Patch from Inaba Hiroto:

- canonical UTF-8 hash keys: if a key string for a hash is
  UTF8-on, try downgrade the string and use it if
  unicode::distinct is not in effect.
  For the task, I added a function bytes_from_utf8() to utf8.c.
  It might resemble utf8_to_bytes() but it is not convenient
  to the task.
  Made a test for it and added to t/op/each.t
- Changed do_print in doio.c to apply sv_utf8_(downgrade|upgrade) to
  the mortal copy of the argument SV.
  And changed t/io/utf8.t test 18 which expects print() to
  upgrade its argument.
- re-implement sv_eq with bytes_from_utf8()
- some bug fixes
  - tr/// does not handle UTF8 range (\x{}-\x{})
  - \ before raw UTF8 character produced
    "Malformed UTF-8 character" warning.
  - "\x{100}\N{CENT SIGN}" is Malformed.
    Added tests for these 3.
  - and one silly bug (by me) with qu operator.

p4raw-id: //depot/perl@8583

1 files changed, 5 insertions, 6 deletions

diff --git a/doio.c b/doio.c
index 6056ea704c..a1d0e466ea 100644
--- a/doio.c
+++ b/doio.c
@@ -1169,13 +1169,12 @@ Perl_do_print(pTHX_ register SV *sv, PerlIO *fp)
 	/* FALL THROUGH */
     default:
 	if (PerlIO_isutf8(fp)) {
-	    tmps = SvPVutf8(sv, len);
-	}
-	else {
-	    if (DO_UTF8(sv))
-		sv_utf8_downgrade(sv, FALSE);
-	    tmps = SvPV(sv, len);
+	    if (!SvUTF8(sv))
+		sv_utf8_upgrade(sv = sv_mortalcopy(sv));
 	}
+	else if (DO_UTF8(sv))
+	    sv_utf8_downgrade((sv = sv_mortalcopy(sv)), FALSE);
+	tmps = SvPV(sv, len);
 	break;
     }
     /* To detect whether the process is about to overstep its