faster */ |
153 | && ref < ip /* the next test will actually take care of this, but this is faster */ |
143 | #endif |
154 | #endif |
144 | && (off = ip - ref - 1) < MAX_OFF |
155 | && (off = ip - ref - 1) < MAX_OFF |
145 | && ip + 4 < in_end |
156 | && ip + 4 < in_end |
146 | && ref > (u8 *)in_data |
157 | && ref > (u8 *)in_data |
… | |
… | |
157 | /* match found at *ref++ */ |
168 | /* match found at *ref++ */ |
158 | unsigned int len = 2; |
169 | unsigned int len = 2; |
159 | unsigned int maxlen = in_end - ip - len; |
170 | unsigned int maxlen = in_end - ip - len; |
160 | maxlen = maxlen > MAX_REF ? MAX_REF : maxlen; |
171 | maxlen = maxlen > MAX_REF ? MAX_REF : maxlen; |
161 | |
172 | |
162 | if (expect_false (op + 1 + 3 >= out_end)) |
|
|
163 | return 0; |
|
|
164 | |
|
|
165 | op [- lit - 1] = lit - 1; /* stop run */ |
173 | op [- lit - 1] = lit - 1; /* stop run */ |
166 | op -= !lit; /* undo run if length is zero */ |
174 | op -= !lit; /* undo run if length is zero */ |
|
|
175 | |
|
|
176 | if (expect_false (op + 3 + 1 >= out_end)) |
|
|
177 | return 0; |
167 | |
178 | |
168 | for (;;) |
179 | for (;;) |
169 | { |
180 | { |
170 | if (expect_true (maxlen > 16)) |
181 | if (expect_true (maxlen > 16)) |
171 | { |
182 | { |
… | |
… | |
195 | while (len < maxlen && ref[len] == ip[len]); |
206 | while (len < maxlen && ref[len] == ip[len]); |
196 | |
207 | |
197 | break; |
208 | break; |
198 | } |
209 | } |
199 | |
210 | |
200 | len -= 2; |
211 | len -= 2; /* len is now #octets - 1 */ |
201 | ip++; |
212 | ip++; |
202 | |
213 | |
203 | if (len < 7) |
214 | if (len < 7) |
204 | { |
215 | { |
205 | *op++ = (off >> 8) + (len << 5); |
216 | *op++ = (off >> 8) + (len << 5); |
… | |
… | |
210 | *op++ = len - 7; |
221 | *op++ = len - 7; |
211 | } |
222 | } |
212 | |
223 | |
213 | *op++ = off; |
224 | *op++ = off; |
214 | |
225 | |
|
|
226 | if (expect_true (ip + len < in_end - 2)) |
|
|
227 | { |
215 | #if ULTRA_FAST || VERY_FAST |
228 | #if ULTRA_FAST || VERY_FAST |
216 | ip += len; |
229 | ip += len; |
217 | #if VERY_FAST && !ULTRA_FAST |
230 | # if VERY_FAST && !ULTRA_FAST |
218 | --ip; |
231 | --ip; |
219 | #endif |
232 | # endif |
220 | hval = FRST (ip); |
233 | hval = FRST (ip); |
221 | |
234 | |
222 | hval = NEXT (hval, ip); |
|
|
223 | htab[IDX (hval)] = ip; |
|
|
224 | ip++; |
|
|
225 | |
|
|
226 | #if VERY_FAST && !ULTRA_FAST |
|
|
227 | hval = NEXT (hval, ip); |
|
|
228 | htab[IDX (hval)] = ip; |
|
|
229 | ip++; |
|
|
230 | #endif |
|
|
231 | #else |
|
|
232 | do |
|
|
233 | { |
|
|
234 | hval = NEXT (hval, ip); |
235 | hval = NEXT (hval, ip); |
235 | htab[IDX (hval)] = ip; |
236 | htab[IDX (hval)] = ip; |
236 | ip++; |
237 | ip++; |
|
|
238 | |
|
|
239 | # if VERY_FAST && !ULTRA_FAST |
|
|
240 | hval = NEXT (hval, ip); |
|
|
241 | htab[IDX (hval)] = ip; |
237 | } |
242 | ip++; |
238 | while (len--); |
|
|
239 | #endif |
243 | # endif |
|
|
244 | #else |
|
|
245 | do |
|
|
246 | { |
|
|
247 | hval = NEXT (hval, ip); |
|
|
248 | htab[IDX (hval)] = ip; |
|
|
249 | ip++; |
|
|
250 | } |
|
|
251 | while (len--); |
|
|
252 | #endif |
|
|
253 | } |
|
|
254 | |
240 | lit = 0; op++; /* start run */ |
255 | lit = 0; op++; /* start run */ |
241 | continue; |
|
|
242 | } |
256 | } |
243 | |
257 | else |
|
|
258 | { |
244 | /* one more literal byte we must copy */ |
259 | /* one more literal byte we must copy */ |
245 | |
|
|
246 | if (expect_false (op >= out_end)) |
260 | if (expect_false (op >= out_end)) |
247 | return 0; |
261 | return 0; |
248 | |
262 | |
249 | lit++; |
263 | lit++; *op++ = *ip++; |
|
|
264 | |
|
|
265 | if (expect_false (lit == MAX_LIT)) |
|
|
266 | { |
|
|
267 | op [- lit - 1] = lit - 1; /* stop run */ |
|
|
268 | lit = 0; op++; /* start run */ |
|
|
269 | } |
|
|
270 | } |
|
|
271 | } |
|
|
272 | |
|
|
273 | if (op + 3 > out_end) /* at most 3 bytes can be missing here */ |
|
|
274 | return 0; |
|
|
275 | |
|
|
276 | while (ip < in_end) |
|
|
277 | { |
250 | *op++ = *ip++; |
278 | lit++; *op++ = *ip++; |
251 | |
279 | |
252 | if (expect_false (lit == MAX_LIT)) |
280 | if (expect_false (lit == MAX_LIT)) |
253 | { |
281 | { |
254 | op [- lit - 1] = lit - 1; /* stop run */ |
282 | op [- lit - 1] = lit - 1; /* stop run */ |
255 | lit = 0; op++; /* start run */ |
283 | lit = 0; op++; /* start run */ |
256 | } |
284 | } |
257 | } |
285 | } |
258 | |
286 | |
259 | if (op + lit + 2 >= out_end) |
|
|
260 | return 0; |
|
|
261 | |
|
|
262 | while (ip < in_end) |
|
|
263 | { |
|
|
264 | lit++; |
|
|
265 | *op++ = *ip++; |
|
|
266 | } |
|
|
267 | |
|
|
268 | op [- lit - 1] = lit - 1; |
287 | op [- lit - 1] = lit - 1; /* end run */ |
|
|
288 | op -= !lit; /* undo run if length is zero */ |
269 | |
289 | |
270 | return op - (u8 *)out_data; |
290 | return op - (u8 *)out_data; |
271 | } |
291 | } |
272 | |
292 | |