Update libregexp with recent bug fixes from Plan 9.

12 files changed, 111 insertions, 81 deletions

diff --git a/src/libregexp/lib9.h b/src/libregexp/lib9.h
index d022656d..07a1f976 100644
--- a/src/libregexp/lib9.h
+++ b/src/libregexp/lib9.h
@@ -1,6 +1,2 @@
-#include <fmt.h>
-#include <setjmp.h>
-#include <string.h>
-#include <stdlib.h>
-#include <unistd.h>
-
+#include <u.h>
+#include <libc.h>
diff --git a/src/libregexp/lib9.h.std b/src/libregexp/lib9.h.std
new file mode 100644
index 00000000..d022656d
--- /dev/null
+++ b/src/libregexp/lib9.h.std
@@ -0,0 +1,6 @@
+#include <fmt.h>
+#include <setjmp.h>
+#include <string.h>
+#include <stdlib.h>
+#include <unistd.h>
+
diff --git a/src/libregexp/mkfile b/src/libregexp/mkfile
index 8cb740cc..ad482240 100644
--- a/src/libregexp/mkfile
+++ b/src/libregexp/mkfile
@@ -9,7 +9,6 @@ OFILES=\
 	regexec.$O\
 	regsub.$O\
 	regaux.$O\
-	rregaux.$O\
 	rregexec.$O\
 	rregsub.$O\
 
diff --git a/src/libregexp/regaux.c b/src/libregexp/regaux.c
index 956c1eb0..b854b5ac 100644
--- a/src/libregexp/regaux.c
+++ b/src/libregexp/regaux.c
@@ -30,19 +30,27 @@ _renewmatch(Resub *mp, int ms, Resublist *sp)
 extern Relist*
 _renewthread(Relist *lp,	/* _relist to add to */
 	Reinst *ip,		/* instruction to add */
+	int ms,
 	Resublist *sep)		/* pointers to subexpressions */
 {
 	Relist *p;
 
 	for(p=lp; p->inst; p++){
 		if(p->inst == ip){
-			if((sep)->m[0].s.sp < p->se.m[0].s.sp)
-				p->se = *sep;
+			if(sep->m[0].s.sp < p->se.m[0].s.sp){
+				if(ms > 1)
+					p->se = *sep;
+				else
+					p->se.m[0] = sep->m[0];
+			}
 			return 0;
 		}
 	}
 	p->inst = ip;
-	p->se = *sep;
+	if(ms > 1)
+		p->se = *sep;
+	else
+		p->se.m[0] = sep->m[0];
 	(++p)->inst = 0;
 	return p;
 }
@@ -54,6 +62,7 @@ _renewthread(Relist *lp,	/* _relist to add to */
 extern Relist*
 _renewemptythread(Relist *lp,	/* _relist to add to */
 	Reinst *ip,		/* instruction to add */
+	int ms,
 	char *sp)		/* pointers to subexpressions */
 {
 	Relist *p;
@@ -61,16 +70,43 @@ _renewemptythread(Relist *lp,	/* _relist to add to */
 	for(p=lp; p->inst; p++){
 		if(p->inst == ip){
 			if(sp < p->se.m[0].s.sp) {
-				memset((void *)&p->se, 0, sizeof(p->se));
+				if(ms > 1)
+					memset(&p->se, 0, sizeof(p->se));
 				p->se.m[0].s.sp = sp;
 			}
 			return 0;
 		}
 	}
 	p->inst = ip;
-	memset((void *)&p->se, 0, sizeof(p->se));
+	if(ms > 1)
+		memset(&p->se, 0, sizeof(p->se));
 	p->se.m[0].s.sp = sp;
 	(++p)->inst = 0;
 	return p;
 }
 
+extern Relist*
+_rrenewemptythread(Relist *lp,	/* _relist to add to */
+	Reinst *ip,		/* instruction to add */
+	int ms,
+	Rune *rsp)		/* pointers to subexpressions */
+{
+	Relist *p;
+
+	for(p=lp; p->inst; p++){
+		if(p->inst == ip){
+			if(rsp < p->se.m[0].s.rsp) {
+				if(ms > 1)
+					memset(&p->se, 0, sizeof(p->se));
+				p->se.m[0].s.rsp = rsp;
+			}
+			return 0;
+		}
+	}
+	p->inst = ip;
+	if(ms > 1)
+		memset(&p->se, 0, sizeof(p->se));
+	p->se.m[0].s.rsp = rsp;
+	(++p)->inst = 0;
+	return p;
+}
diff --git a/src/libregexp/regcomp.c b/src/libregexp/regcomp.c
index 6c6939c6..796efc41 100644
--- a/src/libregexp/regcomp.c
+++ b/src/libregexp/regcomp.c
@@ -15,8 +15,6 @@ struct Node
 	Reinst*	last;
 }Node;
 
-Reprog	RePrOg;
-
 #define	NSTACK	20
 static	Node	andstack[NSTACK];
 static	Node	*andp;
@@ -252,7 +250,7 @@ optimize(Reprog *pp)
 	 *  and then relocate the code.
 	 */
 	size = sizeof(Reprog) + (freep - pp->firstinst)*sizeof(Reinst);
-	npp = (Reprog *)realloc(pp, size);
+	npp = realloc(pp, size);
 	if(npp==0 || npp==pp)
 		return pp;
 	diff = (char *)npp - (char *)pp;
@@ -303,12 +301,12 @@ dump(Reprog *pp)
 		print("%d:\t0%o\t%d\t%d", l-pp->firstinst, l->type,
 			l->u2.left-pp->firstinst, l->u1.right-pp->firstinst);
 		if(l->type == RUNE)
-			print("\t%C\n", l->r);
+			print("\t%C\n", l->u1.r);
 		else if(l->type == CCLASS || l->type == NCCLASS){
 			print("\t[");
 			if(l->type == NCCLASS)
 				print("^");
-			for(p = l->cp->spans; p < l->cp->end; p += 2)
+			for(p = l->u1.cp->spans; p < l->u1.cp->end; p += 2)
 				if(p[0] == p[1])
 					print("%C", p[0]);
 				else
@@ -477,7 +475,7 @@ regcomp1(char *s, int literal, int dot_type)
 	Reprog *pp;
 
 	/* get memory for the program */
-	pp = (Reprog *)malloc(sizeof(Reprog) + 6*sizeof(Reinst)*strlen(s));
+	pp = malloc(sizeof(Reprog) + 6*sizeof(Reinst)*strlen(s));
 	if(pp == 0){
 		regerror("out of memory");
 		return 0;
diff --git a/src/libregexp/regcomp.h b/src/libregexp/regcomp.h
index a6728b0e..6c88cd09 100644
--- a/src/libregexp/regcomp.h
+++ b/src/libregexp/regcomp.h
@@ -68,7 +68,7 @@ struct	Reljunk
 	Rune*	reol;
 };
 
-extern Relist*	_renewthread(Relist*, Reinst*, Resublist*);
+extern Relist*	_renewthread(Relist*, Reinst*, int, Resublist*);
 extern void	_renewmatch(Resub*, int, Resublist*);
-extern Relist*	_renewemptythread(Relist*, Reinst*, char*);
-extern Relist*	_rrenewemptythread(Relist*, Reinst*, Rune*);
+extern Relist*	_renewemptythread(Relist*, Reinst*, int, char*);
+extern Relist*	_rrenewemptythread(Relist*, Reinst*, int, Rune*);
diff --git a/src/libregexp/regerror.c b/src/libregexp/regerror.c
index 2cd1e3e2..99ff0c74 100644
--- a/src/libregexp/regerror.c
+++ b/src/libregexp/regerror.c
@@ -10,5 +10,5 @@ regerror(char *s)
 	strcat(buf, s);
 	strcat(buf, "\n");
 	write(2, buf, strlen(buf));
-	exit(1);
+	exits("regerr");
 }
diff --git a/src/libregexp/regexec.c b/src/libregexp/regexec.c
index c9f1eba2..62ab1a31 100644
--- a/src/libregexp/regexec.c
+++ b/src/libregexp/regexec.c
@@ -48,7 +48,7 @@ regexec1(Reprog *progp,	/* program to run */
 			switch(j->starttype) {
 			case RUNE:
 				p = utfrune(s, j->startchar);
-				if(p == 0)
+				if(p == 0 || s == j->eol)
 					return match;
 				s = p;
 				break;
@@ -56,14 +56,14 @@ regexec1(Reprog *progp,	/* program to run */
 				if(s == bol)
 					break;
 				p = utfrune(s, '\n');
-				if(p == 0)
+				if(p == 0 || s == j->eol)
 					return match;
 				s = p;
 				break;
 			}
 		}
 		r = *(uchar*)s;
-		if(r < (Rune)Runeself)
+		if(r < Runeself)
 			n = 1;
 		else
 			n = chartorune(&r, s);
@@ -77,7 +77,7 @@ regexec1(Reprog *progp,	/* program to run */
 
 		/* Add first instruction to current list */
 		if(match == 0)
-			_renewemptythread(tl, progp->startinst, s);
+			_renewemptythread(tl, progp->startinst, ms, s);
 
 		/* Execute machine until current list is empty */
 		for(tlp=tl; tlp->inst; tlp++){	/* assignment = */
@@ -85,7 +85,7 @@ regexec1(Reprog *progp,	/* program to run */
 				switch(inst->type){
 				case RUNE:	/* regular character */
 					if(inst->u1.r == r){
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					}
 					break;
@@ -97,11 +97,11 @@ regexec1(Reprog *progp,	/* program to run */
 					continue;
 				case ANY:
 					if(r != '\n')
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case ANYNL:
-					if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+					if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case BOL:
@@ -116,7 +116,7 @@ regexec1(Reprog *progp,	/* program to run */
 					ep = inst->u1.cp->end;
 					for(rp = inst->u1.cp->spans; rp < ep; rp += 2)
 						if(r >= rp[0] && r <= rp[1]){
-							if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+							if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 								return -1;
 							break;
 						}
@@ -127,12 +127,12 @@ regexec1(Reprog *progp,	/* program to run */
 						if(r >= rp[0] && r <= rp[1])
 							break;
 					if(rp == ep)
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case OR:
 					/* evaluate right choice later */
-					if(_renewthread(tlp, inst->u1.right, &tlp->se) == tle)
+					if(_renewthread(tlp, inst->u1.right, ms, &tlp->se) == tle)
 						return -1;
 					/* efficiency: advance and re-evaluate */
 					continue;
@@ -162,15 +162,27 @@ regexec2(Reprog *progp,	/* program to run */
 	Reljunk *j
 )
 {
-	Relist relist0[BIGLISTSIZE], relist1[BIGLISTSIZE];
+	int rv;
+	Relist *relist0, *relist1;
 
 	/* mark space */
+	relist0 = malloc(BIGLISTSIZE*sizeof(Relist));
+	if(relist0 == nil)
+		return -1;
+	relist1 = malloc(BIGLISTSIZE*sizeof(Relist));
+	if(relist1 == nil){
+		free(relist1);
+		return -1;
+	}
 	j->relist[0] = relist0;
 	j->relist[1] = relist1;
-	j->reliste[0] = relist0 + nelem(relist0) - 2;
-	j->reliste[1] = relist1 + nelem(relist1) - 2;
+	j->reliste[0] = relist0 + BIGLISTSIZE - 2;
+	j->reliste[1] = relist1 + BIGLISTSIZE - 2;
 
-	return regexec1(progp, bol, mp, ms, j);
+	rv = regexec1(progp, bol, mp, ms, j);
+	free(relist0);
+	free(relist1);
+	return rv;
 }
 
 extern int
@@ -196,7 +208,7 @@ regexec(Reprog *progp,	/* program to run */
 	}
 	j.starttype = 0;
 	j.startchar = 0;
-	if(progp->startinst->type == RUNE && progp->startinst->u1.r < (Rune)Runeself) {
+	if(progp->startinst->type == RUNE && progp->startinst->u1.r < Runeself) {
 		j.starttype = RUNE;
 		j.startchar = progp->startinst->u1.r;
 	}
diff --git a/src/libregexp/regsub.c b/src/libregexp/regsub.c
index 6de2c957..ab5ad99d 100644
--- a/src/libregexp/regsub.c
+++ b/src/libregexp/regsub.c
@@ -53,9 +53,10 @@ regsub(char *sp,	/* source string */
 				     ssp < mp[0].e.ep; ssp++)
 					if(dp < ep)
 						*dp++ = *ssp;
-		}else
+		}else{
 			if(dp < ep)
 				*dp++ = *sp;
+		}
 		sp++;
 	}
 	*dp = '\0';
diff --git a/src/libregexp/rregaux.c b/src/libregexp/rregaux.c
deleted file mode 100644
index f4cb0068..00000000
--- a/src/libregexp/rregaux.c
+++ /dev/null
@@ -1,26 +0,0 @@
-#include "lib9.h"
-#include "regexp9.h"
-#include "regcomp.h"
-
-extern Relist*
-_rrenewemptythread(Relist *lp,	/* _relist to add to */
-	Reinst *ip,		/* instruction to add */
-	Rune *rsp)		/* pointers to subexpressions */
-{
-	Relist *p;
-
-	for(p=lp; p->inst; p++){
-		if(p->inst == ip){
-			if(rsp < p->se.m[0].s.rsp) {
-				memset((void *)&p->se, 0, sizeof(p->se));
-				p->se.m[0].s.rsp = rsp;
-			}
-			return 0;
-		}
-	}
-	p->inst = ip;
-	memset((void *)&p->se, 0, sizeof(p->se));
-	p->se.m[0].s.rsp = rsp;
-	(++p)->inst = 0;
-	return p;
-}
diff --git a/src/libregexp/rregexec.c b/src/libregexp/rregexec.c
index e96c9721..eece0eb9 100644
--- a/src/libregexp/rregexec.c
+++ b/src/libregexp/rregexec.c
@@ -45,7 +45,7 @@ rregexec1(Reprog *progp,	/* program to run */
 			switch(j->starttype) {
 			case RUNE:
 				while(*s != j->startchar) {
-					if(*s == 0)
+					if(*s == 0 || s == j->reol)
 						return match;
 					s++;
 				}
@@ -54,7 +54,7 @@ rregexec1(Reprog *progp,	/* program to run */
 				if(s == bol)
 					break;
 				while(*s != '\n') {
-					if(*s == 0)
+					if(*s == 0 || s == j->reol)
 						return match;
 					s++;
 				}
@@ -72,7 +72,7 @@ rregexec1(Reprog *progp,	/* program to run */
 		nl->inst = 0;
 
 		/* Add first instruction to current list */
-		_rrenewemptythread(tl, progp->startinst, s);
+		_rrenewemptythread(tl, progp->startinst, ms, s);
 
 		/* Execute machine until current list is empty */
 		for(tlp=tl; tlp->inst; tlp++){
@@ -80,7 +80,7 @@ rregexec1(Reprog *progp,	/* program to run */
 				switch(inst->type){
 				case RUNE:	/* regular character */
 					if(inst->u1.r == r)
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case LBRA:
@@ -91,11 +91,11 @@ rregexec1(Reprog *progp,	/* program to run */
 					continue;
 				case ANY:
 					if(r != '\n')
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case ANYNL:
-					if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+					if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case BOL:
@@ -110,7 +110,7 @@ rregexec1(Reprog *progp,	/* program to run */
 					ep = inst->u1.cp->end;
 					for(rp = inst->u1.cp->spans; rp < ep; rp += 2)
 						if(r >= rp[0] && r <= rp[1]){
-							if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+							if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 								return -1;
 							break;
 						}
@@ -121,12 +121,12 @@ rregexec1(Reprog *progp,	/* program to run */
 						if(r >= rp[0] && r <= rp[1])
 							break;
 					if(rp == ep)
-						if(_renewthread(nl, inst->u2.next, &tlp->se)==nle)
+						if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
 							return -1;
 					break;
 				case OR:
 					/* evaluate right choice later */
-					if(_renewthread(tlp, inst->u1.right, &tlp->se) == tle)
+					if(_renewthread(tlp, inst->u1.right, ms, &tlp->se) == tle)
 						return -1;
 					/* efficiency: advance and re-evaluate */
 					continue;
@@ -190,7 +190,7 @@ rregexec(Reprog *progp,	/* program to run */
 	}
 	j.starttype = 0;
 	j.startchar = 0;
-	if(progp->startinst->type == RUNE && progp->startinst->u1.r < (Rune)Runeself) {
+	if(progp->startinst->type == RUNE && progp->startinst->u1.r < Runeself) {
 		j.starttype = RUNE;
 		j.startchar = progp->startinst->u1.r;
 	}
diff --git a/src/libregexp/rregsub.c b/src/libregexp/rregsub.c
index 15f3c174..5a4a564d 100644
--- a/src/libregexp/rregsub.c
+++ b/src/libregexp/rregsub.c
@@ -5,12 +5,14 @@
 extern	void
 rregsub(Rune *sp,	/* source string */
 	Rune *dp,	/* destination string */
+	int dlen,
 	Resub *mp,	/* subexpression elements */
 	int ms)		/* number of elements pointed to by mp */
 {
-	Rune *ssp;
+	Rune *ssp, *ep;
 	int i;
 
+	ep = dp+(dlen/sizeof(Rune))-1;
 	while(*sp != '\0'){
 		if(*sp == '\\'){
 			switch(*++sp){
@@ -29,16 +31,19 @@ rregsub(Rune *sp,	/* source string */
 					for(ssp = mp[i].s.rsp;
 					     ssp < mp[i].e.rep;
 					     ssp++)
-						*dp++ = *ssp;
+						if(dp < ep)
+							*dp++ = *ssp;
 				break;
 			case '\\':
-				*dp++ = '\\';
+				if(dp < ep)
+					*dp++ = '\\';
 				break;
 			case '\0':
 				sp--;
 				break;
 			default:
-				*dp++ = *sp;
+				if(dp < ep)
+					*dp++ = *sp;
 				break;
 			}
 		}else if(*sp == '&'){				
@@ -46,9 +51,12 @@ rregsub(Rune *sp,	/* source string */
 			if(mp[0].s.rsp != 0)
 				for(ssp = mp[0].s.rsp;
 				     ssp < mp[0].e.rep; ssp++)
-					*dp++ = *ssp;
-		}else
-			*dp++ = *sp;
+					if(dp < ep)
+						*dp++ = *ssp;
+		}else{
+			if(dp < ep)
+				*dp++ = *sp;
+		}
 		sp++;
 	}
 	*dp = '\0';