tr/// UTF-8 tweaking from Inaba Hiroto.

p4raw-id: //depot/perl@9499

4 files changed, 13 insertions, 32 deletions

diff --git a/embed.h b/embed.h
index c8015c5a3b..8ec511b2c0 100644
--- a/embed.h
+++ b/embed.h
@@ -909,7 +909,6 @@
 #define scalarboolean		S_scalarboolean
 #define too_few_arguments	S_too_few_arguments
 #define too_many_arguments	S_too_many_arguments
-#define trlist_upgrade		S_trlist_upgrade
 #define op_clear		S_op_clear
 #define null			S_null
 #define pad_addlex		S_pad_addlex
@@ -2393,7 +2392,6 @@
 #define scalarboolean(a)	S_scalarboolean(aTHX_ a)
 #define too_few_arguments(a,b)	S_too_few_arguments(aTHX_ a,b)
 #define too_many_arguments(a,b)	S_too_many_arguments(aTHX_ a,b)
-#define trlist_upgrade(a,b)	S_trlist_upgrade(aTHX_ a,b)
 #define op_clear(a)		S_op_clear(aTHX_ a)
 #define null(a)			S_null(aTHX_ a)
 #define pad_addlex(a)		S_pad_addlex(aTHX_ a)
@@ -4681,8 +4679,6 @@
 #define too_few_arguments	S_too_few_arguments
 #define S_too_many_arguments	CPerlObj::S_too_many_arguments
 #define too_many_arguments	S_too_many_arguments
-#define S_trlist_upgrade	CPerlObj::S_trlist_upgrade
-#define trlist_upgrade		S_trlist_upgrade
 #define S_op_clear		CPerlObj::S_op_clear
 #define op_clear		S_op_clear
 #define S_null			CPerlObj::S_null
diff --git a/embed.pl b/embed.pl
index 552c0a526f..36821923f4 100755
--- a/embed.pl
+++ b/embed.pl
@@ -2292,7 +2292,6 @@ s	|OP*	|no_fh_allowed	|OP *o
 s	|OP*	|scalarboolean	|OP *o
 s	|OP*	|too_few_arguments|OP *o|char* name
 s	|OP*	|too_many_arguments|OP *o|char* name
-s	|U8*	|trlist_upgrade	|U8** sp|U8** ep
 s	|void	|op_clear	|OP* o
 s	|void	|null		|OP* o
 s	|PADOFFSET|pad_addlex	|SV* name
diff --git a/op.c b/op.c
index 88d773c6af..155f107b1f 100644
--- a/op.c
+++ b/op.c
@@ -103,30 +103,6 @@ S_no_bareword_allowed(pTHX_ OP *o)
 		     SvPV_nolen(cSVOPo_sv)));
 }
 
-STATIC U8*
-S_trlist_upgrade(pTHX_ U8** sp, U8** ep)
-{
-    U8 *s = *sp;
-    U8 *e = *ep;
-    U8 *d;
-
-    Newz(801, d, (e - s) * 2, U8);
-    *sp = d;
-
-    while (s < e) {
-        if (NATIVE_IS_INVARIANT(*s) || NATIVE_TO_UTF(*s) == 0xff)
-            *d++ = *s++;
-	else {
-	    U8 c = NATIVE_TO_ASCII(*s++);
-	    *d++ = UTF8_EIGHT_BIT_HI(c);
-	    *d++ = UTF8_EIGHT_BIT_LO(c);
-        }
-    }
-    *ep = d;
-    return *sp;
-}
-
-
 /* "register" allocation */
 
 PADOFFSET
@@ -2708,8 +2684,19 @@ Perl_pmtrans(pTHX_ OP *o, OP *expr, OP *repl)
 	U32 final;
 	I32 from_utf	= o->op_private & OPpTRANS_FROM_UTF;
 	I32 to_utf	= o->op_private & OPpTRANS_TO_UTF;
-	U8* tsave = from_utf ? NULL : trlist_upgrade(&t, &tend);
-	U8* rsave = (to_utf || !rlen) ? NULL : trlist_upgrade(&r, &rend);
+	U8* tsave = NULL;
+	U8* rsave = NULL;
+
+	if (!from_utf) {
+	    STRLEN len = tlen;
+	    tsave = t = bytes_to_utf8(t, &len);
+	    tend = t + len;
+	}
+	if (!to_utf && rlen) {
+	    STRLEN len = rlen;
+	    rsave = r = bytes_to_utf8(r, &len);
+	    rend = r + len;
+	}
 
 /* There are several snags with this code on EBCDIC:
    1. 0xFF is a legal UTF-EBCDIC byte (there are no illegal bytes).
diff --git a/proto.h b/proto.h
index 9be4cd7931..c8a59a454c 100644
--- a/proto.h
+++ b/proto.h
@@ -1023,7 +1023,6 @@ STATIC OP*	S_no_fh_allowed(pTHX_ OP *o);
 STATIC OP*	S_scalarboolean(pTHX_ OP *o);
 STATIC OP*	S_too_few_arguments(pTHX_ OP *o, char* name);
 STATIC OP*	S_too_many_arguments(pTHX_ OP *o, char* name);
-STATIC U8*	S_trlist_upgrade(pTHX_ U8** sp, U8** ep);
 STATIC void	S_op_clear(pTHX_ OP* o);
 STATIC void	S_null(pTHX_ OP* o);
 STATIC PADOFFSET	S_pad_addlex(pTHX_ SV* name);