master..73d7307 · text

git.iko.soy/text

50 files changed+461−1330

unifiedsplit

.gitignore−1

@@ -1,3 +1,2 @@

11dist

22cabal-dev

3−cabal.sandbox.config

.hgtags−14

@@ -50,17 +50,3 @@ d99cd091cdf71ce807a4255f6cc509c3154f51ea 0.11.3.1

5050fb7299f22449750c653ac64dee4bd63c13ea6e4c 1.1.1.1

5151eb170cb533b63534d166b5fd5344b36676adf140 1.1.1.2

5252e749e05f715d02445803f653cd3d97d5395d7c6f 1.1.1.3

53−b0dff8ca1142595bab489a768ca62e6a6805aac4 1.1.1.4

54−f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

55−6231451979b2d33e28038b75f2dab19c9731f5c3 1.2.0.1

56−99a3503b4ad001fc1894e90286e96d7dea1e9195 1.2.0.2

57−47f138edf15576b9adaeed64784e30811099f202 1.2.0.3

58−0b064b0c69106bbcd13cb0099e72ae5fa2347bf0 1.2.0.4

59−8bbb3e222a0037035d893f731b0a23be449f7346 1.2.0.5

60−18c7f8e19872762b7e28ee1871e72f70ea54e3cd 1.2.0.6

61−42f751b87b4a187db93f595a8166d51746ff57d4 1.2.1.0

62−78227a84740b70dcd4cf7a65f3bcf603d87e48e2 1.2.1.1

63−40f268e6735749ef36f487e1d928eaa957bb431d 1.2.1.2

64−631fd055e27fa10983e17e9ad36d1b285b85c661 1.2.1.3

65−8221a68ed5ec37028357df9c6c25f52320e54ad9 1.2.2.0

66−ce1c5ac65eb9f5580114b4e380092b05a485e95d 1.2.2.1

Data/Text.hs+58−139

@@ -3,9 +3,6 @@

33#if __GLASGOW_HASKELL__ >= 702

44{-# LANGUAGE Trustworthy #-}

55#endif

6−#if __GLASGOW_HASKELL__ >= 708

7−{-# LANGUAGE TypeFamilies #-}

8−#endif

107-- |

118-- Module : Data.Text

@@ -129,7 +126,6 @@ module Data.Text

129126 , drop

130127 , dropEnd

131128 , takeWhile

132− , takeWhileEnd

133129 , dropWhile

134130 , dropWhileEnd

135131 , dropAround

@@ -192,16 +188,15 @@ module Data.Text

192188

193189 -- * Low level operations

194190 , copy

195− , unpackCString#

196191 ) where

197192

198193import Prelude (Char, Bool(..), Int, Maybe(..), String,

199194 Eq(..), Ord(..), Ordering(..), (++),

200− Read(..),

195+ Read(..), Show(..),

201196 (&&), (||), (+), (-), (.), ($), ($!), (>>),

202− not, return, otherwise, quot)

197+ maxBound, not, return, otherwise, quot)

203198#if defined(HAVE_DEEPSEQ)

204−import Control.DeepSeq (NFData(rnf))

199+import Control.DeepSeq (NFData)

205200#endif

206201#if defined(ASSERTS)

207202import Control.Exception (assert)

@@ -210,22 +205,15 @@ import Data.Char (isSpace)

210205import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,

211206 Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))

212207import Control.Monad (foldM)

213−import Control.Monad.ST (ST)

214208import qualified Data.Text.Array as A

215209import qualified Data.List as L

216−import Data.Binary (Binary(get, put))

217210import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

219−import Data.Semigroup (Semigroup(..))

220−#endif

221211import Data.String (IsString(..))

222212import qualified Data.Text.Internal.Fusion as S

223213import qualified Data.Text.Internal.Fusion.Common as S

224−import Data.Text.Encoding (decodeUtf8', encodeUtf8)

225214import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

226215import Data.Text.Internal.Private (span_)

227216import Data.Text.Internal (Text(..), empty, firstf, mul, safe, text)

228−import Data.Text.Show (singleton, unpack, unpackCString#)

229217import qualified Prelude as P

230218import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

231219 reverseIter_, unsafeHead, unsafeTail)

@@ -238,13 +226,12 @@ import Data.ByteString (ByteString)

238226import qualified Data.Text.Lazy as L

239227import Data.Int (Int64)

240228#endif

241−import GHC.Base (eqInt, neInt, gtInt, geInt, ltInt, leInt)

242−#if __GLASGOW_HASKELL__ >= 708

243−import qualified GHC.Exts as Exts

244−#endif

245−#if MIN_VERSION_base(4,7,0)

246−import Text.Printf (PrintfArg, formatArg, formatString)

229+#if __GLASGOW_HASKELL__ >= 702

230+import qualified GHC.CString as GHC

231+#else

232+import qualified GHC.Base as GHC

247233#endif

234+import GHC.Prim (Addr#)

248235

249236-- $strict

250237--

@@ -330,48 +317,24 @@ instance Eq Text where

330317instance Ord Text where

331318 compare = compareText

332319

320+instance Show Text where

321+ showsPrec p ps r = showsPrec p (unpack ps) r

322+

333323instance Read Text where

334324 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]

335325

336−#if MIN_VERSION_base(4,9,0)

337−-- Semigroup orphan instances for older GHCs are provided by

338−-- 'semigroups` package

339−

340−instance Semigroup Text where

341− (<>) = append

342−#endif

343−

344326instance Monoid Text where

345327 mempty = empty

346−#if MIN_VERSION_base(4,9,0)

347− mappend = (<>) -- future-proof definition

348−#else

349328 mappend = append

350−#endif

351329 mconcat = concat

352330

353331instance IsString Text where

354332 fromString = pack

355333

356−#if __GLASGOW_HASKELL__ >= 708

357−instance Exts.IsList Text where

358− type Item Text = Char

359− fromList = pack

360− toList = unpack

361−#endif

362−

363334#if defined(HAVE_DEEPSEQ)

364−instance NFData Text where rnf !_ = ()

335+instance NFData Text

365336#endif

366337

367−instance Binary Text where

368− put t = put (encodeUtf8 t)

369− get = do

370− bs <- get

371− case decodeUtf8' bs of

372− P.Left exn -> P.fail (P.show exn)

373− P.Right a -> P.return a

374−

375338-- | This instance preserves data abstraction at the cost of inefficiency.

376339-- We omit reflection services for the sake of data abstraction.

377340--

@@ -395,12 +358,6 @@ instance Data Text where

395358 _ -> P.error "gunfold"

396359 dataTypeOf _ = textDataType

397360

398−#if MIN_VERSION_base(4,7,0)

399−-- | Only defined for @base-4.7.0.0@ and later

400−instance PrintfArg Text where

401− formatArg txt = formatString $ unpack txt

402−#endif

403−

404361packConstr :: Constr

405362packConstr = mkConstr textDataType "pack" [] Prefix

406363

@@ -430,6 +387,30 @@ pack :: String -> Text

430387pack = unstream . S.map safe . S.streamList

431388{-# INLINE [1] pack #-}

432389

390+-- | /O(n)/ Convert a 'Text' into a 'String'. Subject to fusion.

391+unpack :: Text -> String

392+unpack = S.unstreamList . stream

393+{-# INLINE [1] unpack #-}

394+

395+-- | /O(n)/ Convert a literal string into a Text. Subject to fusion.

396+unpackCString# :: Addr# -> Text

397+unpackCString# addr# = unstream (S.streamCString# addr#)

398+{-# NOINLINE unpackCString# #-}

399+

400+{-# RULES "TEXT literal" forall a.

401+ unstream (S.map safe (S.streamList (GHC.unpackCString# a)))

402+ = unpackCString# a #-}

403+

404+{-# RULES "TEXT literal UTF8" forall a.

405+ unstream (S.map safe (S.streamList (GHC.unpackCStringUtf8# a)))

406+ = unpackCString# a #-}

407+

408+-- | /O(1)/ Convert a character into a Text. Subject to fusion.

409+-- Performs replacement on invalid scalar values.

410+singleton :: Char -> Text

411+singleton = unstream . S.singleton . safe

412+{-# INLINE [1] singleton #-}

413+

433414-- -----------------------------------------------------------------------------

434415-- * Basic functions

435416

@@ -460,7 +441,6 @@ append a@(Text arr1 off1 len1) b@(Text arr2 off2 len2)

460441 | otherwise = overflowError "append"

461442 where

462443 len = len1+len2

463− x :: ST s (A.MArray s)

464444 x = do

465445 arr <- A.new len

466446 A.copyI arr 0 arr1 off1 len1

@@ -486,8 +466,8 @@ head t = S.head (stream t)

486466uncons :: Text -> Maybe (Char, Text)

487467uncons t@(Text arr off len)

488468 | len <= 0 = Nothing

489− | otherwise = Just $ let !(Iter c d) = iter t 0

490− in (c, text arr (off+d) (len-d))

469+ | otherwise = Just (c, text arr (off+d) (len-d))

470+ where Iter c d = iter t 0

491471{-# INLINE [1] uncons #-}

492472

493473-- | Lifted from Control.Arrow and specialized.

@@ -572,9 +552,7 @@ isSingleton = S.isSingleton . stream

572552-- Subject to fusion.

573553length :: Text -> Int

574554length t = S.length (stream t)

575−{-# INLINE [0] length #-}

576−-- length needs to be phased after the compareN/length rules otherwise

577−-- it may inline before the rules have an opportunity to fire.

555+{-# INLINE length #-}

578556

579557-- | /O(n)/ Compare the count of characters in a 'Text' to a number.

580558-- Subject to fusion.

@@ -593,32 +571,32 @@ compareLength t n = S.compareLengthI (stream t) n

593571

594572{-# RULES

595573"TEXT ==N/length -> compareLength/==EQ" [~1] forall t n.

596− eqInt (length t) n = compareLength t n == EQ

574+ (==) (length t) n = compareLength t n == EQ

597575 #-}

598576

599577{-# RULES

600578"TEXT /=N/length -> compareLength//=EQ" [~1] forall t n.

601− neInt (length t) n = compareLength t n /= EQ

579+ (/=) (length t) n = compareLength t n /= EQ

602580 #-}

603581

604582{-# RULES

605583"TEXT <N/length -> compareLength/==LT" [~1] forall t n.

606− ltInt (length t) n = compareLength t n == LT

584+ (<) (length t) n = compareLength t n == LT

607585 #-}

608586

609587{-# RULES

610588"TEXT <=N/length -> compareLength//=GT" [~1] forall t n.

611− leInt (length t) n = compareLength t n /= GT

589+ (<=) (length t) n = compareLength t n /= GT

612590 #-}

613591

614592{-# RULES

615593"TEXT >N/length -> compareLength/==GT" [~1] forall t n.

616− gtInt (length t) n = compareLength t n == GT

594+ (>) (length t) n = compareLength t n == GT

617595 #-}

618596

619597{-# RULES

620598"TEXT >=N/length -> compareLength//=LT" [~1] forall t n.

621− geInt (length t) n = compareLength t n /= LT

599+ (>=) (length t) n = compareLength t n /= LT

622600 #-}

623601

624602-- -----------------------------------------------------------------------------

@@ -649,37 +627,11 @@ reverse :: Text -> Text

649627reverse t = S.reverse (stream t)

650628{-# INLINE reverse #-}

651629

652−-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in

653−-- @haystack@ with @replacement@.

654−--

655−-- This function behaves as though it was defined as follows:

656−--

657−-- @

658−-- replace needle replacement haystack =

659−-- 'intercalate' replacement ('splitOn' needle haystack)

660−-- @

661−--

662−-- As this suggests, each occurrence is replaced exactly once. So if

663−-- @needle@ occurs in @replacement@, that occurrence will /not/ itself

664−-- be replaced recursively:

665−--

666−-- > replace "oo" "foo" "oo" == "foo"

667−--

668−-- In cases where several instances of @needle@ overlap, only the

669−-- first one will be replaced:

670−--

671−-- > replace "ofo" "bar" "ofofo" == "barfo"

630+-- | /O(m+n)/ Replace every occurrence of one substring with another.

672631--

673632-- In (unlikely) bad cases, this function's time complexity degrades

674633-- towards /O(n*m)/.

675−replace :: Text

676− -- ^ @needle@ to search for. If this string is empty, an

677− -- error will occur.

678− -> Text

679− -- ^ @replacement@ to replace @needle@ with.

680− -> Text

681− -- ^ @haystack@ in which to search.

682− -> Text

634+replace :: Text -> Text -> Text -> Text

683635replace needle@(Text _ _ neeLen)

684636 (Text repArr repOff repLen)

685637 haystack@(Text hayArr hayOff hayLen)

@@ -690,7 +642,6 @@ replace needle@(Text _ _ neeLen)

690642 where

691643 ixs = indices needle haystack

692644 len = hayLen - (neeLen - repLen) `mul` L.length ixs

693− x :: ST s (A.MArray s)

694645 x = do

695646 marr <- A.new len

696647 let loop (i:is) o d = do

@@ -741,7 +692,7 @@ replace needle@(Text _ _ neeLen)

741692-- instead of itself.

742693toCaseFold :: Text -> Text

743694toCaseFold t = unstream (S.toCaseFold (stream t))

744−{-# INLINE toCaseFold #-}

695+{-# INLINE [0] toCaseFold #-}

745696

746697-- | /O(n)/ Convert a string to lower case, using simple case

747698-- conversion. Subject to fusion.

@@ -901,7 +852,6 @@ concat ts = case ts' of

901852 where

902853 ts' = L.filter (not . null) ts

903854 len = sumP "concat" $ L.map lengthWord16 ts'

904− go :: ST s (A.MArray s)

905855 go = do

906856 arr <- A.new len

907857 let step i (Text a o l) =

@@ -915,13 +865,13 @@ concatMap f = concat . foldr ((:) . f) []

915865{-# INLINE concatMap #-}

916866

917867-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the

918−-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.

868+-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.

919869any :: (Char -> Bool) -> Text -> Bool

920870any p t = S.any p (stream t)

921871{-# INLINE any #-}

922872

923873-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the

924−-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.

874+-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.

925875all :: (Char -> Bool) -> Text -> Bool

926876all p t = S.all p (stream t)

927877{-# INLINE all #-}

@@ -1015,7 +965,6 @@ replicate n t@(Text a o l)

1015965 | otherwise = Text (A.run x) 0 len

1016966 where

1017967 len = l `mul` n

1018− x :: ST s (A.MArray s)

1019968 x = do

1020969 arr <- A.new len

1021970 let loop !d !i | i >= n = return arr

@@ -1152,27 +1101,6 @@ takeWhile p t@(Text arr off len) = loop 0

11521101 unstream (S.takeWhile p (stream t)) = takeWhile p t

11531102 #-}

11541103

1155−-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',

1156−-- returns the longest suffix (possibly empty) of elements that

1157−-- satisfy @p@. Subject to fusion.

1158−-- Examples:

1159−--

1160−-- > takeWhileEnd (=='o') "foo" == "oo"

1161−takeWhileEnd :: (Char -> Bool) -> Text -> Text

1162−takeWhileEnd p t@(Text arr off len) = loop (len-1) len

1163− where loop !i !l | l <= 0 = t

1164− | p c = loop (i+d) (l+d)

1165− | otherwise = text arr (off+l) (len-l)

1166− where (c,d) = reverseIter t i

1167−{-# INLINE [1] takeWhileEnd #-}

1168−

1169−{-# RULES

1170−"TEXT takeWhileEnd -> fused" [~1] forall p t.

1171− takeWhileEnd p t = S.reverse (S.takeWhile p (S.reverseStream t))

1172−"TEXT takeWhileEnd -> unfused" [1] forall p t.

1173− S.reverse (S.takeWhile p (S.reverseStream t)) = takeWhileEnd p t

1174− #-}

1175−

11761104-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

11771105-- 'takeWhile' @p@ @t@. Subject to fusion.

11781106dropWhile :: (Char -> Bool) -> Text -> Text

@@ -1191,9 +1119,8 @@ dropWhile p t@(Text arr off len) = loop 0 0

11911119 #-}

11921120

11931121-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1194−-- dropping characters that satisfy the predicate @p@ from the end of

1122+-- dropping characters that fail the predicate @p@ from the end of

11951123-- @t@. Subject to fusion.

1196−--

11971124-- Examples:

11981125--

11991126-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1213,7 +1140,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len

12131140 #-}

12141141

12151142-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1216−-- dropping characters that satisfy the predicate @p@ from both the

1143+-- dropping characters that fail the predicate @p@ from both the

12171144-- beginning and end of @t@. Subject to fusion.

12181145dropAround :: (Char -> Bool) -> Text -> Text

12191146dropAround p = dropWhile p . dropWhileEnd p

@@ -1308,9 +1235,9 @@ tails t | null t = [empty]

13081235-- copies to create substrings; they just construct new 'Text's that

13091236-- are slices of the original.

13101237

1311−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'

1312−-- argument (which cannot be empty), consuming the delimiter. An empty

1313−-- delimiter is invalid, and will cause an error to be raised.

1238+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first

1239+-- 'Text' argument, consuming the delimiter. An empty delimiter is

1240+-- invalid, and will cause an error to be raised.

13141241--

13151242-- Examples:

13161243--

@@ -1323,16 +1250,9 @@ tails t | null t = [empty]

13231250-- > intercalate s . splitOn s == id

13241251-- > splitOn (singleton c) == split (==c)

13251252--

1326−-- (Note: the string @s@ to split on above cannot be empty.)

1327−--

13281253-- In (unlikely) bad cases, this function's time complexity degrades

13291254-- towards /O(n*m)/.

1330−splitOn :: Text

1331− -- ^ String to split on. If this string is empty, an error

1332− -- will occur.

1333− -> Text

1334− -- ^ Input text.

1335− -> [Text]

1255+splitOn :: Text -> Text -> [Text]

13361256splitOn pat@(Text _ _ l) src@(Text arr off len)

13371257 | l <= 0 = emptyError "splitOn"

13381258 | isSingleton pat = split (== unsafeHead pat) src

@@ -1540,7 +1460,7 @@ countChar c t = S.countChar c (stream t)

15401460-- equivalent to a pair of 'unpack' operations.

15411461zip :: Text -> Text -> [(Char,Char)]

15421462zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)

1543−{-# INLINE zip #-}

1463+{-# INLINE [0] zip #-}

15441464

15451465-- | /O(n)/ 'zipWith' generalises 'zip' by zipping with the function

15461466-- given as the first argument, instead of a tupling function.

@@ -1548,7 +1468,7 @@ zip a b = S.unstreamList $ S.zipWith (,) (stream a) (stream b)

15481468zipWith :: (Char -> Char -> Char) -> Text -> Text -> Text

15491469zipWith f t1 t2 = unstream (S.zipWith g (stream t1) (stream t2))

15501470 where g a b = safe (f a b)

1551−{-# INLINE zipWith #-}

1471+{-# INLINE [0] zipWith #-}

15521472

15531473-- | /O(n)/ Breaks a 'Text' up into a list of words, delimited by 'Char's

15541474-- representing white space.

@@ -1748,7 +1668,6 @@ overflowError fun = P.error $ "Data.Text." ++ fun ++ ": size overflow"

17481668copy :: Text -> Text

17491669copy (Text arr off len) = Text (A.run go) 0 len

17501670 where

1751− go :: ST s (A.MArray s)

17521671 go = do

17531672 marr <- A.new len

17541673 A.copyI marr 0 arr off len

Data/Text/Array.hs+4−5

@@ -20,7 +20,7 @@

2020-- > import qualified Data.Text.Array as A

2121--

2222-- The names in this module resemble those in the 'Data.Array' family

23−-- of modules, but are shorter due to the assumption of qualified

23+-- of modules, but are shorter due to the assumption of qualifid

2424-- naming.

2525module Data.Text.Array

2626 (

@@ -74,7 +74,7 @@ import Foreign.C.Types (CInt, CSize)

7474#endif

7575import GHC.Base (ByteArray#, MutableByteArray#, Int(..),

7676 indexWord16Array#, newByteArray#,

77− unsafeFreezeByteArray#, writeWord16Array#)

77+ unsafeCoerce#, writeWord16Array#)

7878import GHC.ST (ST(..), runST)

7979import GHC.Word (Word16(..))

8080import Prelude hiding (length, read)

@@ -130,9 +130,8 @@ array_size_error = error "Data.Text.Array.new: size overflow"

130130

131131-- | Freeze a mutable array. Do not mutate the 'MArray' afterwards!

132132unsafeFreeze :: MArray s -> ST s Array

133−unsafeFreeze MArray{..} = ST $ \s1# ->

134− case unsafeFreezeByteArray# maBA s1# of

135− (# s2#, ba# #) -> (# s2#, Array ba#

133+unsafeFreeze MArray{..} = ST $ \s# ->

134+ (# s#, Array (unsafeCoerce# maBA)

136135#if defined(ASSERTS)

137136 maLen

138137#endif

Data/Text/Encoding.hs+23−10

@@ -55,9 +55,13 @@ module Data.Text.Encoding

5555 , encodeUtf32LE

5656 , encodeUtf32BE

5757

58+#if MIN_VERSION_bytestring(0,10,4)

5859 -- * Encoding Text using ByteString Builders

60+ -- | /Note/ that these functions are only available if built against

61+ -- @bytestring >= 0.10.4.0@.

5962 , encodeUtf8Builder

6063 , encodeUtf8BuilderEscaped

64+#endif

6165 ) where

6266

6367#if __GLASGOW_HASKELL__ >= 702

@@ -66,17 +70,26 @@ import Control.Monad.ST.Unsafe (unsafeIOToST, unsafeSTToIO)

6670import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

6771#endif

6872

73+#if MIN_VERSION_bytestring(0,10,4)

74+import Data.Bits ((.&.))

75+import Data.Text.Internal.Unsafe.Char (ord)

76+import qualified Data.ByteString.Builder as B

77+import qualified Data.ByteString.Builder.Internal as B hiding (empty, append)

78+import qualified Data.ByteString.Builder.Prim as BP

79+import qualified Data.ByteString.Builder.Prim.Internal as BP

80+import qualified Data.Text.Internal.Encoding.Utf16 as U16

81+#endif

82+

6983import Control.Exception (evaluate, try)

7084import Control.Monad.ST (runST)

71−import Data.Bits ((.&.))

7285import Data.ByteString as B

7386import Data.ByteString.Internal as B hiding (c2w)

87+import Data.Text ()

7488import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

7589import Data.Text.Internal (Text(..), safe, text)

7690import Data.Text.Internal.Private (runText)

77−import Data.Text.Internal.Unsafe.Char (ord, unsafeWrite)

91+import Data.Text.Internal.Unsafe.Char (unsafeWrite)

7892import Data.Text.Internal.Unsafe.Shift (shiftR)

79−import Data.Text.Show ()

8093import Data.Text.Unsafe (unsafeDupablePerformIO)

8194import Data.Word (Word8, Word32)

8295import Foreign.C.Types (CSize(..))

@@ -85,13 +98,8 @@ import Foreign.Marshal.Utils (with)

8598import Foreign.Ptr (Ptr, minusPtr, nullPtr, plusPtr)

8699import Foreign.Storable (Storable, peek, poke)

87100import GHC.Base (ByteArray#, MutableByteArray#)

88−import qualified Data.ByteString.Builder as B

89−import qualified Data.ByteString.Builder.Internal as B hiding (empty, append)

90−import qualified Data.ByteString.Builder.Prim as BP

91−import qualified Data.ByteString.Builder.Prim.Internal as BP

92101import qualified Data.Text.Array as A

93102import qualified Data.Text.Internal.Encoding.Fusion as E

94−import qualified Data.Text.Internal.Encoding.Utf16 as U16

95103import qualified Data.Text.Internal.Fusion as F

96104

97105#include "text_cbits.h"

@@ -111,6 +119,8 @@ import qualified Data.Text.Internal.Fusion as F

111119

112120-- | /Deprecated/. Decode a 'ByteString' containing 7-bit ASCII

113121-- encoded text.

122+--

123+-- This function is deprecated. Use 'decodeLatin1' instead.

114124decodeASCII :: ByteString -> Text

115125decodeASCII = decodeUtf8

116126{-# DEPRECATED decodeASCII "Use decodeUtf8 instead" #-}

@@ -259,7 +269,6 @@ streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

259269 UTF8_REJECT -> do

260270 -- We encountered an encoding error

261271 x <- peek curPtr'

262− poke statePtr 0

263272 case onErr desc (Just x) of

264273 Nothing -> loop $ curPtr' `plusPtr` 1

265274 Just c -> do

@@ -267,6 +276,7 @@ streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

267276 w <- unsafeSTToIO $

268277 unsafeWrite dest (fromIntegral destOff) (safe c)

269278 poke destOffPtr (destOff + fromIntegral w)

279+ poke statePtr 0

270280 loop $ curPtr' `plusPtr` 1

271281

272282 _ -> do

@@ -278,7 +288,7 @@ streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

278288 return $! text arr 0 (fromIntegral n)

279289 lastPtr <- peek curPtrPtr

280290 let left = lastPtr `minusPtr` curPtr

281− !undecoded = case state of

291+ undecoded = case state of

282292 UTF8_ACCEPT -> B.empty

283293 _ -> B.append undecoded0 (B.drop left bs)

284294 return $ Some chunkText undecoded

@@ -307,6 +317,8 @@ decodeUtf8' :: ByteString -> Either UnicodeException Text

307317decodeUtf8' = unsafeDupablePerformIO . try . evaluate . decodeUtf8With strictDecode

308318{-# INLINE decodeUtf8' #-}

309319

320+#if MIN_VERSION_bytestring(0,10,4)

321+

310322-- | Encode text to a ByteString 'B.Builder' using UTF-8 encoding.

311323encodeUtf8Builder :: Text -> B.Builder

312324encodeUtf8Builder = encodeUtf8BuilderEscaped (BP.liftFixedToBounded BP.word8)

@@ -367,6 +379,7 @@ encodeUtf8BuilderEscaped be =

367379 outerLoop i (B.BufferRange op ope)

368380 where

369381 poke8 j v = poke (op `plusPtr` j) (fromIntegral v :: Word8)

382+#endif

370383

371384-- | Encode text using UTF-8 encoding.

372385encodeUtf8 :: Text -> ByteString

Data/Text/Encoding/Error.hs+1−7

@@ -1,7 +1,5 @@

11{-# LANGUAGE CPP, DeriveDataTypeable #-}

2−#if __GLASGOW_HASKELL__ >= 704

3−{-# LANGUAGE Safe #-}

4−#elif __GLASGOW_HASKELL__ >= 702

2+#if __GLASGOW_HASKELL__ >= 702

53{-# LANGUAGE Trustworthy #-}

64#endif

75-- |

@@ -68,7 +66,6 @@ type OnError a b = String -> Maybe a -> Maybe b

6866type OnDecodeError = OnError Word8 Char

6967

7068-- | A handler for an encoding error.

71−{-# DEPRECATED OnEncodeError "This exception is never used in practice, and will be removed." #-}

7269type OnEncodeError = OnError Char Word8

7370

7471-- | An exception type for representing Unicode encoding errors.

@@ -81,8 +78,6 @@ data UnicodeException =

8178 -- under the given encoding, or ran out of input in mid-encode.

8279 deriving (Eq, Typeable)

8380

84−{-# DEPRECATED EncodeError "This constructor is never used, and will be removed." #-}

85−

8681showUnicodeException :: UnicodeException -> String

8782showUnicodeException (DecodeError desc (Just w))

8883 = "Cannot decode byte '\\x" ++ showHex w ("': " ++ desc)

@@ -112,7 +107,6 @@ lenientDecode :: OnDecodeError

112107lenientDecode _ _ = Just '\xfffd'

113108

114109-- | Throw a 'UnicodeException' if encoding fails.

115−{-# DEPRECATED strictEncode "This function always throws an exception, and will be removed." #-}

116110strictEncode :: OnEncodeError

117111strictEncode desc c = throw (EncodeError desc c)

118112

Data/Text/Internal.hs+1−7

@@ -36,7 +36,6 @@ module Data.Text.Internal

3636 , safe

3737 -- * Code that must be here for accessibility

3838 , empty

39− , empty_

4039 -- * Utilities

4140 , firstf

4241 -- * Checked multiplication

@@ -82,11 +81,6 @@ empty :: Text

8281empty = Text A.empty 0 0

8382{-# INLINE [1] empty #-}

8483

85−-- | A non-inlined version of 'empty'.

86−empty_ :: Text

87−empty_ = Text A.empty 0 0

88−{-# NOINLINE empty_ #-}

89−

9084-- | Construct a 'Text' without invisibly pinning its byte array in

9185-- memory if its length has dwindled to zero.

9286text :: A.Array -> Int -> Int -> Text

@@ -116,7 +110,7 @@ safe :: Char -> Char

116110safe c

117111 | ord c .&. 0x1ff800 /= 0xd800 = c

118112 | otherwise = '\xfffd'

119−{-# INLINE [0] safe #-}

113+{-# INLINE safe #-}

120114

121115-- | Apply a function to the first element of an optional pair.

122116firstf :: (a -> c) -> Maybe (a,b) -> Maybe (c,b)

Data/Text/Internal/Builder.hs+16−19

@@ -6,7 +6,7 @@

66-- Module : Data.Text.Internal.Builder

9−-- License : BSD-style (see LICENSE)

9+-- License : BSD3-style (see LICENSE)

1010--

1111-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1212-- Stability : experimental

@@ -58,14 +58,13 @@ module Data.Text.Internal.Builder

5858 ) where

5959

6060import Control.Monad.ST (ST, runST)

61+import Data.Bits ((.&.))

6162import Data.Monoid (Monoid(..))

62−#if MIN_VERSION_base(4,9,0)

63−import Data.Semigroup (Semigroup(..))

64−#endif

6563import Data.Text.Internal (Text(..))

6664import Data.Text.Internal.Lazy (smallChunkSize)

6765import Data.Text.Unsafe (inlineInterleaveST)

68−import Data.Text.Internal.Unsafe.Char (unsafeWrite)

66+import Data.Text.Internal.Unsafe.Char (ord, unsafeWrite)

67+import Data.Text.Internal.Unsafe.Shift (shiftR)

6968import Prelude hiding (map, putChar)

7069

7170import qualified Data.String as String

@@ -92,22 +91,12 @@ newtype Builder = Builder {

9291 -> ST s [S.Text]

9392 }

9493

95−#if MIN_VERSION_base(4,9,0)

96−instance Semigroup Builder where

97− (<>) = append

98− {-# INLINE (<>) #-}

99−#endif

100−

10194instance Monoid Builder where

10295 mempty = empty

10396 {-# INLINE mempty #-}

104−#if MIN_VERSION_base(4,9,0)

105− mappend = (<>) -- future-proof definition

106−#else

10797 mappend = append

108−#endif

10998 {-# INLINE mappend #-}

110− mconcat = foldr mappend Data.Monoid.mempty

99+ mconcat = foldr mappend mempty

111100 {-# INLINE mconcat #-}

112101

113102instance String.IsString Builder where

@@ -138,7 +127,17 @@ empty = Builder (\ k buf -> k buf)

138127-- * @'toLazyText' ('singleton' c) = 'L.singleton' c@

139128--

140129singleton :: Char -> Builder

141−singleton c = writeAtMost 2 $ \ marr o -> unsafeWrite marr o c

130+singleton c = writeAtMost 2 $ \ marr o ->

131+ if n < 0x10000

132+ then A.unsafeWrite marr o (fromIntegral n) >> return 1

133+ else do

134+ A.unsafeWrite marr o lo

135+ A.unsafeWrite marr (o+1) hi

136+ return 2

137+ where n = ord c

138+ m = n - 0x10000

139+ lo = fromIntegral $ (m `shiftR` 10) + 0xD800

140+ hi = fromIntegral $ (m .&. 0x3FF) + 0xDC00

142141{-# INLINE singleton #-}

143142

144143------------------------------------------------------------------------

@@ -242,8 +241,6 @@ flush = Builder $ \ k buf@(Buffer p o u l) ->

242241 !t = Text arr o u

243242 ts <- inlineInterleaveST (k b)

244243 return $! t : ts

245−{-# INLINE [1] flush #-}

246−-- defer inlining so that flush/flush rule may fire.

247244

248245------------------------------------------------------------------------

249246

Data/Text/Internal/Builder/Functions.hs+1−2

@@ -23,8 +23,7 @@ module Data.Text.Internal.Builder.Functions

2323

2424import Data.Monoid (mappend)

2525import Data.Text.Lazy.Builder (Builder)

26−import GHC.Base (chr#,ord#,(+#),Int(I#),Char(C#))

27−import Prelude ()

26+import GHC.Base

2827

2928-- | Unsafe conversion for decimal digits.

3029{-# INLINE i2d #-}

Data/Text/Internal/Builder/Int/Digits.hs+1−1

@@ -2,7 +2,7 @@

33-- Module: Data.Text.Internal.Builder.Int.Digits

5−-- License: BSD-style

5+-- License: BSD3

66-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

77-- Stability: experimental

88-- Portability: portable

Data/Text/Internal/Builder/RealFloat/Functions.hs−28

@@ -1,5 +1,3 @@

1−{-# LANGUAGE CPP #-}

2−

31-- |

42-- Module: Data.Text.Internal.Builder.RealFloat.Functions

53-- Copyright: (c) The University of Glasgow 1994-2002

@@ -15,30 +13,6 @@ module Data.Text.Internal.Builder.RealFloat.Functions

1513 ) where

1614

1715roundTo :: Int -> [Int] -> (Int,[Int])

18−

19−#if MIN_VERSION_base(4,6,0)

20−

21−roundTo d is =

22− case f d True is of

23− x@(0,_) -> x

24− (1,xs) -> (1, 1:xs)

25− _ -> error "roundTo: bad Value"

26− where

27− b2 = base `quot` 2

28−

29− f n _ [] = (0, replicate n 0)

30− f 0 e (x:xs) | x == b2 && e && all (== 0) xs = (0, []) -- Round to even when at exactly half the base

31− | otherwise = (if x >= b2 then 1 else 0, [])

32− f n _ (i:xs)

33− | i' == base = (1,0:ds)

34− | otherwise = (0,i':ds)

35− where

36− (c,ds) = f (n-1) (even i) xs

37− i' = c + i

38− base = 10

39−

40−#else

41−

4216roundTo d is =

4317 case f d is of

4418 x@(0,_) -> x

@@ -53,5 +27,3 @@ roundTo d is =

5327 where

5428 (c,ds) = f (n-1) xs

5529 i' = c + i

56−

57−#endif

Data/Text/Internal/Fusion.hs+24−34

@@ -98,35 +98,25 @@ reverseStream (Text arr off len) = Stream next (off+len-1) (betweenSize (len `sh

9898-- | /O(n)/ Convert a 'Stream Char' into a 'Text'.

9999unstream :: Stream Char -> Text

100100unstream (Stream next0 s0 len) = runText $ \done -> do

101− -- Before encoding each char we perform a buffer realloc check assuming

102− -- worst case encoding size of two 16-bit units for the char. Just add an

103− -- extra space to the buffer so that we do not end up reallocating even when

104− -- all the chars are encoded as single unit.

105− let mlen = upperBound 4 len + 1

101+ let mlen = upperBound 4 len

106102 arr0 <- A.new mlen

107− let outer !arr !maxi = encode

103+ let outer arr top = loop

108104 where

109− -- keep the common case loop as small as possible

110− encode !si !di =

111− case next0 si of

112− Done -> done arr di

113− Skip si' -> encode si' di

114− Yield c si'

115− -- simply check for the worst case

116− | maxi < di + 1 -> realloc si di

117− | otherwise -> do

118− n <- unsafeWrite arr di c

119− encode si' (di + n)

120−

121− -- keep uncommon case separate from the common case code

122− {-# NOINLINE realloc #-}

123− realloc !si !di = do

124− let newlen = (maxi + 1) * 2

125− arr' <- A.new newlen

126− A.copyM arr' 0 arr 0 di

127− outer arr' (newlen - 1) si di

128−

129− outer arr0 (mlen - 1) s0 0

105+ loop !s !i =

106+ case next0 s of

107+ Done -> done arr i

108+ Skip s' -> loop s' i

109+ Yield x s'

110+ | j >= top -> {-# SCC "unstream/resize" #-} do

111+ let top' = (top + 1) `shiftL` 1

112+ arr' <- A.new top'

113+ A.copyM arr' 0 arr 0 top

114+ outer arr' top' s i

115+ | otherwise -> do d <- unsafeWrite arr i x

116+ loop s' (i+d)

117+ where j | ord x < 0x10000 = i

118+ | otherwise = i + 1

119+ outer arr0 mlen s0 0

130120{-# INLINE [0] unstream #-}

131121{-# RULES "STREAM stream/unstream fusion" forall s. stream (unstream s) = s #-}

132122

@@ -176,15 +166,15 @@ reverse (Stream next s len0)

176166-- | /O(n)/ Perform the equivalent of 'scanr' over a list, only with

177167-- the input and result reversed.

178168reverseScanr :: (Char -> Char -> Char) -> Char -> Stream Char -> Stream Char

179−reverseScanr f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT maybe too low

169+reverseScanr f z0 (Stream next0 s0 len) = Stream next (S1 :*: z0 :*: s0) (len+1) -- HINT maybe too low

180170 where

181171 {-# INLINE next #-}

182− next (Scan1 z s) = Yield z (Scan2 z s)

183− next (Scan2 z s) = case next0 s of

184− Yield x s' -> let !x' = f x z

185− in Yield x' (Scan2 x' s')

186− Skip s' -> Skip (Scan2 z s')

187− Done -> Done

172+ next (S1 :*: z :*: s) = Yield z (S2 :*: z :*: s)

173+ next (S2 :*: z :*: s) = case next0 s of

174+ Yield x s' -> let !x' = f x z

175+ in Yield x' (S2 :*: x' :*: s')

176+ Skip s' -> Skip (S2 :*: z :*: s')

177+ Done -> Done

188178{-# INLINE reverseScanr #-}

189179

190180-- | /O(n)/ Like 'unfoldr', 'unfoldrN' builds a stream from a seed

Data/Text/Internal/Fusion/CaseMapping.hs+19−291

@@ -1,17 +1,17 @@

11{-# LANGUAGE Rank2Types #-}

22-- AUTOMATICALLY GENERATED - DO NOT EDIT

3−-- Generated by scripts/CaseMapping.hs

4−-- CaseFolding-8.0.0.txt

5−-- Date: 2015-01-13, 18:16:36 GMT [MD]

6−-- SpecialCasing-8.0.0.txt

7−-- Date: 2014-12-16, 23:08:04 GMT [MD]

3+-- Generated by scripts/SpecialCasing.hs

4+-- CaseFolding-6.3.0.txt

5+-- Date: 2012-12-20, 22:14:35 GMT [MD]

6+-- SpecialCasing-6.3.0.txt

7+-- Date: 2013-05-08, 13:54:51 GMT [MD]

99module Data.Text.Internal.Fusion.CaseMapping where

1010import Data.Char

1111import Data.Text.Internal.Fusion.Types

1212

1313upperMapping :: forall s. Char -> s -> Step (CC s) Char

14−{-# NOINLINE upperMapping #-}

14+{-# INLINE upperMapping #-}

1515-- LATIN SMALL LETTER SHARP S

1616upperMapping '\x00df' s = Yield '\x0053' (CC s '\x0053' '\x0000')

1717-- LATIN SMALL LIGATURE FF

@@ -218,12 +218,12 @@ upperMapping '\x1fc7' s = Yield '\x0397' (CC s '\x0342' '\x0399')

218218upperMapping '\x1ff7' s = Yield '\x03a9' (CC s '\x0342' '\x0399')

219219upperMapping c s = Yield (toUpper c) (CC s '\0' '\0')

220220lowerMapping :: forall s. Char -> s -> Step (CC s) Char

221−{-# NOINLINE lowerMapping #-}

221+{-# INLINE lowerMapping #-}

222222-- LATIN CAPITAL LETTER I WITH DOT ABOVE

223223lowerMapping '\x0130' s = Yield '\x0069' (CC s '\x0307' '\x0000')

224224lowerMapping c s = Yield (toLower c) (CC s '\0' '\0')

225225titleMapping :: forall s. Char -> s -> Step (CC s) Char

226−{-# NOINLINE titleMapping #-}

226+{-# INLINE titleMapping #-}

227227-- LATIN SMALL LETTER SHARP S

228228titleMapping '\x00df' s = Yield '\x0053' (CC s '\x0073' '\x0000')

229229-- LATIN SMALL LIGATURE FF

@@ -322,7 +322,7 @@ titleMapping '\x1fc7' s = Yield '\x0397' (CC s '\x0342' '\x0345')

322322titleMapping '\x1ff7' s = Yield '\x03a9' (CC s '\x0342' '\x0345')

323323titleMapping c s = Yield (toTitle c) (CC s '\0' '\0')

324324foldMapping :: forall s. Char -> s -> Step (CC s) Char

325−{-# NOINLINE foldMapping #-}

325+{-# INLINE foldMapping #-}

326326-- MICRO SIGN

327327foldMapping '\x00b5' s = Yield '\x03bc' (CC s '\x0000' '\x0000')

328328-- LATIN SMALL LETTER SHARP S

@@ -359,18 +359,10 @@ foldMapping '\x03f1' s = Yield '\x03c1' (CC s '\x0000' '\x0000')

359359foldMapping '\x03f5' s = Yield '\x03b5' (CC s '\x0000' '\x0000')

360360-- ARMENIAN SMALL LIGATURE ECH YIWN

361361foldMapping '\x0587' s = Yield '\x0565' (CC s '\x0582' '\x0000')

362−-- CHEROKEE SMALL LETTER YE

363−foldMapping '\x13f8' s = Yield '\x13f0' (CC s '\x0000' '\x0000')

364−-- CHEROKEE SMALL LETTER YI

365−foldMapping '\x13f9' s = Yield '\x13f1' (CC s '\x0000' '\x0000')

366−-- CHEROKEE SMALL LETTER YO

367−foldMapping '\x13fa' s = Yield '\x13f2' (CC s '\x0000' '\x0000')

368−-- CHEROKEE SMALL LETTER YU

369−foldMapping '\x13fb' s = Yield '\x13f3' (CC s '\x0000' '\x0000')

370−-- CHEROKEE SMALL LETTER YV

371−foldMapping '\x13fc' s = Yield '\x13f4' (CC s '\x0000' '\x0000')

372−-- CHEROKEE SMALL LETTER MV

373−foldMapping '\x13fd' s = Yield '\x13f5' (CC s '\x0000' '\x0000')

362+-- GEORGIAN CAPITAL LETTER YN

363+foldMapping '\x10c7' s = Yield '\x2d27' (CC s '\x0000' '\x0000')

364+-- GEORGIAN CAPITAL LETTER AEN

365+foldMapping '\x10cd' s = Yield '\x2d2d' (CC s '\x0000' '\x0000')

374366-- LATIN SMALL LETTER H WITH LINE BELOW

375367foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

376368-- LATIN SMALL LETTER T WITH DIAERESIS

@@ -545,174 +537,12 @@ foldMapping '\x1ff6' s = Yield '\x03c9' (CC s '\x0342' '\x0000')

545537foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

546538-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

547539foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

548−-- LATIN CAPITAL LETTER J WITH CROSSED-TAIL

549−foldMapping '\xa7b2' s = Yield '\x029d' (CC s '\x0000' '\x0000')

550−-- LATIN CAPITAL LETTER CHI

551−foldMapping '\xa7b3' s = Yield '\xab53' (CC s '\x0000' '\x0000')

552−-- LATIN CAPITAL LETTER BETA

553−foldMapping '\xa7b4' s = Yield '\xa7b5' (CC s '\x0000' '\x0000')

554−-- LATIN CAPITAL LETTER OMEGA

555−foldMapping '\xa7b6' s = Yield '\xa7b7' (CC s '\x0000' '\x0000')

556−-- CHEROKEE SMALL LETTER A

557−foldMapping '\xab70' s = Yield '\x13a0' (CC s '\x0000' '\x0000')

558−-- CHEROKEE SMALL LETTER E

559−foldMapping '\xab71' s = Yield '\x13a1' (CC s '\x0000' '\x0000')

560−-- CHEROKEE SMALL LETTER I

561−foldMapping '\xab72' s = Yield '\x13a2' (CC s '\x0000' '\x0000')

562−-- CHEROKEE SMALL LETTER O

563−foldMapping '\xab73' s = Yield '\x13a3' (CC s '\x0000' '\x0000')

564−-- CHEROKEE SMALL LETTER U

565−foldMapping '\xab74' s = Yield '\x13a4' (CC s '\x0000' '\x0000')

566−-- CHEROKEE SMALL LETTER V

567−foldMapping '\xab75' s = Yield '\x13a5' (CC s '\x0000' '\x0000')

568−-- CHEROKEE SMALL LETTER GA

569−foldMapping '\xab76' s = Yield '\x13a6' (CC s '\x0000' '\x0000')

570−-- CHEROKEE SMALL LETTER KA

571−foldMapping '\xab77' s = Yield '\x13a7' (CC s '\x0000' '\x0000')

572−-- CHEROKEE SMALL LETTER GE

573−foldMapping '\xab78' s = Yield '\x13a8' (CC s '\x0000' '\x0000')

574−-- CHEROKEE SMALL LETTER GI

575−foldMapping '\xab79' s = Yield '\x13a9' (CC s '\x0000' '\x0000')

576−-- CHEROKEE SMALL LETTER GO

577−foldMapping '\xab7a' s = Yield '\x13aa' (CC s '\x0000' '\x0000')

578−-- CHEROKEE SMALL LETTER GU

579−foldMapping '\xab7b' s = Yield '\x13ab' (CC s '\x0000' '\x0000')

580−-- CHEROKEE SMALL LETTER GV

581−foldMapping '\xab7c' s = Yield '\x13ac' (CC s '\x0000' '\x0000')

582−-- CHEROKEE SMALL LETTER HA

583−foldMapping '\xab7d' s = Yield '\x13ad' (CC s '\x0000' '\x0000')

584−-- CHEROKEE SMALL LETTER HE

585−foldMapping '\xab7e' s = Yield '\x13ae' (CC s '\x0000' '\x0000')

586−-- CHEROKEE SMALL LETTER HI

587−foldMapping '\xab7f' s = Yield '\x13af' (CC s '\x0000' '\x0000')

588−-- CHEROKEE SMALL LETTER HO

589−foldMapping '\xab80' s = Yield '\x13b0' (CC s '\x0000' '\x0000')

590−-- CHEROKEE SMALL LETTER HU

591−foldMapping '\xab81' s = Yield '\x13b1' (CC s '\x0000' '\x0000')

592−-- CHEROKEE SMALL LETTER HV

593−foldMapping '\xab82' s = Yield '\x13b2' (CC s '\x0000' '\x0000')

594−-- CHEROKEE SMALL LETTER LA

595−foldMapping '\xab83' s = Yield '\x13b3' (CC s '\x0000' '\x0000')

596−-- CHEROKEE SMALL LETTER LE

597−foldMapping '\xab84' s = Yield '\x13b4' (CC s '\x0000' '\x0000')

598−-- CHEROKEE SMALL LETTER LI

599−foldMapping '\xab85' s = Yield '\x13b5' (CC s '\x0000' '\x0000')

600−-- CHEROKEE SMALL LETTER LO

601−foldMapping '\xab86' s = Yield '\x13b6' (CC s '\x0000' '\x0000')

602−-- CHEROKEE SMALL LETTER LU

603−foldMapping '\xab87' s = Yield '\x13b7' (CC s '\x0000' '\x0000')

604−-- CHEROKEE SMALL LETTER LV

605−foldMapping '\xab88' s = Yield '\x13b8' (CC s '\x0000' '\x0000')

606−-- CHEROKEE SMALL LETTER MA

607−foldMapping '\xab89' s = Yield '\x13b9' (CC s '\x0000' '\x0000')

608−-- CHEROKEE SMALL LETTER ME

609−foldMapping '\xab8a' s = Yield '\x13ba' (CC s '\x0000' '\x0000')

610−-- CHEROKEE SMALL LETTER MI

611−foldMapping '\xab8b' s = Yield '\x13bb' (CC s '\x0000' '\x0000')

612−-- CHEROKEE SMALL LETTER MO

613−foldMapping '\xab8c' s = Yield '\x13bc' (CC s '\x0000' '\x0000')

614−-- CHEROKEE SMALL LETTER MU

615−foldMapping '\xab8d' s = Yield '\x13bd' (CC s '\x0000' '\x0000')

616−-- CHEROKEE SMALL LETTER NA

617−foldMapping '\xab8e' s = Yield '\x13be' (CC s '\x0000' '\x0000')

618−-- CHEROKEE SMALL LETTER HNA

619−foldMapping '\xab8f' s = Yield '\x13bf' (CC s '\x0000' '\x0000')

620−-- CHEROKEE SMALL LETTER NAH

621−foldMapping '\xab90' s = Yield '\x13c0' (CC s '\x0000' '\x0000')

622−-- CHEROKEE SMALL LETTER NE

623−foldMapping '\xab91' s = Yield '\x13c1' (CC s '\x0000' '\x0000')

624−-- CHEROKEE SMALL LETTER NI

625−foldMapping '\xab92' s = Yield '\x13c2' (CC s '\x0000' '\x0000')

626−-- CHEROKEE SMALL LETTER NO

627−foldMapping '\xab93' s = Yield '\x13c3' (CC s '\x0000' '\x0000')

628−-- CHEROKEE SMALL LETTER NU

629−foldMapping '\xab94' s = Yield '\x13c4' (CC s '\x0000' '\x0000')

630−-- CHEROKEE SMALL LETTER NV

631−foldMapping '\xab95' s = Yield '\x13c5' (CC s '\x0000' '\x0000')

632−-- CHEROKEE SMALL LETTER QUA

633−foldMapping '\xab96' s = Yield '\x13c6' (CC s '\x0000' '\x0000')

634−-- CHEROKEE SMALL LETTER QUE

635−foldMapping '\xab97' s = Yield '\x13c7' (CC s '\x0000' '\x0000')

636−-- CHEROKEE SMALL LETTER QUI

637−foldMapping '\xab98' s = Yield '\x13c8' (CC s '\x0000' '\x0000')

638−-- CHEROKEE SMALL LETTER QUO

639−foldMapping '\xab99' s = Yield '\x13c9' (CC s '\x0000' '\x0000')

640−-- CHEROKEE SMALL LETTER QUU

641−foldMapping '\xab9a' s = Yield '\x13ca' (CC s '\x0000' '\x0000')

642−-- CHEROKEE SMALL LETTER QUV

643−foldMapping '\xab9b' s = Yield '\x13cb' (CC s '\x0000' '\x0000')

644−-- CHEROKEE SMALL LETTER SA

645−foldMapping '\xab9c' s = Yield '\x13cc' (CC s '\x0000' '\x0000')

646−-- CHEROKEE SMALL LETTER S

647−foldMapping '\xab9d' s = Yield '\x13cd' (CC s '\x0000' '\x0000')

648−-- CHEROKEE SMALL LETTER SE

649−foldMapping '\xab9e' s = Yield '\x13ce' (CC s '\x0000' '\x0000')

650−-- CHEROKEE SMALL LETTER SI

651−foldMapping '\xab9f' s = Yield '\x13cf' (CC s '\x0000' '\x0000')

652−-- CHEROKEE SMALL LETTER SO

653−foldMapping '\xaba0' s = Yield '\x13d0' (CC s '\x0000' '\x0000')

654−-- CHEROKEE SMALL LETTER SU

655−foldMapping '\xaba1' s = Yield '\x13d1' (CC s '\x0000' '\x0000')

656−-- CHEROKEE SMALL LETTER SV

657−foldMapping '\xaba2' s = Yield '\x13d2' (CC s '\x0000' '\x0000')

658−-- CHEROKEE SMALL LETTER DA

659−foldMapping '\xaba3' s = Yield '\x13d3' (CC s '\x0000' '\x0000')

660−-- CHEROKEE SMALL LETTER TA

661−foldMapping '\xaba4' s = Yield '\x13d4' (CC s '\x0000' '\x0000')

662−-- CHEROKEE SMALL LETTER DE

663−foldMapping '\xaba5' s = Yield '\x13d5' (CC s '\x0000' '\x0000')

664−-- CHEROKEE SMALL LETTER TE

665−foldMapping '\xaba6' s = Yield '\x13d6' (CC s '\x0000' '\x0000')

666−-- CHEROKEE SMALL LETTER DI

667−foldMapping '\xaba7' s = Yield '\x13d7' (CC s '\x0000' '\x0000')

668−-- CHEROKEE SMALL LETTER TI

669−foldMapping '\xaba8' s = Yield '\x13d8' (CC s '\x0000' '\x0000')

670−-- CHEROKEE SMALL LETTER DO

671−foldMapping '\xaba9' s = Yield '\x13d9' (CC s '\x0000' '\x0000')

672−-- CHEROKEE SMALL LETTER DU

673−foldMapping '\xabaa' s = Yield '\x13da' (CC s '\x0000' '\x0000')

674−-- CHEROKEE SMALL LETTER DV

675−foldMapping '\xabab' s = Yield '\x13db' (CC s '\x0000' '\x0000')

676−-- CHEROKEE SMALL LETTER DLA

677−foldMapping '\xabac' s = Yield '\x13dc' (CC s '\x0000' '\x0000')

678−-- CHEROKEE SMALL LETTER TLA

679−foldMapping '\xabad' s = Yield '\x13dd' (CC s '\x0000' '\x0000')

680−-- CHEROKEE SMALL LETTER TLE

681−foldMapping '\xabae' s = Yield '\x13de' (CC s '\x0000' '\x0000')

682−-- CHEROKEE SMALL LETTER TLI

683−foldMapping '\xabaf' s = Yield '\x13df' (CC s '\x0000' '\x0000')

684−-- CHEROKEE SMALL LETTER TLO

685−foldMapping '\xabb0' s = Yield '\x13e0' (CC s '\x0000' '\x0000')

686−-- CHEROKEE SMALL LETTER TLU

687−foldMapping '\xabb1' s = Yield '\x13e1' (CC s '\x0000' '\x0000')

688−-- CHEROKEE SMALL LETTER TLV

689−foldMapping '\xabb2' s = Yield '\x13e2' (CC s '\x0000' '\x0000')

690−-- CHEROKEE SMALL LETTER TSA

691−foldMapping '\xabb3' s = Yield '\x13e3' (CC s '\x0000' '\x0000')

692−-- CHEROKEE SMALL LETTER TSE

693−foldMapping '\xabb4' s = Yield '\x13e4' (CC s '\x0000' '\x0000')

694−-- CHEROKEE SMALL LETTER TSI

695−foldMapping '\xabb5' s = Yield '\x13e5' (CC s '\x0000' '\x0000')

696−-- CHEROKEE SMALL LETTER TSO

697−foldMapping '\xabb6' s = Yield '\x13e6' (CC s '\x0000' '\x0000')

698−-- CHEROKEE SMALL LETTER TSU

699−foldMapping '\xabb7' s = Yield '\x13e7' (CC s '\x0000' '\x0000')

700−-- CHEROKEE SMALL LETTER TSV

701−foldMapping '\xabb8' s = Yield '\x13e8' (CC s '\x0000' '\x0000')

702−-- CHEROKEE SMALL LETTER WA

703−foldMapping '\xabb9' s = Yield '\x13e9' (CC s '\x0000' '\x0000')

704−-- CHEROKEE SMALL LETTER WE

705−foldMapping '\xabba' s = Yield '\x13ea' (CC s '\x0000' '\x0000')

706−-- CHEROKEE SMALL LETTER WI

707−foldMapping '\xabbb' s = Yield '\x13eb' (CC s '\x0000' '\x0000')

708−-- CHEROKEE SMALL LETTER WO

709−foldMapping '\xabbc' s = Yield '\x13ec' (CC s '\x0000' '\x0000')

710−-- CHEROKEE SMALL LETTER WU

711−foldMapping '\xabbd' s = Yield '\x13ed' (CC s '\x0000' '\x0000')

712−-- CHEROKEE SMALL LETTER WV

713−foldMapping '\xabbe' s = Yield '\x13ee' (CC s '\x0000' '\x0000')

714−-- CHEROKEE SMALL LETTER YA

715−foldMapping '\xabbf' s = Yield '\x13ef' (CC s '\x0000' '\x0000')

540+-- COPTIC CAPITAL LETTER BOHAIRIC KHEI

541+foldMapping '\x2cf2' s = Yield '\x2cf3' (CC s '\x0000' '\x0000')

542+-- LATIN CAPITAL LETTER C WITH BAR

543+foldMapping '\xa792' s = Yield '\xa793' (CC s '\x0000' '\x0000')

544+-- LATIN CAPITAL LETTER H WITH HOOK

545+foldMapping '\xa7aa' s = Yield '\x0266' (CC s '\x0000' '\x0000')

716546-- LATIN SMALL LIGATURE FF

717547foldMapping '\xfb00' s = Yield '\x0066' (CC s '\x0066' '\x0000')

718548-- LATIN SMALL LIGATURE FI

@@ -737,106 +567,4 @@ foldMapping '\xfb15' s = Yield '\x0574' (CC s '\x056b' '\x0000')

737567foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

738568-- ARMENIAN SMALL LIGATURE MEN XEH

739569foldMapping '\xfb17' s = Yield '\x0574' (CC s '\x056d' '\x0000')

740−-- OLD HUNGARIAN CAPITAL LETTER A

741−foldMapping '\x10c80' s = Yield '\x10cc0' (CC s '\x0000' '\x0000')

742−-- OLD HUNGARIAN CAPITAL LETTER AA

743−foldMapping '\x10c81' s = Yield '\x10cc1' (CC s '\x0000' '\x0000')

744−-- OLD HUNGARIAN CAPITAL LETTER EB

745−foldMapping '\x10c82' s = Yield '\x10cc2' (CC s '\x0000' '\x0000')

746−-- OLD HUNGARIAN CAPITAL LETTER AMB

747−foldMapping '\x10c83' s = Yield '\x10cc3' (CC s '\x0000' '\x0000')

748−-- OLD HUNGARIAN CAPITAL LETTER EC

749−foldMapping '\x10c84' s = Yield '\x10cc4' (CC s '\x0000' '\x0000')

750−-- OLD HUNGARIAN CAPITAL LETTER ENC

751−foldMapping '\x10c85' s = Yield '\x10cc5' (CC s '\x0000' '\x0000')

752−-- OLD HUNGARIAN CAPITAL LETTER ECS

753−foldMapping '\x10c86' s = Yield '\x10cc6' (CC s '\x0000' '\x0000')

754−-- OLD HUNGARIAN CAPITAL LETTER ED

755−foldMapping '\x10c87' s = Yield '\x10cc7' (CC s '\x0000' '\x0000')

756−-- OLD HUNGARIAN CAPITAL LETTER AND

757−foldMapping '\x10c88' s = Yield '\x10cc8' (CC s '\x0000' '\x0000')

758−-- OLD HUNGARIAN CAPITAL LETTER E

759−foldMapping '\x10c89' s = Yield '\x10cc9' (CC s '\x0000' '\x0000')

760−-- OLD HUNGARIAN CAPITAL LETTER CLOSE E

761−foldMapping '\x10c8a' s = Yield '\x10cca' (CC s '\x0000' '\x0000')

762−-- OLD HUNGARIAN CAPITAL LETTER EE

763−foldMapping '\x10c8b' s = Yield '\x10ccb' (CC s '\x0000' '\x0000')

764−-- OLD HUNGARIAN CAPITAL LETTER EF

765−foldMapping '\x10c8c' s = Yield '\x10ccc' (CC s '\x0000' '\x0000')

766−-- OLD HUNGARIAN CAPITAL LETTER EG

767−foldMapping '\x10c8d' s = Yield '\x10ccd' (CC s '\x0000' '\x0000')

768−-- OLD HUNGARIAN CAPITAL LETTER EGY

769−foldMapping '\x10c8e' s = Yield '\x10cce' (CC s '\x0000' '\x0000')

770−-- OLD HUNGARIAN CAPITAL LETTER EH

771−foldMapping '\x10c8f' s = Yield '\x10ccf' (CC s '\x0000' '\x0000')

772−-- OLD HUNGARIAN CAPITAL LETTER I

773−foldMapping '\x10c90' s = Yield '\x10cd0' (CC s '\x0000' '\x0000')

774−-- OLD HUNGARIAN CAPITAL LETTER II

775−foldMapping '\x10c91' s = Yield '\x10cd1' (CC s '\x0000' '\x0000')

776−-- OLD HUNGARIAN CAPITAL LETTER EJ

777−foldMapping '\x10c92' s = Yield '\x10cd2' (CC s '\x0000' '\x0000')

778−-- OLD HUNGARIAN CAPITAL LETTER EK

779−foldMapping '\x10c93' s = Yield '\x10cd3' (CC s '\x0000' '\x0000')

780−-- OLD HUNGARIAN CAPITAL LETTER AK

781−foldMapping '\x10c94' s = Yield '\x10cd4' (CC s '\x0000' '\x0000')

782−-- OLD HUNGARIAN CAPITAL LETTER UNK

783−foldMapping '\x10c95' s = Yield '\x10cd5' (CC s '\x0000' '\x0000')

784−-- OLD HUNGARIAN CAPITAL LETTER EL

785−foldMapping '\x10c96' s = Yield '\x10cd6' (CC s '\x0000' '\x0000')

786−-- OLD HUNGARIAN CAPITAL LETTER ELY

787−foldMapping '\x10c97' s = Yield '\x10cd7' (CC s '\x0000' '\x0000')

788−-- OLD HUNGARIAN CAPITAL LETTER EM

789−foldMapping '\x10c98' s = Yield '\x10cd8' (CC s '\x0000' '\x0000')

790−-- OLD HUNGARIAN CAPITAL LETTER EN

791−foldMapping '\x10c99' s = Yield '\x10cd9' (CC s '\x0000' '\x0000')

792−-- OLD HUNGARIAN CAPITAL LETTER ENY

793−foldMapping '\x10c9a' s = Yield '\x10cda' (CC s '\x0000' '\x0000')

794−-- OLD HUNGARIAN CAPITAL LETTER O

795−foldMapping '\x10c9b' s = Yield '\x10cdb' (CC s '\x0000' '\x0000')

796−-- OLD HUNGARIAN CAPITAL LETTER OO

797−foldMapping '\x10c9c' s = Yield '\x10cdc' (CC s '\x0000' '\x0000')

798−-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG OE

799−foldMapping '\x10c9d' s = Yield '\x10cdd' (CC s '\x0000' '\x0000')

800−-- OLD HUNGARIAN CAPITAL LETTER RUDIMENTA OE

801−foldMapping '\x10c9e' s = Yield '\x10cde' (CC s '\x0000' '\x0000')

802−-- OLD HUNGARIAN CAPITAL LETTER OEE

803−foldMapping '\x10c9f' s = Yield '\x10cdf' (CC s '\x0000' '\x0000')

804−-- OLD HUNGARIAN CAPITAL LETTER EP

805−foldMapping '\x10ca0' s = Yield '\x10ce0' (CC s '\x0000' '\x0000')

806−-- OLD HUNGARIAN CAPITAL LETTER EMP

807−foldMapping '\x10ca1' s = Yield '\x10ce1' (CC s '\x0000' '\x0000')

808−-- OLD HUNGARIAN CAPITAL LETTER ER

809−foldMapping '\x10ca2' s = Yield '\x10ce2' (CC s '\x0000' '\x0000')

810−-- OLD HUNGARIAN CAPITAL LETTER SHORT ER

811−foldMapping '\x10ca3' s = Yield '\x10ce3' (CC s '\x0000' '\x0000')

812−-- OLD HUNGARIAN CAPITAL LETTER ES

813−foldMapping '\x10ca4' s = Yield '\x10ce4' (CC s '\x0000' '\x0000')

814−-- OLD HUNGARIAN CAPITAL LETTER ESZ

815−foldMapping '\x10ca5' s = Yield '\x10ce5' (CC s '\x0000' '\x0000')

816−-- OLD HUNGARIAN CAPITAL LETTER ET

817−foldMapping '\x10ca6' s = Yield '\x10ce6' (CC s '\x0000' '\x0000')

818−-- OLD HUNGARIAN CAPITAL LETTER ENT

819−foldMapping '\x10ca7' s = Yield '\x10ce7' (CC s '\x0000' '\x0000')

820−-- OLD HUNGARIAN CAPITAL LETTER ETY

821−foldMapping '\x10ca8' s = Yield '\x10ce8' (CC s '\x0000' '\x0000')

822−-- OLD HUNGARIAN CAPITAL LETTER ECH

823−foldMapping '\x10ca9' s = Yield '\x10ce9' (CC s '\x0000' '\x0000')

824−-- OLD HUNGARIAN CAPITAL LETTER U

825−foldMapping '\x10caa' s = Yield '\x10cea' (CC s '\x0000' '\x0000')

826−-- OLD HUNGARIAN CAPITAL LETTER UU

827−foldMapping '\x10cab' s = Yield '\x10ceb' (CC s '\x0000' '\x0000')

828−-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG UE

829−foldMapping '\x10cac' s = Yield '\x10cec' (CC s '\x0000' '\x0000')

830−-- OLD HUNGARIAN CAPITAL LETTER RUDIMENTA UE

831−foldMapping '\x10cad' s = Yield '\x10ced' (CC s '\x0000' '\x0000')

832−-- OLD HUNGARIAN CAPITAL LETTER EV

833−foldMapping '\x10cae' s = Yield '\x10cee' (CC s '\x0000' '\x0000')

834−-- OLD HUNGARIAN CAPITAL LETTER EZ

835−foldMapping '\x10caf' s = Yield '\x10cef' (CC s '\x0000' '\x0000')

836−-- OLD HUNGARIAN CAPITAL LETTER EZS

837−foldMapping '\x10cb0' s = Yield '\x10cf0' (CC s '\x0000' '\x0000')

838−-- OLD HUNGARIAN CAPITAL LETTER ENT-SHAPED SIGN

839−foldMapping '\x10cb1' s = Yield '\x10cf1' (CC s '\x0000' '\x0000')

840−-- OLD HUNGARIAN CAPITAL LETTER US

841−foldMapping '\x10cb2' s = Yield '\x10cf2' (CC s '\x0000' '\x0000')

842570foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

Data/Text/Internal/Fusion/Common.hs+72−59

@@ -71,6 +71,9 @@ module Data.Text.Internal.Fusion.Common

7171 -- ** Scans

7272 , scanl

7373

74+ -- ** Accumulating maps

75+ -- , mapAccumL

76+

7477 -- ** Generation and unfolding

7578 , replicateCharI

7679 , replicateI

@@ -120,7 +123,7 @@ singleton :: Char -> Stream Char

120123singleton c = Stream next False 1

121124 where next False = Yield c True

122125 next True = Done

123−{-# INLINE [0] singleton #-}

126+{-# INLINE singleton #-}

124127

125128streamList :: [a] -> Stream a

126129{-# INLINE [0] streamList #-}

@@ -184,9 +187,6 @@ cons !w (Stream next0 s0 len) = Stream next (C1 s0) (len+1)

184187 Yield x s' -> Yield x (C0 s')

185188{-# INLINE [0] cons #-}

186189

187−data Snoc a = N

188− | J !a

189−

190190-- | /O(n)/ Adds a character to the end of a stream.

191191snoc :: Stream Char -> Char -> Stream Char

192192snoc (Stream next0 xs0 len) w = Stream next (J xs0) (len+1)

@@ -473,20 +473,17 @@ toTitle (Stream next0 s0 len) = Stream next (CC (False :*: s0) '\0' '\0') len

473473 next (CC s a b) = Yield a (CC s b '\0')

474474{-# INLINE [0] toTitle #-}

475475

476−data Justify i s = Just1 !i !s

477− | Just2 !i !s

478−

479476justifyLeftI :: Integral a => a -> Char -> Stream Char -> Stream Char

480477justifyLeftI k c (Stream next0 s0 len) =

481− Stream next (Just1 0 s0) (larger (fromIntegral k) len)

478+ Stream next (s0 :*: S1 :*: 0) (larger (fromIntegral k) len)

482479 where

483− next (Just1 n s) =

480+ next (s :*: S1 :*: n) =

484481 case next0 s of

485− Done -> next (Just2 n s)

486− Skip s' -> Skip (Just1 n s')

487− Yield x s' -> Yield x (Just1 (n+1) s')

488− next (Just2 n s)

489− | n < k = Yield c (Just2 (n+1) s)

482+ Done -> next (s :*: S2 :*: n)

483+ Skip s' -> Skip (s' :*: S1 :*: n)

484+ Yield x s' -> Yield x (s' :*: S1 :*: n+1)

485+ next (s :*: S2 :*: n)

486+ | n < k = Yield c (s :*: S2 :*: n+1)

490487 | otherwise = Done

491488 {-# INLINE next #-}

492489{-# INLINE [0] justifyLeftI #-}

@@ -593,7 +590,7 @@ concatMap f = foldr (append . f) empty

593590{-# INLINE [0] concatMap #-}

594591

595592-- | /O(n)/ any @p @xs determines if any character in the stream

596−-- @xs@ satisfies the predicate @p@.

593+-- @xs@ satisifes the predicate @p@.

597594any :: (Char -> Bool) -> Stream Char -> Bool

598595any p (Stream next0 s0 _len) = loop_any s0

599596 where

@@ -605,7 +602,7 @@ any p (Stream next0 s0 _len) = loop_any s0

605602{-# INLINE [0] any #-}

606603

607604-- | /O(n)/ all @p @xs determines if all characters in the 'Text'

608−-- @xs@ satisfy the predicate @p@.

605+-- @xs@ satisify the predicate @p@.

609606all :: (Char -> Bool) -> Stream Char -> Bool

610607all p (Stream next0 s0 _len) = loop_all s0

611608 where

@@ -654,27 +651,50 @@ minimum (Stream next0 s0 _len) = loop0_minimum s0

654651-- * Building streams

655652

656653scanl :: (Char -> Char -> Char) -> Char -> Stream Char -> Stream Char

657−scanl f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT maybe too low

654+scanl f z0 (Stream next0 s0 len) = Stream next (S1 :*: z0 :*: s0) (len+1) -- HINT maybe too low

658655 where

659656 {-# INLINE next #-}

660− next (Scan1 z s) = Yield z (Scan2 z s)

661− next (Scan2 z s) = case next0 s of

662− Yield x s' -> let !x' = f z x

663− in Yield x' (Scan2 x' s')

664− Skip s' -> Skip (Scan2 z s')

665− Done -> Done

657+ next (S1 :*: z :*: s) = Yield z (S2 :*: z :*: s)

658+ next (S2 :*: z :*: s) = case next0 s of

659+ Yield x s' -> let !x' = f z x

660+ in Yield x' (S2 :*: x' :*: s')

661+ Skip s' -> Skip (S2 :*: z :*: s')

662+ Done -> Done

666663{-# INLINE [0] scanl #-}

667664

665+-- -----------------------------------------------------------------------------

666+-- ** Accumulating maps

667+

668+{-

669+-- | /O(n)/ Like a combination of 'map' and 'foldl'. Applies a

670+-- function to each element of a stream, passing an accumulating

671+-- parameter from left to right, and returns a final stream.

672+--

673+-- /Note/: Unlike the version over lists, this function does not

674+-- return a final value for the accumulator, because the nature of

675+-- streams precludes it.

676+mapAccumL :: (a -> b -> (a,b)) -> a -> Stream b -> Stream b

677+mapAccumL f z0 (Stream next0 s0 len) = Stream next (s0 :*: z0) len -- HINT depends on f

678+ where

679+ {-# INLINE next #-}

680+ next (s :*: z) = case next0 s of

681+ Yield x s' -> let (z',y) = f z x

682+ in Yield y (s' :*: z')

683+ Skip s' -> Skip (s' :*: z)

684+ Done -> Done

685+{-# INLINE [0] mapAccumL #-}

686+-}

687+

668688-- -----------------------------------------------------------------------------

669689-- ** Generating and unfolding streams

670690

671691replicateCharI :: Integral a => a -> Char -> Stream Char

672−replicateCharI !n !c

692+replicateCharI n c

673693 | n < 0 = empty

674694 | otherwise = Stream next 0 (fromIntegral n) -- HINT maybe too low

675695 where

676− next !i | i >= n = Done

677− | otherwise = Yield c (i + 1)

696+ next i | i >= n = Done

697+ | otherwise = Yield c (i + 1)

678698{-# INLINE [0] replicateCharI #-}

679699

680700data RI s = RI !s {-# UNPACK #-} !Int64

@@ -739,27 +759,24 @@ take n0 (Stream next0 s0 len) =

739759 Yield x s' -> Yield x ((n-1) :*: s')

740760{-# INLINE [0] take #-}

741761

742−data Drop a s = NS !s

743− | JS !a !s

744−

745762-- | /O(n)/ drop n, applied to a stream, returns the suffix of the

746763-- stream after the first @n@ characters, or the empty stream if @n@

747764-- is greater than the length of the stream.

748765drop :: Integral a => a -> Stream Char -> Stream Char

749766drop n0 (Stream next0 s0 len) =

750− Stream next (JS n0 s0) (len - fromIntegral (max 0 n0))

767+ Stream next (J n0 :*: s0) (len - fromIntegral (max 0 n0))

751768 where

752769 {-# INLINE next #-}

753− next (JS n s)

754− | n <= 0 = Skip (NS s)

770+ next (J n :*: s)

771+ | n <= 0 = Skip (N :*: s)

755772 | otherwise = case next0 s of

756773 Done -> Done

757− Skip s' -> Skip (JS n s')

758− Yield _ s' -> Skip (JS (n-1) s')

759− next (NS s) = case next0 s of

774+ Skip s' -> Skip (J n :*: s')

775+ Yield _ s' -> Skip (J (n-1) :*: s')

776+ next (N :*: s) = case next0 s of

760777 Done -> Done

761− Skip s' -> Skip (NS s')

762− Yield x s' -> Yield x (NS s')

778+ Skip s' -> Skip (N :*: s')

779+ Yield x s' -> Yield x (N :*: s')

763780{-# INLINE [0] drop #-}

764781

765782-- | takeWhile, applied to a predicate @p@ and a stream, returns the

@@ -777,18 +794,18 @@ takeWhile p (Stream next0 s0 len) = Stream next s0 len -- HINT maybe too high

777794

778795-- | dropWhile @p @xs returns the suffix remaining after takeWhile @p @xs.

779796dropWhile :: (Char -> Bool) -> Stream Char -> Stream Char

780−dropWhile p (Stream next0 s0 len) = Stream next (L s0) len -- HINT maybe too high

797+dropWhile p (Stream next0 s0 len) = Stream next (S1 :*: s0) len -- HINT maybe too high

781798 where

782799 {-# INLINE next #-}

783− next (L s) = case next0 s of

800+ next (S1 :*: s) = case next0 s of

784801 Done -> Done

785− Skip s' -> Skip (L s')

786− Yield x s' | p x -> Skip (L s')

787− | otherwise -> Yield x (R s')

788− next (R s) = case next0 s of

802+ Skip s' -> Skip (S1 :*: s')

803+ Yield x s' | p x -> Skip (S1 :*: s')

804+ | otherwise -> Yield x (S2 :*: s')

805+ next (S2 :*: s) = case next0 s of

789806 Done -> Done

790− Skip s' -> Skip (R s')

791− Yield x s' -> Yield x (R s')

807+ Skip s' -> Skip (S2 :*: s')

808+ Yield x s' -> Yield x (S2 :*: s')

792809{-# INLINE [0] dropWhile #-}

793810

794811-- | /O(n)/ The 'isPrefixOf' function takes two 'Stream's and returns

@@ -895,25 +912,21 @@ findIndicesI p (Stream next s0 _len) = loop_findIndex 0 s0

895912-------------------------------------------------------------------------------

896913-- * Zipping

897914

898−-- | Strict triple.

899−data Zip a b m = Z1 !a !b

900− | Z2 !a !b !m

901−

902915-- | zipWith generalises 'zip' by zipping with the function given as

903916-- the first argument, instead of a tupling function.

904917zipWith :: (a -> a -> b) -> Stream a -> Stream a -> Stream b

905918zipWith f (Stream next0 sa0 len1) (Stream next1 sb0 len2) =

906− Stream next (Z1 sa0 sb0) (smaller len1 len2)

919+ Stream next (sa0 :*: sb0 :*: N) (smaller len1 len2)

907920 where

908− next (Z1 sa sb) = case next0 sa of

909− Done -> Done

910− Skip sa' -> Skip (Z1 sa' sb)

911− Yield a sa' -> Skip (Z2 sa' sb a)

912−

913− next (Z2 sa' sb a) = case next1 sb of

914− Done -> Done

915− Skip sb' -> Skip (Z2 sa' sb' a)

916− Yield b sb' -> Yield (f a b) (Z1 sa' sb')

921+ next (sa :*: sb :*: N) = case next0 sa of

922+ Done -> Done

923+ Skip sa' -> Skip (sa' :*: sb :*: N)

924+ Yield a sa' -> Skip (sa' :*: sb :*: J a)

925+

926+ next (sa' :*: sb :*: J a) = case next1 sb of

927+ Done -> Done

928+ Skip sb' -> Skip (sa' :*: sb' :*: J a)

929+ Yield b sb' -> Yield (f a b) (sa' :*: sb' :*: N)

917930{-# INLINE [0] zipWith #-}

918931

919932-- | /O(n)/ The 'countCharI' function returns the number of times the

Data/Text/Internal/Fusion/Types.hs+13−8

@@ -20,11 +20,13 @@

2020module Data.Text.Internal.Fusion.Types

2121 (

2222 CC(..)

23+ , M(..)

24+ , M8

2325 , PairS(..)

24− , Scan(..)

2526 , RS(..)

2627 , Step(..)

2728 , Stream(..)

29+ , Switch(..)

2830 , empty

2931 ) where

3032

@@ -34,23 +36,26 @@ import Data.Word (Word8)

3436-- | Specialised tuple for case conversion.

3537data CC s = CC !s {-# UNPACK #-} !Char {-# UNPACK #-} !Char

3638

37−-- | Restreaming state.

39+-- | Specialised, strict Maybe-like type.

40+data M a = N

41+ | J !a

42+

43+type M8 = M Word8

44+

45+-- Restreaming state.

3846data RS s

3947 = RS0 !s

4048 | RS1 !s {-# UNPACK #-} !Word8

4149 | RS2 !s {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8

4250 | RS3 !s {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8 {-# UNPACK #-} !Word8

4351

44−-- | Strict pair.

52+infixl 2 :*:

4553data PairS a b = !a :*: !b

4654 -- deriving (Eq, Ord, Show)

47−infixl 2 :*:

4855

49−-- | An intermediate result in a scan.

50−data Scan s = Scan1 {-# UNPACK #-} !Char !s

51− | Scan2 {-# UNPACK #-} !Char !s

56+-- | Allow a function over a stream to switch between two states.

57+data Switch = S1 | S2

5258

53−-- | Intermediate result in a processing pipeline.

5459data Step s a = Done

5560 | Skip !s

5661 | Yield !a !s

Data/Text/Internal/Read.hs+3−3

@@ -7,7 +7,7 @@

77-- Stability : experimental

88-- Portability : GHC

99--

10−-- Common internal functions for reading textual data.

10+-- Common internal functiopns for reading textual data.

1111module Data.Text.Internal.Read

1212 (

1313 IReader

@@ -18,7 +18,7 @@ module Data.Text.Internal.Read

1818 , perhaps

1919 ) where

2020

21−import Control.Applicative as App (Applicative(..))

21+import Control.Applicative (Applicative(..))

2222import Control.Arrow (first)

2323import Control.Monad (ap)

2424import Data.Char (ord)

@@ -38,7 +38,7 @@ instance Applicative (IParser t) where

3838 (<*>) = ap

3939

4040instance Monad (IParser t) where

41− return = App.pure

41+ return = pure

4242 m >>= k = P $ \t -> case runP m t of

4343 Left err -> Left err

4444 Right (a,t') -> runP (k a) t'

Data/Text/Lazy.hs+21−161

@@ -3,10 +3,6 @@

33#if __GLASGOW_HASKELL__ >= 702

44{-# LANGUAGE Trustworthy #-}

55#endif

6−#if __GLASGOW_HASKELL__ >= 708

7−{-# LANGUAGE TypeFamilies #-}

8−#endif

9−

106-- |

117-- Module : Data.Text.Lazy

@@ -125,10 +121,7 @@ module Data.Text.Lazy

125121 , mapAccumR

126122

127123 -- ** Generation and unfolding

128− , repeat

129124 , replicate

130− , cycle

131− , iterate

132125 , unfoldr

133126 , unfoldrN

134127

@@ -140,7 +133,6 @@ module Data.Text.Lazy

140133 , drop

141134 , dropEnd

142135 , takeWhile

143− , takeWhileEnd

144136 , dropWhile

145137 , dropWhileEnd

146138 , dropAround

@@ -211,13 +203,9 @@ import Control.DeepSeq (NFData(..))

211203import Data.Int (Int64)

212204import qualified Data.List as L

213205import Data.Char (isSpace)

214−import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,

215− Constr, mkConstr, DataType, mkDataType, Fixity(Prefix))

216−import Data.Binary (Binary(get, put))

206+import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf))

207+import Data.Data (mkNoRepType)

217208import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

219−import Data.Semigroup (Semigroup(..))

220−#endif

221209import Data.String (IsString(..))

222210import qualified Data.Text as T

223211import qualified Data.Text.Internal as T

@@ -226,10 +214,8 @@ import qualified Data.Text.Unsafe as T

226214import qualified Data.Text.Internal.Lazy.Fusion as S

227215import Data.Text.Internal.Fusion.Types (PairS(..))

228216import Data.Text.Internal.Lazy.Fusion (stream, unstream)

229−import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldlChunks,

230− foldrChunks, smallChunkSize)

217+import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldlChunks, foldrChunks)

231218import Data.Text.Internal (firstf, safe, text)

232−import Data.Text.Lazy.Encoding (decodeUtf8', encodeUtf8)

233219import qualified Data.Text.Internal.Functions as F

234220import Data.Text.Internal.Lazy.Search (indices)

235221#if __GLASGOW_HASKELL__ >= 702

@@ -237,13 +223,7 @@ import qualified GHC.CString as GHC

237223#else

238224import qualified GHC.Base as GHC

239225#endif

240−#if __GLASGOW_HASKELL__ >= 708

241−import qualified GHC.Exts as Exts

242−#endif

243226import GHC.Prim (Addr#)

244−#if MIN_VERSION_base(4,7,0)

245−import Text.Printf (PrintfArg, formatArg, formatString)

246−#endif

247227

248228-- $fusion

249229--

@@ -341,71 +321,25 @@ instance Show Text where

341321instance Read Text where

342322 readsPrec p str = [(pack x,y) | (x,y) <- readsPrec p str]

343323

344−#if MIN_VERSION_base(4,9,0)

345−-- Semigroup orphan instances for older GHCs are provided by

346−-- 'semigroups` package

347−

348−instance Semigroup Text where

349− (<>) = append

350−#endif

351−

352324instance Monoid Text where

353325 mempty = empty

354−#if MIN_VERSION_base(4,9,0)

355− mappend = (<>) -- future-proof definition

356−#else

357326 mappend = append

358−#endif

359327 mconcat = concat

360328

361329instance IsString Text where

362330 fromString = pack

363331

364−#if __GLASGOW_HASKELL__ >= 708

365−instance Exts.IsList Text where

366− type Item Text = Char

367− fromList = pack

368− toList = unpack

369−#endif

370−

371332#if defined(HAVE_DEEPSEQ)

372333instance NFData Text where

373334 rnf Empty = ()

374335 rnf (Chunk _ ts) = rnf ts

375336#endif

376337

377−instance Binary Text where

378− put t = put (encodeUtf8 t)

379− get = do

380− bs <- get

381− case decodeUtf8' bs of

382− P.Left exn -> P.fail (P.show exn)

383− P.Right a -> P.return a

384−

385−-- | This instance preserves data abstraction at the cost of inefficiency.

386−-- We omit reflection services for the sake of data abstraction.

387−--

388−-- This instance was created by copying the updated behavior of

389−-- @"Data.Text".@'Data.Text.Text'

390338instance Data Text where

391339 gfoldl f z txt = z pack `f` (unpack txt)

392− toConstr _ = packConstr

393− gunfold k z c = case constrIndex c of

394− 1 -> k (z pack)

395− _ -> error "Data.Text.Lazy.Text.gunfold"

396− dataTypeOf _ = textDataType

397−

398−#if MIN_VERSION_base(4,7,0)

399−-- | Only defined for @base-4.7.0.0@ and later

400−instance PrintfArg Text where

401− formatArg txt = formatString $ unpack txt

402−#endif

403−

404−packConstr :: Constr

405−packConstr = mkConstr textDataType "pack" [] Prefix

406−

407−textDataType :: DataType

408−textDataType = mkDataType "Data.Text.Lazy.Text" [packConstr]

340+ toConstr _ = error "Data.Text.Lazy.Text.toConstr"

341+ gunfold _ _ = error "Data.Text.Lazy.Text.gunfold"

342+ dataTypeOf _ = mkNoRepType "Data.Text.Lazy.Text"

409343

410344-- | /O(n)/ Convert a 'String' into a 'Text'.

411345--

@@ -433,14 +367,6 @@ unpackCString# addr# = unstream (S.streamCString# addr#)

433367 unstream (S.streamList (L.map safe (GHC.unpackCStringUtf8# a)))

434368 = unpackCString# a #-}

435369

436−{-# RULES "LAZY TEXT empty literal"

437− unstream (S.streamList (L.map safe []))

438− = Empty #-}

439−

440−{-# RULES "LAZY TEXT empty literal" forall a.

441− unstream (S.streamList (L.map safe [a]))

442− = Chunk (T.singleton a) Empty #-}

443−

444370-- | /O(1)/ Convert a character into a Text. Subject to fusion.

445371-- Performs replacement on invalid scalar values.

446372singleton :: Char -> Text

@@ -545,7 +471,7 @@ tail Empty = emptyError "tail"

545471 unstream (S.tail (stream t)) = tail t

546472 #-}

547473

548−-- | /O(n\/c)/ Returns all but the last character of a 'Text', which must

474+-- | /O(1)/ Returns all but the last character of a 'Text', which must

549475-- be non-empty. Subject to fusion.

550476init :: Text -> Text

551477init (Chunk t0 ts0) = go t0 ts0

@@ -581,7 +507,7 @@ isSingleton :: Text -> Bool

581507isSingleton = S.isSingleton . stream

582508{-# INLINE isSingleton #-}

583509

584−-- | /O(n\/c)/ Returns the last character of a 'Text', which must be

510+-- | /O(1)/ Returns the last character of a 'Text', which must be

585511-- non-empty. Subject to fusion.

586512last :: Text -> Char

587513last Empty = emptyError "last"

@@ -715,36 +641,13 @@ reverse = rev Empty

715641 where rev a Empty = a

716642 rev a (Chunk t ts) = rev (Chunk (T.reverse t) a) ts

717643

718−-- | /O(m+n)/ Replace every non-overlapping occurrence of @needle@ in

719−-- @haystack@ with @replacement@.

720−--

721−-- This function behaves as though it was defined as follows:

722−--

723−-- @

724−-- replace needle replacement haystack =

725−-- 'intercalate' replacement ('splitOn' needle haystack)

726−-- @

727−--

728−-- As this suggests, each occurrence is replaced exactly once. So if

729−-- @needle@ occurs in @replacement@, that occurrence will /not/ itself

730−-- be replaced recursively:

731−--

732−-- > replace "oo" "foo" "oo" == "foo"

733−--

734−-- In cases where several instances of @needle@ overlap, only the

735−-- first one will be replaced:

736−--

737−-- > replace "ofo" "bar" "ofofo" == "barfo"

644+-- | /O(m+n)/ Replace every occurrence of one substring with another.

738645--

739646-- In (unlikely) bad cases, this function's time complexity degrades

740647-- towards /O(n*m)/.

741−replace :: Text

742− -- ^ @needle@ to search for. If this string is empty, an

743− -- error will occur.

744− -> Text

745− -- ^ @replacement@ to replace @needle@ with.

746− -> Text

747− -- ^ @haystack@ in which to search.

648+replace :: Text -- ^ Text to search for

649+ -> Text -- ^ Replacement text

650+ -> Text -- ^ Input text

748651 -> Text

749652replace s d = intercalate d . splitOn s

750653{-# INLINE replace #-}

@@ -881,13 +784,13 @@ concatMap f = concat . foldr ((:) . f) []

881784{-# INLINE concatMap #-}

882785

883786-- | /O(n)/ 'any' @p@ @t@ determines whether any character in the

884−-- 'Text' @t@ satisfies the predicate @p@. Subject to fusion.

787+-- 'Text' @t@ satisifes the predicate @p@. Subject to fusion.

885788any :: (Char -> Bool) -> Text -> Bool

886789any p t = S.any p (stream t)

887790{-# INLINE any #-}

888791

889792-- | /O(n)/ 'all' @p@ @t@ determines whether all characters in the

890−-- 'Text' @t@ satisfy the predicate @p@. Subject to fusion.

793+-- 'Text' @t@ satisify the predicate @p@. Subject to fusion.

891794all :: (Char -> Bool) -> Text -> Bool

892795all p t = S.all p (stream t)

893796{-# INLINE all #-}

@@ -970,12 +873,6 @@ mapAccumR f = go

970873 go z Empty = (z, Empty)

971874{-# INLINE mapAccumR #-}

972875

973−-- | @'repeat' x@ is an infinite 'Text', with @x@ the value of every

974−-- element.

975−repeat :: Char -> Text

976−repeat c = let t = Chunk (T.replicate smallChunkSize (T.singleton c)) t

977− in t

978−

979876-- | /O(n*m)/ 'replicate' @n@ @t@ is a 'Text' consisting of the input

980877-- @t@ repeated @n@ times.

981878replicate :: Int64 -> Text -> Text

@@ -987,21 +884,6 @@ replicate n t

987884 | otherwise = t : rep (i+1)

988885{-# INLINE [1] replicate #-}

989886

990−-- | 'cycle' ties a finite, non-empty 'Text' into a circular one, or

991−-- equivalently, the infinite repetition of the original 'Text'.

992−cycle :: Text -> Text

993−cycle Empty = emptyError "cycle"

994−cycle t = let t' = foldrChunks Chunk t' t

995− in t'

996−

997−-- | @'iterate' f x@ returns an infinite 'Text' of repeated applications

998−-- of @f@ to @x@:

999−--

1000−-- > iterate f x == [x, f x, f (f x), ...]

1001−iterate :: (Char -> Char) -> Char -> Text

1002−iterate f c = let t c' = Chunk (T.singleton c') (t (f c'))

1003− in t c

1004−

1005887-- | /O(n)/ 'replicateChar' @n@ @c@ is a 'Text' of length @n@ with @c@ the

1006888-- value of every element. Subject to fusion.

1007889replicateChar :: Int64 -> Char -> Text

@@ -1144,20 +1026,6 @@ takeWhile p t0 = takeWhile' t0

11441026"LAZY TEXT takeWhile -> unfused" [1] forall p t.

11451027 unstream (S.takeWhile p (stream t)) = takeWhile p t

11461028 #-}

1147−-- | /O(n)/ 'takeWhileEnd', applied to a predicate @p@ and a 'Text',

1148−-- returns the longest suffix (possibly empty) of elements that

1149−-- satisfy @p@.

1150−-- Examples:

1151−--

1152−-- > takeWhileEnd (=='o') "foo" == "oo"

1153−takeWhileEnd :: (Char -> Bool) -> Text -> Text

1154−takeWhileEnd p = takeChunk empty . L.reverse . toChunks

1155− where takeChunk acc [] = acc

1156− takeChunk acc (t:ts) = if T.length t' < T.length t

1157− then (Chunk t' acc)

1158− else takeChunk (Chunk t' acc) ts

1159− where t' = T.takeWhileEnd p t

1160−{-# INLINE takeWhileEnd #-}

11611029

11621030-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

11631031-- 'takeWhile' @p@ @t@. Subject to fusion.

@@ -1176,11 +1044,9 @@ dropWhile p t0 = dropWhile' t0

11761044"LAZY TEXT dropWhile -> unfused" [1] forall p t.

11771045 unstream (S.dropWhile p (stream t)) = dropWhile p t

11781046 #-}

1179−

11801047-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1181−-- dropping characters that satisfy the predicate @p@ from the end of

1048+-- dropping characters that fail the predicate @p@ from the end of

11821049-- @t@.

1183−--

11841050-- Examples:

11851051--

11861052-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1197,7 +1063,7 @@ dropWhileEnd p = go

11971063{-# INLINE dropWhileEnd #-}

11981064

11991065-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1200−-- dropping characters that satisfy the predicate @p@ from both the

1066+-- dropping characters that fail the predicate @p@ from both the

12011067-- beginning and end of @t@. Subject to fusion.

12021068dropAround :: (Char -> Bool) -> Text -> Text

12031069dropAround p = dropWhile p . dropWhileEnd p

@@ -1392,10 +1258,9 @@ tails ts@(Chunk t ts')

13921258-- copies to create substrings; they just construct new 'Text's that

13931259-- are slices of the original.

13941260

1395−-- | /O(m+n)/ Break a 'Text' into pieces separated by the first 'Text'

1396−-- argument (which cannot be an empty string), consuming the

1397−-- delimiter. An empty delimiter is invalid, and will cause an error

1398−-- to be raised.

1261+-- | /O(m+n)/ Break a 'Text' into pieces separated by the first

1262+-- 'Text' argument, consuming the delimiter. An empty delimiter is

1263+-- invalid, and will cause an error to be raised.

13991264--

14001265-- Examples:

14011266--

@@ -1408,18 +1273,13 @@ tails ts@(Chunk t ts')

14081273-- > intercalate s . splitOn s == id

14091274-- > splitOn (singleton c) == split (==c)

14101275--

1411−-- (Note: the string @s@ to split on above cannot be empty.)

1412−--

14131276-- This function is strict in its first argument, and lazy in its

14141277-- second.

14151278--

14161279-- In (unlikely) bad cases, this function's time complexity degrades

14171280-- towards /O(n*m)/.

1418−splitOn :: Text

1419− -- ^ String to split on. If this string is empty, an error

1420− -- will occur.

1421− -> Text

1422− -- ^ Input text.

1281+splitOn :: Text -- ^ Text to split on

1282+ -> Text -- ^ Input text

14231283 -> [Text]

14241284splitOn pat src

14251285 | null pat = emptyError "splitOn"

Data/Text/Lazy/Builder.hs+1−1

@@ -8,7 +8,7 @@

88-- Module : Data.Text.Lazy.Builder

11−-- License : BSD-style (see LICENSE)

11+-- License : BSD3-style (see LICENSE)

1212--

1313-- Maintainer : Johan Tibell <johan.tibell@gmail.com>

1414-- Stability : experimental

Data/Text/Lazy/Builder/Int.hs+14−19

@@ -1,5 +1,4 @@

1−{-# LANGUAGE BangPatterns, CPP, MagicHash, RankNTypes, ScopedTypeVariables,

2− UnboxedTuples #-}

1+{-# LANGUAGE BangPatterns, CPP, MagicHash, RankNTypes, UnboxedTuples #-}

32#if __GLASGOW_HASKELL__ >= 702

43{-# LANGUAGE Trustworthy #-}

54#endif

@@ -7,7 +6,7 @@

76-- Module: Data.Text.Lazy.Builder.Int

10−-- License: BSD-style

9+-- License: BSD3

1110-- Maintainer: Bryan O'Sullivan <bos@serpentine.com>

1211-- Stability: experimental

1312-- Portability: portable

@@ -35,7 +34,7 @@ import Control.Monad.ST

3534

3635#ifdef __GLASGOW_HASKELL__

3736# if defined(INTEGER_GMP)

38−import GHC.Integer.GMP.Internals (Integer(S#))

37+import GHC.Integer.GMP.Internals

3938# elif defined(INTEGER_SIMPLE)

4039import GHC.Integer

4140# else

@@ -55,7 +54,7 @@ decimal :: Integral a => a -> Builder

5554{-# RULES "decimal/Int16" decimal = boundedDecimal :: Int16 -> Builder #-}

5655{-# RULES "decimal/Int32" decimal = boundedDecimal :: Int32 -> Builder #-}

5756{-# RULES "decimal/Int64" decimal = boundedDecimal :: Int64 -> Builder #-}

58−{-# RULES "decimal/Word" decimal = positive :: Data.Word.Word -> Builder #-}

57+{-# RULES "decimal/Word" decimal = positive :: Word -> Builder #-}

5958{-# RULES "decimal/Word8" decimal = positive :: Word8 -> Builder #-}

6059{-# RULES "decimal/Word16" decimal = positive :: Word16 -> Builder #-}

6160{-# RULES "decimal/Word32" decimal = positive :: Word32 -> Builder #-}

@@ -134,15 +133,8 @@ i2w v = zero + fromIntegral v

134133

135134countDigits :: (Integral a) => a -> Int

136135{-# INLINE countDigits #-}

137−countDigits v0

138− | fromIntegral v64 == v0 = go 1 v64

139− | otherwise = goBig 1 (fromIntegral v0)

140− where v64 = fromIntegral v0

141− goBig !k (v :: Integer)

142− | v > big = goBig (k + 19) (v `quot` big)

143− | otherwise = go k (fromIntegral v)

144− big = 10000000000000000000

145− go !k (v :: Word64)

136+countDigits v0 = go 1 (fromIntegral v0 :: Word64)

137+ where go !k v

146138 | v < 10 = k

147139 | v < 100 = k + 1

148140 | v < 1000 = k + 2

@@ -194,12 +186,19 @@ hexDigit n

194186 | otherwise = singleton $! toEnum (fromIntegral n + 87)

195187{-# INLINE hexDigit #-}

196188

189+int :: Int -> Builder

190+int = decimal

191+{-# INLINE int #-}

192+

197193data T = T !Integer !Int

198194

199195integer :: Int -> Integer -> Builder

200196#ifdef INTEGER_GMP

201197integer 10 (S# i#) = decimal (I# i#)

202198integer 16 (S# i#) = hexadecimal (I# i#)

199+#else

200+integer 10 i = decimal i

201+integer 16 i = hexadecimal i

203202#endif

204203integer base i

205204 | i < 0 = singleton '-' <> go (-i)

@@ -247,11 +246,7 @@ integer base i

247246 PAIR(x,y) -> pblock q <> pblock r <> putB ns

248247 where q = fromInteger x

249248 r = fromInteger y

250− putB _ = Data.Monoid.mempty

251−

252− int :: Int -> Builder

253− int x | base == 10 = decimal x

254− | otherwise = hexadecimal x

249+ putB _ = mempty

255250

256251 pblock = loop maxDigits

257252 where

Data/Text/Lazy/Encoding.hs+21−9

@@ -46,25 +46,30 @@ module Data.Text.Lazy.Encoding

4646 , encodeUtf32LE

4747 , encodeUtf32BE

4848

49+#if MIN_VERSION_bytestring(0,10,4)

4950 -- * Encoding Text using ByteString Builders

5051 , encodeUtf8Builder

5152 , encodeUtf8BuilderEscaped

53+#endif

5254 ) where

5355

5456import Control.Exception (evaluate, try)

55−import Data.Monoid (Monoid(..))

5657import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

5758import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldrChunks)

58−import Data.Word (Word8)

5959import qualified Data.ByteString as S

60−import qualified Data.ByteString.Builder as B

61−import qualified Data.ByteString.Builder.Extra as B (safeStrategy, toLazyByteStringWith)

62−import qualified Data.ByteString.Builder.Prim as BP

6360import qualified Data.ByteString.Lazy as B

6461import qualified Data.ByteString.Lazy.Internal as B

6562import qualified Data.ByteString.Unsafe as B

63+#if MIN_VERSION_bytestring(0,10,4)

64+import Data.Word (Word8)

65+import Data.Monoid (mempty, (<>))

66+import qualified Data.ByteString.Builder as B

67+import qualified Data.ByteString.Builder.Extra as B (safeStrategy, toLazyByteStringWith)

68+import qualified Data.ByteString.Builder.Prim as BP

6669import qualified Data.Text as T

70+#endif

6771import qualified Data.Text.Encoding as TE

72+import qualified Data.Text.Lazy as L

6873import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

6974import qualified Data.Text.Internal.Lazy.Fusion as F

7075import Data.Text.Unsafe (unsafeDupablePerformIO)

@@ -84,6 +89,8 @@ import Data.Text.Unsafe (unsafeDupablePerformIO)

8489

8590-- | /Deprecated/. Decode a 'ByteString' containing 7-bit ASCII

8691-- encoded text.

92+--

93+-- This function is deprecated. Use 'decodeLatin1' instead.

8794decodeASCII :: B.ByteString -> Text

8895decodeASCII = decodeUtf8

8996{-# DEPRECATED decodeASCII "Use decodeUtf8 instead" #-}

@@ -105,7 +112,7 @@ decodeUtf8With onErr (B.Chunk b0 bs0) =

105112 | S.null l = empty

106113 | otherwise = case onErr desc (Just (B.unsafeHead l)) of

107114 Nothing -> empty

108− Just c -> Chunk (T.singleton c) Empty

115+ Just c -> L.singleton c

109116 desc = "Data.Text.Lazy.Encoding.decodeUtf8With: Invalid UTF-8 stream"

110117decodeUtf8With _ _ = empty

111118

@@ -142,6 +149,7 @@ decodeUtf8' bs = unsafeDupablePerformIO $ do

142149{-# INLINE decodeUtf8' #-}

143150

144151encodeUtf8 :: Text -> B.ByteString

152+#if MIN_VERSION_bytestring(0,10,4)

145153encodeUtf8 Empty = B.empty

146154encodeUtf8 lt@(Chunk t _) =

147155 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

@@ -155,13 +163,17 @@ encodeUtf8 lt@(Chunk t _) =

155163 strategy = B.safeStrategy firstChunkSize B.defaultChunkSize

156164

157165encodeUtf8Builder :: Text -> B.Builder

158−encodeUtf8Builder =

159− foldrChunks (\c b -> TE.encodeUtf8Builder c `mappend` b) Data.Monoid.mempty

166+encodeUtf8Builder = foldrChunks (\c b -> TE.encodeUtf8Builder c <> b) mempty

160167

161168{-# INLINE encodeUtf8BuilderEscaped #-}

162169encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

163170encodeUtf8BuilderEscaped prim =

164− foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c `mappend` b) mempty

171+ foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c <> b) mempty

172+

173+#else

174+encodeUtf8 (Chunk c cs) = B.Chunk (TE.encodeUtf8 c) (encodeUtf8 cs)

175+encodeUtf8 Empty = B.Empty

176+#endif

165177

166178-- | Decode text from little endian UTF-16 encoding.

167179decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

Data/Text/Lazy/Read.hs+2−4

@@ -1,7 +1,5 @@

11{-# LANGUAGE OverloadedStrings, CPP #-}

2−#if __GLASGOW_HASKELL__ >= 704

3−{-# LANGUAGE Safe #-}

4−#elif __GLASGOW_HASKELL__ >= 702

2+#if __GLASGOW_HASKELL__ >= 702

53{-# LANGUAGE Trustworthy #-}

64#endif

@@ -56,7 +54,7 @@ decimal :: Integral a => Reader a

5654{-# SPECIALIZE decimal :: Reader Int32 #-}

5755{-# SPECIALIZE decimal :: Reader Int64 #-}

5856{-# SPECIALIZE decimal :: Reader Integer #-}

59−{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

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

6058{-# SPECIALIZE decimal :: Reader Word8 #-}

6159{-# SPECIALIZE decimal :: Reader Word16 #-}

6260{-# SPECIALIZE decimal :: Reader Word32 #-}

Data/Text/Read.hs+1−1

@@ -55,7 +55,7 @@ decimal :: Integral a => Reader a

5555{-# SPECIALIZE decimal :: Reader Int32 #-}

5656{-# SPECIALIZE decimal :: Reader Int64 #-}

5757{-# SPECIALIZE decimal :: Reader Integer #-}

58−{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

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

5959{-# SPECIALIZE decimal :: Reader Word8 #-}

6060{-# SPECIALIZE decimal :: Reader Word16 #-}

6161{-# SPECIALIZE decimal :: Reader Word32 #-}

Data/Text/Show.hsdeleted−89

@@ -1,89 +0,0 @@

1−{-# LANGUAGE CPP, MagicHash #-}

2−{-# OPTIONS_GHC -fno-warn-orphans #-}

3−#if __GLASGOW_HASKELL__ >= 702

4−{-# LANGUAGE Trustworthy #-}

5−#endif

6−

7−-- |

8−-- Module : Data.Text.Show

10−--

11−-- License : BSD-style

12−-- Maintainer : bos@serpentine.com

13−-- Stability : experimental

14−-- Portability : GHC

15−

16−module Data.Text.Show

17− (

18− singleton

19− , unpack

20− , unpackCString#

21− ) where

22−

23−import Control.Monad.ST (ST)

24−import Data.Text.Internal (Text(..), empty_, safe)

25−import Data.Text.Internal.Fusion (stream, unstream)

26−import Data.Text.Internal.Unsafe.Char (unsafeWrite)

27−import GHC.Prim (Addr#)

28−import qualified Data.Text.Array as A

29−import qualified Data.Text.Internal.Fusion.Common as S

30−

31−#if __GLASGOW_HASKELL__ >= 702

32−import qualified GHC.CString as GHC

33−#else

34−import qualified GHC.Base as GHC

35−#endif

36−

37−instance Show Text where

38− showsPrec p ps r = showsPrec p (unpack ps) r

39−

40−-- | /O(n)/ Convert a 'Text' into a 'String'. Subject to fusion.

41−unpack :: Text -> String

42−unpack = S.unstreamList . stream

43−{-# INLINE [1] unpack #-}

44−

45−-- | /O(n)/ Convert a literal string into a 'Text'. Subject to

46−-- fusion.

47−--

48−-- This is exposed solely for people writing GHC rewrite rules.

49−unpackCString# :: Addr# -> Text

50−unpackCString# addr# = unstream (S.streamCString# addr#)

51−{-# NOINLINE unpackCString# #-}

52−

53−{-# RULES "TEXT literal" [1] forall a.

54− unstream (S.map safe (S.streamList (GHC.unpackCString# a)))

55− = unpackCString# a #-}

56−

57−{-# RULES "TEXT literal UTF8" [1] forall a.

58− unstream (S.map safe (S.streamList (GHC.unpackCStringUtf8# a)))

59− = unpackCString# a #-}

60−

61−{-# RULES "TEXT empty literal" [1]

62− unstream (S.map safe (S.streamList []))

63− = empty_ #-}

64−

65−{-# RULES "TEXT singleton literal" [1] forall a.

66− unstream (S.map safe (S.streamList [a]))

67− = singleton_ a #-}

68−

69−-- | /O(1)/ Convert a character into a Text. Subject to fusion.

70−-- Performs replacement on invalid scalar values.

71−singleton :: Char -> Text

72−singleton = unstream . S.singleton . safe

73−{-# INLINE [1] singleton #-}

74−

75−{-# RULES "TEXT singleton" forall a.

76− unstream (S.singleton (safe a))

77− = singleton_ a #-}

78−

79−-- This is intended to reduce inlining bloat.

80−singleton_ :: Char -> Text

81−singleton_ c = Text (A.run x) 0 len

82− where x :: ST s (A.MArray s)

83− x = do arr <- A.new len

84− _ <- unsafeWrite arr 0 d

85− return arr

86− len | d < '\x10000' = 1

87− | otherwise = 2

88− d = safe c

89−{-# NOINLINE singleton_ #-}

benchmarks/haskell/Benchmarks.hs+1−1

@@ -53,7 +53,7 @@ benchmarks = do

5353 , Mul.benchmark

5454 , Pure.benchmark "tiny" (tf "tiny.txt")

5555 , Pure.benchmark "ascii" (tf "ascii-small.txt")

56− -- , Pure.benchmark "france" (tf "france.html")

56+ , Pure.benchmark "france" (tf "france.html")

5757 , Pure.benchmark "russian" (tf "russian-small.txt")

5858 , Pure.benchmark "japanese" (tf "japanese.txt")

5959 , ReadNumbers.benchmark (tf "numbers.txt")

benchmarks/haskell/Benchmarks/DecodeUtf8.hs+2−2

@@ -25,7 +25,7 @@ import Foreign.Ptr (Ptr, plusPtr)

2525import Foreign.ForeignPtr (withForeignPtr)

2626import Data.Word (Word8)

2727import qualified Criterion as C

28−import Criterion (Benchmark, bgroup, nf, whnfIO)

28+import Criterion (Benchmark, bgroup, nf)

2929import qualified Codec.Binary.UTF8.Generic as U8

3030import qualified Data.ByteString as B

3131import qualified Data.ByteString.Lazy as BL

@@ -48,7 +48,7 @@ benchmark kind fp = do

4848 return $ bgroup "DecodeUtf8"

4949 [ bench "Strict" $ nf T.decodeUtf8 bs

5050 , bench "Stream" $ nf decodeStream lbs

51− , bench "IConv" $ whnfIO $ iconv bs

51+ , bench "IConv" $ iconv bs

5252 , bench "StrictLength" $ nf (T.length . T.decodeUtf8) bs

5353 , bench "StrictInitLength" $ nf (T.length . T.init . T.decodeUtf8) bs

5454 , bench "Lazy" $ nf TL.decodeUtf8 lbs

benchmarks/haskell/Benchmarks/FileRead.hs+11−11

@@ -8,8 +8,8 @@ module Benchmarks.FileRead

88 ( benchmark

99 ) where

1010

11−import Control.Applicative ((<$>))

12−import Criterion (Benchmark, bgroup, bench, whnfIO)

11+import Control.Exception (evaluate)

12+import Criterion (Benchmark, bgroup, bench)

1313import qualified Data.ByteString as SB

1414import qualified Data.ByteString.Lazy as LB

1515import qualified Data.Text as T

@@ -21,13 +21,13 @@ import qualified Data.Text.Lazy.IO as LT

2121

2222benchmark :: FilePath -> IO Benchmark

2323benchmark p = return $ bgroup "FileRead"

24− [ bench "String" $ whnfIO $ length <$> readFile p

25− , bench "ByteString" $ whnfIO $ SB.length <$> SB.readFile p

26− , bench "LazyByteString" $ whnfIO $ LB.length <$> LB.readFile p

27− , bench "Text" $ whnfIO $ T.length <$> T.readFile p

28− , bench "LazyText" $ whnfIO $ LT.length <$> LT.readFile p

29− , bench "TextByteString" $ whnfIO $

30− (T.length . T.decodeUtf8) <$> SB.readFile p

31− , bench "LazyTextByteString" $ whnfIO $

32− (LT.length . LT.decodeUtf8) <$> LB.readFile p

24+ [ bench "String" $ readFile p >>= evaluate . length

25+ , bench "ByteString" $ SB.readFile p >>= evaluate . SB.length

26+ , bench "LazyByteString" $ LB.readFile p >>= evaluate . LB.length

27+ , bench "Text" $ T.readFile p >>= evaluate . T.length

28+ , bench "LazyText" $ LT.readFile p >>= evaluate . LT.length

29+ , bench "TextByteString" $

30+ SB.readFile p >>= evaluate . T.length . T.decodeUtf8

31+ , bench "LazyTextByteString" $

32+ LB.readFile p >>= evaluate . LT.length . LT.decodeUtf8

3333 ]

benchmarks/haskell/Benchmarks/FoldLines.hs+2−2

@@ -10,7 +10,7 @@ module Benchmarks.FoldLines

1010 ( benchmark

1111 ) where

1212

13−import Criterion (Benchmark, bgroup, bench, whnfIO)

13+import Criterion (Benchmark, bgroup, bench)

1414import System.IO

1515import qualified Data.ByteString as B

1616import qualified Data.Text as T

@@ -22,7 +22,7 @@ benchmark fp = return $ bgroup "ReadLines"

2222 , bench "ByteString" $ withHandle $ foldLinesB (\n _ -> n + 1) (0 :: Int)

2323 ]

2424 where

25− withHandle f = whnfIO $ do

25+ withHandle f = do

2626 h <- openFile fp ReadMode

2727 hSetBuffering h (BlockBuffering (Just 16384))

2828 x <- f h

benchmarks/haskell/Benchmarks/Programs/BigTable.hs+2−2

@@ -11,7 +11,7 @@ module Benchmarks.Programs.BigTable

1111 ( benchmark

1212 ) where

1313

14−import Criterion (Benchmark, bench, whnfIO)

14+import Criterion (Benchmark, bench)

1515import Data.Monoid (mappend, mconcat)

1616import Data.Text.Lazy.Builder (Builder, fromText, toLazyText)

1717import Data.Text.Lazy.IO (hPutStr)

@@ -19,7 +19,7 @@ import System.IO (Handle)

1919import qualified Data.Text as T

2020

2121benchmark :: Handle -> IO Benchmark

22−benchmark sink = return $ bench "BigTable" $ whnfIO $ do

22+benchmark sink = return $ bench "BigTable" $ do

2323 hPutStr sink "Content-Type: text/html\n\n<table>"

2424 hPutStr sink . toLazyText . makeTable =<< rows

2525 hPutStr sink "</table>"

benchmarks/haskell/Benchmarks/Programs/Cut.hs+2−2

@@ -16,7 +16,7 @@ module Benchmarks.Programs.Cut

1616 ( benchmark

1717 ) where

1818

19−import Criterion (Benchmark, bgroup, bench, whnfIO)

19+import Criterion (Benchmark, bgroup, bench)

2020import System.IO (Handle, hPutStr)

2121import qualified Data.ByteString as B

2222import qualified Data.ByteString.Char8 as BC

@@ -40,7 +40,7 @@ benchmark p sink from to = return $ bgroup "Cut"

4040 , bench' "LazyTextByteString" lazyTextByteString

4141 ]

4242 where

43− bench' n s = bench n $ whnfIO (s p sink from to)

43+ bench' n s = bench n (s p sink from to)

4444

4545string :: FilePath -> Handle -> Int -> Int -> IO ()

4646string fp sink from to = do

benchmarks/haskell/Benchmarks/Programs/Fold.hs+2−2

@@ -21,7 +21,7 @@ import Data.List (foldl')

2121import Data.List (intersperse)

2222import Data.Monoid (mempty, mappend, mconcat)

2323import System.IO (Handle)

24−import Criterion (Benchmark, bench, whnfIO)

24+import Criterion (Benchmark, bench)

2525import qualified Data.Text as T

2626import qualified Data.Text.IO as T

2727import qualified Data.Text.Lazy.Builder as TLB

@@ -30,7 +30,7 @@ import qualified Data.Text.Lazy.IO as TL

3030

3131benchmark :: FilePath -> Handle -> IO Benchmark

3232benchmark i o = return $

33− bench "Fold" $ whnfIO $ T.readFile i >>= TL.hPutStr o . fold 80

33+ bench "Fold" $ T.readFile i >>= TL.hPutStr o . fold 80

3434

3535-- | We represent a paragraph by a word list

3636--

benchmarks/haskell/Benchmarks/Programs/Sort.hs+9−10

@@ -17,7 +17,7 @@ module Benchmarks.Programs.Sort

1717 ( benchmark

1818 ) where

1919

20−import Criterion (Benchmark, bgroup, bench, whnfIO)

20+import Criterion (Benchmark, bgroup, bench)

2121import Data.Monoid (mconcat)

2222import System.IO (Handle, hPutStr)

2323import qualified Data.ByteString as B

@@ -35,17 +35,16 @@ import qualified Data.Text.Lazy.IO as TL

3535

3636benchmark :: FilePath -> Handle -> IO Benchmark

3737benchmark i o = return $ bgroup "Sort"

38− [ bench "String" $ whnfIO $ readFile i >>= hPutStr o . string

39− , bench "ByteString" $ whnfIO $ B.readFile i >>= B.hPutStr o . byteString

40− , bench "LazyByteString" $ whnfIO $

41− BL.readFile i >>= BL.hPutStr o . lazyByteString

42− , bench "Text" $ whnfIO $ T.readFile i >>= T.hPutStr o . text

43− , bench "LazyText" $ whnfIO $ TL.readFile i >>= TL.hPutStr o . lazyText

44− , bench "TextByteString" $ whnfIO $ B.readFile i >>=

38+ [ bench "String" $ readFile i >>= hPutStr o . string

39+ , bench "ByteString" $ B.readFile i >>= B.hPutStr o . byteString

40+ , bench "LazyByteString" $ BL.readFile i >>= BL.hPutStr o . lazyByteString

41+ , bench "Text" $ T.readFile i >>= T.hPutStr o . text

42+ , bench "LazyText" $ TL.readFile i >>= TL.hPutStr o . lazyText

43+ , bench "TextByteString" $ B.readFile i >>=

4544 B.hPutStr o . T.encodeUtf8 . text . T.decodeUtf8

46− , bench "LazyTextByteString" $ whnfIO $ BL.readFile i >>=

45+ , bench "LazyTextByteString" $ BL.readFile i >>=

4746 BL.hPutStr o . TL.encodeUtf8 . lazyText . TL.decodeUtf8

48− , bench "TextBuilder" $ whnfIO $ B.readFile i >>=

47+ , bench "TextBuilder" $ B.readFile i >>=

4948 BL.hPutStr o . TL.encodeUtf8 . textBuilder . T.decodeUtf8

5049 ]

5150

benchmarks/haskell/Benchmarks/Programs/StripTags.hs+5−5

@@ -15,7 +15,7 @@ module Benchmarks.Programs.StripTags

1515 ( benchmark

1616 ) where

1717

18−import Criterion (Benchmark, bgroup, bench, whnfIO)

18+import Criterion (Benchmark, bgroup, bench)

1919import Data.List (mapAccumL)

2020import System.IO (Handle, hPutStr)

2121import qualified Data.ByteString as B

@@ -26,10 +26,10 @@ import qualified Data.Text.IO as T

2626

2727benchmark :: FilePath -> Handle -> IO Benchmark

2828benchmark i o = return $ bgroup "StripTags"

29− [ bench "String" $ whnfIO $ readFile i >>= hPutStr o . string

30− , bench "ByteString" $ whnfIO $ B.readFile i >>= B.hPutStr o . byteString

31− , bench "Text" $ whnfIO $ T.readFile i >>= T.hPutStr o . text

32− , bench "TextByteString" $ whnfIO $

29+ [ bench "String" $ readFile i >>= hPutStr o . string

30+ , bench "ByteString" $ B.readFile i >>= B.hPutStr o . byteString

31+ , bench "Text" $ T.readFile i >>= T.hPutStr o . text

32+ , bench "TextByteString" $

3333 B.readFile i >>= B.hPutStr o . T.encodeUtf8 . text . T.decodeUtf8

3434 ]

3535

benchmarks/haskell/Benchmarks/Programs/Throughput.hs+8−8

@@ -18,7 +18,7 @@ module Benchmarks.Programs.Throughput

1818 ( benchmark

1919 ) where

2020

21−import Criterion (Benchmark, bgroup, bench, whnfIO)

21+import Criterion (Benchmark, bgroup, bench)

2222import System.IO (Handle, hPutStr)

2323import qualified Data.ByteString as B

2424import qualified Data.ByteString.Lazy as BL

@@ -29,13 +29,13 @@ import qualified Data.Text.Lazy.IO as TL

2929

3030benchmark :: FilePath -> Handle -> IO Benchmark

3131benchmark fp sink = return $ bgroup "Throughput"

32− [ bench "String" $ whnfIO $ readFile fp >>= hPutStr sink

33− , bench "ByteString" $ whnfIO $ B.readFile fp >>= B.hPutStr sink

34− , bench "LazyByteString" $ whnfIO $ BL.readFile fp >>= BL.hPutStr sink

35− , bench "Text" $ whnfIO $ T.readFile fp >>= T.hPutStr sink

36− , bench "LazyText" $ whnfIO $ TL.readFile fp >>= TL.hPutStr sink

37− , bench "TextByteString" $ whnfIO $

32+ [ bench "String" $ readFile fp >>= hPutStr sink

33+ , bench "ByteString" $ B.readFile fp >>= B.hPutStr sink

34+ , bench "LazyByteString" $ BL.readFile fp >>= BL.hPutStr sink

35+ , bench "Text" $ T.readFile fp >>= T.hPutStr sink

36+ , bench "LazyText" $ TL.readFile fp >>= TL.hPutStr sink

37+ , bench "TextByteString" $

3838 B.readFile fp >>= B.hPutStr sink . T.encodeUtf8 . T.decodeUtf8

39− , bench "LazyTextByteString" $ whnfIO $

39+ , bench "LazyTextByteString" $

4040 BL.readFile fp >>= BL.hPutStr sink . TL.encodeUtf8 . TL.decodeUtf8

4141 ]

benchmarks/haskell/Benchmarks/Pure.hs−7

@@ -262,13 +262,6 @@ benchmark kind fp = do

262262 , benchBSL $ nf (BL.map toUpper) bla

263263 , benchS $ nf (L.map toUpper) sa

264264 ]

265− , bgroup "uncons"

266− [ benchT $ nf T.uncons ta

267− , benchTL $ nf TL.uncons tla

268− , benchBS $ nf BS.uncons bsa

269− , benchBSL $ nf BL.uncons bla

270− , benchS $ nf L.uncons sa

271− ]

272265 , bgroup "words"

273266 [ benchT $ nf T.words ta

274267 , benchTL $ nf TL.words tla

benchmarks/haskell/Benchmarks/ReadNumbers.hs+5−2

@@ -25,7 +25,8 @@ import Data.List (foldl')

2525import Numeric (readDec, readFloat, readHex)

2626import qualified Data.ByteString.Char8 as B

2727import qualified Data.ByteString.Lazy.Char8 as BL

28−import qualified Data.ByteString.Lex.Fractional as B

28+import qualified Data.ByteString.Lex.Double as B

29+import qualified Data.ByteString.Lex.Lazy.Double as BL

2930import qualified Data.Text as T

3031import qualified Data.Text.IO as T

3132import qualified Data.Text.Lazy as TL

@@ -62,10 +63,12 @@ benchmark fp = do

6263 whnf (double . text TL.rational) tl

6364

6465 , bench "DecimalByteString" $ whnf (int . byteString B.readInt) b

65− , bench "DoubleByteString" $ whnf (double . byteString B.readDecimal) b

66+ , bench "DoubleByteString" $ whnf (double . byteString B.readDouble) b

6667

6768 , bench "DecimalLazyByteString" $

6869 whnf (int . byteString BL.readInt) bl

70+ , bench "DoubleLazyByteString" $

71+ whnf (double . byteString BL.readDouble) bl

6972 ]

7073 where

7174 -- Used for fixing types

benchmarks/haskell/Benchmarks/Stream.hs+5−16

@@ -18,11 +18,11 @@ import qualified Data.Text.Encoding as T

1818import qualified Data.Text.Encoding.Error as E

1919import qualified Data.Text.Internal.Encoding.Fusion as T

2020import qualified Data.Text.Internal.Encoding.Fusion.Common as F

21−import qualified Data.Text.Internal.Fusion as F

21+import qualified Data.Text.Internal.Fusion as T

2222import qualified Data.Text.IO as T

2323import qualified Data.Text.Lazy.Encoding as TL

2424import qualified Data.Text.Internal.Lazy.Encoding.Fusion as TL

25−import qualified Data.Text.Internal.Lazy.Fusion as FL

25+import qualified Data.Text.Internal.Lazy.Fusion as TL

2626import qualified Data.Text.Lazy.IO as TL

2727

2828instance NFData a => NFData (Stream a) where

@@ -53,25 +53,14 @@ benchmark fp = do

5353 !utf32beL = TL.encodeUtf32BE tl

5454

5555 -- For the functions which operate on streams

56− let !s = F.stream t

56+ let !s = T.stream t

5757

5858 return $ bgroup "Stream"

5959

6060 -- Fusion

6161 [ bgroup "stream" $

62− [ bench "Text" $ nf F.stream t

63− , bench "LazyText" $ nf FL.stream tl

64− ]

65− -- must perform exactly the same as stream above due to

66− -- stream/unstream (i.e. stream after unstream) fusion

67− , bgroup "stream-fusion" $

68− [ bench "Text" $ nf (F.stream . F.unstream . F.stream) t

69− , bench "LazyText" $ nf (FL.stream . FL.unstream . FL.stream) tl

70− ]

71− -- measure the overhead of unstream after stream

72− , bgroup "stream-unstream" $

73− [ bench "Text" $ nf (F.unstream . F.stream) t

74− , bench "LazyText" $ nf (FL.unstream . FL.stream) tl

62+ [ bench "Text" $ nf T.stream t

63+ , bench "LazyText" $ nf TL.stream tl

7564 ]

7665

7766 -- Encoding.Fusion

benchmarks/text-benchmarks.cabal+4−14

@@ -3,7 +3,7 @@ version: 0.0.0.0

33synopsis: Benchmarks for the text package

44description: Benchmarks for the text package

55homepage: https://bitbucket.org/bos/text

6−license: BSD2

6+license: BSD3

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

@@ -15,11 +15,6 @@ build-type: Simple

1515

1616cabal-version: >=1.2

1717

18−flag bytestring-builder

19− description: Depend on the bytestring-builder package for backwards compatibility.

20− default: False

21− manual: False

22−

2318flag llvm

2419 description: use LLVM

2520 default: False

@@ -38,9 +33,10 @@ executable text-benchmarks

3833 build-depends: base == 4.*,

3934 binary,

4035 blaze-builder,

41− bytestring-lexing >= 0.5.0,

36+ bytestring,

37+ bytestring-lexing,

4238 containers,

43− criterion >= 0.10.0.0,

39+ criterion >= 0.6.0.1,

4440 deepseq,

4541 directory,

4642 filepath,

@@ -50,12 +46,6 @@ executable text-benchmarks

5046 utf8-string,

5147 vector

5248

53− if flag(bytestring-builder)

54− build-depends: bytestring >= 0.9 && < 0.10.4,

55− bytestring-builder >= 0.10.4

56− else

57− build-depends: bytestring >= 0.10.4

58−

5949executable text-multilang

6050 hs-source-dirs: haskell

6151 main-is: Multilang.hs

changelog.md−109

@@ -1,112 +1,3 @@

1−1.2.2.1

2−

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

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

5− maintainers apologizes for the churn.

6−

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

8− (updated from 7.0)

9−

10−* An STG lint error has been fixed

11−

12−1.2.2.0

13−

14−* The `integer-simple` package, upon which this package optionally

15− depended, has been replaced with `integer-pure`. The build flag has

16− been renamed accordingly.

17−

18−* Bug fix: For the `Binary` instance, If UTF-8 decoding fails during a

19− `get`, the error is propagated via `fail` instead of an uncatchable

20− crash.

21−

22−* New function: `takeWhileEnd`

23−

24−* New instances for the `Text` types:

25− * if `base` >= 4.7: `PrintfArg`

26− * if `base` >= 4.9: `Semigroup`

27−

28−1.2.1.3

29−

30−* Bug fix: As it turns out, moving the literal rewrite rules to simplifier

31− phase 2 does not prevent competition with the `unpack` rule, which is

32− also active in this phase. Unfortunately this was hidden due to a silly

33− test environment mistake. Moving literal rules back to phase 1 finally

34− fixes GHC Trac #10528 correctly.

35−

36−1.2.1.2

37−

38−* Bug fix: Run literal rewrite rules in simplifier phase 2.

39− The behavior of the simplifier changed in GHC 7.10.2,

40− causing these rules to fail to fire, leading to poor code generation

41− and long compilation times. See

42− [GHC Trac #10528](https://ghc.haskell.org/trac/ghc/ticket/10528).

43−

44−1.2.1.1

45−

46−* Expose unpackCString#, which you should never use.

47−

48−1.2.1.0

49−

50−* Added Binary instances for both Text types. (If you have previously

51− been using the text-binary package to get a Binary instance, it is

52− now obsolete.)

53−

54−1.2.0.6

55−

56−* Fixed a space leak in UTF-8 decoding

57−

58−1.2.0.5

59−

60−* Feature parity: repeat, cycle, iterate are now implemented for lazy

61− Text, and the Data instance is more complete

62−

63−* Build speed: an inliner space explosion has been fixed with toCaseFold

64−

65−* Bug fix: encoding Int to a Builder would infinite-loop if the

66− integer-simple package was used

67−

68−* Deprecation: OnEncodeError and EncodeError are deprecated, as they

69− are never used

70−

71−* Internals: some types that are used internally in fusion-related

72− functions have moved around, been renamed, or been deleted (we don't

73− bump the major version if .Internal modules change)

74−

75−* Spec compliance: toCaseFold now follows the Unicode 7.0 spec

76− (updated from 6.3)

77−

78−1.2.0.4

79−

80−* Fixed an incompatibility with base < 4.5

81−

82−1.2.0.3

83−

84−* Update formatRealFloat to correspond to the definition in versions

85− of base newer than 4.5 (https://github.com/bos/text/issues/105)

86−

87−1.2.0.2

88−

89−* Bumped lower bound on deepseq to 1.4 for compatibility with the

90− upcoming GHC 7.10

91−

92−1.2.0.1

93−

94−* Fixed a buffer overflow in rendering of large Integers

95− (https://github.com/bos/text/issues/99)

96−

97−1.2.0.0

98−

99−* Fixed an integer overflow in the replace function

100− (https://github.com/bos/text/issues/81)

101−

102−* Fixed a hang in lazy decodeUtf8With

103− (https://github.com/bos/text/issues/87)

104−

105−* Reduced codegen bloat caused by use of empty and single-character

106− literals

107−

108−* Added an instance of IsList for GHC 7.8 and above

109−

11011.1.1.0

1112

1123* The Data.Data instance now allows gunfold to work, via a virtual

scripts/CaseFolding.hs+1−1

@@ -37,7 +37,7 @@ mapCF :: CaseFolding -> [String]

3737mapCF (CF _ ms) = typ ++ (map nice . filter p $ ms) ++ [last]

3838 where

3939 typ = ["foldMapping :: forall s. Char -> s -> Step (CC s) Char"

40− ,"{-# NOINLINE foldMapping #-}"]

40+ ,"{-# INLINE foldMapping #-}"]

4141 last = "foldMapping c s = Yield (toLower c) (CC s '\\0' '\\0')"

4242 nice c = "-- " ++ name c ++ "\n" ++

4343 "foldMapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (CC s " ++ y ++ " " ++ z ++ ")"

scripts/CaseMapping.hs+3−3

@@ -8,7 +8,7 @@ import SpecialCasing

88main = do

99 args <- getArgs

1010 let oname = case args of

11− [] -> "../Data/Text/Internal/Fusion/CaseMapping.hs"

11+ [] -> "../Data/Text/Fusion/CaseMapping.hs"

1212 [o] -> o

1313 psc <- parseSC "SpecialCasing.txt"

1414 pcf <- parseCF "CaseFolding.txt"

@@ -24,10 +24,10 @@ main = do

2424 mapM_ (hPutStrLn h) $

2525 ["{-# LANGUAGE Rank2Types #-}"

2626 ,"-- AUTOMATICALLY GENERATED - DO NOT EDIT"

27− ,"-- Generated by scripts/CaseMapping.hs"] ++

27+ ,"-- Generated by scripts/SpecialCasing.hs"] ++

2828 comments ++

2929 [""

30− ,"module Data.Text.Internal.Fusion.CaseMapping where"

30+ ,"module Data.Text.Fusion.CaseMapping where"

3131 ,"import Data.Char"

3232 ,"import Data.Text.Internal.Fusion.Types"

3333 ,""]

scripts/SpecialCasing.hs+1−1

@@ -43,7 +43,7 @@ mapSC which access twiddle (SC _ ms) =

4343 typ ++ (map nice . filter p $ ms) ++ [last]

4444 where

4545 typ = [which ++ "Mapping :: forall s. Char -> s -> Step (CC s) Char"

46− ,"{-# NOINLINE " ++ which ++ "Mapping #-}"]

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

4747 last = which ++ "Mapping c s = Yield (to" ++ ucFirst which ++ " c) (CC s '\\0' '\\0')"

4848 nice c = "-- " ++ name c ++ "\n" ++

4949 which ++ "Mapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (CC s " ++ y ++ " " ++ z ++ ")"

tests/LiteralRuleTest.hsdeleted−21

@@ -1,21 +0,0 @@

1−{-# LANGUAGE OverloadedStrings #-}

2−

3−module LiteralRuleTest where

4−

5−import Data.Text (Text)

6−

7−-- This should produce 8 firings of the "TEXT literal" rule

8−strings :: [Text]

9−strings = [ "abstime", "aclitem", "bit", "bool", "box", "bpchar", "bytea", "char" ]

10−

11−-- This should produce 7 firings of the "TEXT literal UTF8" rule

12−utf8Strings :: [Text]

13−utf8Strings = [ "\0abstime", "\0aclitem", "\xfefe bit", "\0bool", "\0box", "\0bpchar", "\0bytea" ]

14−

15−-- This should produce 4 firings of the "TEXT empty literal" rule

16−empties :: [Text]

17−empties = [ "", "", "", "" ]

18−

19−-- This should produce 5 firings of the "TEXT empty literal" rule

20−--singletons :: [Text]

21−--singletons = [ "a", "b", "c", "d", "e" ]

tests/Makefile+1−6

@@ -1,10 +1,5 @@

11count = 1000

3−all: coverage literal-rule-test

4−

5−literal-rule-test:

6− ./literal-rule-test.sh

7−

83coverage: build coverage/hpc_index.html

105build: text-test-data

@@ -37,4 +32,4 @@ coverage/hpc_index.html: coverage/coverage.tix

3732clean:

3833 rm -rf dist coverage .hpc

3934

40−.PHONY: build coverage all literal-rule-test

35+.PHONY: build coverage

tests/Tests/Properties.hs+9−57

@@ -1,8 +1,9 @@

1−-- | QuickCheck properties for the text library.

2−

1+-- | General quicktest properties for the text library

2+--

33{-# LANGUAGE BangPatterns, FlexibleInstances, OverloadedStrings,

44 ScopedTypeVariables, TypeSynonymInstances #-}

5−{-# OPTIONS_GHC -fno-enable-rewrite-rules -fno-warn-missing-signatures #-}

5+{-# OPTIONS_GHC -fno-enable-rewrite-rules #-}

6+{-# OPTIONS_GHC -fno-warn-missing-signatures #-}

67module Tests.Properties

78 (

89 tests

@@ -14,7 +15,7 @@ import Data.Bits ((.&.))

1415import Data.Char (chr, isDigit, isHexDigit, isLower, isSpace, isUpper, ord)

1516import Data.Int (Int8, Int16, Int32, Int64)

1617import Data.Monoid (Monoid(..))

17−import Data.String (IsString(fromString))

18+import Data.String (fromString)

1819import Data.Text.Encoding.Error

1920import Data.Text.Foreign

2021import Data.Text.Internal.Encoding.Utf8

@@ -23,7 +24,7 @@ import Data.Text.Internal.Search (indices)

2324import Data.Text.Lazy.Read as TL

2425import Data.Text.Read as T

2526import Data.Word (Word, Word8, Word16, Word32, Word64)

26−import Numeric (showEFloat, showFFloat, showGFloat, showHex)

27+import Numeric (showHex)

2728import Prelude hiding (replicate)

2829import Test.Framework (Test, testGroup)

2930import Test.Framework.Providers.QuickCheck2 (testProperty)

@@ -440,10 +441,6 @@ t_mapAccumR f z = L.mapAccumR f z `eqP` (second unpackS . T.mapAccumR f z)

440441tl_mapAccumR f z = L.mapAccumR f z `eqP` (second unpackS . TL.mapAccumR f z)

441442 where _types = f :: Int -> Char -> (Int,Char)

442443

443−tl_repeat n = (L.take m . L.repeat) `eq`

444− (unpackS . TL.take (fromIntegral m) . TL.repeat)

445− where m = fromIntegral (n :: Word8)

446−

447444replicate n l = concat (L.replicate n l)

448445

449446s_replicate n = replicate m `eq`

@@ -455,14 +452,6 @@ tl_replicate n = replicate m `eq`

455452 (unpackS . TL.replicate (fromIntegral m) . packS)

456453 where m = fromIntegral (n :: Word8)

457454

458−tl_cycle n = (L.take m . L.cycle) `eq`

459− (unpackS . TL.take (fromIntegral m) . TL.cycle . packS)

460− where m = fromIntegral (n :: Word8)

461−

462−tl_iterate f n = (L.take m . L.iterate f) `eq`

463− (unpackS . TL.take (fromIntegral m) . TL.iterate f)

464− where m = fromIntegral (n :: Word8)

465−

466455unf :: Int -> Char -> Maybe (Char, Char)

467456unf n c | fromEnum c * 100 > n = Nothing

468457 | otherwise = Just (c, succ c)

@@ -516,10 +505,6 @@ sf_takeWhile q p = (L.takeWhile p . L.filter q) `eqP`

516505 (unpackS . S.takeWhile p . S.filter q)

517506t_takeWhile p = L.takeWhile p `eqP` (unpackS . T.takeWhile p)

518507tl_takeWhile p = L.takeWhile p `eqP` (unpackS . TL.takeWhile p)

519−t_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

520− (unpackS . T.takeWhileEnd p)

521−tl_takeWhileEnd p = (L.reverse . L.takeWhile p . L.reverse) `eqP`

522− (unpackS . TL.takeWhileEnd p)

523508s_dropWhile p = L.dropWhile p `eqP` (unpackS . S.dropWhile p)

524509s_dropWhile_s p = L.dropWhile p `eqP` (unpackS . S.unstream . S.dropWhile p)

525510sf_dropWhile q p = (L.dropWhile p . L.filter q) `eqP`

@@ -756,7 +741,6 @@ tb_decimal :: (Integral a, Show a) => a -> Bool

756741tb_decimal = (TB.toLazyText . TB.decimal) `eq` (TL.pack . show)

757742

758743tb_decimal_integer (a::Integer) = tb_decimal a

759−tb_decimal_integer_big (Big a) = tb_decimal a

760744tb_decimal_int (a::Int) = tb_decimal a

761745tb_decimal_int8 (a::Int8) = tb_decimal a

762746tb_decimal_int16 (a::Int16) = tb_decimal a

@@ -768,11 +752,6 @@ tb_decimal_word16 (a::Word16) = tb_decimal a

768752tb_decimal_word32 (a::Word32) = tb_decimal a

769753tb_decimal_word64 (a::Word64) = tb_decimal a

770754

771−tb_decimal_big_int (BigBounded (a::Int)) = tb_decimal a

772−tb_decimal_big_int64 (BigBounded (a::Int64)) = tb_decimal a

773−tb_decimal_big_word (BigBounded (a::Word)) = tb_decimal a

774−tb_decimal_big_word64 (BigBounded (a::Word64)) = tb_decimal a

775−

776755tb_hex :: (Integral a, Show a) => a -> Bool

777756tb_hex = (TB.toLazyText . TB.hexadecimal) `eq` (TL.pack . flip showHex "")

778757

@@ -794,21 +773,6 @@ tb_realfloat = (TB.toLazyText . TB.realFloat) `eq` (TL.pack . show)

794773tb_realfloat_float (a::Float) = tb_realfloat a

795774tb_realfloat_double (a::Double) = tb_realfloat a

796775

797−showFloat :: (RealFloat a) => TB.FPFormat -> Maybe Int -> a -> ShowS

798−showFloat TB.Exponent = showEFloat

799−showFloat TB.Fixed = showFFloat

800−showFloat TB.Generic = showGFloat

801−

802−tb_formatRealFloat :: (RealFloat a, Show a) =>

803− a -> TB.FPFormat -> Precision a -> Property

804−tb_formatRealFloat a fmt prec =

805− TB.formatRealFloat fmt p a ===

806− TB.fromString (showFloat fmt p a "")

807− where p = precision a prec

808−

809−tb_formatRealFloat_float (a::Float) = tb_formatRealFloat a

810−tb_formatRealFloat_double (a::Double) = tb_formatRealFloat a

811−

812776-- Reading.

813777

814778t_decimal (n::Int) s =

@@ -828,7 +792,7 @@ tl_hexadecimal m s ox =

828792 p = if ox then "0x" else ""

829793 n = getPositive m :: Int

830794

831−isFloaty c = c `elem` ("+-.0123456789eE" :: String)

795+isFloaty c = c `elem` "+-.0123456789eE"

832796

833797t_read_rational p tol (n::Double) s =

834798 case p (T.pack (show n) `T.append` t) of

@@ -1110,12 +1074,9 @@ tests =

11101074 ],

11111075

11121076 testGroup "unfolds" [

1113− testProperty "tl_repeat" tl_repeat,

11141077 testProperty "s_replicate" s_replicate,

11151078 testProperty "t_replicate" t_replicate,

11161079 testProperty "tl_replicate" tl_replicate,

1117− testProperty "tl_cycle" tl_cycle,

1118− testProperty "tl_iterate" tl_iterate,

11191080 testProperty "t_unfoldr" t_unfoldr,

11201081 testProperty "tl_unfoldr" tl_unfoldr,

11211082 testProperty "t_unfoldrN" t_unfoldrN,

@@ -1146,8 +1107,6 @@ tests =

11461107 testProperty "sf_takeWhile" sf_takeWhile,

11471108 testProperty "t_takeWhile" t_takeWhile,

11481109 testProperty "tl_takeWhile" tl_takeWhile,

1149− testProperty "t_takeWhileEnd" t_takeWhileEnd,

1150− testProperty "tl_takeWhileEnd" tl_takeWhileEnd,

11511110 testProperty "sf_dropWhile" sf_dropWhile,

11521111 testProperty "s_dropWhile" s_dropWhile,

11531112 testProperty "s_dropWhile_s" s_dropWhile_s,

@@ -1293,16 +1252,11 @@ tests =

12931252 testProperty "tb_decimal_int32" tb_decimal_int32,

12941253 testProperty "tb_decimal_int64" tb_decimal_int64,

12951254 testProperty "tb_decimal_integer" tb_decimal_integer,

1296− testProperty "tb_decimal_integer_big" tb_decimal_integer_big,

12971255 testProperty "tb_decimal_word" tb_decimal_word,

12981256 testProperty "tb_decimal_word8" tb_decimal_word8,

12991257 testProperty "tb_decimal_word16" tb_decimal_word16,

13001258 testProperty "tb_decimal_word32" tb_decimal_word32,

1301− testProperty "tb_decimal_word64" tb_decimal_word64,

1302− testProperty "tb_decimal_big_int" tb_decimal_big_int,

1303− testProperty "tb_decimal_big_word" tb_decimal_big_word,

1304− testProperty "tb_decimal_big_int64" tb_decimal_big_int64,

1305− testProperty "tb_decimal_big_word64" tb_decimal_big_word64

1259+ testProperty "tb_decimal_word64" tb_decimal_word64

13061260 ],

13071261 testGroup "hexadecimal" [

13081262 testProperty "tb_hexadecimal_int" tb_hexadecimal_int,

@@ -1319,9 +1273,7 @@ tests =

13191273 ],

13201274 testGroup "realfloat" [

13211275 testProperty "tb_realfloat_double" tb_realfloat_double,

1322− testProperty "tb_realfloat_float" tb_realfloat_float,

1323− testProperty "tb_formatRealFloat_float" tb_formatRealFloat_float,

1324− testProperty "tb_formatRealFloat_double" tb_formatRealFloat_double

1276+ testProperty "tb_realfloat_float" tb_realfloat_float

13251277 ],

13261278 testProperty "tb_fromText" tb_fromText,

13271279 testProperty "tb_singleton" tb_singleton

tests/Tests/Properties/Mul.hs+2−6

@@ -4,8 +4,8 @@ module Tests.Properties.Mul (tests) where

55import Control.Applicative ((<$>), pure)

66import Control.Exception as E (SomeException, catch, evaluate)

7−import Data.Int (Int32, Int64)

8−import Data.Text.Internal (mul, mul32, mul64)

7+import Data.Int (Int64)

8+import Data.Text.Internal (mul, mul64)

99import System.IO.Unsafe (unsafePerformIO)

1010import Test.Framework (Test)

1111import Test.Framework.Providers.QuickCheck2 (testProperty)

@@ -23,9 +23,6 @@ eval :: (a -> b -> c) -> a -> b -> Maybe c

2323eval f a b = unsafePerformIO $

2424 (Just <$> evaluate (f a b)) `E.catch` (\(_::SomeException) -> pure Nothing)

2525

26−t_mul32 :: Int32 -> Int32 -> Property

27−t_mul32 a b = mulRef a b === eval mul32 a b

28−

2926t_mul64 :: Int64 -> Int64 -> Property

3027t_mul64 a b = mulRef a b === eval mul64 a b

3128

@@ -35,6 +32,5 @@ t_mul a b = mulRef a b === eval mul a b

3532tests :: [Test]

3633tests = [

3734 testProperty "t_mul" t_mul

38− , testProperty "t_mul32" t_mul32

3935 , testProperty "t_mul64" t_mul64

4036 ]

tests/Tests/QuickCheckUtils.hs+82−73

@@ -7,46 +7,48 @@

77module Tests.QuickCheckUtils

88 (

99 genUnicode

10+ , genUnicodeWith

11+ , ascii

12+ , plane0

13+ , plane1

14+ , plane2

15+ , plane14

16+ , planes

17+

1018 , unsquare

1119 , smallArbitrary

1220

13− , BigBounded(..)

14− , BigInt(..)

15− , NotEmpty(..)

21+ , NotEmpty (..)

1622

17− , Small(..)

23+ , Small (..)

1824 , small

1925

20− , Precision(..)

21− , precision

22−

2326 , integralRandomR

2427

25− , DecodeErr(..)

28+ , DecodeErr (..)

2629 , genDecodeErr

2730

28− , Stringy(..)

31+ , Stringy (..)

2932 , eq

3033 , eqP

3134

32− , Encoding(..)

35+ , Encoding (..)

3336

3437 , write_read

3538 ) where

3639

37−import Control.Applicative ((<$>))

3840import Control.Arrow (first, (***))

3941import Control.DeepSeq (NFData (..), deepseq)

4042import Control.Exception (bracket)

43+import Data.Bits ((.&.))

44+import Data.Char (chr)

4145import Data.String (IsString, fromString)

4246import Data.Text.Foreign (I16)

43−import Data.Text.Lazy.Builder.RealFloat (FPFormat(..))

4447import Data.Word (Word8, Word16)

4548import Debug.Trace (trace)

46−import System.Random (Random(..), RandomGen)

47−import Test.QuickCheck hiding (Fixed(..), Small (..), (.&.))

49+import System.Random (Random (..), RandomGen)

50+import Test.QuickCheck hiding (Small (..), (.&.))

4851import Test.QuickCheck.Monadic (assert, monadicIO, run)

49−import Test.QuickCheck.Unicode (string)

5052import Tests.Utils

5153import qualified Data.ByteString as B

5254import qualified Data.Text as T

@@ -58,14 +60,6 @@ import qualified Data.Text.Internal.Lazy.Fusion as TLF

5860import qualified Data.Text.Lazy as TL

5961import qualified System.IO as IO

6062

61−#if !MIN_VERSION_base(4,4,0)

62−import Data.Int (Int64)

63−import Data.Word (Word, Word64)

64−#endif

65−

66−genUnicode :: IsString a => Gen a

67−genUnicode = fromString <$> string

68−

6963instance Random I16 where

7064 randomR = integralRandomR

7165 random = randomR (minBound,maxBound)

@@ -79,23 +73,76 @@ instance Arbitrary B.ByteString where

7973 shrink = map B.pack . shrink . B.unpack

8074

8175#if !MIN_VERSION_base(4,4,0)

82−instance Random Int64 where

83− randomR = integralRandomR

84− random = randomR (minBound,maxBound)

85−

86−instance Random Word where

87− randomR = integralRandomR

88− random = randomR (minBound,maxBound)

89−

9076instance Random Word8 where

9177 randomR = integralRandomR

9278 random = randomR (minBound,maxBound)

93−

94−instance Random Word64 where

95− randomR = integralRandomR

96− random = randomR (minBound,maxBound)

9779#endif

9880

81+genUnicode :: IsString a => Gen a

82+genUnicode = genUnicodeWith planes

83+

84+genUnicodeWith :: IsString a => [Gen Int] -> Gen a

85+genUnicodeWith gens = fmap fromString string

86+ where

87+ string = sized $ \n ->

88+ do k <- choose (0,n)

89+ sequence [ char | _ <- [1..k] ]

90+

91+ excluding :: [a -> Bool] -> Gen a -> Gen a

92+ excluding bad gen = loop

93+ where

94+ loop = do

95+ x <- gen

96+ if or (map ($ x) bad)

97+ then loop

98+ else return x

99+

100+ reserved = [lowSurrogate, highSurrogate, noncharacter]

101+ lowSurrogate c = c >= 0xDC00 && c <= 0xDFFF

102+ highSurrogate c = c >= 0xD800 && c <= 0xDBFF

103+ noncharacter c = masked == 0xFFFE || masked == 0xFFFF

104+ where

105+ masked = c .&. 0xFFFF

106+

107+ char = chr `fmap` excluding reserved (oneof gens)

108+

109+ascii :: Gen Int

110+ascii = choose (0,0x7F)

111+

112+plane0 :: Gen Int

113+plane0 = choose (0xF0, 0xFFFF)

114+

115+plane1 :: Gen Int

116+plane1 = oneof [ choose (0x10000, 0x10FFF)

117+ , choose (0x11000, 0x11FFF)

118+ , choose (0x12000, 0x12FFF)

119+ , choose (0x13000, 0x13FFF)

120+ , choose (0x1D000, 0x1DFFF)

121+ , choose (0x1F000, 0x1FFFF)

122+ ]

123+

124+plane2 :: Gen Int

125+plane2 = oneof [ choose (0x20000, 0x20FFF)

126+ , choose (0x21000, 0x21FFF)

127+ , choose (0x22000, 0x22FFF)

128+ , choose (0x23000, 0x23FFF)

129+ , choose (0x24000, 0x24FFF)

130+ , choose (0x25000, 0x25FFF)

131+ , choose (0x26000, 0x26FFF)

132+ , choose (0x27000, 0x27FFF)

133+ , choose (0x28000, 0x28FFF)

134+ , choose (0x29000, 0x29FFF)

135+ , choose (0x2A000, 0x2AFFF)

136+ , choose (0x2B000, 0x2BFFF)

137+ , choose (0x2F000, 0x2FFFF)

138+ ]

139+

140+plane14 :: Gen Int

141+plane14 = choose (0xE0000, 0xE0FFF)

142+

143+planes :: [Gen Int]

144+planes = [ascii, plane0, plane1, plane2, plane14]

145+

99146-- For tests that have O(n^2) running times or input sizes, resize

100147-- their inputs to the square root of the originals.

101148unsquare :: (Arbitrary a, Show a, Testable b) => (a -> b) -> Property

@@ -113,20 +160,6 @@ instance Arbitrary TL.Text where

113160 arbitrary = (TL.fromChunks . map notEmpty) `fmap` smallArbitrary

114161 shrink = map TL.pack . shrink . TL.unpack

115162

116−newtype BigInt = Big Integer

117− deriving (Eq, Show)

118−

119−instance Arbitrary BigInt where

120− arbitrary = choose (1::Int,200) >>= \e -> Big <$> choose (10^(e-1),10^e)

121− shrink (Big a) = [Big (a `div` 2^(l-e)) | e <- shrink l]

122− where l = truncate (log (fromIntegral a) / log 2 :: Double) :: Integer

123−

124−newtype BigBounded a = BigBounded a

125− deriving (Eq, Show)

126−

127−instance (Bounded a, Random a, Arbitrary a) => Arbitrary (BigBounded a) where

128− arbitrary = BigBounded <$> choose (minBound, maxBound)

129−

130163newtype NotEmpty a = NotEmpty { notEmpty :: a }

131164 deriving (Eq, Ord)

132165

@@ -267,30 +300,6 @@ eqP f g s w = eql "orig" (f s) (g t) &&

267300 | a =^= b = True

268301 | otherwise = trace (d ++ ": " ++ show a ++ " /= " ++ show b) False

269302

270−instance Arbitrary FPFormat where

271− arbitrary = elements [Exponent, Fixed, Generic]

272−

273−newtype Precision a = Precision (Maybe Int)

274− deriving (Eq, Show)

275−

276−precision :: a -> Precision a -> Maybe Int

277−precision _ (Precision prec) = prec

278−

279−arbitraryPrecision :: Int -> Gen (Precision a)

280−arbitraryPrecision maxDigits = Precision <$> do

281− n <- choose (-1,maxDigits)

282− return $ if n == -1

283− then Nothing

284− else Just n

285−

286−instance Arbitrary (Precision Float) where

287− arbitrary = arbitraryPrecision 11

288− shrink = map Precision . shrink . precision undefined

289−

290−instance Arbitrary (Precision Double) where

291− arbitrary = arbitraryPrecision 22

292− shrink = map Precision . shrink . precision undefined

293−

294303-- Work around lack of Show instance for TextEncoding.

295304data Encoding = E String IO.TextEncoding

296305

tests/cabal.config+1

@@ -1,6 +1,7 @@

11-- These flags help to speed up building the test suite.

33documentation: False

4+executable-profiling: False

45executable-stripping: False

56flags: developer

67library-profiling: False

tests/literal-rule-test.shdeleted−29

@@ -1,29 +0,0 @@

1−#!/bin/bash -e

2−

3−failed=0

4−

5−function check_firings() {

6− rule=$1

7− expected=$2

8− build="ghc -O -ddump-rule-firings LiteralRuleTest.hs"

9− build="$build -i.. -I../include"

10− touch LiteralRuleTest.hs

11− echo -n "Want to see $expected firings of rule $rule... " >&2

12− firings=$($build 2>&1 | grep "Rule fired: $rule\$" | wc -l)

13− rm -f LiteralRuleTest.{o.hi}

14−

15− if [ $firings != $expected ]; then

16− echo "failed, saw $firings" >&2

17− failed=1

18− else

19− echo "pass" >&2

20− fi

21−}

22−

23−check_firings "TEXT literal" 8

24−check_firings "TEXT literal UTF8" 7

25−check_firings "TEXT empty literal" 4

26−# This is broken at the moment. "TEXT literal" rule fires instead.

27−#check_firings "TEXT singleton literal" 5

28−

29−exit $failed

tests/text-tests.cabal+1−4

@@ -3,7 +3,7 @@ version: 0.0.0.0

33synopsis: Functional tests for the text package

44description: Functional tests for the text package

55homepage: https://github.com/bos/text

6−license: BSD2

6+license: BSD3

77license-file: ../LICENSE

88author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

99 Bryan O'Sullivan <bos@serpentine.com>,

@@ -51,7 +51,6 @@ executable text-tests

5151 bytestring,

5252 deepseq,

5353 directory,

54− quickcheck-unicode,

5554 random,

5655 test-framework >= 0.4,

5756 test-framework-hunit >= 0.2,

@@ -114,7 +113,6 @@ library

114113 Data.Text.Internal.Lazy.Search

115114 Data.Text.Internal.Private

116115 Data.Text.Read

117− Data.Text.Show

118116 Data.Text.Internal.Read

119117 Data.Text.Internal.Search

120118 Data.Text.Unsafe

@@ -136,7 +134,6 @@ library

136134 build-depends:

137135 array,

138136 base == 4.*,

139− binary,

140137 bytestring,

141138 deepseq,

142139 ghc-prim,

text.cabal+6−24

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

11name: text

2−version: 1.2.2.1

2+version: 1.1.1.3

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

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

55synopsis: An efficient packed Unicode text type.

@@ -31,7 +31,7 @@ description:

3131 the @text-icu@ package:

3232 <http://hackage.haskell.org/package/text-icu>

3333

34−license: BSD2

34+license: BSD3

3535license-file: LICENSE

3636author: Bryan O'Sullivan <bos@serpentine.com>

3737maintainer: Bryan O'Sullivan <bos@serpentine.com>

@@ -64,11 +64,6 @@ extra-source-files:

6464 tests/scripts/*.sh

6565 tests/text-tests.cabal

6666

67−flag bytestring-builder

68− description: Depend on the bytestring-builder package for backwards compatibility.

69− default: False

70− manual: False

71−

7267flag developer

7368 description: operate in developer mode

7469 default: False

@@ -77,7 +72,6 @@ flag developer

7772flag integer-simple

7873 description: Use the simple integer library instead of GMP

7974 default: False

80− manual: False

8175

8276library

8377 c-sources: cbits/cbits.c

@@ -128,21 +122,16 @@ library

128122 Data.Text.Read

129123 Data.Text.Unsafe

130124

131− other-modules:

132− Data.Text.Show

133−

134125 build-depends:

135126 array >= 0.3,

136127 base >= 4.2 && < 5,

137− binary,

138128 deepseq >= 1.1.0.0,

139129 ghc-prim >= 0.2

140130

141− if flag(bytestring-builder)

142− build-depends: bytestring >= 0.9 && < 0.10.4,

143− bytestring-builder >= 0.10.4

131+ if impl(ghc >= 7.7)

132+ build-depends: bytestring >= 0.10.4.0

144133 else

145− build-depends: bytestring >= 0.10.4

134+ build-depends: bytestring >= 0.9

146135

147136 cpp-options: -DHAVE_DEEPSEQ

148137 ghc-options: -Wall -fwarn-tabs -funbox-strict-fields -O2

@@ -176,22 +165,15 @@ test-suite tests

176165 QuickCheck >= 2.7,

177166 array,

178167 base,

179− binary,

168+ bytestring,

180169 deepseq,

181170 directory,

182171 ghc-prim,

183− quickcheck-unicode,

184172 random,

185173 test-framework >= 0.4,

186174 test-framework-hunit >= 0.2,

187175 test-framework-quickcheck2 >= 0.2

188176

189− if flag(bytestring-builder)

190− build-depends: bytestring >= 0.9 && < 0.10.4,

191− bytestring-builder >= 0.10.4

192− else

193− build-depends: bytestring >= 0.10.4

194−

195177 if flag(integer-simple)

196178 cpp-options: -DINTEGER_SIMPLE

197179 build-depends: integer-simple >= 0.1 && < 0.5