599c8d1 · text · git.iko.soy

git.iko.soy/text

Add support for toLower.

--HG-- extra : convert_revision : e0a040d08252f34d669f728d2484d640b60ae438

commit599c8d1b039c205dbc8268ddbd714532d229d252 authorBryan O'Sullivan <bos@serpentine.com> dateSun 7 Jun 2009, 01:09 (17 years ago) parent7c53770

5 files changed+63−18

compare with master

unifiedsplit

Data/Text.hs+9 Data/Text/Fusion/CaseMapping.hs+7 Data/Text/Fusion/Common.hs+22−9 Data/Text/Lazy.hs+9 scripts/SpecialCasing.hs+16−9

Data/Text.hs+9

@@ -57,6 +57,7 @@ module Data.Text

5757

5858 -- * Case conversion

5959 , toUpper

60+ , toLower

6061

6162 -- * Folds

6263 , foldl

@@ -392,6 +393,14 @@ toUpper :: Text -> Text

392393toUpper t = unstream (S.toUpper (stream t))

393394{-# INLINE toUpper #-}

394395

396+-- | /O(n)/ Convert a string to lower case, using simple case

397+-- conversion. The result string may be longer than the input string.

398+-- For instance, the German eszett (U+00DF) maps to the two-letter

399+-- sequence SS.

400+toLower :: Text -> Text

401+toLower t = unstream (S.toLower (stream t))

402+{-# INLINE toLower #-}

403+

395404-- | /O(n)/ The 'transpose' function transposes the rows and columns

396405-- of its 'Text' argument. Note that this function uses 'pack',

397406-- 'unpack', and the list version of transpose, and is thus not very

Data/Text/Fusion/CaseMapping.hs+7

@@ -1,4 +1,6 @@

11{-# LANGUAGE Rank2Types #-}

2+-- AUTOMATICALLY GENERATED - DO NOT EDIT

3+-- Generated by scripts/SpecialCasing.hs

24module Data.Text.Fusion.CaseMapping where

35import Data.Char

46import Data.Text.Fusion.Internal

@@ -210,3 +212,8 @@ upperMapping '\x1fc7' s = Yield '\x0397' (s :!: '\x0342' :!: '\x0399')

210212-- GREEK SMALL LETTER OMEGA WITH PERISPOMENI AND YPOGEGRAMMENI

211213upperMapping '\x1ff7' s = Yield '\x03a9' (s :!: '\x0342' :!: '\x0399')

212214upperMapping c s = Yield (toUpper c) (s :!: '\0' :!: '\0')

215+lowerMapping :: forall s. Char -> s -> Step (PairS (PairS s Char) Char) Char

216+{-# INLINE lowerMapping #-}

217+-- LATIN CAPITAL LETTER I WITH DOT ABOVE

218+lowerMapping '\x0130' s = Yield '\x0069' (s :!: '\x0307' :!: '\x0000')

219+lowerMapping c s = Yield (toLower c) (s :!: '\0' :!: '\0')

Data/Text/Fusion/Common.hs+22−9

@@ -1,4 +1,4 @@

1−{-# LANGUAGE BangPatterns #-}

1+{-# LANGUAGE BangPatterns, Rank2Types #-}

22-- |

33-- Module : Data.Text.Fusion.Common

44-- Copyright : (c) Bryan O'Sullivan 2009

@@ -37,6 +37,7 @@ module Data.Text.Fusion.Common

3737

3838 -- ** Case conversion

3939 , toUpper

40+ , toLower

4041

4142 -- * Folds

4243 , foldl

@@ -99,7 +100,7 @@ import Prelude (Bool(..), Char, Either(..), Eq(..), Int, Integral, Maybe(..),

99100import qualified Data.List as L

100101import qualified Prelude as P

101102import Data.Text.Fusion.Internal

102−import Data.Text.Fusion.CaseMapping (upperMapping)

103+import Data.Text.Fusion.CaseMapping (lowerMapping, upperMapping)

103104

104105singleton :: Char -> Stream Char

105106singleton c = Stream next False 1 -- HINT maybe too low

@@ -299,22 +300,34 @@ intersperse c (Stream next0 s0 len) = Stream next (s0 :!: N :!: S1) len -- HINT

299300-- ----------------------------------------------------------------------------

300301-- ** Case conversions (folds)

301302

302−-- | /O(n)/ Convert a string to upper case, using simple case

303−-- conversion. The result string may be longer than the input string.

304−-- For instance, the German eszett (U+00DF) maps to the two-letter

305−-- sequence SS.

306−toUpper :: Stream Char -> Stream Char

307−toUpper (Stream next0 s0 len) = Stream next (s0 :!: '\0' :!: '\0') len

303+caseConvert :: (forall s. Char -> s -> Step (PairS (PairS s Char) Char) Char)

304+ -> Stream Char -> Stream Char

305+caseConvert remap (Stream next0 s0 len) = Stream next (s0 :!: '\0' :!: '\0') len

308306 where

309307 {-# INLINE next #-}

310308 next (s :!: '\0' :!: _) =

311309 case next0 s of

312310 Done -> Done

313311 Skip s' -> Skip (s' :!: '\0' :!: '\0')

314− Yield c s' -> upperMapping c s'

312+ Yield c s' -> remap c s'

315313 next (s :!: a :!: b) = Yield a (s :!: b :!: '\0')

314+

315+-- | /O(n)/ Convert a string to upper case, using simple case

316+-- conversion. The result string may be longer than the input string.

317+-- For instance, the German eszett (U+00DF) maps to the two-letter

318+-- sequence SS.

319+toUpper :: Stream Char -> Stream Char

320+toUpper = caseConvert upperMapping

316321{-# INLINE [0] toUpper #-}

317322

323+-- | /O(n)/ Convert a string to lower case, using simple case

324+-- conversion. The result string may be longer than the input string.

325+-- For instance, the German eszett (U+00DF) maps to the two-letter

326+-- sequence SS.

327+toLower :: Stream Char -> Stream Char

328+toLower = caseConvert lowerMapping

329+{-# INLINE [0] toLower #-}

330+

318331-- ----------------------------------------------------------------------------

319332-- * Reducing Streams (folds)

320333

Data/Text/Lazy.hs+9

@@ -60,6 +60,7 @@ module Data.Text.Lazy

6060

6161 -- ** Case conversion

6262 , toUpper

63+ , toLower

6364

6465 -- * Folds

6566 , foldl

@@ -378,6 +379,14 @@ toUpper :: Text -> Text

378379toUpper t = unstream (S.toUpper (stream t))

379380{-# INLINE toUpper #-}

380381

382+-- | /O(n)/ Convert a string to lower case, using simple case

383+-- conversion. The result string may be longer than the input string.

384+-- For instance, the German eszett (U+00DF) maps to the two-letter

385+-- sequence SS.

386+toLower :: Text -> Text

387+toLower t = unstream (S.toLower (stream t))

388+{-# INLINE toLower #-}

389+

381390-- | /O(n)/ 'foldl', applied to a binary operator, a starting value

382391-- (typically the left-identity of the operator), and a 'Text',

383392-- reduces the 'Text' using the binary operator, from left to right.

scripts/SpecialCasing.hs+16−9

@@ -4,7 +4,7 @@ import Data.Char

44import Numeric

55import System.Environment

66import System.IO

7−import Text.ParserCombinators.Parsec hiding (many, optional, upper, (<|>))

7+import Text.ParserCombinators.Parsec hiding (many, optional, upper, lower, (<|>))

88import Text.ParserCombinators.Parsec.Combinator

1010instance Applicative (GenParser s a) where

@@ -41,18 +41,24 @@ entries = many comment *> many (entry <* many comment)

4141parseFile :: FilePath -> IO (Either ParseError [Case])

4242parseFile name = parse entries name <$> readFile name

4343

44−upperFunc ms = typ ++ (map nice . filter p $ ms) ++ [last]

44+mapFunc which access twiddle ms = typ ++ (map nice . filter p $ ms) ++ [last]

4545 where

46− typ = ["upperMapping :: forall s. Char -> s -> Step (PairS (PairS s Char) Char) Char"

47− ,"{-# INLINE upperMapping #-}"]

48− last = "upperMapping c s = Yield (toUpper c) (s :!: '\\0' :!: '\\0')"

49− nice c = "-- " ++ name c ++ "\nupperMapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (s :!: " ++ y ++ " :!: " ++ z ++ ")"

50− where [x,y,z] = (map showC . take 3) (upper c ++ repeat '\0')

51− p c = [code c] /= upper c && null (conditions c)

46+ typ = [which ++ "Mapping :: forall s. Char -> s -> Step (PairS (PairS s Char) Char) Char"

47+ ,"{-# INLINE " ++ which ++ "Mapping #-}"]

48+ last = which ++ "Mapping c s = Yield (to" ++ ucFirst which ++ " c) (s :!: '\\0' :!: '\\0')"

49+ nice c = "-- " ++ name c ++ "\n" ++

50+ which ++ "Mapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (s :!: " ++ y ++ " :!: " ++ z ++ ")"

51+ where [x,y,z] = (map showC . take 3) (access c ++ repeat '\0')

52+ p c = [k] /= a && a /= [twiddle k] && null (conditions c)

53+ where a = access c

54+ k = code c

5255 showC c = "'\\x" ++ d ++ "'"

5356 where h = showHex (ord c) ""

5457 d = replicate (4 - length h) '0' ++ h

5558

59+ucFirst (c:cs) = toUpper c : cs

60+ucFirst [] = []

61+

5662main = do

5763 args <- getArgs

5864 let (iname, oname) = case args of

@@ -71,5 +77,6 @@ main = do

7177 ,"import Data.Char"

7278 ,"import Data.Text.Fusion.Internal"

7379 ,""]

74− mapM_ (hPutStrLn h) (upperFunc ms)

80+ mapM_ (hPutStrLn h) (mapFunc "upper" upper toUpper ms)

81+ mapM_ (hPutStrLn h) (mapFunc "lower" lower toLower ms)

7582 hClose h