xref: /PHP-5.4/ext/ereg/regex/regexec.c (revision 39b58500)
1 /*
2  * the outer shell of regexec()
3  *
4  * This file includes engine.c *twice*, after muchos fiddling with the
5  * macros that code uses.  This lets the same code operate on two different
6  * representations for state sets.
7  */
8 #include <sys/types.h>
9 #include <stdio.h>
10 #include <stdlib.h>
11 #include <string.h>
12 #include <limits.h>
13 #include <ctype.h>
14 
15 #include "regex.h"
16 #include "utils.h"
17 #include "regex2.h"
18 
19 #define PHP_REGEX_NOPE 0;		/* for use in asserts; shuts lint up */
20 
21 /* macros for manipulating states, small version */
22 #define	states	unsigned
23 #define	states1	unsigned	/* for later use in regexec() decision */
24 #define	CLEAR(v)	((v) = 0)
25 #define	SET0(v, n)	((v) &= ~((unsigned)1 << (n)))
26 #define	SET1(v, n)	((v) |= (unsigned)1 << (n))
27 #define	ISSET(v, n)	((v) & ((unsigned)1 << (n)))
28 #define	ASSIGN(d, s)	((d) = (s))
29 #define	EQ(a, b)	((a) == (b))
30 #define	STATEVARS	int dummy	/* dummy version */
31 #define	STATESETUP(m, n)	/* nothing */
32 #define	STATETEARDOWN(m)	/* nothing */
33 #define	SETUP(v)	((v) = 0)
34 #define	onestate	unsigned
35 #define	INIT(o, n)	((o) = (unsigned)1 << (n))
36 #define	INC(o)	((o) <<= 1)
37 #define	ISSTATEIN(v, o)	((v) & (o))
38 /* some abbreviations; note that some of these know variable names! */
39 /* do "if I'm here, I can also be there" etc without branches */
40 #define	FWD(dst, src, n)	((dst) |= ((unsigned)(src)&(here)) << (n))
41 #define	BACK(dst, src, n)	((dst) |= ((unsigned)(src)&(here)) >> (n))
42 #define	ISSETBACK(v, n)	((v) & ((unsigned)here >> (n)))
43 /* function names */
44 #define SNAMES			/* engine.c looks after details */
45 
46 #include "engine.c"
47 
48 /* now undo things */
49 #undef	states
50 #undef	CLEAR
51 #undef	SET0
52 #undef	SET1
53 #undef	ISSET
54 #undef	ASSIGN
55 #undef	EQ
56 #undef	STATEVARS
57 #undef	STATESETUP
58 #undef	STATETEARDOWN
59 #undef	SETUP
60 #undef	onestate
61 #undef	INIT
62 #undef	INC
63 #undef	ISSTATEIN
64 #undef	FWD
65 #undef	BACK
66 #undef	ISSETBACK
67 #undef	SNAMES
68 
69 /* macros for manipulating states, large version */
70 #define	states	unsigned char *
71 #define	CLEAR(v)	memset(v, 0, m->g->nstates)
72 #define	SET0(v, n)	((v)[n] = 0)
73 #define	SET1(v, n)	((v)[n] = 1)
74 #define	ISSET(v, n)	((v)[n])
75 #define	ASSIGN(d, s)	memcpy(d, s, m->g->nstates)
76 #define	EQ(a, b)	(memcmp(a, b, m->g->nstates) == 0)
77 #define	STATEVARS	int vn; unsigned char *space
78 #define	STATESETUP(m, nv)	{ (m)->space = malloc((nv)*(m)->g->nstates); \
79 				if ((m)->space == NULL) return(REG_ESPACE); \
80 				(m)->vn = 0; }
81 #define	STATETEARDOWN(m)	{ free((m)->space); }
82 #define	SETUP(v)	((v) = &m->space[m->vn++ * m->g->nstates])
83 #define	onestate	int
84 #define	INIT(o, n)	((o) = (n))
85 #define	INC(o)	((o)++)
86 #define	ISSTATEIN(v, o)	((v)[o])
87 /* some abbreviations; note that some of these know variable names! */
88 /* do "if I'm here, I can also be there" etc without branches */
89 #define	FWD(dst, src, n)	((dst)[here+(n)] |= (src)[here])
90 #define	BACK(dst, src, n)	((dst)[here-(n)] |= (src)[here])
91 #define	ISSETBACK(v, n)	((v)[here - (n)])
92 /* function names */
93 #define	LNAMES			/* flag */
94 
95 #include "engine.c"
96 
97 /*
98  - regexec - interface for matching
99  = API_EXPORT(int) regexec(const regex_t *, const char *, size_t, \
100  =					regmatch_t [], int);
101  = #define	REG_NOTBOL	00001
102  = #define	REG_NOTEOL	00002
103  = #define	REG_STARTEND	00004
104  = #define	REG_TRACE	00400	// tracing of execution
105  = #define	REG_LARGE	01000	// force large representation
106  = #define	REG_BACKR	02000	// force use of backref code
107  *
108  * We put this here so we can exploit knowledge of the state representation
109  * when choosing which matcher to call.  Also, by this point the matchers
110  * have been prototyped.
111  */
112 API_EXPORT(int)				/* 0 success, REG_NOMATCH failure */
113 regexec(preg, string, nmatch, pmatch, eflags)
114 const regex_t *preg;
115 const char *string;
116 size_t nmatch;
117 regmatch_t pmatch[];
118 int eflags;
119 {
120 	register struct re_guts *g = preg->re_g;
121 #ifdef REDEBUG
122 #	define	GOODFLAGS(f)	(f)
123 #else
124 #	define	GOODFLAGS(f)	((f)&(REG_NOTBOL|REG_NOTEOL|REG_STARTEND))
125 #endif
126 
127 	if (preg->re_magic != MAGIC1 || g->magic != MAGIC2)
128 		return(REG_BADPAT);
129 	assert(!(g->iflags&BAD));
130 	if (g->iflags&BAD)		/* backstop for no-debug case */
131 		return(REG_BADPAT);
132 	eflags = GOODFLAGS(eflags);
133 
134 	if (g->nstates <= CHAR_BIT*sizeof(states1) && !(eflags&REG_LARGE))
135 		return(smatcher(g, (unsigned char *)string, nmatch, pmatch, eflags));
136 	else
137 		return(lmatcher(g, (unsigned char *)string, nmatch, pmatch, eflags));
138 }
139