o5r/plan9port
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

libregexp: fix ambiguous match selection

Browse source 
echo SYSSYSR1 | sed 's/SYS.+/sysr1/'

was producing SYSsysr1 instead of sysr1.
Bug was introduced during overflow cleanup earlier this year.

Also bring regexec.c and rregexec.c into sync again.
Also allocate large enough lists in the regexec2/rregexec2 case.
This commit is contained in:
Russ Cox 2007-12-07 15:32:45 -05:00
parent 6c6117397f
commit a7511dd43d
5 changed files with 117 additions and 113 deletions
Download patch file Download diff file Expand all files Collapse all files

48
src/libregexp/regexec.c
View file

@ -2,7 +2,6 @@
#include "regexp9.h"
#include "regcomp.h"
/*
* return 0 if no match
* >0 if a match
@ -13,16 +12,14 @@ regexec1(Reprog *progp, /* program to run */
char *bol, /* string to run machine on */
Resub *mp, /* subexpression elements */
int ms, /* number of elements at mp */
Reljunk *j
)
Reljunk *j)
{
int flag=0;
Reinst *inst;
Relist *tlp;
char *s;
int i, checkstart;
int i, checkstart, n;
Rune r, *rp, *ep;
int n;
Relist* tl; /* This list, next list */
Relist* nl;
Relist* tle; /* ends of this and next list */
@ -48,7 +45,7 @@ regexec1(Reprog *progp, /* program to run */
switch(j->starttype) {
case RUNE:
p = utfrune(s, j->startchar);
if(p == 0 || s == j->eol)
if(p == 0 || (j->eol && p >= j->eol))
return match;
s = p;
break;
@ -56,7 +53,7 @@ regexec1(Reprog *progp, /* program to run */
if(s == bol)
break;
p = utfrune(s, '\n');
if(p == 0 || s == j->eol)
if(p == 0 || (j->eol && p+1 >= j->eol))
return match;
s = p+1;
break;
@ -77,17 +74,16 @@ regexec1(Reprog *progp, /* program to run */
/* Add first instruction to current list */
if(match == 0)
_renewemptythread(tl, progp->startinst, ms, s);
_renewemptythread(tl, tle, progp->startinst, ms, s);
/* Execute machine until current list is empty */
for(tlp=tl; tlp->inst; tlp++){ /* assignment = */
for(tlp=tl; tlp->inst; tlp++){
for(inst = tlp->inst; ; inst = inst->u2.next){
switch(inst->type){
case RUNE: /* regular character */
if(inst->u1.r == r){
if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
if(inst->u1.r == r)
if(_renewthread(nl, nle, inst->u2.next, ms, &tlp->se) < 0)
return -1;
}
break;
case LBRA:
tlp->se.m[inst->u1.subid].s.sp = s;
@ -97,11 +93,11 @@ regexec1(Reprog *progp, /* program to run */
continue;
case ANY:
if(r != '\n')
if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
if(_renewthread(nl, nle, inst->u2.next, ms, &tlp->se) < 0)
return -1;
break;
case ANYNL:
if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
if(_renewthread(nl, nle, inst->u2.next, ms, &tlp->se) < 0)
return -1;
break;
case BOL:
@ -116,7 +112,7 @@ 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, ms, &tlp->se)==nle)
if(_renewthread(nl, nle, inst->u2.next, ms, &tlp->se) < 0)
return -1;
break;
}
@ -127,15 +123,12 @@ regexec1(Reprog *progp, /* program to run */
if(r >= rp[0] && r <= rp[1])
break;
if(rp == ep)
if(_renewthread(nl, inst->u2.next, ms, &tlp->se)==nle)
if(_renewthread(nl, nle, inst->u2.next, ms, &tlp->se) < 0)
return -1;
break;
case OR:
/* evaluate right choice later */
if(_renewthread(tl, inst->u1.right, ms, &tlp->se) == tle)
return -1;
/* efficiency: advance and re-evaluate */
continue;
/* expanded during renewthread; just a place holder */
break;
case END: /* Match! */
match = 1;
tlp->se.m[0].e.ep = s;
@ -165,19 +158,18 @@ regexec2(Reprog *progp, /* program to run */
int rv;
Relist *relist0, *relist1;
/* mark space */
relist0 = malloc(BIGLISTSIZE*sizeof(Relist));
relist0 = malloc((progp->proglen+1)*sizeof(Relist));
if(relist0 == nil)
return -1;
relist1 = malloc(BIGLISTSIZE*sizeof(Relist));
relist1 = malloc((progp->proglen+1)*sizeof(Relist));
if(relist1 == nil){
free(relist1);
return -1;
}
j->relist[0] = relist0;
j->relist[1] = relist1;
j->reliste[0] = relist0 + BIGLISTSIZE - 2;
j->reliste[1] = relist1 + BIGLISTSIZE - 2;
j->reliste[0] = relist0 + progp->proglen;
j->reliste[1] = relist1 + progp->proglen;
rv = regexec1(progp, bol, mp, ms, j);
free(relist0);
@ -218,8 +210,8 @@ regexec(Reprog *progp, /* program to run */
/* mark space */
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 + nelem(relist0) - 1;
j.reliste[1] = relist1 + nelem(relist1) - 1;
rv = regexec1(progp, bol, mp, ms, &j);
if(rv >= 0)