X-Git-Url: http://git.megacz.com/?a=blobdiff_plain;f=ghc%2Flib%2Fcompat%2FCompat%2FUnicode.hs;fp=ghc%2Flib%2Fcompat%2FCompat%2FUnicode.hs;h=4765511954783d005b2d7b9c470df5bd42c9d107;hb=8e59ba46e26979cc11fa71e3f67aebbe6da4e8d6;hp=0000000000000000000000000000000000000000;hpb=080c9600d0ca63cc4125c6d2957ed8336b311c76;p=ghc-hetmet.git

diff --git a/ghc/lib/compat/Compat/Unicode.hs b/ghc/lib/compat/Compat/Unicode.hs
new file mode 100644
index 0000000..4765511
--- /dev/null
+++ b/ghc/lib/compat/Compat/Unicode.hs
@@ -0,0 +1,57 @@
+{-# OPTIONS -cpp #-}
+module Compat.Unicode (
+    GeneralCategory(..), generalCategory,
+  ) where
+
+#if __GLASGOW_HASKELL__ > 604
+
+import Data.Char (GeneralCategory(..), generalCategory)
+
+#else
+
+import Foreign.C	( CInt )
+import Data.Char	( ord )
+
+-- | Unicode General Categories (column 2 of the UnicodeData table)
+-- in the order they are listed in the Unicode standard.
+
+data GeneralCategory
+        = UppercaseLetter       -- Lu  Letter, Uppercase
+        | LowercaseLetter       -- Ll  Letter, Lowercase
+        | TitlecaseLetter       -- Lt  Letter, Titlecase
+        | ModifierLetter        -- Lm  Letter, Modifier
+        | OtherLetter           -- Lo  Letter, Other
+        | NonSpacingMark        -- Mn  Mark, Non-Spacing
+        | SpacingCombiningMark  -- Mc  Mark, Spacing Combining
+        | EnclosingMark         -- Me  Mark, Enclosing
+        | DecimalNumber         -- Nd  Number, Decimal
+        | LetterNumber          -- Nl  Number, Letter
+        | OtherNumber           -- No  Number, Other
+        | ConnectorPunctuation  -- Pc  Punctuation, Connector
+        | DashPunctuation       -- Pd  Punctuation, Dash
+        | OpenPunctuation       -- Ps  Punctuation, Open
+        | ClosePunctuation      -- Pe  Punctuation, Close
+        | InitialQuote          -- Pi  Punctuation, Initial quote
+        | FinalQuote            -- Pf  Punctuation, Final quote
+        | OtherPunctuation      -- Po  Punctuation, Other
+        | MathSymbol            -- Sm  Symbol, Math
+        | CurrencySymbol        -- Sc  Symbol, Currency
+        | ModifierSymbol        -- Sk  Symbol, Modifier
+        | OtherSymbol           -- So  Symbol, Other
+        | Space                 -- Zs  Separator, Space
+        | LineSeparator         -- Zl  Separator, Line
+        | ParagraphSeparator    -- Zp  Separator, Paragraph
+        | Control               -- Cc  Other, Control
+        | Format                -- Cf  Other, Format
+        | Surrogate             -- Cs  Other, Surrogate
+        | PrivateUse            -- Co  Other, Private Use
+        | NotAssigned           -- Cn  Other, Not Assigned
+        deriving (Eq, Ord, Enum, Read, Show, Bounded)
+
+-- | Retrieves the general Unicode category of the character.
+generalCategory :: Char -> GeneralCategory
+generalCategory c = toEnum (wgencat (fromIntegral (ord c)))
+
+foreign import ccall unsafe "u_gencat"
+  wgencat :: CInt -> Int
+#endif