1 /*	$OpenBSD: memmem.c,v 1.5 2020/04/16 12:39:28 claudio Exp $ */
2 
3 /*
4  * Copyright (c) 2005-2020 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 
26 #include "../config.h"
27 
28 #include <string.h>
29 #include <stdint.h>
30 
31 static char *
32 twobyte_memmem(const unsigned char *h, size_t k, const unsigned char *n)
33 {
34 	uint16_t nw = n[0]<<8 | n[1], hw = h[0]<<8 | h[1];
35 	for (h+=2, k-=2; k; k--, hw = hw<<8 | *h++)
36 		if (hw == nw) return (char *)h-2;
37 	return hw == nw ? (char *)h-2 : 0;
38 }
39 
40 static char *
41 threebyte_memmem(const unsigned char *h, size_t k, const unsigned char *n)
42 {
43 	uint32_t nw = n[0]<<24 | n[1]<<16 | n[2]<<8;
44 	uint32_t hw = h[0]<<24 | h[1]<<16 | h[2]<<8;
45 	for (h+=3, k-=3; k; k--, hw = (hw|*h++)<<8)
46 		if (hw == nw) return (char *)h-3;
47 	return hw == nw ? (char *)h-3 : 0;
48 }
49 
50 static char *
51 fourbyte_memmem(const unsigned char *h, size_t k, const unsigned char *n)
52 {
53 	uint32_t nw = n[0]<<24 | n[1]<<16 | n[2]<<8 | n[3];
54 	uint32_t hw = h[0]<<24 | h[1]<<16 | h[2]<<8 | h[3];
55 	for (h+=4, k-=4; k; k--, hw = hw<<8 | *h++)
56 		if (hw == nw) return (char *)h-4;
57 	return hw == nw ? (char *)h-4 : 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 (size_t)1<<((size_t)(b)%(8*sizeof *(a))))
65 
66 /*
67  * Maxime Crochemore and Dominique Perrin, Two-way string-matching,
68  * Journal of the ACM, 38(3):651-675, July 1991.
69  */
70 static char *
71 twoway_memmem(const unsigned char *h, const unsigned char *z,
72     const unsigned char *n, size_t l)
73 {
74 	size_t i, ip, jp, k, p, ms, p0, mem, mem0;
75 	size_t byteset[32 / sizeof(size_t)] = { 0 };
76 	size_t shift[256];
77 
78 	/* Computing length of needle and fill shift table */
79 	for (i=0; i<l; i++)
80 		BITOP(byteset, n[i], |=), shift[n[i]] = i+1;
81 
82 	/* Compute maximal suffix */
83 	ip = -1; jp = 0; k = p = 1;
84 	while (jp+k<l) {
85 		if (n[ip+k] == n[jp+k]) {
86 			if (k == p) {
87 				jp += p;
88 				k = 1;
89 			} else k++;
90 		} else if (n[ip+k] > n[jp+k]) {
91 			jp += k;
92 			k = 1;
93 			p = jp - ip;
94 		} else {
95 			ip = jp++;
96 			k = p = 1;
97 		}
98 	}
99 	ms = ip;
100 	p0 = p;
101 
102 	/* And with the opposite comparison */
103 	ip = -1; jp = 0; k = p = 1;
104 	while (jp+k<l) {
105 		if (n[ip+k] == n[jp+k]) {
106 			if (k == p) {
107 				jp += p;
108 				k = 1;
109 			} else k++;
110 		} else if (n[ip+k] < n[jp+k]) {
111 			jp += k;
112 			k = 1;
113 			p = jp - ip;
114 		} else {
115 			ip = jp++;
116 			k = p = 1;
117 		}
118 	}
119 	if (ip+1 > ms+1) ms = ip;
120 	else p = p0;
121 
122 	/* Periodic needle? */
123 	if (memcmp(n, n+p, ms+1)) {
124 		mem0 = 0;
125 		p = MAX(ms, l-ms-1) + 1;
126 	} else mem0 = l-p;
127 	mem = 0;
128 
129 	/* Search loop */
130 	for (;;) {
131 		/* If remainder of haystack is shorter than needle, done */
132 		if (z-h < l) return 0;
133 
134 		/* Check last byte first; advance by shift on mismatch */
135 		if (BITOP(byteset, h[l-1], &)) {
136 			k = l-shift[h[l-1]];
137 			if (k) {
138 				if (k < mem) k = mem;
139 				h += k;
140 				mem = 0;
141 				continue;
142 			}
143 		} else {
144 			h += l;
145 			mem = 0;
146 			continue;
147 		}
148 
149 		/* Compare right half */
150 		for (k=MAX(ms+1,mem); k<l && n[k] == h[k]; k++);
151 		if (k < l) {
152 			h += k-ms;
153 			mem = 0;
154 			continue;
155 		}
156 		/* Compare left half */
157 		for (k=ms+1; k>mem && n[k-1] == h[k-1]; k--);
158 		if (k <= mem) return (char *)h;
159 		h += p;
160 		mem = mem0;
161 	}
162 }
163 
164 void *
165 memmem(const void *h0, size_t k, const void *n0, size_t l)
166 {
167 	const unsigned char *h = h0, *n = n0;
168 
169 	/* Return immediately on empty needle */
170 	if (!l) return (void *)h;
171 
172 	/* Return immediately when needle is longer than haystack */
173 	if (k<l) return 0;
174 
175 	/* Use faster algorithms for short needles */
176 	h = memchr(h0, *n, k);
177 	if (!h || l==1) return (void *)h;
178 	k -= h - (const unsigned char *)h0;
179 	if (k<l) return 0;
180 	if (l==2) return twobyte_memmem(h, k, n);
181 	if (l==3) return threebyte_memmem(h, k, n);
182 	if (l==4) return fourbyte_memmem(h, k, n);
183 
184 	return twoway_memmem(h, h+k, n, l);
185 }