1 b2cfc4e2 2003-09-30 devnull #include "lib9.h"
  2 b2cfc4e2 2003-09-30 devnull #include "regexp9.h"
  3 b2cfc4e2 2003-09-30 devnull #include "regcomp.h"
  4 b2cfc4e2 2003-09-30 devnull 
  5 b2cfc4e2 2003-09-30 devnull /*
  6 b2cfc4e2 2003-09-30 devnull  *  return	0 if no match
  7 b2cfc4e2 2003-09-30 devnull  *		>0 if a match
  8 b2cfc4e2 2003-09-30 devnull  *		<0 if we ran out of _relist space
  9 b2cfc4e2 2003-09-30 devnull  */
 10 b2cfc4e2 2003-09-30 devnull static int
 11 b2cfc4e2 2003-09-30 devnull rregexec1(Reprog *progp,	/* program to run */
 12 6d08a0f5 2007-12-07 rsc 	Rune *bol,		/* string to run machine on */
 13 6d08a0f5 2007-12-07 rsc 	Resub *mp,		/* subexpression elements */
 14 6d08a0f5 2007-12-07 rsc 	int ms,			/* number of elements at mp */
 15 b2cfc4e2 2003-09-30 devnull 	Reljunk *j)
 16 b2cfc4e2 2003-09-30 devnull {
 17 b2cfc4e2 2003-09-30 devnull 	int flag=0;
 18 b2cfc4e2 2003-09-30 devnull 	Reinst *inst;
 19 b2cfc4e2 2003-09-30 devnull 	Relist *tlp;
 20 b2cfc4e2 2003-09-30 devnull 	Rune *s;
 21 b2cfc4e2 2003-09-30 devnull 	int i, checkstart;
 22 b2cfc4e2 2003-09-30 devnull 	Rune r, *rp, *ep;
 23 b2cfc4e2 2003-09-30 devnull 	Relist* tl;		/* This list, next list */
 24 b2cfc4e2 2003-09-30 devnull 	Relist* nl;
 25 b2cfc4e2 2003-09-30 devnull 	Relist* tle;		/* ends of this and next list */
 26 b2cfc4e2 2003-09-30 devnull 	Relist* nle;
 27 b2cfc4e2 2003-09-30 devnull 	int match;
 28 da7f7882 2007-05-18 devnull 	Rune *p;
 29 b2cfc4e2 2003-09-30 devnull 
 30 b2cfc4e2 2003-09-30 devnull 	match = 0;
 31 6d08a0f5 2007-12-07 rsc 	checkstart = j->startchar;
 32 b2cfc4e2 2003-09-30 devnull 	if(mp)
 33 b2cfc4e2 2003-09-30 devnull 		for(i=0; i<ms; i++) {
 34 b2cfc4e2 2003-09-30 devnull 			mp[i].s.rsp = 0;
 35 b2cfc4e2 2003-09-30 devnull 			mp[i].e.rep = 0;
 36 b2cfc4e2 2003-09-30 devnull 		}
 37 b2cfc4e2 2003-09-30 devnull 	j->relist[0][0].inst = 0;
 38 b2cfc4e2 2003-09-30 devnull 	j->relist[1][0].inst = 0;
 39 b2cfc4e2 2003-09-30 devnull 
 40 b2cfc4e2 2003-09-30 devnull 	/* Execute machine once for each character, including terminal NUL */
 41 b2cfc4e2 2003-09-30 devnull 	s = j->rstarts;
 42 b2cfc4e2 2003-09-30 devnull 	do{
 43 b2cfc4e2 2003-09-30 devnull 
 44 b2cfc4e2 2003-09-30 devnull 		/* fast check for first char */
 45 b2cfc4e2 2003-09-30 devnull 		if(checkstart) {
 46 b2cfc4e2 2003-09-30 devnull 			switch(j->starttype) {
 47 b2cfc4e2 2003-09-30 devnull 			case RUNE:
 48 da7f7882 2007-05-18 devnull 				p = runestrchr(s, j->startchar);
 49 6d08a0f5 2007-12-07 rsc 				if(p == 0 || p == j->reol)
 50 da7f7882 2007-05-18 devnull 					return match;
 51 da7f7882 2007-05-18 devnull 				s = p;
 52 b2cfc4e2 2003-09-30 devnull 				break;
 53 b2cfc4e2 2003-09-30 devnull 			case BOL:
 54 b2cfc4e2 2003-09-30 devnull 				if(s == bol)
 55 b2cfc4e2 2003-09-30 devnull 					break;
 56 da7f7882 2007-05-18 devnull 				p = runestrchr(s, '\n');
 57 6d08a0f5 2007-12-07 rsc 				if(p == 0 || s == j->reol)
 58 da7f7882 2007-05-18 devnull 					return match;
 59 da7f7882 2007-05-18 devnull 				s = p+1;
 60 b2cfc4e2 2003-09-30 devnull 				break;
 61 b2cfc4e2 2003-09-30 devnull 			}
 62 b2cfc4e2 2003-09-30 devnull 		}
 63 b2cfc4e2 2003-09-30 devnull 
 64 b2cfc4e2 2003-09-30 devnull 		r = *s;
 65 b2cfc4e2 2003-09-30 devnull 
 66 b2cfc4e2 2003-09-30 devnull 		/* switch run lists */
 67 b2cfc4e2 2003-09-30 devnull 		tl = j->relist[flag];
 68 b2cfc4e2 2003-09-30 devnull 		tle = j->reliste[flag];
 69 b2cfc4e2 2003-09-30 devnull 		nl = j->relist[flag^=1];
 70 b2cfc4e2 2003-09-30 devnull 		nle = j->reliste[flag];
 71 b2cfc4e2 2003-09-30 devnull 		nl->inst = 0;
 72 b2cfc4e2 2003-09-30 devnull 
 73 b2cfc4e2 2003-09-30 devnull 		/* Add first instruction to current list */
 74 6d08a0f5 2007-12-07 rsc 		_rrenewemptythread(tl, progp->startinst, ms, s);
 75 b2cfc4e2 2003-09-30 devnull 
 76 b2cfc4e2 2003-09-30 devnull 		/* Execute machine until current list is empty */
 77 b2cfc4e2 2003-09-30 devnull 		for(tlp=tl; tlp->inst; tlp++){
 78 6d08a0f5 2007-12-07 rsc 			for(inst=tlp->inst; ; inst = inst->u2.next){
 79 b2cfc4e2 2003-09-30 devnull 				switch(inst->type){
 80 b2cfc4e2 2003-09-30 devnull 				case RUNE:	/* regular character */
 81 b2cfc4e2 2003-09-30 devnull 					if(inst->u1.r == r)
 82 6d08a0f5 2007-12-07 rsc 						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 83 b2cfc4e2 2003-09-30 devnull 							return -1;
 84 b2cfc4e2 2003-09-30 devnull 					break;
 85 b2cfc4e2 2003-09-30 devnull 				case LBRA:
 86 b2cfc4e2 2003-09-30 devnull 					tlp->se.m[inst->u1.subid].s.rsp = s;
 87 b2cfc4e2 2003-09-30 devnull 					continue;
 88 b2cfc4e2 2003-09-30 devnull 				case RBRA:
 89 b2cfc4e2 2003-09-30 devnull 					tlp->se.m[inst->u1.subid].e.rep = s;
 90 b2cfc4e2 2003-09-30 devnull 					continue;
 91 b2cfc4e2 2003-09-30 devnull 				case ANY:
 92 b2cfc4e2 2003-09-30 devnull 					if(r != '\n')
 93 6d08a0f5 2007-12-07 rsc 						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 94 b2cfc4e2 2003-09-30 devnull 							return -1;
 95 b2cfc4e2 2003-09-30 devnull 					break;
 96 b2cfc4e2 2003-09-30 devnull 				case ANYNL:
 97 6d08a0f5 2007-12-07 rsc 					if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 98 b2cfc4e2 2003-09-30 devnull 							return -1;
 99 b2cfc4e2 2003-09-30 devnull 					break;
100 b2cfc4e2 2003-09-30 devnull 				case BOL:
101 b2cfc4e2 2003-09-30 devnull 					if(s == bol || *(s-1) == '\n')
102 b2cfc4e2 2003-09-30 devnull 						continue;
103 b2cfc4e2 2003-09-30 devnull 					break;
104 b2cfc4e2 2003-09-30 devnull 				case EOL:
105 b2cfc4e2 2003-09-30 devnull 					if(s == j->reol || r == 0 || r == '\n')
106 b2cfc4e2 2003-09-30 devnull 						continue;
107 b2cfc4e2 2003-09-30 devnull 					break;
108 b2cfc4e2 2003-09-30 devnull 				case CCLASS:
109 b2cfc4e2 2003-09-30 devnull 					ep = inst->u1.cp->end;
110 b2cfc4e2 2003-09-30 devnull 					for(rp = inst->u1.cp->spans; rp < ep; rp += 2)
111 b2cfc4e2 2003-09-30 devnull 						if(r >= rp[0] && r <= rp[1]){
112 6d08a0f5 2007-12-07 rsc 							if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
113 b2cfc4e2 2003-09-30 devnull 								return -1;
114 b2cfc4e2 2003-09-30 devnull 							break;
115 b2cfc4e2 2003-09-30 devnull 						}
116 b2cfc4e2 2003-09-30 devnull 					break;
117 b2cfc4e2 2003-09-30 devnull 				case NCCLASS:
118 b2cfc4e2 2003-09-30 devnull 					ep = inst->u1.cp->end;
119 b2cfc4e2 2003-09-30 devnull 					for(rp = inst->u1.cp->spans; rp < ep; rp += 2)
120 b2cfc4e2 2003-09-30 devnull 						if(r >= rp[0] && r <= rp[1])
121 b2cfc4e2 2003-09-30 devnull 							break;
122 b2cfc4e2 2003-09-30 devnull 					if(rp == ep)
123 6d08a0f5 2007-12-07 rsc 						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
124 b2cfc4e2 2003-09-30 devnull 							return -1;
125 b2cfc4e2 2003-09-30 devnull 					break;
126 b2cfc4e2 2003-09-30 devnull 				case OR:
127 6d08a0f5 2007-12-07 rsc 					/* evaluate right choice later */
128 27589754 2008-01-10 rsc 					if(_renewthread(tlp, inst->u1.right, ms, &tlp->se) == tle)
129 6d08a0f5 2007-12-07 rsc 						return -1;
130 6d08a0f5 2007-12-07 rsc 					/* efficiency: advance and re-evaluate */
131 6d08a0f5 2007-12-07 rsc 					continue;
132 b2cfc4e2 2003-09-30 devnull 				case END:	/* Match! */
133 b2cfc4e2 2003-09-30 devnull 					match = 1;
134 b2cfc4e2 2003-09-30 devnull 					tlp->se.m[0].e.rep = s;
135 b2cfc4e2 2003-09-30 devnull 					if(mp != 0)
136 b2cfc4e2 2003-09-30 devnull 						_renewmatch(mp, ms, &tlp->se);
137 b2cfc4e2 2003-09-30 devnull 					break;
138 b2cfc4e2 2003-09-30 devnull 				}
139 b2cfc4e2 2003-09-30 devnull 				break;
140 b2cfc4e2 2003-09-30 devnull 			}
141 b2cfc4e2 2003-09-30 devnull 		}
142 b2cfc4e2 2003-09-30 devnull 		if(s == j->reol)
143 b2cfc4e2 2003-09-30 devnull 			break;
144 6d08a0f5 2007-12-07 rsc 		checkstart = j->startchar && nl->inst==0;
145 b2cfc4e2 2003-09-30 devnull 		s++;
146 b2cfc4e2 2003-09-30 devnull 	}while(r);
147 b2cfc4e2 2003-09-30 devnull 	return match;
148 b2cfc4e2 2003-09-30 devnull }
149 b2cfc4e2 2003-09-30 devnull 
150 b2cfc4e2 2003-09-30 devnull static int
151 b2cfc4e2 2003-09-30 devnull rregexec2(Reprog *progp,	/* program to run */
152 b2cfc4e2 2003-09-30 devnull 	Rune *bol,	/* string to run machine on */
153 b2cfc4e2 2003-09-30 devnull 	Resub *mp,	/* subexpression elements */
154 b2cfc4e2 2003-09-30 devnull 	int ms,		/* number of elements at mp */
155 b2cfc4e2 2003-09-30 devnull 	Reljunk *j
156 b2cfc4e2 2003-09-30 devnull )
157 b2cfc4e2 2003-09-30 devnull {
158 6d08a0f5 2007-12-07 rsc 	Relist relist0[5*LISTSIZE], relist1[5*LISTSIZE];
159 b2cfc4e2 2003-09-30 devnull 
160 6d08a0f5 2007-12-07 rsc 	/* mark space */
161 b2cfc4e2 2003-09-30 devnull 	j->relist[0] = relist0;
162 b2cfc4e2 2003-09-30 devnull 	j->relist[1] = relist1;
163 6d08a0f5 2007-12-07 rsc 	j->reliste[0] = relist0 + nelem(relist0) - 2;
164 6d08a0f5 2007-12-07 rsc 	j->reliste[1] = relist1 + nelem(relist1) - 2;
165 b2cfc4e2 2003-09-30 devnull 
166 6d08a0f5 2007-12-07 rsc 	return rregexec1(progp, bol, mp, ms, j);
167 b2cfc4e2 2003-09-30 devnull }
168 b2cfc4e2 2003-09-30 devnull 
169 b2cfc4e2 2003-09-30 devnull extern int
170 b2cfc4e2 2003-09-30 devnull rregexec(Reprog *progp,	/* program to run */
171 b2cfc4e2 2003-09-30 devnull 	Rune *bol,	/* string to run machine on */
172 b2cfc4e2 2003-09-30 devnull 	Resub *mp,	/* subexpression elements */
173 b2cfc4e2 2003-09-30 devnull 	int ms)		/* number of elements at mp */
174 b2cfc4e2 2003-09-30 devnull {
175 b2cfc4e2 2003-09-30 devnull 	Reljunk j;
176 b2cfc4e2 2003-09-30 devnull 	Relist relist0[LISTSIZE], relist1[LISTSIZE];
177 b2cfc4e2 2003-09-30 devnull 	int rv;
178 b2cfc4e2 2003-09-30 devnull 
179 b2cfc4e2 2003-09-30 devnull 	/*
180 b2cfc4e2 2003-09-30 devnull  	 *  use user-specified starting/ending location if specified
181 b2cfc4e2 2003-09-30 devnull 	 */
182 b2cfc4e2 2003-09-30 devnull 	j.rstarts = bol;
183 b2cfc4e2 2003-09-30 devnull 	j.reol = 0;
184 b2cfc4e2 2003-09-30 devnull 	if(mp && ms>0){
185 b2cfc4e2 2003-09-30 devnull 		if(mp->s.sp)
186 b2cfc4e2 2003-09-30 devnull 			j.rstarts = mp->s.rsp;
187 b2cfc4e2 2003-09-30 devnull 		if(mp->e.ep)
188 b2cfc4e2 2003-09-30 devnull 			j.reol = mp->e.rep;
189 b2cfc4e2 2003-09-30 devnull 	}
190 b2cfc4e2 2003-09-30 devnull 	j.starttype = 0;
191 b2cfc4e2 2003-09-30 devnull 	j.startchar = 0;
192 62390091 2004-03-05 devnull 	if(progp->startinst->type == RUNE && progp->startinst->u1.r < Runeself) {
193 b2cfc4e2 2003-09-30 devnull 		j.starttype = RUNE;
194 b2cfc4e2 2003-09-30 devnull 		j.startchar = progp->startinst->u1.r;
195 b2cfc4e2 2003-09-30 devnull 	}
196 b2cfc4e2 2003-09-30 devnull 	if(progp->startinst->type == BOL)
197 b2cfc4e2 2003-09-30 devnull 		j.starttype = BOL;
198 b2cfc4e2 2003-09-30 devnull 
199 b2cfc4e2 2003-09-30 devnull 	/* mark space */
200 b2cfc4e2 2003-09-30 devnull 	j.relist[0] = relist0;
201 b2cfc4e2 2003-09-30 devnull 	j.relist[1] = relist1;
202 6d08a0f5 2007-12-07 rsc 	j.reliste[0] = relist0 + nelem(relist0) - 2;
203 6d08a0f5 2007-12-07 rsc 	j.reliste[1] = relist1 + nelem(relist1) - 2;
204 b2cfc4e2 2003-09-30 devnull 
205 b2cfc4e2 2003-09-30 devnull 	rv = rregexec1(progp, bol, mp, ms, &j);
206 b2cfc4e2 2003-09-30 devnull 	if(rv >= 0)
207 b2cfc4e2 2003-09-30 devnull 		return rv;
208 b2cfc4e2 2003-09-30 devnull 	rv = rregexec2(progp, bol, mp, ms, &j);
209 b2cfc4e2 2003-09-30 devnull 	if(rv >= 0)
210 b2cfc4e2 2003-09-30 devnull 		return rv;
211 b2cfc4e2 2003-09-30 devnull 	return -1;
212 b2cfc4e2 2003-09-30 devnull }