feat(lexer): implement numericEntity lexer

src/Lexer.hs

@@ -8,10 +8,11 @@ module Lexer (
 where
 
 import Control.Monad (mfilter, void)
+import Data.Char (ord, toLower)
 import Data.Functor (($>))
 import Data.Text (Text, intercalate)
 import Data.Text qualified as Text
-import GHC.Unicode (isAlphaNum, isControl, isPrint, isSpace, isUpper)
+import GHC.Unicode (isAlphaNum, isControl, isDigit, isPrint, isSpace, isUpper)
 import ParserMonad (Parser, initialParserState)
 import Text.Parsec
 import Text.Parsec qualified as Parsec
@@ -94,6 +95,7 @@ lexText = go
                   , labeledLink
                   , module_
                   , anchor
+                  , numericEntity
                   , textElement
                   , quotes
                   , birdTrack
@@ -255,6 +257,30 @@ bold = delimitedSymmetric "__" BoldOpen BoldClose
 monospace :: Lexer
 monospace = delimitedSymmetric "@" MonospaceOpen MonospaceClose
 
+decimal :: Parser Int
+decimal = read . Text.unpack <$> takeWhile1_ isDigit
+
+hexadecimal :: Parser Int
+hexadecimal = "x" *> (convert 0 . fmap (normalise . toLower) <$> many1 hexDigit)
+ where
+  normalise :: Char -> Int
+  normalise c
+    | ord '0' <= n && n <= ord '9' = n - ord '0'
+    | ord 'A' <= n && n <= ord 'F' = n - ord 'A' + 10
+    | ord 'a' <= n && n <= ord 'f' = n - ord 'a' + 10
+    | otherwise = error "unexpected: invalid hex number"
+   where
+    n = ord c
+
+  convert :: Int -> [Int] -> Int
+  convert acc [] = acc
+  convert acc (x : xs) = convert (acc * 16 + x) xs
+
+numericEntity :: Lexer
+numericEntity = do
+  x <- located $ between "&#" ";" (NumericEntity <$> (hexadecimal <|> decimal))
+  pure [x]
+
 other :: Lexer
 other = do
   pos <- getPosition