This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
[perl #23769] Unicode regex broken on simple example
[perl5.git] / pp_hot.c
index c70b227..0ad2fcf 100644 (file)
--- a/pp_hot.c
+++ b/pp_hot.c
@@ -1234,7 +1234,7 @@ PP(pp_match)
     if (SvSCREAM(TARG))
        r_flags |= REXEC_SCREAM;
 
-    if ((bool)(pm->op_pmflags & PMf_MULTILINE) != (bool)PL_multiline) {
+    if ((int)(pm->op_pmflags & PMf_MULTILINE) != PL_multiline) {
        SAVEINT(PL_multiline);
        PL_multiline = pm->op_pmflags & PMf_MULTILINE;
     }
@@ -1593,6 +1593,17 @@ Perl_do_readline(pTHX)
                (void)POPs;             /* Unmatched wildcard?  Chuck it... */
                continue;
            }
+       } else if (SvUTF8(sv)) { /* OP_READLINE, OP_RCATLINE */
+            U8 *s = (U8*)SvPVX(sv) + offset;
+            STRLEN len = SvCUR(sv) - offset;
+            U8 *f;
+            
+            if (ckWARN(WARN_UTF8) &&
+                !Perl_is_utf8_string_loc(aTHX_ s, len, &f))
+                 /* Emulate :encoding(utf8) warning in the same case. */
+                 Perl_warner(aTHX_ packWARN(WARN_UTF8),
+                             "utf8 \"\\x%02X\" does not map to Unicode",
+                             f < (U8*)SvEND(sv) ? *f : 0);
        }
        if (gimme == G_ARRAY) {
            if (SvLEN(sv) - SvCUR(sv) > 20) {
@@ -1979,7 +1990,7 @@ PP(pp_subst)
               ? REXEC_COPY_STR : 0;
     if (SvSCREAM(TARG))
        r_flags |= REXEC_SCREAM;
-    if ((bool)(pm->op_pmflags & PMf_MULTILINE) != (bool)PL_multiline) {
+    if ((int)(pm->op_pmflags & PMf_MULTILINE) != PL_multiline) {
        SAVEINT(PL_multiline);
        PL_multiline = pm->op_pmflags & PMf_MULTILINE;
     }