master..a324702 · text

Data/Text.hs+7−10

since splitsnapshots

unifiedsplit

@@ -238,7 +238,6 @@ import Data.ByteString (ByteString)

238import qualified Data.Text.Lazy as L

239import Data.Int (Int64)

240#endif

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

242#if __GLASGOW_HASKELL__ >= 708

241#if __GLASGOW_HASKELL__ >= 708

243import qualified GHC.Exts as Exts

242import qualified GHC.Exts as Exts

244#endif

243#endif

@@ -572,9 +571,7 @@ isSingleton = S.isSingleton . stream

572-- Subject to fusion.

571-- Subject to fusion.

573length :: Text -> Int

572length :: Text -> Int

574length t = S.length (stream t)

573length t = S.length (stream t)

575−{-# INLINE [0] length #-}

574+{-# 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

575

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

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

580-- Subject to fusion.

577-- Subject to fusion.

@@ -593,32 +590,32 @@ compareLength t n = S.compareLengthI (stream t) n

593

590

594{-# RULES

591{-# RULES

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

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

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

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

597 #-}

594 #-}

598

595

599{-# RULES

596{-# RULES

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

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

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

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

602 #-}

599 #-}

603

600

604{-# RULES

601{-# RULES

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

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

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

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

607 #-}

604 #-}

608

605

609{-# RULES

606{-# RULES

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

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

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

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

612 #-}

609 #-}

613

610

614{-# RULES

611{-# RULES

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

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

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

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

617 #-}

614 #-}

618

615

619{-# RULES

616{-# RULES

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

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

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

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

622 #-}

619 #-}

623

620

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

621-- -----------------------------------------------------------------------------

Data/Text/Array.hs+1−1

@@ -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 (

Data/Text/Encoding.hs+18−7

@@ -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

@@ -66,15 +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

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)

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

96

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+1−3

@@ -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

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

242 !t = Text arr o u

243 ts <- inlineInterleaveST (k b)

244 return $! t : ts

245−{-# INLINE [1] flush #-}

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

247

245

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

246------------------------------------------------------------------------

249

247

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/Lazy.hs+2−2

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

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

546 #-}

547

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

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

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

550init :: Text -> Text

551init (Chunk t0 ts0) = go t0 ts0

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

581isSingleton = S.isSingleton . stream

582{-# INLINE isSingleton #-}

583

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

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

585-- non-empty. Subject to fusion.

586last :: Text -> Char

587last Empty = emptyError "last"

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

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 (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/text-benchmarks.cabal+2−12

@@ -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

@@ -38,6 +33,7 @@ executable text-benchmarks

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

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

39 binary,

34 binary,

40 blaze-builder,

35 blaze-builder,

36+ bytestring,

41 bytestring-lexing >= 0.5.0,

37 bytestring-lexing >= 0.5.0,

42 containers,

38 containers,

43 criterion >= 0.10.0.0,

39 criterion >= 0.10.0.0,

@@ -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

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+5−16

@@ -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

@@ -138,11 +133,10 @@ library

138 deepseq >= 1.1.0.0,

133 deepseq >= 1.1.0.0,

139 ghc-prim >= 0.2

134 ghc-prim >= 0.2

140

135

141− if flag(bytestring-builder)

136+ if impl(ghc >= 7.7)

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

137+ build-depends: bytestring >= 0.10.4.0

143− bytestring-builder >= 0.10.4

144 else

138 else

145− build-depends: bytestring >= 0.10.4

139+ build-depends: bytestring >= 0.9

146

140

147 cpp-options: -DHAVE_DEEPSEQ

141 cpp-options: -DHAVE_DEEPSEQ

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

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

@@ -177,6 +171,7 @@ test-suite tests

177 array,

171 array,

178 base,

172 base,

179 binary,

173 binary,

174+ bytestring,

180 deepseq,

175 deepseq,

181 directory,

176 directory,

182 ghc-prim,

177 ghc-prim,

@@ -186,12 +181,6 @@ test-suite tests

186 test-framework-hunit >= 0.2,

181 test-framework-hunit >= 0.2,

187 test-framework-quickcheck2 >= 0.2

182 test-framework-quickcheck2 >= 0.2

188

183

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)

184 if flag(integer-simple)

196 cpp-options: -DINTEGER_SIMPLE

185 cpp-options: -DINTEGER_SIMPLE

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

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