Re: Regex-Unicode bugs

Message-Id: <200203271239.g2RCdvL31355@crypt.compulink.co.uk>

p4raw-id: //depot/perl@15573

diff --git a/pod/perldiag.pod b/pod/perldiag.pod
index 7808f88..d140b0a 100644 (file)
--- a/pod/perldiag.pod
+++ b/pod/perldiag.pod
@@ -1328,7 +1328,9 @@ unlikely to be what you want.
 
 =item Empty %s
 
-(F) Empty C<\p{}> or C<\P{}>.
+(F) C<\p> and C<\P> are used to introduce a named Unicode property, as
+described in L<perlunicode> and L<perlre>. You used C<\p> or C<\P> in
+a regular expression without specifying the property name.
 
 =item entering effective %s failed
 

diff --git a/regcomp.c b/regcomp.c
index 12f9016..c124d48 100644 (file)
--- a/regcomp.c
+++ b/regcomp.c
@@ -2970,8 +2970,11 @@ tryagain:
                    }
                    RExC_end++;
                }
-               else
+               else {
                    RExC_end = RExC_parse + 2;
+                   if (RExC_end > oldregxend)
+                       RExC_end = oldregxend;
+               }
                RExC_parse--;
 
                ret = regclass(pRExC_state);
@@ -3593,6 +3596,8 @@ S_regclass(pTHX_ RExC_state_t *pRExC_state)
            case 'D':   namedclass = ANYOF_NDIGIT;      break;
            case 'p':
            case 'P':
+               if (RExC_parse >= RExC_end)
+                   vFAIL("Empty \\p{}");
                if (*RExC_parse == '{') {
                    U8 c = (U8)value;
                    e = strchr(RExC_parse++, '}');