1 /*-
2 * SPDX-License-Identifier: MIT
3 *
4 * Copyright (c) 2005-2014 Rich Felker, et al.
5 *
6 * Permission is hereby granted, free of charge, to any person obtaining
7 * a copy of this software and associated documentation files (the
8 * "Software"), to deal in the Software without restriction, including
9 * without limitation the rights to use, copy, modify, merge, publish,
10 * distribute, sublicense, and/or sell copies of the Software, and to
11 * permit persons to whom the Software is furnished to do so, subject to
12 * the following conditions:
13 *
14 * The above copyright notice and this permission notice shall be
15 * included in all copies or substantial portions of the Software.
16 *
17 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
18 * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
19 * MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.
20 * IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY
21 * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT,
22 * TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE
23 * SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
24 */
25 #include <sys/cdefs.h>
26 __FBSDID("$FreeBSD$");
27
28 #include <stdint.h>
29 #include <string.h>
30
31 static char *
twobyte_strstr(const unsigned char * h,const unsigned char * n)32 twobyte_strstr(const unsigned char *h, const unsigned char *n)
33 {
34 uint16_t nw = n[0] << 8 | n[1], hw = h[0] << 8 | h[1];
35 for (h++; *h && hw != nw; hw = hw << 8 | *++h)
36 ;
37 return *h ? (char *)h - 1 : 0;
38 }
39
40 static char *
threebyte_strstr(const unsigned char * h,const unsigned char * n)41 threebyte_strstr(const unsigned char *h, const unsigned char *n)
42 {
43 uint32_t nw = (uint32_t)n[0] << 24 | n[1] << 16 | n[2] << 8;
44 uint32_t hw = (uint32_t)h[0] << 24 | h[1] << 16 | h[2] << 8;
45 for (h += 2; *h && hw != nw; hw = (hw | *++h) << 8)
46 ;
47 return *h ? (char *)h - 2 : 0;
48 }
49
50 static char *
fourbyte_strstr(const unsigned char * h,const unsigned char * n)51 fourbyte_strstr(const unsigned char *h, const unsigned char *n)
52 {
53 uint32_t nw = (uint32_t)n[0] << 24 | n[1] << 16 | n[2] << 8 | n[3];
54 uint32_t hw = (uint32_t)h[0] << 24 | h[1] << 16 | h[2] << 8 | h[3];
55 for (h += 3; *h && hw != nw; hw = hw << 8 | *++h)
56 ;
57 return *h ? (char *)h - 3 : 0;
58 }
59
60 #define MAX(a, b) ((a) > (b) ? (a) : (b))
61 #define MIN(a, b) ((a) < (b) ? (a) : (b))
62
63 #define BITOP(a, b, op) \
64 ((a)[(size_t)(b) / (8 * sizeof *(a))] op \
65 (size_t)1 << ((size_t)(b) % (8 * sizeof *(a))))
66
67 /*
68 * Two Way string search algorithm, with a bad shift table applied to the last
69 * byte of the window. A bit array marks which entries in the shift table are
70 * initialized to avoid fully initializing a 1kb/2kb table.
71 *
72 * Reference: CROCHEMORE M., PERRIN D., 1991, Two-way string-matching,
73 * Journal of the ACM 38(3):651-675
74 */
75 static char *
twoway_strstr(const unsigned char * h,const unsigned char * n)76 twoway_strstr(const unsigned char *h, const unsigned char *n)
77 {
78 const unsigned char *z;
79 size_t l, ip, jp, k, p, ms, p0, mem, mem0;
80 size_t byteset[32 / sizeof(size_t)] = { 0 };
81 size_t shift[256];
82
83 /* Computing length of needle and fill shift table */
84 for (l = 0; n[l] && h[l]; l++)
85 BITOP(byteset, n[l], |=), shift[n[l]] = l + 1;
86 if (n[l])
87 return 0; /* hit the end of h */
88
89 /* Compute maximal suffix */
90 ip = -1;
91 jp = 0;
92 k = p = 1;
93 while (jp + k < l) {
94 if (n[ip + k] == n[jp + k]) {
95 if (k == p) {
96 jp += p;
97 k = 1;
98 } else
99 k++;
100 } else if (n[ip + k] > n[jp + k]) {
101 jp += k;
102 k = 1;
103 p = jp - ip;
104 } else {
105 ip = jp++;
106 k = p = 1;
107 }
108 }
109 ms = ip;
110 p0 = p;
111
112 /* And with the opposite comparison */
113 ip = -1;
114 jp = 0;
115 k = p = 1;
116 while (jp + k < l) {
117 if (n[ip + k] == n[jp + k]) {
118 if (k == p) {
119 jp += p;
120 k = 1;
121 } else
122 k++;
123 } else if (n[ip + k] < n[jp + k]) {
124 jp += k;
125 k = 1;
126 p = jp - ip;
127 } else {
128 ip = jp++;
129 k = p = 1;
130 }
131 }
132 if (ip + 1 > ms + 1)
133 ms = ip;
134 else
135 p = p0;
136
137 /* Periodic needle? */
138 if (memcmp(n, n + p, ms + 1)) {
139 mem0 = 0;
140 p = MAX(ms, l - ms - 1) + 1;
141 } else
142 mem0 = l - p;
143 mem = 0;
144
145 /* Initialize incremental end-of-haystack pointer */
146 z = h;
147
148 /* Search loop */
149 for (;;) {
150 /* Update incremental end-of-haystack pointer */
151 if (z - h < l) {
152 /* Fast estimate for MAX(l,63) */
153 size_t grow = l | 63;
154 const unsigned char *z2 = memchr(z, 0, grow);
155 if (z2) {
156 z = z2;
157 if (z - h < l)
158 return 0;
159 } else
160 z += grow;
161 }
162
163 /* Check last byte first; advance by shift on mismatch */
164 if (BITOP(byteset, h[l - 1], &)) {
165 k = l - shift[h[l - 1]];
166 if (k) {
167 if (k < mem)
168 k = mem;
169 h += k;
170 mem = 0;
171 continue;
172 }
173 } else {
174 h += l;
175 mem = 0;
176 continue;
177 }
178
179 /* Compare right half */
180 for (k = MAX(ms + 1, mem); n[k] && n[k] == h[k]; k++)
181 ;
182 if (n[k]) {
183 h += k - ms;
184 mem = 0;
185 continue;
186 }
187 /* Compare left half */
188 for (k = ms + 1; k > mem && n[k - 1] == h[k - 1]; k--)
189 ;
190 if (k <= mem)
191 return (char *)h;
192 h += p;
193 mem = mem0;
194 }
195 }
196
197 char *
strstr(const char * h,const char * n)198 strstr(const char *h, const char *n)
199 {
200 /* Return immediately on empty needle */
201 if (!n[0])
202 return (char *)h;
203
204 /* Use faster algorithms for short needles */
205 h = strchr(h, *n);
206 if (!h || !n[1])
207 return (char *)h;
208 if (!h[1])
209 return 0;
210 if (!n[2])
211 return twobyte_strstr((void *)h, (void *)n);
212 if (!h[2])
213 return 0;
214 if (!n[3])
215 return threebyte_strstr((void *)h, (void *)n);
216 if (!h[3])
217 return 0;
218 if (!n[4])
219 return fourbyte_strstr((void *)h, (void *)n);
220
221 return twoway_strstr((void *)h, (void *)n);
222 }
223