master...40a3168 · text

git.iko.soy/text

8 files changed+89−20

since splitsnapshots

unifiedsplit

Data/Text.hs+12 Data/Text/Internal/Fusion/Common.hs+10−9 Data/Text/Lazy.hs+10 Data/Text/Lazy/Read.hs+3−3 Data/Text/Read.hs+3−3 changelog.md+11 tests/Tests/Properties.hs+23−3 tests/text-tests.cabal+17−2

Data/Text.hs+12

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

6060 , snoc

6161 , append

6262 , uncons

63+ , unsnoc

6364 , head

6465 , last

6566 , tail

@@ -545,6 +546,17 @@ init (Text arr off len) | len <= 0 = emptyError "init"

545546 unstream (S.init (stream t)) = init t

546547 #-}

547548

549+-- | /O(1)/ Returns all but the last character and the last character of a

550+-- 'Text', or 'Nothing' if empty.

551+unsnoc :: Text -> Maybe (Text, Char)

552+unsnoc (Text arr off len)

553+ | len <= 0 = Nothing

554+ | n < 0xDC00 || n > 0xDFFF = Just (text arr off (len-1), unsafeChr n)

555+ | otherwise = Just (text arr off (len-2), U16.chr2 n0 n)

556+ where n = A.unsafeIndex arr (off+len-1)

557+ n0 = A.unsafeIndex arr (off+len-2)

558+{-# INLINE [1] unsnoc #-}

559+

548560-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to

549561-- fusion.

550562null :: Text -> Bool

Data/Text/Internal/Fusion/Common.hs+10−9

@@ -107,7 +107,7 @@ import Prelude (Bool(..), Char, Eq(..), Int, Integral, Maybe(..),

107107import qualified Data.List as L

108108import qualified Prelude as P

109109import Data.Bits (shiftL)

110−import Data.Char (isLetter)

110+import Data.Char (isLetter, isSpace)

111111import Data.Int (Int64)

112112import Data.Text.Internal.Fusion.Types

113113import Data.Text.Internal.Fusion.CaseMapping (foldMapping, lowerMapping, titleMapping,

@@ -462,15 +462,16 @@ toTitle (Stream next0 s0 len) = Stream next (CC (False :*: s0) '\0' '\0') len

462462 where

463463 next (CC (letter :*: s) '\0' _) =

464464 case next0 s of

465− Done -> Done

466− Skip s' -> Skip (CC (letter :*: s') '\0' '\0')

465+ Done -> Done

466+ Skip s' -> Skip (CC (letter :*: s') '\0' '\0')

467467 Yield c s'

468− | letter' -> if letter

469− then lowerMapping c (letter' :*: s')

470− else titleMapping c (letter' :*: s')

471− | otherwise -> Yield c (CC (letter' :*: s') '\0' '\0')

472− where letter' = isLetter c

473− next (CC s a b) = Yield a (CC s b '\0')

468+ | nonSpace -> if letter

469+ then lowerMapping c (nonSpace :*: s')

470+ else titleMapping c (letter' :*: s')

471+ | otherwise -> Yield c (CC (letter' :*: s') '\0' '\0')

472+ where nonSpace = P.not (isSpace c)

473+ letter' = isLetter c

474+ next (CC s a b) = Yield a (CC s b '\0')

474475{-# INLINE [0] toTitle #-}

475476

476477data Justify i s = Just1 !i !s

Data/Text/Lazy.hs+10

@@ -68,6 +68,7 @@ module Data.Text.Lazy

6868 , snoc

6969 , append

7070 , uncons

71+ , unsnoc

7172 , head

7273 , last

7374 , tail

@@ -561,6 +562,15 @@ init Empty = emptyError "init"

561562 unstream (S.init (stream t)) = init t

562563 #-}

563564

565+-- | /O(n\/c)/ Returns the 'init' and 'last' of a 'Text', or 'Nothing' if

566+-- empty.

567+--

568+-- * It is no faster than using 'init' and 'last'.

569+unsnoc :: Text -> Maybe (Text, Char)

570+unsnoc Empty = Nothing

571+unsnoc ts@(Chunk _ _) = Just (init ts, last ts)

572+{-# INLINE unsnoc #-}

573+

564574-- | /O(1)/ Tests whether a 'Text' is empty or not. Subject to

565575-- fusion.

566576null :: Text -> Bool

Data/Text/Lazy/Read.hs+3−3

@@ -68,9 +68,9 @@ decimal txt

6868 go n d = (n * 10 + fromIntegral (digitToInt d))

6969

7070-- | Read a hexadecimal integer, consisting of an optional leading

71−-- @\"0x\"@ followed by at least one decimal digit. Input is consumed

72−-- until a non-hex-digit or end of string is reached. This function

73−-- is case insensitive.

71+-- @\"0x\"@ followed by at least one hexadecimal digit. Input is

72+-- consumed until a non-hex-digit or end of string is reached.

73+-- This function is case insensitive.

7474--

7575-- This function does not handle leading sign characters. If you need

7676-- to handle signed input, use @'signed' 'hexadecimal'@.

Data/Text/Read.hs+3−3

@@ -67,9 +67,9 @@ decimal txt

6767 go n d = (n * 10 + fromIntegral (digitToInt d))

6868

6969-- | Read a hexadecimal integer, consisting of an optional leading

70−-- @\"0x\"@ followed by at least one decimal digit. Input is consumed

71−-- until a non-hex-digit or end of string is reached. This function

72−-- is case insensitive.

70+-- @\"0x\"@ followed by at least one hexadecimal digit. Input is

71+-- consumed until a non-hex-digit or end of string is reached.

72+-- This function is case insensitive.

7373--

7474-- This function does not handle leading sign characters. If you need

7575-- to handle signed input, use @'signed' 'hexadecimal'@.

changelog.md+11

@@ -1,3 +1,14 @@

1+* The `toTitle` function now correctly handles letters that

2+ immediately follow puncutation. Before, `"there's"` would turn into

3+ `"There'S"`. Now, it becomes `"There's"`.

5+* The implementation of unstreaming is faster, resulting in operations

6+ such as `map` and `intersperse` speeding up by up to 30%, with

7+ smaller code generated.

9+* The optimised length comparison function is now more likely to be

10+ used after some rewrite rule tweaking.

11+

1121.2.2.1

213

314* The switch to `integer-pure` in 1.2.2.0 was apparently mistaken.

tests/Tests/Properties.hs+23−3

@@ -9,9 +9,9 @@ module Tests.Properties

99 ) where

1010

1111import Control.Applicative ((<$>), (<*>))

12−import Control.Arrow ((***), second)

12+import Control.Arrow ((***), first, second)

1313import Data.Bits ((.&.))

14−import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isUpper, ord)

14+import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isLetter, isUpper, ord)

1515import Data.Int (Int8, Int16, Int32, Int64)

1616import Data.Monoid (Monoid(..))

1717import Data.String (IsString(fromString))

@@ -23,6 +23,7 @@ import Data.Text.Internal.Search (indices)

2323import Data.Text.Lazy.Read as TL

2424import Data.Text.Read as T

2525import Data.Word (Word, Word8, Word16, Word32, Word64)

26+import Data.Maybe (mapMaybe)

2627import Numeric (showEFloat, showFFloat, showGFloat, showHex)

2728import Prelude hiding (replicate)

2829import Test.Framework (Test, testGroup)

@@ -229,6 +230,13 @@ sf_uncons p = (uncons . L.filter p) `eqP`

229230 (fmap (second unpackS) . S.uncons . S.filter p)

230231t_uncons = uncons `eqP` (fmap (second unpackS) . T.uncons)

231232tl_uncons = uncons `eqP` (fmap (second unpackS) . TL.uncons)

233+

234+unsnoc xs@(_:_) = Just (init xs, last xs)

235+unsnoc [] = Nothing

236+

237+t_unsnoc = unsnoc `eqP` (fmap (first unpackS) . T.unsnoc)

238+tl_unsnoc = unsnoc `eqP` (fmap (first unpackS) . TL.unsnoc)

239+

232240s_head = head `eqP` S.head

233241sf_head p = (head . L.filter p) `eqP` (S.head . S.filter p)

234242t_head = head `eqP` T.head

@@ -326,6 +334,14 @@ t_toUpper_upper t = p (T.toUpper t) >= p t

326334 where p = T.length . T.filter isUpper

327335tl_toUpper_upper t = p (TL.toUpper t) >= p t

328336 where p = TL.length . TL.filter isUpper

337+t_toTitle_title t = all (<= 1) (caps t)

338+ where caps = fmap (T.length . T.filter isUpper) . T.words . T.toTitle

339+t_toTitle_1stNotLower = and . notLow . T.toTitle . T.filter stable

340+ where notLow = mapMaybe (fmap (not . isLower) . (T.find isLetter)) . T.words

341+ -- Surprise! The Spanish/Portuguese ordinal indicators changed

342+ -- from category Ll (letter, lowercase) to Lo (letter, other)

343+ -- in Unicode 7.0

344+ stable c = c /= '\170' && c /= '\186'

329345

330346justifyLeft k c xs = xs ++ L.replicate (k - length xs) c

331347justifyRight m n xs = L.replicate (m - length xs) n ++ xs

@@ -967,6 +983,8 @@ tests =

967983 testProperty "sf_uncons" sf_uncons,

968984 testProperty "t_uncons" t_uncons,

969985 testProperty "tl_uncons" tl_uncons,

986+ testProperty "t_unsnoc" t_unsnoc,

987+ testProperty "tl_unsnoc" tl_unsnoc,

970988 testProperty "s_head" s_head,

971989 testProperty "sf_head" sf_head,

972990 testProperty "t_head" t_head,

@@ -1030,7 +1048,9 @@ tests =

10301048 testProperty "tl_toLower_lower" tl_toLower_lower,

10311049 testProperty "t_toUpper_length" t_toUpper_length,

10321050 testProperty "t_toUpper_upper" t_toUpper_upper,

1033− testProperty "tl_toUpper_upper" tl_toUpper_upper

1051+ testProperty "tl_toUpper_upper" tl_toUpper_upper,

1052+ testProperty "t_toTitle_title" t_toTitle_title,

1053+ testProperty "t_toTitle_1stNotLower" t_toTitle_1stNotLower

10341054 ],

10351055

10361056 testGroup "justification" [

tests/text-tests.cabal+17−2

@@ -20,6 +20,11 @@ flag hpc

2020 default: False

2121 manual: True

2222

23+flag bytestring-builder

24+ description: Depend on the bytestring-builder package for backwards compatibility.

25+ default: False

26+ manual: False

27+

2328executable text-tests

2429 main-is: Tests.hs

2530

@@ -48,7 +53,6 @@ executable text-tests

4853 HUnit >= 1.2,

4954 QuickCheck >= 2.7,

5055 base == 4.*,

51− bytestring,

5256 deepseq,

5357 directory,

5458 quickcheck-unicode,

@@ -58,6 +62,12 @@ executable text-tests

5862 test-framework-quickcheck2 >= 0.2,

5963 text-tests

6064

65+ if flag(bytestring-builder)

66+ build-depends: bytestring >= 0.9 && < 0.10.4,

67+ bytestring-builder >= 0.10.4

68+ else

69+ build-depends: bytestring >= 0.10.4

70+

6171executable text-tests-stdio

6272 main-is: Tests/IO.hs

6373

@@ -137,7 +147,12 @@ library

137147 array,

138148 base == 4.*,

139149 binary,

140− bytestring,

141150 deepseq,

142151 ghc-prim,

143152 integer-gmp

153+

154+ if flag(bytestring-builder)

155+ build-depends: bytestring >= 0.9 && < 0.10.4,

156+ bytestring-builder >= 0.10.4

157+ else

158+ build-depends: bytestring >= 0.10.4