master..620a3af · text

git.iko.soy/text

on 620a3af help sign in

.gitignore−1

since splitsnapshots

unifiedsplit

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

1dist

2cabal-dev

3−cabal.sandbox.config

.hgtags−2

@@ -62,5 +62,3 @@ f8a60279237c1c28e094868740d9188dd94fcc9e 1.2.0.0

6278227a84740b70dcd4cf7a65f3bcf603d87e48e2 1.2.1.1

6340f268e6735749ef36f487e1d928eaa957bb431d 1.2.1.2

64631fd055e27fa10983e17e9ad36d1b285b85c661 1.2.1.3

65−8221a68ed5ec37028357df9c6c25f52320e54ad9 1.2.2.0

66−ce1c5ac65eb9f5580114b4e380092b05a485e95d 1.2.2.1

Data/Text.hs+15−65

@@ -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

@@ -215,13 +214,10 @@ import qualified Data.Text.Array as A

215import qualified Data.List as L

214import qualified Data.List as L

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

215import Data.Binary (Binary(get, put))

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

216import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

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

220−#endif

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

217import Data.String (IsString(..))

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

218import qualified Data.Text.Internal.Fusion as S

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

219import qualified Data.Text.Internal.Fusion.Common as S

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

220+import Data.Text.Encoding (decodeUtf8, encodeUtf8)

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

221import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

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

222import Data.Text.Internal.Private (span_)

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

223import Data.Text.Internal (Text(..), empty, firstf, mul, safe, text)

@@ -238,13 +234,10 @@ import Data.ByteString (ByteString)

238import qualified Data.Text.Lazy as L

234import qualified Data.Text.Lazy as L

239import Data.Int (Int64)

235import Data.Int (Int64)

240#endif

236#endif

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

242#if __GLASGOW_HASKELL__ >= 708

237#if __GLASGOW_HASKELL__ >= 708

243import qualified GHC.Exts as Exts

238import qualified GHC.Exts as Exts

244#endif

239#endif

245−#if MIN_VERSION_base(4,7,0)

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

240import Text.Printf (PrintfArg, formatArg, formatString)

247−#endif

248

241

249-- $strict

242-- $strict

250--

243--

@@ -333,21 +326,9 @@ instance Ord Text where

333instance Read Text where

326instance Read Text where

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

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

335

328

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

329instance Monoid Text where

345 mempty = empty

330 mempty = empty

346−#if MIN_VERSION_base(4,9,0)

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

348−#else

349 mappend = append

331 mappend = append

350−#endif

351 mconcat = concat

332 mconcat = concat

352

333

353instance IsString Text where

334instance IsString Text where

@@ -366,11 +347,7 @@ instance NFData Text where rnf !_ = ()

366

347

367instance Binary Text where

348instance Binary Text where

368 put t = put (encodeUtf8 t)

349 put t = put (encodeUtf8 t)

369− get = do

370− bs <- get

371− case decodeUtf8' bs of

350+ get = P.fmap decodeUtf8 get

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

373− P.Right a -> P.return a

374

351

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.

@@ -395,11 +372,8 @@ 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

400instance PrintfArg Text where

375instance PrintfArg Text where

401 formatArg txt = formatString $ unpack txt

376 formatArg txt = formatString $ unpack txt

402−#endif

403

377

404packConstr :: Constr

378packConstr :: Constr

405packConstr = mkConstr textDataType "pack" [] Prefix

379packConstr = mkConstr textDataType "pack" [] Prefix

@@ -486,8 +460,8 @@ head t = S.head (stream t)

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

460uncons :: Text -> Maybe (Char, Text)

487uncons t@(Text arr off len)

461uncons t@(Text arr off len)

488 | len <= 0 = Nothing

462 | len <= 0 = Nothing

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

490− in (c, text arr (off+d) (len-d))

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

464+ where Iter c d = iter t 0

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

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

492

466

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

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

@@ -572,9 +546,7 @@ isSingleton = S.isSingleton . stream

572-- Subject to fusion.

546-- Subject to fusion.

573length :: Text -> Int

547length :: Text -> Int

574length t = S.length (stream t)

548length t = S.length (stream t)

575−{-# INLINE [0] length #-}

549+{-# 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

550

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

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

580-- Subject to fusion.

552-- Subject to fusion.

@@ -593,32 +565,32 @@ compareLength t n = S.compareLengthI (stream t) n

593

565

594{-# RULES

566{-# RULES

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

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

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

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

597 #-}

569 #-}

598

570

599{-# RULES

571{-# RULES

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

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

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

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

602 #-}

574 #-}

603

575

604{-# RULES

576{-# RULES

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

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

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

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

607 #-}

579 #-}

608

580

609{-# RULES

581{-# RULES

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

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

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

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

612 #-}

584 #-}

613

585

614{-# RULES

586{-# RULES

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

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

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

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

617 #-}

589 #-}

618

590

619{-# RULES

591{-# RULES

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

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

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

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

622 #-}

594 #-}

623

595

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

596-- -----------------------------------------------------------------------------

@@ -915,13 +887,13 @@ concatMap f = concat . foldr ((:) . f) []

915{-# INLINE concatMap #-}

887{-# INLINE concatMap #-}

916

888

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

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

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

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

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

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

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

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

921{-# INLINE any #-}

893{-# INLINE any #-}

922

894

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

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

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

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

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

897all :: (Char -> Bool) -> Text -> Bool

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

898all p t = S.all p (stream t)

927{-# INLINE all #-}

899{-# INLINE all #-}

@@ -1152,27 +1124,6 @@ takeWhile p t@(Text arr off len) = loop 0

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

1124 unstream (S.takeWhile p (stream t)) = takeWhile p t

1153 #-}

1125 #-}

1154

1126

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

1127-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

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

1128-- 'takeWhile' @p@ @t@. Subject to fusion.

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

1129dropWhile :: (Char -> Bool) -> Text -> Text

@@ -1191,9 +1142,8 @@ dropWhile p t@(Text arr off len) = loop 0 0

1191 #-}

1142 #-}

1192

1143

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

1144-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

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

1145+-- dropping characters that fail the predicate @p@ from the end of

1195-- @t@. Subject to fusion.

1146-- @t@. Subject to fusion.

1196−--

1197-- Examples:

1147-- Examples:

1198--

1148--

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

1149-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1213,7 +1163,7 @@ dropWhileEnd p t@(Text arr off len) = loop (len-1) len

1213 #-}

1163 #-}

1214

1164

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

1165-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

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

1166+-- dropping characters that fail the predicate @p@ from both the

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

1167-- beginning and end of @t@. Subject to fusion.

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

1168dropAround :: (Char -> Bool) -> Text -> Text

1219dropAround p = dropWhile p . dropWhileEnd p

1169dropAround 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+18−7

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

55 , encodeUtf32LE

56 , encodeUtf32BE

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

63#if __GLASGOW_HASKELL__ >= 702

67#if __GLASGOW_HASKELL__ >= 702

@@ -66,15 +70,24 @@ import Control.Monad.ST.Unsafe (unsafeIOToST, unsafeSTToIO)

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

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

67#endif

71#endif

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)

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

87import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

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

88import Data.Text.Internal (Text(..), safe, text)

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

89import Data.Text.Internal.Private (runText)

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

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

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

91import Data.Text.Internal.Unsafe.Shift (shiftR)

79import Data.Text.Show ()

92import Data.Text.Show ()

80import Data.Text.Unsafe (unsafeDupablePerformIO)

93import Data.Text.Unsafe (unsafeDupablePerformIO)

@@ -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

104

97#include "text_cbits.h"

105#include "text_cbits.h"

@@ -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/Internal/Builder.hs+1−16

@@ -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

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)

@@ -92,20 +89,10 @@ newtype Builder = Builder {

92 -> ST s [S.Text]

89 -> ST s [S.Text]

93 }

90 }

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 Data.Monoid.mempty

111 {-# INLINE mconcat #-}

98 {-# INLINE mconcat #-}

@@ -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/Int/Digits.hs+1−1

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

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+112−286

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

1{-# LANGUAGE Rank2Types #-}

2-- AUTOMATICALLY GENERATED - DO NOT EDIT

3-- Generated by scripts/CaseMapping.hs

4−-- CaseFolding-8.0.0.txt

4+-- CaseFolding-7.0.0.txt

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

5+-- Date: 2014-04-09, 20:00:56 GMT [MD]

6−-- SpecialCasing-8.0.0.txt

6+-- SpecialCasing-7.0.0.txt

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

7+-- Date: 2014-03-18, 07:18:02 GMT [MD]

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

10import Data.Char

@@ -337,6 +337,8 @@ foldMapping '\x017f' s = Yield '\x0073' (CC s '\x0000' '\x0000')

337foldMapping '\x01f0' s = Yield '\x006a' (CC s '\x030c' '\x0000')

338-- COMBINING GREEK YPOGEGRAMMENI

339foldMapping '\x0345' s = Yield '\x03b9' (CC s '\x0000' '\x0000')

340+-- GREEK CAPITAL LETTER YOT

341+foldMapping '\x037f' s = Yield '\x03f3' (CC s '\x0000' '\x0000')

340-- GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS

342-- GREEK SMALL LETTER IOTA WITH DIALYTIKA AND TONOS

341foldMapping '\x0390' s = Yield '\x03b9' (CC s '\x0308' '\x0301')

343foldMapping '\x0390' s = Yield '\x03b9' (CC s '\x0308' '\x0301')

342-- GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS

344-- GREEK SMALL LETTER UPSILON WITH DIALYTIKA AND TONOS

@@ -357,20 +359,20 @@ foldMapping '\x03f0' s = Yield '\x03ba' (CC s '\x0000' '\x0000')

357foldMapping '\x03f1' s = Yield '\x03c1' (CC s '\x0000' '\x0000')

359foldMapping '\x03f1' s = Yield '\x03c1' (CC s '\x0000' '\x0000')

358-- GREEK LUNATE EPSILON SYMBOL

360-- GREEK LUNATE EPSILON SYMBOL

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

361foldMapping '\x03f5' s = Yield '\x03b5' (CC s '\x0000' '\x0000')

362+-- CYRILLIC CAPITAL LETTER EN WITH LEFT HOOK

363+foldMapping '\x0528' s = Yield '\x0529' (CC s '\x0000' '\x0000')

364+-- CYRILLIC CAPITAL LETTER DZZHE

365+foldMapping '\x052a' s = Yield '\x052b' (CC s '\x0000' '\x0000')

366+-- CYRILLIC CAPITAL LETTER DCHE

367+foldMapping '\x052c' s = Yield '\x052d' (CC s '\x0000' '\x0000')

368+-- CYRILLIC CAPITAL LETTER EL WITH DESCENDER

369+foldMapping '\x052e' s = Yield '\x052f' (CC s '\x0000' '\x0000')

360-- ARMENIAN SMALL LIGATURE ECH YIWN

370-- ARMENIAN SMALL LIGATURE ECH YIWN

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

371foldMapping '\x0587' s = Yield '\x0565' (CC s '\x0582' '\x0000')

362−-- CHEROKEE SMALL LETTER YE

372+-- GEORGIAN CAPITAL LETTER YN

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

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

364−-- CHEROKEE SMALL LETTER YI

374+-- GEORGIAN CAPITAL LETTER AEN

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

375+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

376-- LATIN SMALL LETTER H WITH LINE BELOW

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

377foldMapping '\x1e96' s = Yield '\x0068' (CC s '\x0331' '\x0000')

376-- LATIN SMALL LETTER T WITH DIAERESIS

378-- LATIN SMALL LETTER T WITH DIAERESIS

@@ -545,174 +547,36 @@ foldMapping '\x1ff6' s = Yield '\x03c9' (CC s '\x0342' '\x0000')

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

547foldMapping '\x1ff7' s = Yield '\x03c9' (CC s '\x0342' '\x03b9')

546-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

548-- GREEK CAPITAL LETTER OMEGA WITH PROSGEGRAMMENI

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

549foldMapping '\x1ffc' s = Yield '\x03c9' (CC s '\x03b9' '\x0000')

550+-- COPTIC CAPITAL LETTER BOHAIRIC KHEI

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

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

552+-- CYRILLIC CAPITAL LETTER DOUBLE O

553+foldMapping '\xa698' s = Yield '\xa699' (CC s '\x0000' '\x0000')

554+-- CYRILLIC CAPITAL LETTER CROSSED O

555+foldMapping '\xa69a' s = Yield '\xa69b' (CC s '\x0000' '\x0000')

556+-- LATIN CAPITAL LETTER C WITH BAR

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

558+-- LATIN CAPITAL LETTER B WITH FLOURISH

559+foldMapping '\xa796' s = Yield '\xa797' (CC s '\x0000' '\x0000')

560+-- LATIN CAPITAL LETTER F WITH STROKE

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

561+foldMapping '\xa798' s = Yield '\xa799' (CC s '\x0000' '\x0000')

550−-- LATIN CAPITAL LETTER CHI

562+-- LATIN CAPITAL LETTER VOLAPUK AE

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

563+foldMapping '\xa79a' s = Yield '\xa79b' (CC s '\x0000' '\x0000')

552−-- LATIN CAPITAL LETTER BETA

564+-- LATIN CAPITAL LETTER VOLAPUK OE

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

565+foldMapping '\xa79c' s = Yield '\xa79d' (CC s '\x0000' '\x0000')

554−-- LATIN CAPITAL LETTER OMEGA

566+-- LATIN CAPITAL LETTER VOLAPUK UE

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

567+foldMapping '\xa79e' s = Yield '\xa79f' (CC s '\x0000' '\x0000')

556−-- CHEROKEE SMALL LETTER A

568+-- LATIN CAPITAL LETTER H WITH HOOK

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

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

558−-- CHEROKEE SMALL LETTER E

570+-- LATIN CAPITAL LETTER REVERSED OPEN 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')

571+foldMapping '\xa7ab' s = Yield '\x025c' (CC s '\x0000' '\x0000')

700−-- CHEROKEE SMALL LETTER TSV

572+-- LATIN CAPITAL LETTER SCRIPT G

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')

573+foldMapping '\xa7ac' s = Yield '\x0261' (CC s '\x0000' '\x0000')

704−-- CHEROKEE SMALL LETTER WE

574+-- LATIN CAPITAL LETTER L WITH BELT

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')

575+foldMapping '\xa7ad' s = Yield '\x026c' (CC s '\x0000' '\x0000')

708−-- CHEROKEE SMALL LETTER WO

576+-- LATIN CAPITAL LETTER TURNED K

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')

577+foldMapping '\xa7b0' s = Yield '\x029e' (CC s '\x0000' '\x0000')

712−-- CHEROKEE SMALL LETTER WV

578+-- LATIN CAPITAL LETTER TURNED T

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')

579+foldMapping '\xa7b1' s = Yield '\x0287' (CC s '\x0000' '\x0000')

716-- LATIN SMALL LIGATURE FF

580-- LATIN SMALL LIGATURE FF

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

581foldMapping '\xfb00' s = Yield '\x0066' (CC s '\x0066' '\x0000')

718-- LATIN SMALL LIGATURE FI

582-- LATIN SMALL LIGATURE FI

@@ -737,106 +601,68 @@ foldMapping '\xfb15' s = Yield '\x0574' (CC s '\x056b' '\x0000')

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

601foldMapping '\xfb16' s = Yield '\x057e' (CC s '\x0576' '\x0000')

738-- ARMENIAN SMALL LIGATURE MEN XEH

602-- ARMENIAN SMALL LIGATURE MEN XEH

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

603foldMapping '\xfb17' s = Yield '\x0574' (CC s '\x056d' '\x0000')

604+-- WARANG CITI CAPITAL LETTER NGAA

605+foldMapping '\x118a0' s = Yield '\x118c0' (CC s '\x0000' '\x0000')

740−-- OLD HUNGARIAN CAPITAL LETTER A

606+-- WARANG CITI 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')

607+foldMapping '\x118a1' s = Yield '\x118c1' (CC s '\x0000' '\x0000')

758−-- OLD HUNGARIAN CAPITAL LETTER E

608+-- WARANG CITI CAPITAL LETTER WI

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

609+foldMapping '\x118a2' s = Yield '\x118c2' (CC s '\x0000' '\x0000')

760−-- OLD HUNGARIAN CAPITAL LETTER CLOSE E

610+-- WARANG CITI CAPITAL LETTER YU

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

611+foldMapping '\x118a3' s = Yield '\x118c3' (CC s '\x0000' '\x0000')

762−-- OLD HUNGARIAN CAPITAL LETTER EE

612+-- WARANG CITI CAPITAL LETTER YA

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

613+foldMapping '\x118a4' s = Yield '\x118c4' (CC s '\x0000' '\x0000')

764−-- OLD HUNGARIAN CAPITAL LETTER EF

614+-- WARANG CITI CAPITAL LETTER YO

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

615+foldMapping '\x118a5' s = Yield '\x118c5' (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

616+-- WARANG CITI CAPITAL LETTER II

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

617+foldMapping '\x118a6' s = Yield '\x118c6' (CC s '\x0000' '\x0000')

776−-- OLD HUNGARIAN CAPITAL LETTER EJ

618+-- WARANG CITI CAPITAL LETTER UU

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

619+foldMapping '\x118a7' s = Yield '\x118c7' (CC s '\x0000' '\x0000')

778−-- OLD HUNGARIAN CAPITAL LETTER EK

620+-- WARANG CITI CAPITAL LETTER E

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

621+foldMapping '\x118a8' s = Yield '\x118c8' (CC s '\x0000' '\x0000')

780−-- OLD HUNGARIAN CAPITAL LETTER AK

622+-- WARANG CITI CAPITAL LETTER O

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

623+foldMapping '\x118a9' s = Yield '\x118c9' (CC s '\x0000' '\x0000')

782−-- OLD HUNGARIAN CAPITAL LETTER UNK

624+-- WARANG CITI CAPITAL LETTER ANG

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

625+foldMapping '\x118aa' s = Yield '\x118ca' (CC s '\x0000' '\x0000')

784−-- OLD HUNGARIAN CAPITAL LETTER EL

626+-- WARANG CITI CAPITAL LETTER GA

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

627+foldMapping '\x118ab' s = Yield '\x118cb' (CC s '\x0000' '\x0000')

786−-- OLD HUNGARIAN CAPITAL LETTER ELY

628+-- WARANG CITI CAPITAL LETTER KO

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

629+foldMapping '\x118ac' s = Yield '\x118cc' (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

630+-- WARANG CITI CAPITAL LETTER ENY

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

631+foldMapping '\x118ad' s = Yield '\x118cd' (CC s '\x0000' '\x0000')

794−-- OLD HUNGARIAN CAPITAL LETTER O

632+-- WARANG CITI CAPITAL LETTER YUJ

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

633+foldMapping '\x118ae' s = Yield '\x118ce' (CC s '\x0000' '\x0000')

796−-- OLD HUNGARIAN CAPITAL LETTER OO

634+-- WARANG CITI CAPITAL LETTER UC

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

635+foldMapping '\x118af' s = Yield '\x118cf' (CC s '\x0000' '\x0000')

798−-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG OE

636+-- WARANG CITI CAPITAL LETTER ENN

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')

637+foldMapping '\x118b0' s = Yield '\x118d0' (CC s '\x0000' '\x0000')

810−-- OLD HUNGARIAN CAPITAL LETTER SHORT ER

638+-- WARANG CITI CAPITAL LETTER ODD

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

639+foldMapping '\x118b1' s = Yield '\x118d1' (CC s '\x0000' '\x0000')

812−-- OLD HUNGARIAN CAPITAL LETTER ES

640+-- WARANG CITI CAPITAL LETTER TTE

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

641+foldMapping '\x118b2' s = Yield '\x118d2' (CC s '\x0000' '\x0000')

814−-- OLD HUNGARIAN CAPITAL LETTER ESZ

642+-- WARANG CITI CAPITAL LETTER NUNG

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

643+foldMapping '\x118b3' s = Yield '\x118d3' (CC s '\x0000' '\x0000')

816−-- OLD HUNGARIAN CAPITAL LETTER ET

644+-- WARANG CITI CAPITAL LETTER DA

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

645+foldMapping '\x118b4' s = Yield '\x118d4' (CC s '\x0000' '\x0000')

818−-- OLD HUNGARIAN CAPITAL LETTER ENT

646+-- WARANG CITI CAPITAL LETTER AT

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

647+foldMapping '\x118b5' s = Yield '\x118d5' (CC s '\x0000' '\x0000')

820−-- OLD HUNGARIAN CAPITAL LETTER ETY

648+-- WARANG CITI CAPITAL LETTER AM

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

649+foldMapping '\x118b6' s = Yield '\x118d6' (CC s '\x0000' '\x0000')

822−-- OLD HUNGARIAN CAPITAL LETTER ECH

650+-- WARANG CITI CAPITAL LETTER BU

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

651+foldMapping '\x118b7' s = Yield '\x118d7' (CC s '\x0000' '\x0000')

824−-- OLD HUNGARIAN CAPITAL LETTER U

652+-- WARANG CITI CAPITAL LETTER PU

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

653+foldMapping '\x118b8' s = Yield '\x118d8' (CC s '\x0000' '\x0000')

826−-- OLD HUNGARIAN CAPITAL LETTER UU

654+-- WARANG CITI CAPITAL LETTER HIYO

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

655+foldMapping '\x118b9' s = Yield '\x118d9' (CC s '\x0000' '\x0000')

828−-- OLD HUNGARIAN CAPITAL LETTER NIKOLSBURG UE

656+-- WARANG CITI CAPITAL LETTER HOLO

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

657+foldMapping '\x118ba' s = Yield '\x118da' (CC s '\x0000' '\x0000')

830−-- OLD HUNGARIAN CAPITAL LETTER RUDIMENTA UE

658+-- WARANG CITI CAPITAL LETTER HORR

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

659+foldMapping '\x118bb' s = Yield '\x118db' (CC s '\x0000' '\x0000')

832−-- OLD HUNGARIAN CAPITAL LETTER EV

660+-- WARANG CITI CAPITAL LETTER HAR

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

661+foldMapping '\x118bc' s = Yield '\x118dc' (CC s '\x0000' '\x0000')

834−-- OLD HUNGARIAN CAPITAL LETTER EZ

662+-- WARANG CITI CAPITAL LETTER SSUU

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

663+foldMapping '\x118bd' s = Yield '\x118dd' (CC s '\x0000' '\x0000')

836−-- OLD HUNGARIAN CAPITAL LETTER EZS

664+-- WARANG CITI CAPITAL LETTER SII

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')

665+foldMapping '\x118be' s = Yield '\x118de' (CC s '\x0000' '\x0000')

840−-- OLD HUNGARIAN CAPITAL LETTER US

666+-- WARANG CITI CAPITAL LETTER VIYO

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

667+foldMapping '\x118bf' s = Yield '\x118df' (CC s '\x0000' '\x0000')

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

668foldMapping c s = Yield (toLower c) (CC s '\0' '\0')

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

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

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

594

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

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

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

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

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

599 where

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

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

606

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

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

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

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

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

611 where

@@ -669,12 +669,12 @@ scanl f z0 (Stream next0 s0 len) = Stream next (Scan1 z0 s0) (len+1) -- HINT may

669-- ** Generating and unfolding streams

670

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

672−replicateCharI !n !c

672+replicateCharI n c

673 | n < 0 = empty

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

675 where

676− next !i | i >= n = Done

676+ next i | i >= n = Done

677− | otherwise = Yield c (i + 1)

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

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

679

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

Data/Text/Internal/Read.hs+1−1

@@ -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

Data/Text/Lazy.hs+8−49

@@ -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

@@ -215,9 +214,6 @@ import Data.Data (Data(gfoldl, toConstr, gunfold, dataTypeOf), constrIndex,

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

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

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

215import Data.Binary (Binary(get, put))

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

216import Data.Monoid (Monoid(..))

218−#if MIN_VERSION_base(4,9,0)

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

220−#endif

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

217import Data.String (IsString(..))

222import qualified Data.Text as T

218import qualified Data.Text as T

223import qualified Data.Text.Internal as T

219import qualified Data.Text.Internal as T

@@ -229,7 +225,7 @@ import Data.Text.Internal.Lazy.Fusion (stream, unstream)

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

225import Data.Text.Internal.Lazy (Text(..), chunk, empty, foldlChunks,

230 foldrChunks, smallChunkSize)

226 foldrChunks, smallChunkSize)

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

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

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

228+import Data.Text.Lazy.Encoding (decodeUtf8, encodeUtf8)

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

229import qualified Data.Text.Internal.Functions as F

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

230import Data.Text.Internal.Lazy.Search (indices)

235#if __GLASGOW_HASKELL__ >= 702

231#if __GLASGOW_HASKELL__ >= 702

@@ -241,9 +237,7 @@ import qualified GHC.Base as GHC

241import qualified GHC.Exts as Exts

237import qualified GHC.Exts as Exts

242#endif

238#endif

243import GHC.Prim (Addr#)

239import GHC.Prim (Addr#)

244−#if MIN_VERSION_base(4,7,0)

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

240import Text.Printf (PrintfArg, formatArg, formatString)

246−#endif

247

241

248-- $fusion

242-- $fusion

249--

243--

@@ -341,21 +335,9 @@ instance Show Text where

341instance Read Text where

335instance Read Text where

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

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

343

337

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

338instance Monoid Text where

353 mempty = empty

339 mempty = empty

354−#if MIN_VERSION_base(4,9,0)

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

356−#else

357 mappend = append

340 mappend = append

358−#endif

359 mconcat = concat

341 mconcat = concat

360

342

361instance IsString Text where

343instance IsString Text where

@@ -376,11 +358,7 @@ instance NFData Text where

376

358

377instance Binary Text where

359instance Binary Text where

378 put t = put (encodeUtf8 t)

360 put t = put (encodeUtf8 t)

379− get = do

380− bs <- get

381− case decodeUtf8' bs of

361+ get = P.fmap decodeUtf8 get

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

383− P.Right a -> P.return a

384

362

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

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

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

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

@@ -395,11 +373,8 @@ instance Data Text where

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

373 _ -> error "Data.Text.Lazy.Text.gunfold"

396 dataTypeOf _ = textDataType

374 dataTypeOf _ = textDataType

397

375

398−#if MIN_VERSION_base(4,7,0)

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

400instance PrintfArg Text where

376instance PrintfArg Text where

401 formatArg txt = formatString $ unpack txt

377 formatArg txt = formatString $ unpack txt

402−#endif

403

378

404packConstr :: Constr

379packConstr :: Constr

405packConstr = mkConstr textDataType "pack" [] Prefix

380packConstr = mkConstr textDataType "pack" [] Prefix

@@ -545,7 +520,7 @@ tail Empty = emptyError "tail"

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

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

546 #-}

521 #-}

547

522

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

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

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

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

550init :: Text -> Text

525init :: Text -> Text

551init (Chunk t0 ts0) = go t0 ts0

526init (Chunk t0 ts0) = go t0 ts0

@@ -581,7 +556,7 @@ isSingleton :: Text -> Bool

581isSingleton = S.isSingleton . stream

556isSingleton = S.isSingleton . stream

582{-# INLINE isSingleton #-}

557{-# INLINE isSingleton #-}

583

558

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

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

585-- non-empty. Subject to fusion.

560-- non-empty. Subject to fusion.

586last :: Text -> Char

561last :: Text -> Char

587last Empty = emptyError "last"

562last Empty = emptyError "last"

@@ -881,13 +856,13 @@ concatMap f = concat . foldr ((:) . f) []

881{-# INLINE concatMap #-}

856{-# INLINE concatMap #-}

882

857

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

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

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

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

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

860any :: (Char -> Bool) -> Text -> Bool

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

861any p t = S.any p (stream t)

887{-# INLINE any #-}

862{-# INLINE any #-}

888

863

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

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

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

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

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

866all :: (Char -> Bool) -> Text -> Bool

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

867all p t = S.all p (stream t)

893{-# INLINE all #-}

868{-# INLINE all #-}

@@ -1144,20 +1119,6 @@ takeWhile p t0 = takeWhile' t0

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

1119"LAZY TEXT takeWhile -> unfused" [1] forall p t.

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

1120 unstream (S.takeWhile p (stream t)) = takeWhile p t

1146 #-}

1121 #-}

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

1122

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

1123-- | /O(n)/ 'dropWhile' @p@ @t@ returns the suffix remaining after

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

1124-- 'takeWhile' @p@ @t@. Subject to fusion.

@@ -1176,11 +1137,9 @@ dropWhile p t0 = dropWhile' t0

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

1137"LAZY TEXT dropWhile -> unfused" [1] forall p t.

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

1138 unstream (S.dropWhile p (stream t)) = dropWhile p t

1178 #-}

1139 #-}

1179−

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

1140-- | /O(n)/ 'dropWhileEnd' @p@ @t@ returns the prefix remaining after

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

1141+-- dropping characters that fail the predicate @p@ from the end of

1182-- @t@.

1142-- @t@.

1183−--

1184-- Examples:

1143-- Examples:

1185--

1144--

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

1145-- > dropWhileEnd (=='.') "foo..." == "foo"

@@ -1197,7 +1156,7 @@ dropWhileEnd p = go

1197{-# INLINE dropWhileEnd #-}

1156{-# INLINE dropWhileEnd #-}

1198

1157

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

1158-- | /O(n)/ 'dropAround' @p@ @t@ returns the substring remaining after

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

1159+-- dropping characters that fail the predicate @p@ from both the

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

1160-- beginning and end of @t@. Subject to fusion.

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

1161dropAround :: (Char -> Bool) -> Text -> Text

1203dropAround p = dropWhile p . dropWhileEnd p

1162dropAround 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+1−1

@@ -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

Data/Text/Lazy/Encoding.hs+15−5

@@ -46,23 +46,27 @@ module Data.Text.Lazy.Encoding

46 , encodeUtf32LE

47 , encodeUtf32BE

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

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 (Monoid(..))

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

69+#endif

66import qualified Data.Text as T

70import qualified Data.Text as T

67import qualified Data.Text.Encoding as TE

71import qualified Data.Text.Encoding as TE

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

72import qualified Data.Text.Internal.Lazy.Encoding.Fusion as E

@@ -142,6 +146,7 @@ decodeUtf8' bs = unsafeDupablePerformIO $ do

142{-# INLINE decodeUtf8' #-}

146{-# INLINE decodeUtf8' #-}

143

147

144encodeUtf8 :: Text -> B.ByteString

148encodeUtf8 :: Text -> B.ByteString

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

145encodeUtf8 Empty = B.empty

150encodeUtf8 Empty = B.empty

146encodeUtf8 lt@(Chunk t _) =

151encodeUtf8 lt@(Chunk t _) =

147 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

152 B.toLazyByteStringWith strategy B.empty $ encodeUtf8Builder lt

@@ -163,6 +168,11 @@ encodeUtf8BuilderEscaped :: BP.BoundedPrim Word8 -> Text -> B.Builder

163encodeUtf8BuilderEscaped prim =

168encodeUtf8BuilderEscaped prim =

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

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

165

170

171+#else

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

173+encodeUtf8 Empty = B.Empty

174+#endif

175+

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

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

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

177decodeUtf16LEWith :: OnDecodeError -> B.ByteString -> Text

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

178decodeUtf16LEWith onErr bs = F.unstream (E.streamUtf16LE onErr bs)

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

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

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

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

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

53 !utf32beL = TL.encodeUtf32BE tl

55 -- For the functions which operate on streams

56− let !s = F.stream t

56+ let !s = T.stream t

58 return $ bgroup "Stream"

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 ]

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

16cabal-version: >=1.2

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

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+1−28

@@ -1,33 +1,6 @@

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−

281.2.1.3

11.2.1.3

30−* Bug fix: As it turns out, moving the literal rewrite rules to simplifier

3+* 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

4 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

5 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

6 test environment mistake. Moving literal rules back to phase 1 finally

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.sh+1−1

@@ -6,7 +6,7 @@ function check_firings() {

6 rule=$1

7 expected=$2

8 build="ghc -O -ddump-rule-firings LiteralRuleTest.hs"

9− build="$build -i.. -I../include"

9+ build="$build -i.. -I../include -DMIN_VERSION_bytestring(a,b,c)=1"

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)

tests/text-tests.cabal+1−1

@@ -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>,

text.cabal+6−19

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

1name: text

2−version: 1.2.2.1

2+version: 1.2.1.3

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>

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

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

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

82library

75library

83 c-sources: cbits/cbits.c

76 c-sources: cbits/cbits.c

@@ -138,11 +131,10 @@ library

138 deepseq >= 1.1.0.0,

131 deepseq >= 1.1.0.0,

139 ghc-prim >= 0.2

132 ghc-prim >= 0.2

140

133

141− if flag(bytestring-builder)

134+ if impl(ghc >= 7.7)

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

135+ build-depends: bytestring >= 0.10.4.0

143− bytestring-builder >= 0.10.4

144 else

136 else

145− build-depends: bytestring >= 0.10.4

137+ build-depends: bytestring >= 0.9

146

138

147 cpp-options: -DHAVE_DEEPSEQ

139 cpp-options: -DHAVE_DEEPSEQ

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

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

@@ -177,6 +169,7 @@ test-suite tests

177 array,

169 array,

178 base,

170 base,

179 binary,

171 binary,

172+ bytestring,

180 deepseq,

173 deepseq,

181 directory,

174 directory,

182 ghc-prim,

175 ghc-prim,

@@ -186,12 +179,6 @@ test-suite tests

186 test-framework-hunit >= 0.2,

179 test-framework-hunit >= 0.2,

187 test-framework-quickcheck2 >= 0.2

180 test-framework-quickcheck2 >= 0.2

188

181

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)

182 if flag(integer-simple)

196 cpp-options: -DINTEGER_SIMPLE

183 cpp-options: -DINTEGER_SIMPLE

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

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