core: Add isNoncharacter

Author: wismill

experimental/icu/cbits/icu.c

@@ -7,6 +7,14 @@ void __hs_u_getUnicodeVersion(UVersionInfo versionArray) {
     u_getUnicodeVersion(versionArray);
 }
 
+/*******************************************************************************
+ * Properties
+ ******************************************************************************/
+
+bool __hs_u_hasBinaryProperty(UChar32 c, UProperty which) {
+    return u_hasBinaryProperty(c, which);
+}
+
 /*******************************************************************************
  * Names
  ******************************************************************************/

experimental/icu/cbits/icu.h

@@ -7,6 +7,12 @@
 
 void __hs_u_getUnicodeVersion(UVersionInfo versionArray);
 
+/*******************************************************************************
+ * Properties
+ ******************************************************************************/
+
+bool __hs_u_hasBinaryProperty(UChar32 c, UProperty which);
+
 /*******************************************************************************
  * Names
  ******************************************************************************/

experimental/icu/icu.cabal

@@ -73,4 +73,4 @@ library
   pkgconfig-depends:
     icu-uc >= 72.1
   build-tool-depends:
-      c2hs:c2hs
+      c2hs:c2hs >= 0.28.8

experimental/icu/lib/ICU/Char.chs

@@ -15,6 +15,7 @@ module ICU.Char
     , UGeneralCategory(..)
     , toGeneralCategory
     , charType
+    , isNoncharacter
     ) where
 
 #include <unicode/uchar.h>
@@ -134,3 +135,21 @@ toGeneralCategory = \case
     OtherSymbol -> Char.OtherSymbol
     InitialPunctuation -> Char.InitialQuote
     FinalPunctuation -> Char.FinalQuote
+
+{#enum define UProperty {
+    UCHAR_NONCHARACTER_CODE_POINT as NoncharacterCodePoint
+    }
+    deriving (Bounded, Eq, Ord, Show) #}
+
+foreign import ccall safe "icu.h __hs_u_hasBinaryProperty" u_hasBinaryProperty
+    :: UChar32 -> Int -> Bool
+
+-- hasBinaryProperty :: UChar32 -> Int -> Bool
+-- hasBinaryProperty = {#call pure u_hasBinaryProperty as __hs_u_hasBinaryProperty#}
+-- {#fun pure u_hasBinaryProperty as hasBinaryProperty
+--     {`UChar32', `Int'} -> `Bool' #}
+
+isNoncharacter :: Char -> Bool
+isNoncharacter c = u_hasBinaryProperty
+    (fromIntegral (ord c))
+    (fromEnum NoncharacterCodePoint)

unicode-data/lib/Unicode/Char/General.hs

@@ -24,6 +24,7 @@ module Unicode.Char.General
     , isSeparator
     , isSymbol
     , isWhiteSpace
+    , isNoncharacter
     , isLetter
     , isSpace
     -- ** Re-export
@@ -414,6 +415,22 @@ isSymbol :: Char -> Bool
 isSymbol c = UC.MathSymbol <= gc && gc <= UC.OtherSymbol
     where gc = UC.generalCategory c
 
+-- | Returns 'True' for any /noncharacter/.
+--
+-- A /noncharacter/ is a code point that is permanently reserved for internal
+-- use (see definition D14 in the section
+-- [3.4 “Characters and Encoding”](https://www.unicode.org/versions/Unicode15.0.0/ch03.pdf#G2212)
+-- of the Unicode Standard).
+--
+-- Noncharacters consist of the values @U+nFFFE@ and @U+nFFFF@ (where @n@
+-- is from 0 to 10₁₆) and the values @U+FDD0..U+FDEF@.
+--
+-- @since 0.6.0
+isNoncharacter :: Char -> Bool
+isNoncharacter c
+    = ('\xFDD0' <= c && c <= '\xFDEF')
+    || (ord c .&. 0xFFFF) >= 0xFFFE
+
 -- | Returns 'True' for alphabetic Unicode characters (lower-case, upper-case
 -- and title-case letters, plus letters of caseless scripts and modifiers
 -- letters).

unicode-data/test/ICU/CharSpec.hs

@@ -26,6 +26,10 @@ spec = do
             "charType"
             (GeneralCategory . U.generalCategory)
             (GeneralCategory . ICU.toGeneralCategory . ICU.charType)
+        checkAndGatherErrors
+            "isNoncharacter"
+            (GeneralCategory . U.isNoncharacter)
+            (GeneralCategory . ICU.isNoncharacter)
     -- TODO: other functions
     where
     ourUnicodeVersion = versionBranch U.unicodeVersion