master...7d92943 · text

git.iko.soy/text

11 files changed+95−44

since splitsnapshots

unifiedsplit

.hgtags+1 Data/Text.hs+1−1 Data/Text/Foreign.hs+1−1 Data/Text/Internal/Fusion/Common.hs+10−9 Data/Text/Lazy/Read.hs+3−3 Data/Text/Read.hs+3−3 changelog.md+18−1 tests/Tests/Properties.hs+37−20 tests/Tests/QuickCheckUtils.hs+1−1 tests/text-tests.cabal+18−3 text.cabal+2−2

.hgtags+1

@@ -64,3 +64,4 @@ f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

6464631fd055e27fa10983e17e9ad36d1b285b85c661 1.2.1.3

65658221a68ed5ec37028357df9c6c25f52320e54ad9 1.2.2.0

6666ce1c5ac65eb9f5580114b4e380092b05a485e95d 1.2.2.1

67+d72c634a91ffe27ec5cd03921dd53efc8129d8f6 1.2.2.2

Data/Text.hs+1−1

@@ -1099,8 +1099,8 @@ takeEnd n t@(Text arr off len)

10991099iterNEnd :: Int -> Text -> Int

11001100iterNEnd n t@(Text _arr _off len) = loop (len-1) n

11011101 where loop i !m

1102+ | m <= 0 = i+1

11021103 | i <= 0 = 0

1103− | m <= 1 = i

11041104 | otherwise = loop (i+d) (m-1)

11051105 where d = reverseIter_ t i

11061106

Data/Text/Foreign.hs+1−1

@@ -108,7 +108,7 @@ takeWord16 (I16 n) t@(Text arr off len)

108108 | n >= len || m >= len = t

109109 | otherwise = Text arr off m

110110 where

111− m | w < 0xDB00 || w > 0xD8FF = n

111+ m | w < 0xD800 || w > 0xDBFF = n

112112 | otherwise = n+1

113113 w = A.unsafeIndex arr (off+n-1)

114114

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/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+18−1

@@ -1,10 +1,27 @@

1+1.2.2.2

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

4+ immediately follow punctuation. Before, `"there's"` would turn into

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

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

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

9+ smaller code generated.

10+

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

12+ used after some rewrite rule tweaking.

13+

14+* Bug fix: an off-by-one bug in `takeEnd` is fixed.

15+

16+* Bug fix: a logic error in `takeWord16` is fixed.

17+

1181.2.2.1

219

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

421 The build flag has been renamed accordingly. Your army of diligent

522 maintainers apologizes for the churn.

623

7−* Spec compliance: toCaseFold now follows the Unicode 8.0 spec

24+* Spec compliance: `toCaseFold` now follows the Unicode 8.0 spec

825 (updated from 7.0)

926

1027* An STG lint error has been fixed

tests/Tests/Properties.hs+37−20

@@ -11,7 +11,7 @@ module Tests.Properties

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

1212import Control.Arrow ((***), 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)

@@ -37,6 +38,7 @@ import qualified Control.Exception as Exception

3738import qualified Data.Bits as Bits (shiftL, shiftR)

3839import qualified Data.ByteString as B

3940import qualified Data.ByteString.Lazy as BL

41+import qualified Data.Char as C

4042import qualified Data.List as L

4143import qualified Data.Text as T

4244import qualified Data.Text.Encoding as E

@@ -60,10 +62,10 @@ t_pack_unpack = (T.unpack . T.pack) `eq` id

6062tl_pack_unpack = (TL.unpack . TL.pack) `eq` id

6163t_stream_unstream = (S.unstream . S.stream) `eq` id

6264tl_stream_unstream = (SL.unstream . SL.stream) `eq` id

63−t_reverse_stream t = (S.reverse . S.reverseStream) t == t

64−t_singleton c = [c] == (T.unpack . T.singleton) c

65−tl_singleton c = [c] == (TL.unpack . TL.singleton) c

66−tl_unstreamChunks x = f 11 x == f 1000 x

65+t_reverse_stream t = (S.reverse . S.reverseStream) t === t

66+t_singleton c = [c] === (T.unpack . T.singleton) c

67+tl_singleton c = [c] === (TL.unpack . TL.singleton) c

68+tl_unstreamChunks x = f 11 x === f 1000 x

6769 where f n = SL.unstreamChunks n . S.streamList

6870tl_chunk_unchunk = (TL.fromChunks . TL.toChunks) `eq` id

6971tl_from_to_strict = (TL.fromStrict . TL.toStrict) `eq` id

@@ -74,13 +76,13 @@ encodeL1 = B.pack . map (fromIntegral . fromEnum) . T.unpack

7476encodeLazyL1 :: TL.Text -> BL.ByteString

7577encodeLazyL1 = BL.fromChunks . map encodeL1 . TL.toChunks

7678

77−t_ascii t = E.decodeASCII (E.encodeUtf8 a) == a

79+t_ascii t = E.decodeASCII (E.encodeUtf8 a) === a

7880 where a = T.map (\c -> chr (ord c `mod` 128)) t

79−tl_ascii t = EL.decodeASCII (EL.encodeUtf8 a) == a

81+tl_ascii t = EL.decodeASCII (EL.encodeUtf8 a) === a

8082 where a = TL.map (\c -> chr (ord c `mod` 128)) t

81−t_latin1 t = E.decodeLatin1 (encodeL1 a) == a

83+t_latin1 t = E.decodeLatin1 (encodeL1 a) === a

8284 where a = T.map (\c -> chr (ord c `mod` 256)) t

83−tl_latin1 t = EL.decodeLatin1 (encodeLazyL1 a) == a

85+tl_latin1 t = EL.decodeLatin1 (encodeLazyL1 a) === a

8486 where a = TL.map (\c -> chr (ord c `mod` 256)) t

8587t_utf8 = forAll genUnicode $ (E.decodeUtf8 . E.encodeUtf8) `eq` id

8688t_utf8' = forAll genUnicode $ (E.decodeUtf8' . E.encodeUtf8) `eq` (id . Right)

@@ -111,7 +113,7 @@ t_utf8_undecoded = forAll genUnicode $ \t ->

111113 let b = E.encodeUtf8 t

112114 ls = concatMap (leftover . E.encodeUtf8 . T.singleton) . T.unpack $ t

113115 leftover = (++ [B.empty]) . init . tail . B.inits

114− in (map snd . feedChunksOf 1 E.streamDecodeUtf8) b == ls

116+ in (map snd . feedChunksOf 1 E.streamDecodeUtf8) b === ls

115117

116118data Badness = Solo | Leading | Trailing

117119 deriving (Eq, Show)

@@ -201,8 +203,8 @@ t_mconcat = unsquare $

201203 mconcat `eq` (unpackS . mconcat . L.map T.pack)

202204tl_mconcat = unsquare $

203205 mconcat `eq` (unpackS . mconcat . L.map TL.pack)

204−t_mempty = mempty == (unpackS (mempty :: T.Text))

205−tl_mempty = mempty == (unpackS (mempty :: TL.Text))

206+t_mempty = mempty === (unpackS (mempty :: T.Text))

207+tl_mempty = mempty === (unpackS (mempty :: TL.Text))

206208t_IsString = fromString `eqP` (T.unpack . fromString)

207209tl_IsString = fromString `eqP` (TL.unpack . fromString)

208210

@@ -326,6 +328,19 @@ t_toUpper_upper t = p (T.toUpper t) >= p t

326328 where p = T.length . T.filter isUpper

327329tl_toUpper_upper t = p (TL.toUpper t) >= p t

328330 where p = TL.length . TL.filter isUpper

331+t_toTitle_title t = all (<= 1) (caps w)

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

333+ -- TIL: there exist uppercase-only letters

334+ w = T.filter (\c -> if C.isUpper c then C.toLower c /= c else True) t

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

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

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

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

339+ -- in Unicode 7.0

340+ -- Oh, and there exist lowercase-only letters (see previous test)

341+ stable c = if isLower c

342+ then C.toUpper c /= c

343+ else c /= '\170' && c /= '\186'

329344

330345justifyLeft k c xs = xs ++ L.replicate (k - length xs) c

331346justifyRight m n xs = L.replicate (m - length xs) n ++ xs

@@ -720,7 +735,7 @@ tl_indices (NotEmpty s) = lazyIndices s `eq` S.indices s

720735 conc = T.concat . TL.toChunks

721736t_indices_occurs = unsquare $ \(NotEmpty t) ts ->

722737 let s = T.intercalate t ts

723− in Slow.indices t s == indices t s

738+ in Slow.indices t s === indices t s

724739

725740-- Bit shifts.

726741shiftL w = forAll (choose (0,width-1)) $ \k -> Bits.shiftL w k == U.shiftL w k

@@ -812,18 +827,18 @@ tb_formatRealFloat_double (a::Double) = tb_formatRealFloat a

812827-- Reading.

813828

814829t_decimal (n::Int) s =

815− T.signed T.decimal (T.pack (show n) `T.append` t) == Right (n,t)

830+ T.signed T.decimal (T.pack (show n) `T.append` t) === Right (n,t)

816831 where t = T.dropWhile isDigit s

817832tl_decimal (n::Int) s =

818− TL.signed TL.decimal (TL.pack (show n) `TL.append` t) == Right (n,t)

833+ TL.signed TL.decimal (TL.pack (show n) `TL.append` t) === Right (n,t)

819834 where t = TL.dropWhile isDigit s

820835t_hexadecimal m s ox =

821− T.hexadecimal (T.concat [p, T.pack (showHex n ""), t]) == Right (n,t)

836+ T.hexadecimal (T.concat [p, T.pack (showHex n ""), t]) === Right (n,t)

822837 where t = T.dropWhile isHexDigit s

823838 p = if ox then "0x" else ""

824839 n = getPositive m :: Int

825840tl_hexadecimal m s ox =

826− TL.hexadecimal (TL.concat [p, TL.pack (showHex n ""), t]) == Right (n,t)

841+ TL.hexadecimal (TL.concat [p, TL.pack (showHex n ""), t]) === Right (n,t)

827842 where t = TL.dropWhile isHexDigit s

828843 p = if ox then "0x" else ""

829844 n = getPositive m :: Int

@@ -867,11 +882,11 @@ tl_write_read_line e m b t = write_read head TL.filter TL.hPutStrLn

867882

868883t_dropWord16 m t = dropWord16 m t `T.isSuffixOf` t

869884t_takeWord16 m t = takeWord16 m t `T.isPrefixOf` t

870−t_take_drop_16 m t = T.append (takeWord16 n t) (dropWord16 n t) == t

885+t_take_drop_16 m t = T.append (takeWord16 n t) (dropWord16 n t) === t

871886 where n = small m

872887t_use_from t = monadicIO $ assert . (==t) =<< run (useAsPtr t fromPtr)

873888

874−t_copy t = T.copy t == t

889+t_copy t = T.copy t === t

875890

876891-- Regression tests.

877892s_filter_eq s = S.filter p t == S.streamList (filter p s)

@@ -1030,7 +1045,9 @@ tests =

10301045 testProperty "tl_toLower_lower" tl_toLower_lower,

10311046 testProperty "t_toUpper_length" t_toUpper_length,

10321047 testProperty "t_toUpper_upper" t_toUpper_upper,

1033− testProperty "tl_toUpper_upper" tl_toUpper_upper

1048+ testProperty "tl_toUpper_upper" tl_toUpper_upper,

1049+ testProperty "t_toTitle_title" t_toTitle_title,

1050+ testProperty "t_toTitle_1stNotLower" t_toTitle_1stNotLower

10341051 ],

10351052

10361053 testGroup "justification" [

tests/Tests/QuickCheckUtils.hs+1−1

@@ -106,7 +106,7 @@ smallArbitrary = sized $ \n -> resize (smallish n) arbitrary

106106 where smallish = round . (sqrt :: Double -> Double) . fromIntegral . abs

107107

108108instance Arbitrary T.Text where

109− arbitrary = T.pack `fmap` arbitrary

109+ arbitrary = T.pack `fmap` string

110110 shrink = map T.pack . shrink . T.unpack

111111

112112instance Arbitrary TL.Text where

tests/text-tests.cabal+18−3

@@ -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

@@ -48,16 +53,21 @@ executable text-tests

4853 HUnit >= 1.2,

4954 QuickCheck >= 2.7,

5055 base == 4.*,

51− bytestring,

5256 deepseq,

5357 directory,

54− quickcheck-unicode,

58+ quickcheck-unicode >= 1.0.1.0,

5559 random,

5660 test-framework >= 0.4,

5761 test-framework-hunit >= 0.2,

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

text.cabal+2−2

@@ -1,5 +1,5 @@

11name: text

2−version: 1.2.2.1

2+version: 1.2.2.2

33homepage: https://github.com/bos/text

44bug-reports: https://github.com/bos/text/issues

55synopsis: An efficient packed Unicode text type.

@@ -180,7 +180,7 @@ test-suite tests

180180 deepseq,

181181 directory,

182182 ghc-prim,

183− quickcheck-unicode,

183+ quickcheck-unicode >= 1.0.1.0,

184184 random,

185185 test-framework >= 0.4,

186186 test-framework-hunit >= 0.2,