on master help sign in
Record leading comments from Unicode data files
f0437f2Bryan O'Sullivan
scripts/Arsec.hs+5−2
@@ -1,6 +1,7 @@
1module Arsec
1module Arsec
2 (
2 (
3+ Comment
3− comment
4+ , comment
4 , semi
5 , semi
5 , showC
6 , showC
6 , unichar
7 , unichar
@@ -23,6 +24,8 @@ import Text.ParserCombinators.Parsec.Combinator hiding (optional)
23import Text.ParserCombinators.Parsec.Error
24import Text.ParserCombinators.Parsec.Error
24import Text.ParserCombinators.Parsec.Prim hiding ((<|>), many)
25import Text.ParserCombinators.Parsec.Prim hiding ((<|>), many)
25
26
27+type Comment = String
28+
26unichar :: Parser Char
29unichar :: Parser Char
27unichar = chr . fst . head . readHex <$> many1 hexDigit
30unichar = chr . fst . head . readHex <$> many1 hexDigit
28
31
@@ -32,7 +35,7 @@ unichars = manyTill (unichar <* spaces) semi
32semi :: Parser ()
35semi :: Parser ()
33semi = char ';' *> spaces *> pure ()
36semi = char ';' *> spaces *> pure ()
34
37
35−comment :: Parser String
38+comment :: Parser Comment
36comment = (char '#' *> manyTill anyToken (char '\n')) <|> string "\n"
39comment = (char '#' *> manyTill anyToken (char '\n')) <|> string "\n"
37
40
38showC :: Char -> String
41showC :: Char -> String
scripts/CaseFolding.hs+10−6
@@ -4,7 +4,8 @@
4
4
5module CaseFolding
5module CaseFolding
6 (
6 (
7+ CaseFolding(..)
7− Fold(..)
8+ , Fold(..)
8 , parseCF
9 , parseCF
9 , mapCF
10 , mapCF
10 ) where
11 ) where
@@ -18,19 +19,22 @@ data Fold = Fold {
18 , name :: String
19 , name :: String
19 } deriving (Eq, Ord, Show)
20 } deriving (Eq, Ord, Show)
20
21
22+data CaseFolding = CF { cfComments :: [Comment], cfFolding :: [Fold] }
23+ deriving (Show)
24+
21−entries :: Parser [Fold]
25+entries :: Parser CaseFolding
22−entries = many comment *> many (entry <* many comment)
26+entries = CF <$> many comment <*> many (entry <* many comment)
23 where
27 where
24 entry = Fold <$> unichar <* semi
28 entry = Fold <$> unichar <* semi
25 <*> oneOf "CFST" <* semi
29 <*> oneOf "CFST" <* semi
26 <*> unichars
30 <*> unichars
27 <*> (string "# " *> manyTill anyToken (char '\n'))
31 <*> (string "# " *> manyTill anyToken (char '\n'))
28
32
29−parseCF :: FilePath -> IO (Either ParseError [Fold])
33+parseCF :: FilePath -> IO (Either ParseError CaseFolding)
30parseCF name = parse entries name <$> readFile name
34parseCF name = parse entries name <$> readFile name
31
35
32−mapCF :: [Fold] -> [String]
36+mapCF :: CaseFolding -> [String]
33−mapCF ms = typ ++ (map nice . filter p $ ms) ++ [last]
37+mapCF (CF _ ms) = typ ++ (map nice . filter p $ ms) ++ [last]
34 where
38 where
35 typ = ["foldMapping :: forall s. Char -> s -> Step (CC s) Char"
39 typ = ["foldMapping :: forall s. Char -> s -> Step (CC s) Char"
36 ,"{-# INLINE foldMapping #-}"]
40 ,"{-# INLINE foldMapping #-}"]
scripts/CaseMapping.hs+7−2
@@ -19,9 +19,14 @@ main = do
19 Left err -> print err >> return undefined
19 Left err -> print err >> return undefined
20 Right ms -> return ms
20 Right ms -> return ms
21 h <- openFile oname WriteMode
21 h <- openFile oname WriteMode
22+ let comments = map ("--" ++) $
23+ take 2 (cfComments cfs) ++ take 2 (scComments scs)
24+ mapM_ (hPutStrLn h) $
22− mapM_ (hPutStrLn h) ["{-# LANGUAGE Rank2Types #-}"
25+ ["{-# LANGUAGE Rank2Types #-}"
23 ,"-- AUTOMATICALLY GENERATED - DO NOT EDIT"
26 ,"-- AUTOMATICALLY GENERATED - DO NOT EDIT"
24− ,"-- Generated by scripts/SpecialCasing.hs"
27+ ,"-- Generated by scripts/SpecialCasing.hs"] ++
28+ comments ++
29+ [""
25 ,"module Data.Text.Fusion.CaseMapping where"
30 ,"module Data.Text.Fusion.CaseMapping where"
26 ,"import Data.Char"
31 ,"import Data.Char"
27 ,"import Data.Text.Fusion.Internal"
32 ,"import Data.Text.Fusion.Internal"
scripts/SpecialCasing.hs+12−6
@@ -4,13 +4,17 @@
4
4
5module SpecialCasing
5module SpecialCasing
6 (
6 (
7+ SpecialCasing(..)
7− Case(..)
8+ , Case(..)
8 , parseSC
9 , parseSC
9 , mapSC
10 , mapSC
10 ) where
11 ) where
11
12
12import Arsec
13import Arsec
13
14
15+data SpecialCasing = SC { scComments :: [Comment], scCasing :: [Case] }
16+ deriving (Show)
17+
14data Case = Case {
18data Case = Case {
15 code :: Char
19 code :: Char
16 , lower :: [Char]
20 , lower :: [Char]
@@ -20,8 +24,8 @@ data Case = Case {
20 , name :: String
24 , name :: String
21 } deriving (Eq, Ord, Show)
25 } deriving (Eq, Ord, Show)
22
26
23−entries :: Parser [Case]
27+entries :: Parser SpecialCasing
24−entries = many comment *> many (entry <* many comment)
28+entries = SC <$> many comment <*> many (entry <* many comment)
25 where
29 where
26 entry = Case <$> unichar <* semi
30 entry = Case <$> unichar <* semi
27 <*> unichars
31 <*> unichars
@@ -30,11 +34,13 @@ entries = many comment *> many (entry <* many comment)
30 <*> manyTill anyToken (string "# ")
34 <*> manyTill anyToken (string "# ")
31 <*> manyTill anyToken (char '\n')
35 <*> manyTill anyToken (char '\n')
32
36
33−parseSC :: FilePath -> IO (Either ParseError [Case])
37+parseSC :: FilePath -> IO (Either ParseError SpecialCasing)
34parseSC name = parse entries name <$> readFile name
38parseSC name = parse entries name <$> readFile name
35
39
36−mapSC :: String -> (Case -> String) -> (Char -> Char) -> [Case] -> [String]
40+mapSC :: String -> (Case -> String) -> (Char -> Char) -> SpecialCasing
41+ -> [String]
37−mapSC which access twiddle ms = typ ++ (map nice . filter p $ ms) ++ [last]
42+mapSC which access twiddle (SC _ ms) =
43+ typ ++ (map nice . filter p $ ms) ++ [last]
38 where
44 where
39 typ = [which ++ "Mapping :: forall s. Char -> s -> Step (CC s) Char"
45 typ = [which ++ "Mapping :: forall s. Char -> s -> Step (CC s) Char"
40 ,"{-# INLINE " ++ which ++ "Mapping #-}"]
46 ,"{-# INLINE " ++ which ++ "Mapping #-}"]