From 64650d8500e99b88fcbb19570537a2232fab77da Mon Sep 17 00:00:00 2001
From: Daniel Ehrenberg <ehrenbed@carleton.edu>
Date: Fri, 1 Feb 2008 22:50:30 -0600
Subject: [PATCH] Fixing UTF-8 to put the replacement character for malformed
 stuff

---
 core/io/utf8/utf8-tests.factor | 12 ++++++------
 core/io/utf8/utf8.factor       |  2 +-
 2 files changed, 7 insertions(+), 7 deletions(-)

diff --git a/core/io/utf8/utf8-tests.factor b/core/io/utf8/utf8-tests.factor
index d120b6243d..3576471586 100644
--- a/core/io/utf8/utf8-tests.factor
+++ b/core/io/utf8/utf8-tests.factor
@@ -1,16 +1,16 @@
-USING: io.utf8 tools.test strings ;
+USING: io.utf8 tools.test strings arrays unicode.syntax ;
 
-[ { BIN: 11110101 BIN: 10111111 BIN: 10000000 BIN: 11111111 } decode-utf8 ] unit-test-fails
+[ { UNICHAR: replacement-character } ] [ { BIN: 11110101 BIN: 10111111 BIN: 10000000 BIN: 11111111 } decode-utf8 >array ] unit-test
 
-[ { BIN: 101111111000000111111 } ] [ { BIN: 11110101 BIN: 10111111 BIN: 10000000 BIN: 10111111 } decode-utf8 ] unit-test
+[ { BIN: 101111111000000111111 } ] [ { BIN: 11110101 BIN: 10111111 BIN: 10000000 BIN: 10111111 } decode-utf8 >array ] unit-test
 
 [ "x" ] [ "x" decode-utf8 >string ] unit-test
 
-[ { BIN: 11111000000 } ] [ { BIN: 11011111 BIN: 10000000 } decode-utf8 ] unit-test
+[ { BIN: 11111000000 } ] [ { BIN: 11011111 BIN: 10000000 } decode-utf8 >array ] unit-test
 
-[ { BIN: 10000000 } decode-utf8 ] unit-test-fails
+[ { UNICHAR: replacement-character } ] [ { BIN: 10000000 } decode-utf8 >array ] unit-test
 
-[ { BIN: 1111000000111111 } ] [ { BIN: 11101111 BIN: 10000000 BIN: 10111111 } decode-utf8 ] unit-test
+[ { BIN: 1111000000111111 } ] [ { BIN: 11101111 BIN: 10000000 BIN: 10111111 } decode-utf8 >array ] unit-test
 
 [ B{ BIN: 11110101 BIN: 10111111 BIN: 10000000 BIN: 10111111 BIN: 11101111 BIN: 10000000 BIN: 10111111 BIN: 11011111 BIN: 10000000 CHAR: x } ]
 [ { BIN: 101111111000000111111 BIN: 1111000000111111 BIN: 11111000000 CHAR: x } encode-utf8 ] unit-test
diff --git a/core/io/utf8/utf8.factor b/core/io/utf8/utf8.factor
index 321469378d..213afb6eae 100644
--- a/core/io/utf8/utf8.factor
+++ b/core/io/utf8/utf8.factor
@@ -29,7 +29,7 @@ SYMBOL: quad3
     } cond ;
 
 : end-multibyte ( buf byte ch -- buf ch state )
-    begin append-nums decoded ;
+    f append-nums [ decoded ] unless* ;
 
 : (decode-utf8) ( buf byte ch state -- buf ch state )
     {