master..ac2644b · text

.gitignore−1

since splitsnapshots

unifiedsplit

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

1dist

2cabal-dev

3−cabal.sandbox.config

.hgtags−9

@@ -50,17 +50,8 @@ d99cd091cdf71ce807a4255f6cc509c3154f51ea 0.11.3.1

50fb7299f22449750c653ac64dee4bd63c13ea6e4c 1.1.1.1

51eb170cb533b63534d166b5fd5344b36676adf140 1.1.1.2

52e749e05f715d02445803f653cd3d97d5395d7c6f 1.1.1.3

53−b0dff8ca1142595bab489a768ca62e6a6805aac4 1.1.1.4

54f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

53f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

556231451979b2d33e28038b75f2dab19c9731f5c3 1.2.0.1

546231451979b2d33e28038b75f2dab19c9731f5c3 1.2.0.1

5699a3503b4ad001fc1894e90286e96d7dea1e9195 1.2.0.2

5599a3503b4ad001fc1894e90286e96d7dea1e9195 1.2.0.2

5747f138edf15576b9adaeed64784e30811099f202 1.2.0.3

5647f138edf15576b9adaeed64784e30811099f202 1.2.0.3

580b064b0c69106bbcd13cb0099e72ae5fa2347bf0 1.2.0.4

570b064b0c69106bbcd13cb0099e72ae5fa2347bf0 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+73−78

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

129 , drop

130 , dropEnd

131 , takeWhile

132− , takeWhileEnd

133 , dropWhile

132 , dropWhile

134 , dropWhileEnd

133 , dropWhileEnd

135 , dropAround

134 , dropAround

@@ -192,12 +191,11 @@ module Data.Text

192

191

193 -- * Low level operations

192 -- * Low level operations

194 , copy

193 , copy

195− , unpackCString#

196 ) where

194 ) where

197

195

198import Prelude (Char, Bool(..), Int, Maybe(..), String,

196import Prelude (Char, Bool(..), Int, Maybe(..), String,

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

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

200− Read(..),

198+ Read(..), Show(..),

201 (&&), (||), (+), (-), (.), ($), ($!), (>>),

199 (&&), (||), (+), (-), (.), ($), ($!), (>>),

202 not, return, otherwise, quot)

200 not, return, otherwise, quot)

203#if defined(HAVE_DEEPSEQ)

201#if defined(HAVE_DEEPSEQ)

@@ -213,23 +211,17 @@ import Control.Monad (foldM)

213import Control.Monad.ST (ST)

211import Control.Monad.ST (ST)

214import qualified Data.Text.Array as A

212import qualified Data.Text.Array as A

215import qualified Data.List as L

213import qualified Data.List as L

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

217import Data.Monoid (Monoid(..))

214import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

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

220−#endif

221import Data.String (IsString(..))

215import Data.String (IsString(..))

222import qualified Data.Text.Internal.Fusion as S

216import qualified Data.Text.Internal.Fusion as S

223import qualified Data.Text.Internal.Fusion.Common as S

217import qualified Data.Text.Internal.Fusion.Common as S

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

225import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

218import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

226import Data.Text.Internal.Private (span_)

219import Data.Text.Internal.Private (span_)

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

220+import Data.Text.Internal (Text(..), empty, empty_, firstf, mul, safe, text)

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

229import qualified Prelude as P

221import qualified Prelude as P

230import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

222import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

231 reverseIter_, unsafeHead, unsafeTail)

223 reverseIter_, unsafeHead, unsafeTail)

232−import Data.Text.Internal.Unsafe.Char (unsafeChr)

224+import Data.Text.Internal.Unsafe.Char (unsafeChr, unsafeWrite)

233import qualified Data.Text.Internal.Functions as F

225import qualified Data.Text.Internal.Functions as F

234import qualified Data.Text.Internal.Encoding.Utf16 as U16

226import qualified Data.Text.Internal.Encoding.Utf16 as U16

235import Data.Text.Internal.Search (indices)

227import Data.Text.Internal.Search (indices)

@@ -238,13 +230,15 @@ import Data.ByteString (ByteString)

238import qualified Data.Text.Lazy as L

230import qualified Data.Text.Lazy as L

239import Data.Int (Int64)

231import Data.Int (Int64)

240#endif

232#endif

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

233+#if __GLASGOW_HASKELL__ >= 702

234+import qualified GHC.CString as GHC

235+#else

236+import qualified GHC.Base as GHC

237+#endif

242#if __GLASGOW_HASKELL__ >= 708

238#if __GLASGOW_HASKELL__ >= 708

243import qualified GHC.Exts as Exts

239import qualified GHC.Exts as Exts

244#endif

240#endif

245−#if MIN_VERSION_base(4,7,0)

241+import GHC.Prim (Addr#)

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

247−#endif

248

242

249-- $strict

243-- $strict

250--

244--

@@ -330,24 +324,15 @@ instance Eq Text where

330instance Ord Text where

324instance Ord Text where

331 compare = compareText

325 compare = compareText

332

326

327+instance Show Text where

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

329+

333instance Read Text where

330instance Read Text where

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

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

335

332

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−

344instance Monoid Text where

333instance Monoid Text where

345 mempty = empty

334 mempty = empty

346−#if MIN_VERSION_base(4,9,0)

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

348−#else

349 mappend = append

335 mappend = append

350−#endif

351 mconcat = concat

336 mconcat = concat

352

337

353instance IsString Text where

338instance IsString Text where

@@ -364,14 +349,6 @@ instance Exts.IsList Text where

364instance NFData Text where rnf !_ = ()

349instance NFData Text where rnf !_ = ()

365#endif

350#endif

366

351

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−

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

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

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

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

377--

354--

@@ -395,12 +372,6 @@ instance Data Text where

395 _ -> P.error "gunfold"

372 _ -> P.error "gunfold"

396 dataTypeOf _ = textDataType

373 dataTypeOf _ = textDataType

397

374

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−

404packConstr :: Constr

375packConstr :: Constr

405packConstr = mkConstr textDataType "pack" [] Prefix

376packConstr = mkConstr textDataType "pack" [] Prefix

406

377

@@ -430,6 +401,54 @@ pack :: String -> Text

430pack = unstream . S.map safe . S.streamList

401pack = unstream . S.map safe . S.streamList

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

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

432

403

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

405+unpack :: Text -> String

406+unpack = S.unstreamList . stream

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

408+

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

410+unpackCString# :: Addr# -> Text

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

412+{-# NOINLINE unpackCString# #-}

413+

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

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

416+ = unpackCString# a #-}

417+

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

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

420+ = unpackCString# a #-}

421+

422+{-# RULES "TEXT empty literal"

423+ unstream (S.map safe (S.streamList []))

424+ = empty_ #-}

425+

426+{-# RULES "TEXT singleton literal" forall a.

427+ unstream (S.map safe (S.streamList [a]))

428+ = singleton_ a #-}

429+

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

431+-- Performs replacement on invalid scalar values.

432+singleton :: Char -> Text

433+singleton = unstream . S.singleton . safe

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

435+

436+{-# RULES "TEXT singleton" forall a.

437+ unstream (S.singleton (safe a))

438+ = singleton_ a #-}

439+

440+-- This is intended to reduce inlining bloat.

441+singleton_ :: Char -> Text

442+singleton_ c = Text (A.run x) 0 len

443+ where x :: ST s (A.MArray s)

444+ x = do arr <- A.new len

445+ _ <- unsafeWrite arr 0 d

446+ return arr

447+ len | d < '\x10000' = 1

448+ | otherwise = 2

449+ d = safe c

450+{-# NOINLINE singleton_ #-}

451+

433-- -----------------------------------------------------------------------------

452-- -----------------------------------------------------------------------------

434-- * Basic functions

453-- * Basic functions

435

454

@@ -486,8 +505,8 @@ head t = S.head (stream t)

486uncons :: Text -> Maybe (Char, Text)

505uncons :: Text -> Maybe (Char, Text)

487uncons t@(Text arr off len)

506uncons t@(Text arr off len)

488 | len <= 0 = Nothing

507 | len <= 0 = Nothing

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

490− in (c, text arr (off+d) (len-d))

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

509+ where Iter c d = iter t 0

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

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

492

511

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

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

@@ -572,9 +591,7 @@ isSingleton = S.isSingleton . stream

572-- Subject to fusion.

591-- Subject to fusion.

573length :: Text -> Int

592length :: Text -> Int

574length t = S.length (stream t)

593length t = S.length (stream t)

575−{-# INLINE [0] length #-}

594+{-# INLINE 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.

578

595

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

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

580-- Subject to fusion.

597-- Subject to fusion.

@@ -593,32 +610,32 @@ compareLength t n = S.compareLengthI (stream t) n

593

610

594{-# RULES

611{-# RULES

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

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

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

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

597 #-}

614 #-}

598

615

599{-# RULES

616{-# RULES

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

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

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

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

602 #-}

619 #-}

603

620

604{-# RULES

621{-# RULES

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

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

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

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

607 #-}

624 #-}

608

625

609{-# RULES

626{-# RULES

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

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

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

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

612 #-}

629 #-}

613

630

614{-# RULES

631{-# RULES

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

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

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

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

617 #-}

634 #-}

618

635

619{-# RULES

636{-# RULES

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

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

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

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

622 #-}

639 #-}

623

640

624-- -----------------------------------------------------------------------------

641-- -----------------------------------------------------------------------------

@@ -915,13 +932,13 @@ concatMap f = concat . foldr ((:) . f) []

915{-# INLINE concatMap #-}

932{-# INLINE concatMap #-}

916

933

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

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

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

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

919any :: (Char -> Bool) -> Text -> Bool

936any :: (Char -> Bool) -> Text -> Bool

920any p t = S.any p (stream t)

937any p t = S.any p (stream t)

921{-# INLINE any #-}

938{-# INLINE any #-}

922

939

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

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

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

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

925all :: (Char -> Bool) -> Text -> Bool

942all :: (Char -> Bool) -> Text -> Bool

926all p t = S.all p (stream t)

943all p t = S.all p (stream t)

927{-# INLINE all #-}

944{-# INLINE all #-}

@@ -1152,27 +1169,6 @@ takeWhile p t@(Text arr off len) = loop 0

1152 unstream (S.takeWhile p (stream t)) = takeWhile p t

1169 unstream (S.takeWhile p (stream t)) = takeWhile p t

1153 #-}

1170 #-}

1154

1171

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−

1176-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

1172-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

1177-- 'takeWhile' @p@ @t@. Subject to fusion.

1173-- 'takeWhile' @p@ @t@. Subject to fusion.

1178dropWhile :: (Char -> Bool) -> Text -> Text

1174dropWhile :: (Char -> Bool) -> Text -> Text

@@ -1191,9 +1187,8 @@ dropWhile p t@(Text arr off len) = loop 0 0

1191 #-}

1187 #-}

1192

1188

1193-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1189-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

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

1190+-- dropping characters that fail the predicate @p@ from the end of

1195-- @t@. Subject to fusion.

1191-- @t@. Subject to fusion.

1196−--

1197-- Examples:

1192-- Examples:

1198--

1193--

1199-- > dropWhileEnd (=='.') "foo..." == "foo"

1194-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1213,7 +1208,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len

1213 #-}

1208 #-}

1214

1209

1215-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1210-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

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

1211+-- dropping characters that fail the predicate @p@ from both the

1217-- beginning and end of @t@. Subject to fusion.

1212-- beginning and end of @t@. Subject to fusion.

1218dropAround :: (Char -> Bool) -> Text -> Text

1213dropAround :: (Char -> Bool) -> Text -> Text

1219dropAround p = dropWhile p . dropWhileEnd p

1214dropAround p = dropWhile p . dropWhileEnd p

Data/Text/Array.hs+4−5

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

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

21--

22-- 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

24-- naming.

25module Data.Text.Array

26 (

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

74#endif

75import GHC.Base (ByteArray#, MutableByteArray#, Int(..),

76 indexWord16Array#, newByteArray#,

77− unsafeFreezeByteArray#, writeWord16Array#)

77+ unsafeCoerce#, writeWord16Array#)

78import GHC.ST (ST(..), runST)

79import GHC.Word (Word16(..))

80import Prelude hiding (length, read)

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

130

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

132unsafeFreeze :: MArray s -> ST s Array

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

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

134− case unsafeFreezeByteArray# maBA s1# of

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

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

136#if defined(ASSERTS)

135#if defined(ASSERTS)

137 maLen

136 maLen

138#endif

137#endif

Data/Text/Encoding.hs+20−9

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

55 , encodeUtf32LE

56 , encodeUtf32BE

57

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

58 -- * Encoding Text using ByteString Builders

59 -- * Encoding Text using ByteString Builders

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

61+ -- @bytestring >= 0.10.4.0@.

59 , encodeUtf8Builder

62 , encodeUtf8Builder

60 , encodeUtf8BuilderEscaped

63 , encodeUtf8BuilderEscaped

64+#endif

61 ) where

65 ) where

62

66

63#if __GLASGOW_HASKELL__ >= 702

67#if __GLASGOW_HASKELL__ >= 702

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

66import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

70import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

67#endif

71#endif

68

72

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+

69import Control.Exception (evaluate, try)

83import Control.Exception (evaluate, try)

70import Control.Monad.ST (runST)

84import Control.Monad.ST (runST)

71−import Data.Bits ((.&.))

72import Data.ByteString as B

85import Data.ByteString as B

73import Data.ByteString.Internal as B hiding (c2w)

86import Data.ByteString.Internal as B hiding (c2w)

87+import Data.Text ()

74import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

88import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

75import Data.Text.Internal (Text(..), safe, text)

89import Data.Text.Internal (Text(..), safe, text)

76import Data.Text.Internal.Private (runText)

90import Data.Text.Internal.Private (runText)

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

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

78import Data.Text.Internal.Unsafe.Shift (shiftR)

92import Data.Text.Internal.Unsafe.Shift (shiftR)

79−import Data.Text.Show ()

80import Data.Text.Unsafe (unsafeDupablePerformIO)

93import Data.Text.Unsafe (unsafeDupablePerformIO)

81import Data.Word (Word8, Word32)

94import Data.Word (Word8, Word32)

82import Foreign.C.Types (CSize(..))

95import Foreign.C.Types (CSize(..))

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

85import Foreign.Ptr (Ptr, minusPtr, nullPtr, plusPtr)

98import Foreign.Ptr (Ptr, minusPtr, nullPtr, plusPtr)

86import Foreign.Storable (Storable, peek, poke)

99import Foreign.Storable (Storable, peek, poke)

87import GHC.Base (ByteArray#, MutableByteArray#)

100import 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

92import qualified Data.Text.Array as A

101import qualified Data.Text.Array as A

93import qualified Data.Text.Internal.Encoding.Fusion as E

102import qualified Data.Text.Internal.Encoding.Fusion as E

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

95import qualified Data.Text.Internal.Fusion as F

103import qualified Data.Text.Internal.Fusion as F

96

104

97#include "text_cbits.h"

105#include "text_cbits.h"

@@ -278,7 +286,7 @@ streamDecodeUtf8With onErr = decodeChunk B.empty 0 0

278 return $! text arr 0 (fromIntegral n)

286 return $! text arr 0 (fromIntegral n)

279 lastPtr <- peek curPtrPtr

287 lastPtr <- peek curPtrPtr

280 let left = lastPtr `minusPtr` curPtr

288 let left = lastPtr `minusPtr` curPtr

281− !undecoded = case state of

289+ undecoded = case state of

282 UTF8_ACCEPT -> B.empty

290 UTF8_ACCEPT -> B.empty

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

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

284 return $ Some chunkText undecoded

292 return $ Some chunkText undecoded

@@ -307,6 +315,8 @@ decodeUtf8' :: ByteString -> Either UnicodeException Text

307decodeUtf8' = unsafeDupablePerformIO . try . evaluate . decodeUtf8With strictDecode

315decodeUtf8' = unsafeDupablePerformIO . try . evaluate . decodeUtf8With strictDecode

308{-# INLINE decodeUtf8' #-}

316{-# INLINE decodeUtf8' #-}

309

317

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

319+

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

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

311encodeUtf8Builder :: Text -> B.Builder

321encodeUtf8Builder :: Text -> B.Builder

312encodeUtf8Builder = encodeUtf8BuilderEscaped (BP.liftFixedToBounded BP.word8)

322encodeUtf8Builder = encodeUtf8BuilderEscaped (BP.liftFixedToBounded BP.word8)

@@ -367,6 +377,7 @@ encodeUtf8BuilderEscaped be =

367 outerLoop i (B.BufferRange op ope)

377 outerLoop i (B.BufferRange op ope)

368 where

378 where

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

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

380+#endif

370

381

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

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

372encodeUtf8 :: Text -> ByteString

383encodeUtf8 :: Text -> ByteString

Data/Text/Encoding/Error.hs+1−3

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

1{-# LANGUAGE CPP, DeriveDataTypeable #-}

2−#if __GLASGOW_HASKELL__ >= 704

2+#if __GLASGOW_HASKELL__ >= 702

3−{-# LANGUAGE Safe #-}

4−#elif __GLASGOW_HASKELL__ >= 702

5{-# LANGUAGE Trustworthy #-}

3{-# LANGUAGE Trustworthy #-}

6#endif

4#endif

7-- |

5-- |

Data/Text/Internal/Builder.hs+2−17

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

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

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

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

10--

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

12-- Stability : experimental

@@ -59,9 +59,6 @@ module Data.Text.Internal.Builder

59

60import Control.Monad.ST (ST, runST)

61import Data.Monoid (Monoid(..))

62−#if MIN_VERSION_base(4,9,0)

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

64−#endif

65import Data.Text.Internal (Text(..))

62import Data.Text.Internal (Text(..))

66import Data.Text.Internal.Lazy (smallChunkSize)

63import Data.Text.Internal.Lazy (smallChunkSize)

67import Data.Text.Unsafe (inlineInterleaveST)

64import Data.Text.Unsafe (inlineInterleaveST)

@@ -92,22 +89,12 @@ newtype Builder = Builder {

92 -> ST s [S.Text]

89 -> ST s [S.Text]

93 }

90 }

94

91

95−#if MIN_VERSION_base(4,9,0)

96−instance Semigroup Builder where

97− (<>) = append

98− {-# INLINE (<>) #-}

99−#endif

100−

101instance Monoid Builder where

92instance Monoid Builder where

102 mempty = empty

93 mempty = empty

103 {-# INLINE mempty #-}

94 {-# INLINE mempty #-}

104−#if MIN_VERSION_base(4,9,0)

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

106−#else

107 mappend = append

95 mappend = append

108−#endif

109 {-# INLINE mappend #-}

96 {-# INLINE mappend #-}

110− mconcat = foldr mappend Data.Monoid.mempty

97+ mconcat = foldr mappend mempty

111 {-# INLINE mconcat #-}

98 {-# INLINE mconcat #-}

112

99

113instance String.IsString Builder where

100instance String.IsString Builder where

@@ -242,8 +229,6 @@ flush = Builder $ \ k buf@(Buffer p o u l) ->

242 !t = Text arr o u

229 !t = Text arr o u

243 ts <- inlineInterleaveST (k b)

230 ts <- inlineInterleaveST (k b)

244 return $! t : ts

231 return $! t : ts

245−{-# INLINE [1] flush #-}

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

247

232

248------------------------------------------------------------------------

233------------------------------------------------------------------------

249

234

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

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

23

24import Data.Monoid (mappend)

25import Data.Text.Lazy.Builder (Builder)

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

26+import GHC.Base

27−import Prelude ()

28

27

29-- | Unsafe conversion for decimal digits.

28-- | Unsafe conversion for decimal digits.

30{-# INLINE i2d #-}

29{-# INLINE i2d #-}

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

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

2

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

5−-- License: BSD-style

5+-- License: BSD3

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

7-- Stability: experimental

8-- Portability: portable

Data/Text/Internal/Fusion.hs+17−27

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

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

99unstream :: Stream Char -> Text

100unstream (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

106 arr0 <- A.new mlen

102 arr0 <- A.new mlen

107− let outer !arr !maxi = encode

103+ let outer arr top = loop

108 where

104 where

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

105+ loop !s !i =

110− encode !si !di =

111− case next0 si of

106+ case next0 s of

112− Done -> done arr di

107+ Done -> done arr i

113− Skip si' -> encode si' di

108+ Skip s' -> loop s' i

114− Yield c si'

109+ Yield x s'

115− -- simply check for the worst case

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

116− | maxi < di + 1 -> realloc si di

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

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

112+ arr' <- A.new top'

126− A.copyM arr' 0 arr 0 di

113+ A.copyM arr' 0 arr 0 top

127− outer arr' (newlen - 1) si di

114+ outer arr' top' s i

128−

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

116+ loop s' (i+d)

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

118+ | otherwise = i + 1

129− outer arr0 (mlen - 1) s0 0

119+ outer arr0 mlen s0 0

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

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

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

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

132

122

Data/Text/Internal/Fusion/CaseMapping.hs+15−287

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

1{-# LANGUAGE Rank2Types #-}

2-- AUTOMATICALLY GENERATED - DO NOT EDIT

3−-- Generated by scripts/CaseMapping.hs

3+-- Generated by scripts/SpecialCasing.hs

4−-- CaseFolding-8.0.0.txt

4+-- CaseFolding-6.3.0.txt

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

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

6−-- SpecialCasing-8.0.0.txt

6+-- SpecialCasing-6.3.0.txt

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

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

8

9module Data.Text.Internal.Fusion.CaseMapping where

10import Data.Char

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

359foldMapping '\x03f5' s = Yield '\x03b5' (CC s '\x0000' '\x0000')

360-- ARMENIAN SMALL LIGATURE ECH YIWN

361foldMapping '\x0587' s = Yield '\x0565' (CC s '\x0582' '\x0000')

362−-- CHEROKEE SMALL LETTER YE

362+-- GEORGIAN CAPITAL LETTER YN

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

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

364−-- CHEROKEE SMALL LETTER YI

364+-- GEORGIAN CAPITAL LETTER AEN

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

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

374-- LATIN SMALL LETTER H WITH LINE BELOW

366-- LATIN SMALL LETTER H WITH LINE BELOW

375foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

367foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

376-- LATIN SMALL LETTER T WITH DIAERESIS

368-- LATIN SMALL LETTER T WITH DIAERESIS

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

545foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

537foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

546-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

538-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

547foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

539foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

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

540+-- COPTIC CAPITAL LETTER BOHAIRIC KHEI

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

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

550−-- LATIN CAPITAL LETTER CHI

542+-- LATIN CAPITAL LETTER C WITH BAR

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')

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

554−-- LATIN CAPITAL LETTER OMEGA

544+-- LATIN CAPITAL LETTER H WITH HOOK

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

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

716-- LATIN SMALL LIGATURE FF

546-- LATIN SMALL LIGATURE FF

717foldMapping '\xfb00' s = Yield '\x0066' (CC s '\x0066' '\x0000')

547foldMapping '\xfb00' s = Yield '\x0066' (CC s '\x0066' '\x0000')

718-- LATIN SMALL LIGATURE FI

548-- LATIN SMALL LIGATURE FI

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

737foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

567foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

738-- ARMENIAN SMALL LIGATURE MEN XEH

568-- ARMENIAN SMALL LIGATURE MEN XEH

739foldMapping '\xfb17' s = Yield '\x0574' (CC s '\x056d' '\x0000')

569foldMapping '\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')

842foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

570foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

Data/Text/Internal/Fusion/Common.hs+31−5

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

71 -- ** Scans

72 , scanl

73

74+ -- ** Accumulating maps

75+ -- , mapAccumL

76+

74 -- ** Generation and unfolding

77 -- ** Generation and unfolding

75 , replicateCharI

78 , replicateCharI

76 , replicateI

79 , replicateI

@@ -593,7 +596,7 @@ concatMap f = foldr (append . f) empty

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

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

594

597

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

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

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

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

597any :: (Char -> Bool) -> Stream Char -> Bool

600any :: (Char -> Bool) -> Stream Char -> Bool

598any p (Stream next0 s0 _len) = loop_any s0

601any p (Stream next0 s0 _len) = loop_any s0

599 where

602 where

@@ -605,7 +608,7 @@ any p (Stream next0 s0 _len) = loop_any s0

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

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

606

609

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

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

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

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

609all :: (Char -> Bool) -> Stream Char -> Bool

612all :: (Char -> Bool) -> Stream Char -> Bool

610all p (Stream next0 s0 _len) = loop_all s0

613all p (Stream next0 s0 _len) = loop_all s0

611 where

614 where

@@ -665,16 +668,39 @@ scanl f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT may

665 Done -> Done

668 Done -> Done

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

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

667

670

671+-- -----------------------------------------------------------------------------

672+-- ** Accumulating maps

673+

674+{-

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

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

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

678+--

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

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

681+-- streams precludes it.

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

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

684+ where

685+ {-# INLINE next #-}

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

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

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

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

690+ Done -> Done

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

692+-}

693+

668-- -----------------------------------------------------------------------------

694-- -----------------------------------------------------------------------------

669-- ** Generating and unfolding streams

695-- ** Generating and unfolding streams

670

696

671replicateCharI :: Integral a => a -> Char -> Stream Char

697replicateCharI :: Integral a => a -> Char -> Stream Char

672−replicateCharI !n !c

698+replicateCharI n c

673 | n < 0 = empty

699 | n < 0 = empty

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

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

675 where

701 where

676− next !i | i >= n = Done

702+ next i | i >= n = Done

677− | otherwise = Yield c (i + 1)

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

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

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

679

705

680data RI s = RI !s {-# UNPACK #-} !Int64

706data RI s = RI !s {-# UNPACK #-} !Int64

Data/Text/Internal/Read.hs+3−3

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

7-- Stability : experimental

8-- Portability : GHC

9--

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

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

11module Data.Text.Internal.Read

12 (

13 IReader

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

18 , perhaps

19 ) where

20

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

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

22import Control.Arrow (first)

23import Control.Monad (ap)

24import Data.Char (ord)

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

38 (<*>) = ap

39

40instance Monad (IParser t) where

41− return = App.pure

41+ return = pure

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

43 Left err -> Left err

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

Data/Text/Lazy.hs+12−76

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

140 , drop

141 , dropEnd

142 , takeWhile

143− , takeWhileEnd

144 , dropWhile

143 , dropWhile

145 , dropWhileEnd

144 , dropWhileEnd

146 , dropAround

145 , dropAround

@@ -211,13 +210,9 @@ import Control.DeepSeq (NFData(..))

211import Data.Int (Int64)

210import Data.Int (Int64)

212import qualified Data.List as L

211import qualified Data.List as L

213import Data.Char (isSpace)

212import Data.Char (isSpace)

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

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

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

214+import Data.Data (mkNoRepType)

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

217import Data.Monoid (Monoid(..))

215import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

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

220−#endif

221import Data.String (IsString(..))

216import Data.String (IsString(..))

222import qualified Data.Text as T

217import qualified Data.Text as T

223import qualified Data.Text.Internal as T

218import qualified Data.Text.Internal as T

@@ -229,7 +224,6 @@ import Data.Text.Internal.Lazy.Fusion (stream, unstream)

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

224import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldlChunks,

230 foldrChunks, smallChunkSize)

225 foldrChunks, smallChunkSize)

231import Data.Text.Internal (firstf, safe, text)

226import Data.Text.Internal (firstf, safe, text)

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

233import qualified Data.Text.Internal.Functions as F

227import qualified Data.Text.Internal.Functions as F

234import Data.Text.Internal.Lazy.Search (indices)

228import Data.Text.Internal.Lazy.Search (indices)

235#if __GLASGOW_HASKELL__ >= 702

229#if __GLASGOW_HASKELL__ >= 702

@@ -241,9 +235,6 @@ import qualified GHC.Base as GHC

241import qualified GHC.Exts as Exts

235import qualified GHC.Exts as Exts

242#endif

236#endif

243import GHC.Prim (Addr#)

237import GHC.Prim (Addr#)

244−#if MIN_VERSION_base(4,7,0)

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

246−#endif

247

238

248-- $fusion

239-- $fusion

249--

240--

@@ -341,21 +332,9 @@ instance Show Text where

341instance Read Text where

332instance Read Text where

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

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

343

334

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−

352instance Monoid Text where

335instance Monoid Text where

353 mempty = empty

336 mempty = empty

354−#if MIN_VERSION_base(4,9,0)

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

356−#else

357 mappend = append

337 mappend = append

358−#endif

359 mconcat = concat

338 mconcat = concat

360

339

361instance IsString Text where

340instance IsString Text where

@@ -374,38 +353,11 @@ instance NFData Text where

374 rnf (Chunk _ ts) = rnf ts

353 rnf (Chunk _ ts) = rnf ts

375#endif

354#endif

376

355

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'

390instance Data Text where

356instance Data Text where

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

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

392− toConstr _ = packConstr

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

393− gunfold k z c = case constrIndex c of

394− 1 -> k (z pack)

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

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

396− dataTypeOf _ = textDataType

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

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]

409

361

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

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

411--

363--

@@ -545,7 +497,7 @@ tail Empty = emptyError "tail"

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

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

546 #-}

498 #-}

547

499

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

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

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

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

550init :: Text -> Text

502init :: Text -> Text

551init (Chunk t0 ts0) = go t0 ts0

503init (Chunk t0 ts0) = go t0 ts0

@@ -581,7 +533,7 @@ isSingleton :: Text -> Bool

581isSingleton = S.isSingleton . stream

533isSingleton = S.isSingleton . stream

582{-# INLINE isSingleton #-}

534{-# INLINE isSingleton #-}

583

535

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

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

585-- non-empty. Subject to fusion.

537-- non-empty. Subject to fusion.

586last :: Text -> Char

538last :: Text -> Char

587last Empty = emptyError "last"

539last Empty = emptyError "last"

@@ -881,13 +833,13 @@ concatMap f = concat . foldr ((:) . f) []

881{-# INLINE concatMap #-}

833{-# INLINE concatMap #-}

882

834

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

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

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

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

885any :: (Char -> Bool) -> Text -> Bool

837any :: (Char -> Bool) -> Text -> Bool

886any p t = S.any p (stream t)

838any p t = S.any p (stream t)

887{-# INLINE any #-}

839{-# INLINE any #-}

888

840

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

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

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

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

891all :: (Char -> Bool) -> Text -> Bool

843all :: (Char -> Bool) -> Text -> Bool

892all p t = S.all p (stream t)

844all p t = S.all p (stream t)

893{-# INLINE all #-}

845{-# INLINE all #-}

@@ -996,7 +948,7 @@ cycle t = let t' = foldrChunks Chunk t' t

996

948

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

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

998-- of @f@ to @x@:

950-- of @f@ to @x@:

999−--

951+--

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

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

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

953iterate :: (Char -> Char) -> Char -> Text

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

954iterate f c = let t c' = Chunk (T.singleton c') (t (f c'))

@@ -1144,20 +1096,6 @@ takeWhile p t0 = takeWhile' t0

1144"LAZY TEXT takeWhile -> unfused" [1] forall p t.

1096"LAZY TEXT takeWhile -> unfused" [1] forall p t.

1145 unstream (S.takeWhile p (stream t)) = takeWhile p t

1097 unstream (S.takeWhile p (stream t)) = takeWhile p t

1146 #-}

1098 #-}

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 #-}

1161

1099

1162-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

1100-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

1163-- 'takeWhile' @p@ @t@. Subject to fusion.

1101-- 'takeWhile' @p@ @t@. Subject to fusion.

@@ -1176,11 +1114,9 @@ dropWhile p t0 = dropWhile' t0

1176"LAZY TEXT dropWhile -> unfused" [1] forall p t.

1114"LAZY TEXT dropWhile -> unfused" [1] forall p t.

1177 unstream (S.dropWhile p (stream t)) = dropWhile p t

1115 unstream (S.dropWhile p (stream t)) = dropWhile p t

1178 #-}

1116 #-}

1179−

1180-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

1117-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

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

1118+-- dropping characters that fail the predicate @p@ from the end of

1182-- @t@.

1119-- @t@.

1183−--

1184-- Examples:

1120-- Examples:

1185--

1121--

1186-- > dropWhileEnd (=='.') "foo..." == "foo"

1122-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1197,7 +1133,7 @@ dropWhileEnd p = go

1197{-# INLINE dropWhileEnd #-}

1133{-# INLINE dropWhileEnd #-}

1198

1134

1199-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

1135-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

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

1136+-- dropping characters that fail the predicate @p@ from both the

1201-- beginning and end of @t@. Subject to fusion.

1137-- beginning and end of @t@. Subject to fusion.

1202dropAround :: (Char -> Bool) -> Text -> Text

1138dropAround :: (Char -> Bool) -> Text -> Text

1203dropAround p = dropWhile p . dropWhileEnd p

1139dropAround p = dropWhile p . dropWhileEnd p

Data/Text/Lazy/Builder.hs+1−1

@@ -8,7 +8,7 @@

8-- Module : Data.Text.Lazy.Builder

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

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

12--

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

14-- Stability : experimental

Data/Text/Lazy/Builder/Int.hs+4−4

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

7-- Module: Data.Text.Lazy.Builder.Int

10−-- License: BSD-style

10+-- License: BSD3

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

12-- Stability: experimental

13-- Portability: portable

@@ -35,7 +35,7 @@ import Control.Monad.ST

35

36#ifdef __GLASGOW_HASKELL__

37# if defined(INTEGER_GMP)

38−import GHC.Integer.GMP.Internals (Integer(S#))

38+import GHC.Integer.GMP.Internals

39# elif defined(INTEGER_SIMPLE)

40import GHC.Integer

41# else

@@ -55,7 +55,7 @@ decimal :: Integral a => a -> Builder

55{-# RULES "decimal/Int16" decimal = boundedDecimal :: Int16 -> Builder #-}

56{-# RULES "decimal/Int32" decimal = boundedDecimal :: Int32 -> Builder #-}

57{-# RULES "decimal/Int64" decimal = boundedDecimal :: Int64 -> Builder #-}

58−{-# RULES "decimal/Word" decimal = positive :: Data.Word.Word -> Builder #-}

58+{-# RULES "decimal/Word" decimal = positive :: Word -> Builder #-}

59{-# RULES "decimal/Word8" decimal = positive :: Word8 -> Builder #-}

60{-# RULES "decimal/Word16" decimal = positive :: Word16 -> Builder #-}

61{-# RULES "decimal/Word32" decimal = positive :: Word32 -> Builder #-}

@@ -247,7 +247,7 @@ integer base i

247 PAIR(x,y) -> pblock q <> pblock r <> putB ns

248 where q = fromInteger x

249 r = fromInteger y

250− putB _ = Data.Monoid.mempty

250+ putB _ = mempty

251

252 int :: Int -> Builder

253 int x | base == 10 = decimal x

Data/Text/Lazy/Encoding.hs+18−7

@@ -46,25 +46,30 @@ module Data.Text.Lazy.Encoding

46 , encodeUtf32LE

47 , encodeUtf32BE

48

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

49 -- * Encoding Text using ByteString Builders

50 -- * Encoding Text using ByteString Builders

50 , encodeUtf8Builder

51 , encodeUtf8Builder

51 , encodeUtf8BuilderEscaped

52 , encodeUtf8BuilderEscaped

53+#endif

52 ) where

54 ) where

53

55

54import Control.Exception (evaluate, try)

56import Control.Exception (evaluate, try)

55−import Data.Monoid (Monoid(..))

56import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

57import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

57import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldrChunks)

58import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldrChunks)

58−import Data.Word (Word8)

59import 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

63import qualified Data.ByteString.Lazy as B

60import qualified Data.ByteString.Lazy as B

64import qualified Data.ByteString.Lazy.Internal as B

61import qualified Data.ByteString.Lazy.Internal as B

65import qualified Data.ByteString.Unsafe as B

62import qualified Data.ByteString.Unsafe as B

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

64+import Data.Word (Word8)

65+import Data.Monoid (mappend, 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

66import qualified Data.Text as T

69import qualified Data.Text as T

70+#endif

67import qualified Data.Text.Encoding as TE

71import qualified Data.Text.Encoding as TE

72+import qualified Data.Text.Lazy as L

68import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

73import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

69import qualified Data.Text.Internal.Lazy.Fusion as F

74import qualified Data.Text.Internal.Lazy.Fusion as F

70import Data.Text.Unsafe (unsafeDupablePerformIO)

75import Data.Text.Unsafe (unsafeDupablePerformIO)

@@ -105,7 +110,7 @@ decodeUtf8With onErr (B.Chunk b0 bs0) =

105 | S.null l = empty

110 | S.null l = empty

106 | otherwise = case onErr desc (Just (B.unsafeHead l)) of

111 | otherwise = case onErr desc (Just (B.unsafeHead l)) of

107 Nothing -> empty

112 Nothing -> empty

108− Just c -> Chunk (T.singleton c) Empty

113+ Just c -> L.singleton c

109 desc = "Data.Text.Lazy.Encoding.decodeUtf8With: Invalid UTF-8 stream"

114 desc = "Data.Text.Lazy.Encoding.decodeUtf8With: Invalid UTF-8 stream"

110decodeUtf8With _ _ = empty

115decodeUtf8With _ _ = empty

111

116

@@ -142,6 +147,7 @@ decodeUtf8' bs = unsafeDupablePerformIO $ do

142{-# INLINE decodeUtf8' #-}

147{-# INLINE decodeUtf8' #-}

143

148

144encodeUtf8 :: Text -> B.ByteString

149encodeUtf8 :: Text -> B.ByteString

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

145encodeUtf8 Empty = B.empty

151encodeUtf8 Empty = B.empty

146encodeUtf8 lt@(Chunk t _) =

152encodeUtf8 lt@(Chunk t _) =

147 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

153 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

@@ -156,13 +162,18 @@ encodeUtf8 lt@(Chunk t _) =

156

162

157encodeUtf8Builder :: Text -> B.Builder

163encodeUtf8Builder :: Text -> B.Builder

158encodeUtf8Builder =

164encodeUtf8Builder =

159− foldrChunks (\c b -> TE.encodeUtf8Builder c `mappend` b) Data.Monoid.mempty

165+ foldrChunks (\c b -> TE.encodeUtf8Builder c `mappend` b) mempty

160

166

161{-# INLINE encodeUtf8BuilderEscaped #-}

167{-# INLINE encodeUtf8BuilderEscaped #-}

162encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

168encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

163encodeUtf8BuilderEscaped prim =

169encodeUtf8BuilderEscaped prim =

164 foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c `mappend` b) mempty

170 foldrChunks (\c b -> TE.encodeUtf8BuilderEscaped prim c `mappend` b) mempty

165

171

172+#else

173+encodeUtf8 (Chunk c cs) = B.Chunk (TE.encodeUtf8 c) (encodeUtf8 cs)

174+encodeUtf8 Empty = B.Empty

175+#endif

176+

166-- | Decode text from little endian UTF-16 encoding.

177-- | Decode text from little endian UTF-16 encoding.

167decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

178decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

168decodeUtf16LEWith onErr bs = F.unstream (E.streamUtf16LE onErr bs)

179decodeUtf16LEWith onErr bs = F.unstream (E.streamUtf16LE onErr bs)

Data/Text/Lazy/Read.hs+2−4

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

1{-# LANGUAGE OverloadedStrings, CPP #-}

2−#if __GLASGOW_HASKELL__ >= 704

2+#if __GLASGOW_HASKELL__ >= 702

3−{-# LANGUAGE Safe #-}

4−#elif __GLASGOW_HASKELL__ >= 702

5{-# LANGUAGE Trustworthy #-}

3{-# LANGUAGE Trustworthy #-}

6#endif

4#endif

7

5

@@ -56,7 +54,7 @@ decimal :: Integral a => Reader a

56{-# SPECIALIZE decimal :: Reader Int32 #-}

54{-# SPECIALIZE decimal :: Reader Int32 #-}

57{-# SPECIALIZE decimal :: Reader Int64 #-}

55{-# SPECIALIZE decimal :: Reader Int64 #-}

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

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

59−{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

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

60{-# SPECIALIZE decimal :: Reader Word8 #-}

58{-# SPECIALIZE decimal :: Reader Word8 #-}

61{-# SPECIALIZE decimal :: Reader Word16 #-}

59{-# SPECIALIZE decimal :: Reader Word16 #-}

62{-# SPECIALIZE decimal :: Reader Word32 #-}

60{-# SPECIALIZE decimal :: Reader Word32 #-}

Data/Text/Read.hs+1−1

@@ -55,7 +55,7 @@ decimal :: Integral a => Reader a

55{-# SPECIALIZE decimal :: Reader Int32 #-}

56{-# SPECIALIZE decimal :: Reader Int64 #-}

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

58−{-# SPECIALIZE decimal :: Reader Data.Word.Word #-}

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

59{-# SPECIALIZE decimal :: Reader Word8 #-}

60{-# SPECIALIZE decimal :: Reader Word16 #-}

61{-# 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

53 , Mul.benchmark

54 , Pure.benchmark "tiny" (tf "tiny.txt")

55 , Pure.benchmark "ascii" (tf "ascii-small.txt")

56− -- , Pure.benchmark "france" (tf "france.html")

56+ , Pure.benchmark "france" (tf "france.html")

57 , Pure.benchmark "russian" (tf "russian-small.txt")

58 , Pure.benchmark "japanese" (tf "japanese.txt")

59 , ReadNumbers.benchmark (tf "numbers.txt")

benchmarks/haskell/Benchmarks/Pure.hs−7

@@ -262,13 +262,6 @@ benchmark kind fp = do

262 , benchBSL $ nf (BL.map toUpper) bla

263 , benchS $ nf (L.map toUpper) sa

264 ]

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− ]

272 , bgroup "words"

265 , bgroup "words"

273 [ benchT $ nf T.words ta

266 [ benchT $ nf T.words ta

274 , benchTL $ nf TL.words tla

267 , benchTL $ nf TL.words tla

benchmarks/haskell/Benchmarks/ReadNumbers.hs+5−2

@@ -25,7 +25,8 @@ import Data.List (foldl')

25import Numeric (readDec, readFloat, readHex)

26import qualified Data.ByteString.Char8 as B

27import 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

29import qualified Data.Text as T

30import qualified Data.Text as T

30import qualified Data.Text.IO as T

31import qualified Data.Text.IO as T

31import qualified Data.Text.Lazy as TL

32import qualified Data.Text.Lazy as TL

@@ -62,10 +63,12 @@ benchmark fp = do

62 whnf (double . text TL.rational) tl

63 whnf (double . text TL.rational) tl

63

64

64 , bench "DecimalByteString" $ whnf (int . byteString B.readInt) b

65 , 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

66

67

67 , bench "DecimalLazyByteString" $

68 , bench "DecimalLazyByteString" $

68 whnf (int . byteString BL.readInt) bl

69 whnf (int . byteString BL.readInt) bl

70+ , bench "DoubleLazyByteString" $

71+ whnf (double . byteString BL.readDouble) bl

69 ]

72 ]

70 where

73 where

71 -- Used for fixing types

74 -- Used for fixing types

benchmarks/haskell/Benchmarks/Stream.hs+5−16

@@ -18,11 +18,11 @@ import qualified Data.Text.Encoding as T

18import qualified Data.Text.Encoding.Error as E

19import qualified Data.Text.Internal.Encoding.Fusion as T

20import 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

22import qualified Data.Text.IO as T

23import qualified Data.Text.Lazy.Encoding as TL

24import 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

26import qualified Data.Text.Lazy.IO as TL

27

28instance NFData a => NFData (Stream a) where

@@ -53,25 +53,14 @@ benchmark fp = do

53 !utf32beL = TL.encodeUtf32BE tl

54

55 -- For the functions which operate on streams

56− let !s = F.stream t

56+ let !s = T.stream t

57

58 return $ bgroup "Stream"

59

60 -- Fusion

61 [ bgroup "stream" $

62− [ bench "Text" $ nf F.stream t

62+ [ bench "Text" $ nf T.stream t

63− , bench "LazyText" $ nf FL.stream tl

63+ , bench "LazyText" $ nf TL.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

75 ]

64 ]

76

65

77 -- Encoding.Fusion

66 -- Encoding.Fusion

benchmarks/text-benchmarks.cabal+3−13

@@ -3,7 +3,7 @@ version: 0.0.0.0

3synopsis: Benchmarks for the text package

4description: Benchmarks for the text package

5homepage: https://bitbucket.org/bos/text

6−license: BSD2

6+license: BSD3

7license-file: ../LICENSE

8author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

9 Bryan O'Sullivan <bos@serpentine.com>,

@@ -15,11 +15,6 @@ build-type: Simple

15

16cabal-version: >=1.2

17

18−flag bytestring-builder

19− description: Depend on the bytestring-builder package for backwards compatibility.

20− default: False

21− manual: False

22−

23flag llvm

18flag llvm

24 description: use LLVM

19 description: use LLVM

25 default: False

20 default: False

@@ -38,7 +33,8 @@ executable text-benchmarks

38 build-depends: base == 4.*,

33 build-depends: base == 4.*,

39 binary,

34 binary,

40 blaze-builder,

35 blaze-builder,

41− bytestring-lexing >= 0.5.0,

36+ bytestring,

37+ bytestring-lexing,

42 containers,

38 containers,

43 criterion >= 0.10.0.0,

39 criterion >= 0.10.0.0,

44 deepseq,

40 deepseq,

@@ -50,12 +46,6 @@ executable text-benchmarks

50 utf8-string,

46 utf8-string,

51 vector

47 vector

52

48

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−

59executable text-multilang

49executable text-multilang

60 hs-source-dirs: haskell

50 hs-source-dirs: haskell

61 main-is: Multilang.hs

51 main-is: Multilang.hs

changelog.md−81

@@ -1,84 +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−

821.2.0.3

11.2.0.3

83

2

84* Update formatRealFloat to correspond to the definition in versions

3* Update formatRealFloat to correspond to the definition in versions

scripts/CaseFolding.hs+1−1

@@ -37,7 +37,7 @@ mapCF :: CaseFolding -> [String]

37mapCF (CF _ ms) = typ ++ (map nice . filter p $ ms) ++ [last]

38 where

39 typ = ["foldMapping :: forall s. Char -> s -> Step (CC s) Char"

40− ,"{-# NOINLINE foldMapping #-}"]

40+ ,"{-# INLINE foldMapping #-}"]

41 last = "foldMapping c s = Yield (toLower c) (CC s '\\0' '\\0')"

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

43 "foldMapping " ++ showC (code c) ++ " s = Yield " ++ x ++ " (CC s " ++ y ++ " " ++ z ++ ")"

scripts/CaseMapping.hs+3−3

@@ -8,7 +8,7 @@ import SpecialCasing

8main = do

9 args <- getArgs

10 let oname = case args of

11− [] -> "../Data/Text/Internal/Fusion/CaseMapping.hs"

11+ [] -> "../Data/Text/Fusion/CaseMapping.hs"

12 [o] -> o

13 psc <- parseSC "SpecialCasing.txt"

14 pcf <- parseCF "CaseFolding.txt"

@@ -24,10 +24,10 @@ main = do

24 mapM_ (hPutStrLn h) $

25 ["{-# LANGUAGE Rank2Types #-}"

26 ,"-- AUTOMATICALLY GENERATED - DO NOT EDIT"

27− ,"-- Generated by scripts/CaseMapping.hs"] ++

27+ ,"-- Generated by scripts/SpecialCasing.hs"] ++

28 comments ++

29 [""

30− ,"module Data.Text.Internal.Fusion.CaseMapping where"

30+ ,"module Data.Text.Fusion.CaseMapping where"

31 ,"import Data.Char"

32 ,"import Data.Text.Internal.Fusion.Types"

33 ,""]

scripts/SpecialCasing.hs+1−1

@@ -43,7 +43,7 @@ mapSC which access twiddle (SC _ ms) =

43 typ ++ (map nice . filter p $ ms) ++ [last]

44 where

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

46− ,"{-# NOINLINE " ++ which ++ "Mapping #-}"]

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

47 last = which ++ "Mapping c s = Yield (to" ++ ucFirst which ++ " c) (CC s '\\0' '\\0')"

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

49 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 @@

1count = 1000

2

3−all: coverage literal-rule-test

4−

5−literal-rule-test:

6− ./literal-rule-test.sh

7−

8coverage: build coverage/hpc_index.html

3coverage: build coverage/hpc_index.html

9

4

10build: text-test-data

5build: text-test-data

@@ -37,4 +32,4 @@ coverage/hpc_index.html: coverage/coverage.tix

37clean:

32clean:

38 rm -rf dist coverage .hpc

33 rm -rf dist coverage .hpc

39

34

40−.PHONY: build coverage all literal-rule-test

35+.PHONY: build coverage

tests/Tests/Properties.hs−6

@@ -516,10 +516,6 @@ sf_takeWhile q p = (L.takeWhile p . L.filter q) `eqP`

516 (unpackS . S.takeWhile p . S.filter q)

517t_takeWhile p = L.takeWhile p `eqP` (unpackS . T.takeWhile p)

518tl_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)

523s_dropWhile p = L.dropWhile p `eqP` (unpackS . S.dropWhile p)

519s_dropWhile p = L.dropWhile p `eqP` (unpackS . S.dropWhile p)

524s_dropWhile_s p = L.dropWhile p `eqP` (unpackS . S.unstream . S.dropWhile p)

520s_dropWhile_s p = L.dropWhile p `eqP` (unpackS . S.unstream . S.dropWhile p)

525sf_dropWhile q p = (L.dropWhile p . L.filter q) `eqP`

521sf_dropWhile q p = (L.dropWhile p . L.filter q) `eqP`

@@ -1146,8 +1142,6 @@ tests =

1146 testProperty "sf_takeWhile" sf_takeWhile,

1142 testProperty "sf_takeWhile" sf_takeWhile,

1147 testProperty "t_takeWhile" t_takeWhile,

1143 testProperty "t_takeWhile" t_takeWhile,

1148 testProperty "tl_takeWhile" tl_takeWhile,

1144 testProperty "tl_takeWhile" tl_takeWhile,

1149− testProperty "t_takeWhileEnd" t_takeWhileEnd,

1150− testProperty "tl_takeWhileEnd" tl_takeWhileEnd,

1151 testProperty "sf_dropWhile" sf_dropWhile,

1145 testProperty "sf_dropWhile" sf_dropWhile,

1152 testProperty "s_dropWhile" s_dropWhile,

1146 testProperty "s_dropWhile" s_dropWhile,

1153 testProperty "s_dropWhile_s" s_dropWhile_s,

1147 testProperty "s_dropWhile_s" s_dropWhile_s,

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+1−3

@@ -3,7 +3,7 @@ version: 0.0.0.0

3synopsis: Functional tests for the text package

4description: Functional tests for the text package

5homepage: https://github.com/bos/text

6−license: BSD2

6+license: BSD3

7license-file: ../LICENSE

8author: Jasper Van der Jeugt <jaspervdj@gmail.com>,

9 Bryan O'Sullivan <bos@serpentine.com>,

@@ -114,7 +114,6 @@ library

114 Data.Text.Internal.Lazy.Search

115 Data.Text.Internal.Private

116 Data.Text.Read

117− Data.Text.Show

118 Data.Text.Internal.Read

117 Data.Text.Internal.Read

119 Data.Text.Internal.Search

118 Data.Text.Internal.Search

120 Data.Text.Unsafe

119 Data.Text.Unsafe

@@ -136,7 +135,6 @@ library

136 build-depends:

135 build-depends:

137 array,

136 array,

138 base == 4.*,

137 base == 4.*,

139− binary,

140 bytestring,

138 bytestring,

141 deepseq,

139 deepseq,

142 ghc-prim,

140 ghc-prim,

text.cabal+6−24

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

1name: text

2−version: 1.2.2.1

2+version: 1.2.0.4

3homepage: https://github.com/bos/text

4bug-reports: https://github.com/bos/text/issues

5synopsis: An efficient packed Unicode text type.

@@ -31,7 +31,7 @@ description:

31 the @text-icu@ package:

32 <http://hackage.haskell.org/package/text-icu>

33

34−license: BSD2

34+license: BSD3

35license-file: LICENSE

36author: Bryan O'Sullivan <bos@serpentine.com>

37maintainer: Bryan O'Sullivan <bos@serpentine.com>

@@ -64,11 +64,6 @@ extra-source-files:

64 tests/scripts/*.sh

65 tests/text-tests.cabal

66

67−flag bytestring-builder

68− description: Depend on the bytestring-builder package for backwards compatibility.

69− default: False

70− manual: False

71−

72flag developer

67flag developer

73 description: operate in developer mode

68 description: operate in developer mode

74 default: False

69 default: False

@@ -76,8 +71,6 @@ flag developer

76

71

77flag integer-simple

72flag integer-simple

78 description: Use the simple integer library instead of GMP

73 description: Use the simple integer library instead of GMP

79− default: False

80− manual: False

81

74

82library

75library

83 c-sources: cbits/cbits.c

76 c-sources: cbits/cbits.c

@@ -128,21 +121,16 @@ library

128 Data.Text.Read

121 Data.Text.Read

129 Data.Text.Unsafe

122 Data.Text.Unsafe

130

123

131− other-modules:

132− Data.Text.Show

133−

134 build-depends:

124 build-depends:

135 array >= 0.3,

125 array >= 0.3,

136 base >= 4.2 && < 5,

126 base >= 4.2 && < 5,

137− binary,

138 deepseq >= 1.1.0.0,

127 deepseq >= 1.1.0.0,

139 ghc-prim >= 0.2

128 ghc-prim >= 0.2

140

129

141− if flag(bytestring-builder)

130+ if impl(ghc >= 7.7)

142− build-depends: bytestring >= 0.9 && < 0.10.4,

131+ build-depends: bytestring >= 0.10.4.0

143− bytestring-builder >= 0.10.4

144 else

132 else

145− build-depends: bytestring >= 0.10.4

133+ build-depends: bytestring >= 0.9

146

134

147 cpp-options: -DHAVE_DEEPSEQ

135 cpp-options: -DHAVE_DEEPSEQ

148 ghc-options: -Wall -fwarn-tabs -funbox-strict-fields -O2

136 ghc-options: -Wall -fwarn-tabs -funbox-strict-fields -O2

@@ -176,7 +164,7 @@ test-suite tests

176 QuickCheck >= 2.7,

164 QuickCheck >= 2.7,

177 array,

165 array,

178 base,

166 base,

179− binary,

167+ bytestring,

180 deepseq,

168 deepseq,

181 directory,

169 directory,

182 ghc-prim,

170 ghc-prim,

@@ -186,12 +174,6 @@ test-suite tests

186 test-framework-hunit >= 0.2,

174 test-framework-hunit >= 0.2,

187 test-framework-quickcheck2 >= 0.2

175 test-framework-quickcheck2 >= 0.2

188

176

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−

195 if flag(integer-simple)

177 if flag(integer-simple)

196 cpp-options: -DINTEGER_SIMPLE

178 cpp-options: -DINTEGER_SIMPLE

197 build-depends: integer-simple >= 0.1 && < 0.5

179 build-depends: integer-simple >= 0.1 && < 0.5