X-Git-Url: http://git.megacz.com/?a=blobdiff_plain;f=ghc%2Fcompiler%2Futils%2FUnicodeUtil.lhs;h=56e95a5434a8376b2aa3f928da6445679401c79b;hb=2a2efb720c0fdc06fe749f96f284b00b30f8f3f7;hp=0123e67305f7b99081ffb47804dbdd4c8efea092;hpb=4b17269854ccf10df8b3ca1711410a5ca439ea8a;p=ghc-hetmet.git

diff --git a/ghc/compiler/utils/UnicodeUtil.lhs b/ghc/compiler/utils/UnicodeUtil.lhs
index 0123e67..56e95a5 100644
--- a/ghc/compiler/utils/UnicodeUtil.lhs
+++ b/ghc/compiler/utils/UnicodeUtil.lhs
@@ -2,45 +2,35 @@ Various Unicode-related utilities.
 
 \begin{code}
 module UnicodeUtil(
-    stringToUtf8
+    stringToUtf8, intsToUtf8
   ) where
 
 #include "HsVersions.h"
 
-import Panic (panic)
-import Char  (chr, ord)
+import Panic ( panic )
+import Char  ( chr, ord )
 \end{code}
 
 \begin{code}
-stringToUtf8 :: [Int] -> String
-stringToUtf8 []       = ""
-stringToUtf8 (c:s)
-    | c >= 1 && c <= 0x7F = chr c : stringToUtf8 s
+stringToUtf8 :: String -> String
+stringToUtf8 s = intsToUtf8 (map ord s)
+
+intsToUtf8 :: [Int] -> String
+intsToUtf8 []       = ""
+intsToUtf8 (c:s)
+    | c >= 1 && c <= 0x7F = chr c : intsToUtf8 s
     | c < 0           = panic ("charToUtf8 ("++show c++")")
     | c <= 0x7FF      = chr (0xC0 + c `div`       0x40           ) :
                         chr (0x80 + c                  `mod` 0x40) :
-                        stringToUtf8 s
+                        intsToUtf8 s
     | c <= 0xFFFF     = chr (0xE0 + c `div`     0x1000           ) :
                         chr (0x80 + c `div`       0x40 `mod` 0x40) :
                         chr (0x80 + c                  `mod` 0x40) :
-                        stringToUtf8 s
-    | c <= 0x1FFFFF   = chr (0xF0 + c `div`    0x40000           ) :
-                        chr (0x80 + c `div`     0x1000 `mod` 0x40) :
-                        chr (0x80 + c `div`       0x40 `mod` 0x40) :
-                        chr (0x80 + c                  `mod` 0x40) :
-                        stringToUtf8 s
-    | c <= 0x3FFFFFF  = chr (0xF8 + c `div`  0x1000000           ) :
-                        chr (0x80 + c `div`    0x40000 `mod` 0x40) :
-                        chr (0x80 + c `div`     0x1000 `mod` 0x40) :
-                        chr (0x80 + c `div`       0x40 `mod` 0x40) :
-                        chr (0x80 + c                  `mod` 0x40) :
-                        stringToUtf8 s
-    | c <= 0x7FFFFFFF = chr (0xFC + c `div` 0x40000000           ) :
-                        chr (0x80 + c `div`  0x1000000 `mod` 0x40) :
-                        chr (0x80 + c `div`    0x40000 `mod` 0x40) :
+                        intsToUtf8 s
+    | c <= 0x10FFFF   = chr (0xF0 + c `div`    0x40000           ) :
                         chr (0x80 + c `div`     0x1000 `mod` 0x40) :
                         chr (0x80 + c `div`       0x40 `mod` 0x40) :
                         chr (0x80 + c                  `mod` 0x40) :
-                        stringToUtf8 s
+                        intsToUtf8 s
     | otherwise       = panic ("charToUtf8 "++show c)
 \end{code}