Merge pull request #46 from rintcius/fix-codepoint-anychar

Fix CodePoint.anyChar parser

Author: hdgarrood

src/Text/Parsing/StringParser/CodePoints.purs

@@ -26,12 +26,13 @@ import Prelude
 import Control.Alt ((<|>))
 import Data.Array ((..))
 import Data.Array.NonEmpty as NEA
-import Data.Char (toCharCode)
+import Data.Char (fromCharCode, toCharCode)
 import Data.Either (Either(..))
+import Data.Enum (fromEnum)
 import Data.Foldable (class Foldable, foldMap, elem, notElem)
 import Data.Maybe (Maybe(..))
-import Data.String.CodePoints (drop, length, indexOf', stripPrefix)
-import Data.String.CodeUnits (charAt, singleton)
+import Data.String.CodePoints (codePointAt, drop, indexOf', length, stripPrefix)
+import Data.String.CodeUnits (singleton)
 import Data.String.Pattern (Pattern(..))
 import Data.String.Regex as Regex
 import Data.String.Regex.Flags (noFlags)
@@ -48,9 +49,13 @@ eof = Parser \s ->
 -- | Match any character.
 anyChar :: Parser Char
 anyChar = Parser \{ str, pos } ->
-  case charAt pos str of
-    Just chr -> Right { result: chr, suffix: { str, pos: pos + 1 } }
+  case codePointAt pos str of
+    Just cp -> case toChar cp of
+      Just chr -> Right { result: chr, suffix: { str, pos: pos + 1 } }
+      Nothing -> Left { pos, error: ParseError $ "CodePoint " <> show cp <> " is not a character" }
     Nothing -> Left { pos, error: ParseError "Unexpected EOF" }
+  where
+    toChar = fromCharCode <<< fromEnum
 
 -- | Match any digit.
 anyDigit :: Parser Char

test/CodePoints.purs

@@ -17,7 +17,7 @@ import Test.Assert (assert', assert)
 import Text.Parsing.StringParser (Parser, runParser, try)
 import Text.Parsing.StringParser.Combinators (many1, endBy1, sepBy1, optionMaybe, many, manyTill, many1Till, chainl, fix, between)
 import Text.Parsing.StringParser.Expr (Assoc(..), Operator(..), buildExprParser)
-import Text.Parsing.StringParser.CodePoints (anyDigit, eof, string, anyChar, regex)
+import Text.Parsing.StringParser.CodePoints (anyDigit, char, eof, string, anyChar, regex)
 
 parens :: forall a. Parser a -> Parser a
 parens = between (string "(") (string ")")
@@ -97,3 +97,6 @@ testCodePoints = do
   assert $ canParse (many1Till (string "a") (string "and")) $ (fold <<< take 10000 $ repeat "a") <> "and"
   -- check correct order
   assert $ expectResult (NonEmptyList ('a' :| 'b':'c':Nil)) (many1Till anyChar (string "d")) "abcd"
+  assert $ expectResult "\x458CA" (string "\x458CA" <* char ']' <* eof ) "\x458CA]"
+  assert $ expectResult "\x458CA" (string "\x458CA" <* string ")" <* eof ) "\x458CA)"
+  assert $ expectResult '\xEEE2' (char '\xEEE2' <* eof ) "\xEEE2"