2ceb577 · text · git.iko.soy

git.iko.soy/text

Many improvements, all small.

Add a span_ function, using unboxed tuples, to Text.Private. Use span_ in a few places where it can help a little. Relax the constraint on rational to Fractional. Specialize over many more integral types.

commit2ceb5777ab7b0cbbd03a0c3c4edc3c79fe82063f authorBryan O'Sullivan <bos@serpentine.com> dateFri 23 Dec 2011, 14:12 (15 years ago) parentbd23332

6 files changed+114−32

compare with master

unifiedsplit

Data/Text.hs+9−13 Data/Text/Encoding.hs+5−4 Data/Text/Lazy/Read.hs+30−7 Data/Text/Private.hs+28 Data/Text/Read.hs+41−8 text.cabal+1

Data/Text.hs+9−13

@@ -1,11 +1,11 @@

1−{-# LANGUAGE BangPatterns, CPP, Rank2Types #-}

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

22{-# OPTIONS_GHC -fno-warn-orphans #-}

44-- |

55-- Module : Data.Text

99--

1010-- License : BSD-style

1111-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

@@ -208,6 +208,7 @@ import Data.String (IsString(..))

208208import qualified Data.Text.Fusion as S

209209import qualified Data.Text.Fusion.Common as S

210210import Data.Text.Fusion (stream, reverseStream, unstream)

211+import Data.Text.Private (span_)

211212import Data.Text.Internal (Text(..), empty, firstf, safe, text, textP)

212213import qualified Prelude as P

213214import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

@@ -1081,13 +1082,8 @@ splitAt n t@(Text arr off len)

10811082-- of @t@ of elements that satisfy @p@, and whose second is the

10821083-- remainder of the list.

10831084span :: (Char -> Bool) -> Text -> (Text, Text)

1084−span p t@(Text arr off len) = (hd,tl)

1085− where hd = textP arr off k

1086− tl = textP arr (off+k) (len-k)

1087− !k = loop 0

1088− loop !i | i < len && p c = loop (i+d)

1089− | otherwise = i

1090− where Iter c d = iter t i

1085+span p t = case span_ p t of

1086+ (# hd,tl #) -> (hd,tl)

10911087{-# INLINE span #-}

10921088

10931089-- | /O(n)/ 'break' is like 'span', but the prefix returned is

@@ -1182,7 +1178,7 @@ split _ t@(Text _off _arr 0) = [t]

11821178split p t = loop t

11831179 where loop s | null s' = [l]

11841180 | otherwise = l : loop (unsafeTail s')

1185− where (l, s') = break p s

1181+ where (# l, s' #) = span_ (not . p) s

11861182{-# INLINE split #-}

11871183

11881184-- | /O(n)/ Splits a 'Text' into components of length @k@. The last

@@ -1397,7 +1393,7 @@ lines ps | null ps = []

13971393 | otherwise = h : if null t

13981394 then []

13991395 else lines (unsafeTail t)

1400− where (h,t) = span (/= '\n') ps

1396+ where (# h,t #) = span_ (/= '\n') ps

14011397{-# INLINE lines #-}

14021398

14031399{-

Data/Text/Encoding.hs+5−4

@@ -2,9 +2,9 @@

22 UnliftedFFITypes #-}

33-- |

44-- Module : Data.Text.Encoding

88--

99-- License : BSD-style

1010-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

@@ -175,7 +175,8 @@ encodeUtf8 (Text arr off len) = unsafePerformIO $ do

175175 -- special-case this assumption.

176176 let end = ptr `plusPtr` size

177177 ascii !t !u

178− | t == offLen || u == end || v >= 0x80 = go t (u `minusPtr` ptr)

178+ | t == offLen || u == end || v >= 0x80 =

179+ go t (u `minusPtr` ptr)

179180 | otherwise = do

180181 poke u (fromIntegral v :: Word8)

181182 ascii (t+1) (u `plusPtr` 1)

Data/Text/Lazy/Read.hs+30−7

@@ -2,11 +2,10 @@

33-- |

44-- Module : Data.Text.Lazy.Read

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : GHC

1211--

@@ -23,8 +22,10 @@ module Data.Text.Lazy.Read

2322

2423import Control.Monad (liftM)

2524import Data.Char (isDigit, isHexDigit, ord)

25+import Data.Int (Int8, Int16, Int32, Int64)

2626import Data.Ratio ((%))

2727import Data.Text.Lazy as T

28+import Data.Word (Word, Word8, Word16, Word32, Word64)

2829

2930-- | Read some text. If the read succeeds, return its value and the

3031-- remaining text, otherwise an error message.

@@ -43,7 +44,16 @@ type Reader a = Text -> Either String (a,Text)

4344-- 'Integer' for your result type.

4445decimal :: Integral a => Reader a

4546{-# SPECIALIZE decimal :: Reader Int #-}

47+{-# SPECIALIZE decimal :: Reader Int8 #-}

48+{-# SPECIALIZE decimal :: Reader Int16 #-}

49+{-# SPECIALIZE decimal :: Reader Int32 #-}

50+{-# SPECIALIZE decimal :: Reader Int64 #-}

4651{-# SPECIALIZE decimal :: Reader Integer #-}

52+{-# SPECIALIZE decimal :: Reader Word #-}

53+{-# SPECIALIZE decimal :: Reader Word8 #-}

54+{-# SPECIALIZE decimal :: Reader Word16 #-}

55+{-# SPECIALIZE decimal :: Reader Word32 #-}

56+{-# SPECIALIZE decimal :: Reader Word64 #-}

4757decimal txt

4858 | T.null h = Left "input does not start with a digit"

4959 | otherwise = Right (T.foldl' go 0 h, t)

@@ -71,8 +81,17 @@ hexadecimal txt

7181 where (h,t) = T.splitAt 2 txt

7282

7383hex :: Integral a => Reader a

74−{-# SPECIALIZE hex :: Reader Int #-}

75−{-# SPECIALIZE hex :: Reader Integer #-}

84+{-# SPECIALIZE hexadecimal :: Reader Int #-}

85+{-# SPECIALIZE hexadecimal :: Reader Int8 #-}

86+{-# SPECIALIZE hexadecimal :: Reader Int16 #-}

87+{-# SPECIALIZE hexadecimal :: Reader Int32 #-}

88+{-# SPECIALIZE hexadecimal :: Reader Int64 #-}

89+{-# SPECIALIZE hexadecimal :: Reader Integer #-}

90+{-# SPECIALIZE hexadecimal :: Reader Word #-}

91+{-# SPECIALIZE hexadecimal :: Reader Word8 #-}

92+{-# SPECIALIZE hexadecimal :: Reader Word16 #-}

93+{-# SPECIALIZE hexadecimal :: Reader Word32 #-}

94+{-# SPECIALIZE hexadecimal :: Reader Word64 #-}

7695hex txt

7796 | T.null h = Left "input does not start with a hexadecimal digit"

7897 | otherwise = Right (T.foldl' go 0 h, t)

@@ -114,7 +133,7 @@ signed f = runP (signa (P f))

114133--

115134-- >rational "3.foo" == Right (3.0, ".foo")

116135-- >rational "3e" == Right (3.0, "e")

117−rational :: RealFloat a => Reader a

136+rational :: Fractional a => Reader a

118137{-# SPECIALIZE rational :: Reader Double #-}

119138rational = floaty $ \real frac fracDenom -> fromRational $

120139 real % 1 + frac % fracDenom

@@ -138,6 +157,10 @@ double = floaty $ \real frac fracDenom ->

138157

139158signa :: Num a => Parser a -> Parser a

140159{-# SPECIALIZE signa :: Parser Int -> Parser Int #-}

160+{-# SPECIALIZE signa :: Parser Int8 -> Parser Int8 #-}

161+{-# SPECIALIZE signa :: Parser Int16 -> Parser Int16 #-}

162+{-# SPECIALIZE signa :: Parser Int32 -> Parser Int32 #-}

163+{-# SPECIALIZE signa :: Parser Int64 -> Parser Int64 #-}

141164{-# SPECIALIZE signa :: Parser Integer -> Parser Integer #-}

142165signa p = do

143166 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

@@ -168,7 +191,7 @@ char p = P $ \t -> case T.uncons t of

168191

169192data T = T !Integer !Int

170193

171−floaty :: RealFloat a => (Integer -> Integer -> Integer -> a) -> Reader a

194+floaty :: Fractional a => (Integer -> Integer -> Integer -> a) -> Reader a

172195{-# INLINE floaty #-}

173196floaty f = runP $ do

174197 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

Data/Text/Private.hsnew file+28

@@ -0,0 +1,28 @@

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

3+-- |

4+-- Module : Data.Text.Private

6+--

7+-- License : BSD-style

8+-- Maintainer : bos@serpentine.com

9+-- Stability : experimental

10+-- Portability : GHC

11+

12+module Data.Text.Private

13+ (

14+ span_

15+ ) where

16+

17+import Data.Text.Internal (Text(..), textP)

18+import Data.Text.Unsafe (Iter(..), iter)

19+

20+span_ :: (Char -> Bool) -> Text -> (# Text, Text #)

21+span_ p t@(Text arr off len) = (# hd,tl #)

22+ where hd = textP arr off k

23+ tl = textP arr (off+k) (len-k)

24+ !k = loop 0

25+ loop !i | i < len && p c = loop (i+d)

26+ | otherwise = i

27+ where Iter c d = iter t i

28+{-# INLINE span_ #-}

Data/Text/Read.hs+41−8

@@ -1,12 +1,11 @@

1−{-# LANGUAGE OverloadedStrings #-}

1+{-# LANGUAGE OverloadedStrings, UnboxedTuples #-}

33-- |

44-- Module : Data.Text.Read

66--

77-- License : BSD-style

8−-- Maintainer : bos@serpentine.com, rtomharper@googlemail.com,

9−-- duncan@haskell.org

8+-- Maintainer : bos@serpentine.com

109-- Stability : experimental

1110-- Portability : GHC

1211--

@@ -23,8 +22,11 @@ module Data.Text.Read

2322

2423import Control.Monad (liftM)

2524import Data.Char (isDigit, isHexDigit, ord)

25+import Data.Int (Int8, Int16, Int32, Int64)

2626import Data.Ratio ((%))

2727import Data.Text as T

28+import Data.Text.Private (span_)

29+import Data.Word (Word, Word8, Word16, Word32, Word64)

2830

2931-- | Read some text. If the read succeeds, return its value and the

3032-- remaining text, otherwise an error message.

@@ -43,11 +45,20 @@ type Reader a = Text -> Either String (a,Text)

4345-- 'Integer' for your result type.

4446decimal :: Integral a => Reader a

4547{-# SPECIALIZE decimal :: Reader Int #-}

48+{-# SPECIALIZE decimal :: Reader Int8 #-}

49+{-# SPECIALIZE decimal :: Reader Int16 #-}

50+{-# SPECIALIZE decimal :: Reader Int32 #-}

51+{-# SPECIALIZE decimal :: Reader Int64 #-}

4652{-# SPECIALIZE decimal :: Reader Integer #-}

53+{-# SPECIALIZE decimal :: Reader Word #-}

54+{-# SPECIALIZE decimal :: Reader Word8 #-}

55+{-# SPECIALIZE decimal :: Reader Word16 #-}

56+{-# SPECIALIZE decimal :: Reader Word32 #-}

57+{-# SPECIALIZE decimal :: Reader Word64 #-}

4758decimal txt

4859 | T.null h = Left "input does not start with a digit"

4960 | otherwise = Right (T.foldl' go 0 h, t)

50− where (h,t) = T.span isDigit txt

61+ where (# h,t #) = span_ isDigit txt

5162 go n d = (n * 10 + fromIntegral (digitToInt d))

5263

5364-- | Read a hexadecimal integer, consisting of an optional leading

@@ -64,7 +75,16 @@ decimal txt

6475-- 'Integer' for your result type.

6576hexadecimal :: Integral a => Reader a

6677{-# SPECIALIZE hexadecimal :: Reader Int #-}

78+{-# SPECIALIZE hexadecimal :: Reader Int8 #-}

79+{-# SPECIALIZE hexadecimal :: Reader Int16 #-}

80+{-# SPECIALIZE hexadecimal :: Reader Int32 #-}

81+{-# SPECIALIZE hexadecimal :: Reader Int64 #-}

6782{-# SPECIALIZE hexadecimal :: Reader Integer #-}

83+{-# SPECIALIZE hexadecimal :: Reader Word #-}

84+{-# SPECIALIZE hexadecimal :: Reader Word8 #-}

85+{-# SPECIALIZE hexadecimal :: Reader Word16 #-}

86+{-# SPECIALIZE hexadecimal :: Reader Word32 #-}

87+{-# SPECIALIZE hexadecimal :: Reader Word64 #-}

6888hexadecimal txt

6989 | h == "0x" || h == "0X" = hex t

7090 | otherwise = hex txt

@@ -72,11 +92,20 @@ hexadecimal txt

7292

7393hex :: Integral a => Reader a

7494{-# SPECIALIZE hex :: Reader Int #-}

95+{-# SPECIALIZE hex :: Reader Int8 #-}

96+{-# SPECIALIZE hex :: Reader Int16 #-}

97+{-# SPECIALIZE hex :: Reader Int32 #-}

98+{-# SPECIALIZE hex :: Reader Int64 #-}

7599{-# SPECIALIZE hex :: Reader Integer #-}

100+{-# SPECIALIZE hex :: Reader Word #-}

101+{-# SPECIALIZE hex :: Reader Word8 #-}

102+{-# SPECIALIZE hex :: Reader Word16 #-}

103+{-# SPECIALIZE hex :: Reader Word32 #-}

104+{-# SPECIALIZE hex :: Reader Word64 #-}

76105hex txt

77106 | T.null h = Left "input does not start with a hexadecimal digit"

78107 | otherwise = Right (T.foldl' go 0 h, t)

79− where (h,t) = T.span isHexDigit txt

108+ where (# h,t #) = span_ isHexDigit txt

80109 go n d = (n * 16 + fromIntegral (hexDigitToInt d))

81110

82111hexDigitToInt :: Char -> Int

@@ -114,7 +143,7 @@ signed f = runP (signa (P f))

114143--

115144-- >rational "3.foo" == Right (3.0, ".foo")

116145-- >rational "3e" == Right (3.0, "e")

117−rational :: RealFloat a => Reader a

146+rational :: Fractional a => Reader a

118147{-# SPECIALIZE rational :: Reader Double #-}

119148rational = floaty $ \real frac fracDenom -> fromRational $

120149 real % 1 + frac % fracDenom

@@ -138,6 +167,10 @@ double = floaty $ \real frac fracDenom ->

138167

139168signa :: Num a => Parser a -> Parser a

140169{-# SPECIALIZE signa :: Parser Int -> Parser Int #-}

170+{-# SPECIALIZE signa :: Parser Int8 -> Parser Int8 #-}

171+{-# SPECIALIZE signa :: Parser Int16 -> Parser Int16 #-}

172+{-# SPECIALIZE signa :: Parser Int32 -> Parser Int32 #-}

173+{-# SPECIALIZE signa :: Parser Int64 -> Parser Int64 #-}

141174{-# SPECIALIZE signa :: Parser Integer -> Parser Integer #-}

142175signa p = do

143176 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

@@ -168,7 +201,7 @@ char p = P $ \t -> case T.uncons t of

168201

169202data T = T !Integer !Int

170203

171−floaty :: RealFloat a => (Integer -> Integer -> Integer -> a) -> Reader a

204+floaty :: Fractional a => (Integer -> Integer -> Integer -> a) -> Reader a

172205{-# INLINE floaty #-}

173206floaty f = runP $ do

174207 sign <- perhaps '+' $ char (\c -> c == '-' || c == '+')

text.cabal+1

@@ -100,6 +100,7 @@ library

100100 Data.Text.Lazy.Encoding.Fusion

101101 Data.Text.Lazy.Fusion

102102 Data.Text.Lazy.Search

103+ Data.Text.Private

103104 Data.Text.Search

104105 Data.Text.Unsafe

105106 Data.Text.Unsafe.Base