xref: /illumos-gate/usr/src/contrib/ast/src/lib/libast/comp/regexp.c (revision b30d193948be5a7794d7ae3ba0ed9c2f72c88e0f)
1*b30d1939SAndy Fiddaman /***********************************************************************
2*b30d1939SAndy Fiddaman *                                                                      *
3*b30d1939SAndy Fiddaman *               This software is part of the ast package               *
4*b30d1939SAndy Fiddaman *          Copyright (c) 1985-2011 AT&T Intellectual Property          *
5*b30d1939SAndy Fiddaman *                      and is licensed under the                       *
6*b30d1939SAndy Fiddaman *                 Eclipse Public License, Version 1.0                  *
7*b30d1939SAndy Fiddaman *                    by AT&T Intellectual Property                     *
8*b30d1939SAndy Fiddaman *                                                                      *
9*b30d1939SAndy Fiddaman *                A copy of the License is available at                 *
10*b30d1939SAndy Fiddaman *          http://www.eclipse.org/org/documents/epl-v10.html           *
11*b30d1939SAndy Fiddaman *         (with md5 checksum b35adb5213ca9657e911e9befb180842)         *
12*b30d1939SAndy Fiddaman *                                                                      *
13*b30d1939SAndy Fiddaman *              Information and Software Systems Research               *
14*b30d1939SAndy Fiddaman *                            AT&T Research                             *
15*b30d1939SAndy Fiddaman *                           Florham Park NJ                            *
16*b30d1939SAndy Fiddaman *                                                                      *
17*b30d1939SAndy Fiddaman *                 Glenn Fowler <gsf@research.att.com>                  *
18*b30d1939SAndy Fiddaman *                  David Korn <dgk@research.att.com>                   *
19*b30d1939SAndy Fiddaman *                   Phong Vo <kpv@research.att.com>                    *
20*b30d1939SAndy Fiddaman *                                                                      *
21*b30d1939SAndy Fiddaman ***********************************************************************/
22*b30d1939SAndy Fiddaman #pragma prototyped
23*b30d1939SAndy Fiddaman /*
24*b30d1939SAndy Fiddaman  * <regexp.h> library support
25*b30d1939SAndy Fiddaman  */
26*b30d1939SAndy Fiddaman 
27*b30d1939SAndy Fiddaman #define _REGEXP_DECLARE
28*b30d1939SAndy Fiddaman 
29*b30d1939SAndy Fiddaman #include <ast.h>
30*b30d1939SAndy Fiddaman #include <regexp.h>
31*b30d1939SAndy Fiddaman #include <regex.h>
32*b30d1939SAndy Fiddaman #include <align.h>
33*b30d1939SAndy Fiddaman 
34*b30d1939SAndy Fiddaman typedef struct
35*b30d1939SAndy Fiddaman {
36*b30d1939SAndy Fiddaman 	regex_t		re;
37*b30d1939SAndy Fiddaman 	char*		buf;
38*b30d1939SAndy Fiddaman 	char*		cur;
39*b30d1939SAndy Fiddaman 	unsigned int	size;
40*b30d1939SAndy Fiddaman } Env_t;
41*b30d1939SAndy Fiddaman 
42*b30d1939SAndy Fiddaman static void*
block(void * handle,void * data,size_t size)43*b30d1939SAndy Fiddaman block(void* handle, void* data, size_t size)
44*b30d1939SAndy Fiddaman {
45*b30d1939SAndy Fiddaman 	register Env_t*	env = (Env_t*)handle;
46*b30d1939SAndy Fiddaman 
47*b30d1939SAndy Fiddaman 	if (data || (size = roundof(size, ALIGN_BOUND2)) > (env->buf + env->size - env->cur))
48*b30d1939SAndy Fiddaman 		return 0;
49*b30d1939SAndy Fiddaman 	data = (void*)env->cur;
50*b30d1939SAndy Fiddaman 	env->cur += size;
51*b30d1939SAndy Fiddaman 	return data;
52*b30d1939SAndy Fiddaman }
53*b30d1939SAndy Fiddaman 
54*b30d1939SAndy Fiddaman int
_re_comp(regexp_t * re,const char * pattern,char * handle,unsigned int size)55*b30d1939SAndy Fiddaman _re_comp(regexp_t* re, const char* pattern, char* handle, unsigned int size)
56*b30d1939SAndy Fiddaman {
57*b30d1939SAndy Fiddaman 	register Env_t*	env = (Env_t*)handle;
58*b30d1939SAndy Fiddaman 	register int	n;
59*b30d1939SAndy Fiddaman 
60*b30d1939SAndy Fiddaman 	if (size <= sizeof(Env_t))
61*b30d1939SAndy Fiddaman 		return 50;
62*b30d1939SAndy Fiddaman 	env->buf = env->cur = (char*)env + sizeof(Env_t);
63*b30d1939SAndy Fiddaman 	env->size = size - sizeof(Env_t);
64*b30d1939SAndy Fiddaman 	regalloc(env, block, REG_NOFREE);
65*b30d1939SAndy Fiddaman 	n = regcomp(&env->re, pattern, REG_LENIENT|REG_NULL);
66*b30d1939SAndy Fiddaman 	switch (n)
67*b30d1939SAndy Fiddaman 	{
68*b30d1939SAndy Fiddaman 	case 0:
69*b30d1939SAndy Fiddaman 		break;
70*b30d1939SAndy Fiddaman 	case REG_ERANGE:
71*b30d1939SAndy Fiddaman 		n = 11;
72*b30d1939SAndy Fiddaman 		break;
73*b30d1939SAndy Fiddaman 	case REG_BADBR:
74*b30d1939SAndy Fiddaman 		n = 16;
75*b30d1939SAndy Fiddaman 		break;
76*b30d1939SAndy Fiddaman 	case REG_ESUBREG:
77*b30d1939SAndy Fiddaman 		n = 25;
78*b30d1939SAndy Fiddaman 		break;
79*b30d1939SAndy Fiddaman 	case REG_EPAREN:
80*b30d1939SAndy Fiddaman 		n = 42;
81*b30d1939SAndy Fiddaman 		break;
82*b30d1939SAndy Fiddaman 	case REG_EBRACK:
83*b30d1939SAndy Fiddaman 		n = 49;
84*b30d1939SAndy Fiddaman 		break;
85*b30d1939SAndy Fiddaman 	default:
86*b30d1939SAndy Fiddaman 		n = 50;
87*b30d1939SAndy Fiddaman 		break;
88*b30d1939SAndy Fiddaman 	}
89*b30d1939SAndy Fiddaman 	re->re_nbra = env->re.re_nsub;
90*b30d1939SAndy Fiddaman 	return n;
91*b30d1939SAndy Fiddaman }
92*b30d1939SAndy Fiddaman 
93*b30d1939SAndy Fiddaman int
_re_exec(regexp_t * re,const char * subject,const char * handle,int anchor)94*b30d1939SAndy Fiddaman _re_exec(regexp_t* re, const char* subject, const char* handle, int anchor)
95*b30d1939SAndy Fiddaman {
96*b30d1939SAndy Fiddaman 	register Env_t*	env = (Env_t*)handle;
97*b30d1939SAndy Fiddaman 	register int	n;
98*b30d1939SAndy Fiddaman 	regmatch_t	match[elementsof(re->re_braslist)+1];
99*b30d1939SAndy Fiddaman 
100*b30d1939SAndy Fiddaman 	if (regexec(&env->re, subject, elementsof(match), match, 0) || anchor && match[0].rm_so)
101*b30d1939SAndy Fiddaman 		return 0;
102*b30d1939SAndy Fiddaman 	re->re_loc1 = (char*)subject + match[0].rm_so;
103*b30d1939SAndy Fiddaman 	re->re_loc2 = (char*)subject + match[0].rm_eo;
104*b30d1939SAndy Fiddaman 	for (n = 1; n <= env->re.re_nsub; n++)
105*b30d1939SAndy Fiddaman 	{
106*b30d1939SAndy Fiddaman 		re->re_braslist[n-1] = (char*)subject + match[n].rm_so;
107*b30d1939SAndy Fiddaman 		re->re_braelist[n-1] = (char*)subject + match[n].rm_eo;
108*b30d1939SAndy Fiddaman 	}
109*b30d1939SAndy Fiddaman 	return 1;
110*b30d1939SAndy Fiddaman }
111*b30d1939SAndy Fiddaman 
112*b30d1939SAndy Fiddaman char*
_re_putc(int c)113*b30d1939SAndy Fiddaman _re_putc(int c)
114*b30d1939SAndy Fiddaman {
115*b30d1939SAndy Fiddaman 	static Sfio_t*	sp;
116*b30d1939SAndy Fiddaman 
117*b30d1939SAndy Fiddaman 	if (!sp && !(sp = sfstropen()))
118*b30d1939SAndy Fiddaman 		return 0;
119*b30d1939SAndy Fiddaman 	if (!c)
120*b30d1939SAndy Fiddaman 		return sfstruse(sp);
121*b30d1939SAndy Fiddaman 	sfputc(sp, c);
122*b30d1939SAndy Fiddaman 	return 0;
123*b30d1939SAndy Fiddaman }
124