From c7e378122d41ace44704cb07362f7399e5f3c495 Mon Sep 17 00:00:00 2001
From: Paul Brinkmeier <hallo@pbrinkmeier.de>
Date: Wed, 1 Mar 2023 03:41:39 +0100
Subject: [PATCH] Add ranges to /codepoints

---
 app/Main.hs         |  4 ++--
 src/UToy/Parsers.hs | 23 ++++++++++++++++++-----
 2 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/app/Main.hs b/app/Main.hs
index 580ede2..c68a92b 100644
--- a/app/Main.hs
+++ b/app/Main.hs
@@ -127,8 +127,8 @@ newtype CodepointsModel = CodepointsModel
   { codepoints :: [(Word, Either String Char)]
   }
 
-mkCodepointsModel :: [Word] -> CodepointsModel
-mkCodepointsModel = CodepointsModel . map go
+mkCodepointsModel :: [(Word, Word)] -> CodepointsModel
+mkCodepointsModel = CodepointsModel . map go . concatMap (uncurry enumFromTo)
   where
     go codepoint = (codepoint, toChar codepoint)
 
diff --git a/src/UToy/Parsers.hs b/src/UToy/Parsers.hs
index b956fb6..74c6f13 100644
--- a/src/UToy/Parsers.hs
+++ b/src/UToy/Parsers.hs
@@ -1,3 +1,5 @@
+{-# LANGUAGE OverloadedStrings #-}
+
 module UToy.Parsers
   ( parseHexBytes
   , parseCodepoints
@@ -29,20 +31,31 @@ hexBytes = hexByte `Atto.sepBy` separators
       | 'a' <= c && c <= 'f' = ord c - ord 'a' + 10
       | otherwise = error $ printf "not a hex digit: %c" c
 
-parseCodepoints :: Text -> Either String [Word]
+parseCodepoints :: Text -> Either String [(Word, Word)]
 parseCodepoints = Atto.parseOnly $ codepoints <* Atto.endOfInput
 
-codepoints :: Atto.Parser [Word]
-codepoints = codepoint `Atto.sepBy` separators
-  where
-    codepoint = Atto.choice [literal, decLiteral, hexLiteral]
+codepoints :: Atto.Parser [(Word, Word)]
+codepoints = codepointRange `Atto.sepBy` separators
 
+codepointRange :: Atto.Parser (Word, Word)
+codepointRange = do
+  codepoint1 <- codepoint
+  codepoint2 <- Atto.choice [Atto.skip (== '-') *> codepoint, pure codepoint1]
+
+  pure (codepoint1, codepoint2)
+
+  
+
+codepoint :: Atto.Parser Word
+codepoint = Atto.choice [literal, decLiteral, hexLiteral, uCodepoint]
+  where
     literal = Atto.decimal
 
     decLiteral = Atto.char 'd' *> Atto.decimal
 
     hexLiteral = Atto.char 'x' *> Atto.hexadecimal
 
+    uCodepoint = Atto.string "U+" *> Atto.hexadecimal
 
 -- Common