d70ce27 · text · git.iko.soy

git.iko.soy/text

on master help sign in

Add a Binary instance for strict Text

d70ce27Bryan O'Sullivan11 years ago

Data/Text.hs+10−60

compare with master

unifiedsplit

@@ -195,7 +195,7 @@ module Data.Text

195

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

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

198− Read(..), Show(..),

198+ Read(..),

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

200 not, return, otherwise, quot)

201#if defined(HAVE_DEEPSEQ)

@@ -211,17 +211,20 @@ import Control.Monad (foldM)

211import Control.Monad.ST (ST)

212import qualified Data.Text.Array as A

213import qualified Data.List as L

214+import Data.Binary (Binary(get, put))

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

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

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

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

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

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

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

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

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

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

220import Data.Text.Internal.Fusion (stream, reverseStream, unstream)

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

221import Data.Text.Internal.Private (span_)

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

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

223+import Data.Text.Show (singleton, unpack)

221import qualified Prelude as P

224import qualified Prelude as P

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

225import Data.Text.Unsafe (Iter(..), iter, iter_, lengthWord16, reverseIter,

223 reverseIter_, unsafeHead, unsafeTail)

226 reverseIter_, unsafeHead, unsafeTail)

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

227+import Data.Text.Internal.Unsafe.Char (unsafeChr)

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

228import qualified Data.Text.Internal.Functions as F

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

229import qualified Data.Text.Internal.Encoding.Utf16 as U16

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

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

@@ -230,15 +233,9 @@ import Data.ByteString (ByteString)

230import qualified Data.Text.Lazy as L

233import qualified Data.Text.Lazy as L

231import Data.Int (Int64)

234import Data.Int (Int64)

232#endif

235#endif

233−#if __GLASGOW_HASKELL__ >= 702

234−import qualified GHC.CString as GHC

235−#else

236−import qualified GHC.Base as GHC

237−#endif

238#if __GLASGOW_HASKELL__ >= 708

236#if __GLASGOW_HASKELL__ >= 708

239import qualified GHC.Exts as Exts

237import qualified GHC.Exts as Exts

240#endif

238#endif

241−import GHC.Prim (Addr#)

242

239

243-- $strict

240-- $strict

244--

241--

@@ -324,9 +321,6 @@ instance Eq Text where

324instance Ord Text where

321instance Ord Text where

325 compare = compareText

322 compare = compareText

326

323

327−instance Show Text where

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

329−

330instance Read Text where

324instance Read Text where

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

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

332

326

@@ -349,6 +343,10 @@ instance Exts.IsList Text where

349instance NFData Text where rnf !_ = ()

343instance NFData Text where rnf !_ = ()

350#endif

344#endif

351

345

346+instance Binary Text where

347+ put t = put (encodeUtf8 t)

348+ get = P.fmap decodeUtf8 get

349+

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

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

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

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

354--

352--

@@ -401,54 +399,6 @@ pack :: String -> Text

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

399pack = unstream . S.map safe . S.streamList

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

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

403

401

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−

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

402-- -----------------------------------------------------------------------------

453-- * Basic functions

403-- * Basic functions

454

404

Data/Text/Encoding.hs+1−1

@@ -84,12 +84,12 @@ import Control.Exception (evaluate, try)

84import Control.Monad.ST (runST)

85import Data.ByteString as B

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

87−import Data.Text ()

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

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

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

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

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

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

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

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

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

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

92+import Data.Text.Show ()

93import Data.Text.Unsafe (unsafeDupablePerformIO)

94import Data.Word (Word8, Word32)

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

Data/Text/Show.hsnew file+85

@@ -0,0 +1,85 @@

1+{-# LANGUAGE CPP, MagicHash #-}

2+{-# OPTIONS_GHC -fno-warn-orphans #-}

3+#if __GLASGOW_HASKELL__ >= 702

4+{-# LANGUAGE Trustworthy #-}

5+#endif

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+ ) where

21+

22+import Control.Monad.ST (ST)

23+import Data.Text.Internal (Text(..), empty_, safe)

24+import Data.Text.Internal.Fusion (stream, unstream)

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

26+import GHC.Prim (Addr#)

27+import qualified Data.Text.Array as A

28+import qualified Data.Text.Internal.Fusion.Common as S

29+

30+#if __GLASGOW_HASKELL__ >= 702

31+import qualified GHC.CString as GHC

32+#else

33+import qualified GHC.Base as GHC

34+#endif

35+

36+instance Show Text where

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

38+

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

40+unpack :: Text -> String

41+unpack = S.unstreamList . stream

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

43+

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

45+unpackCString# :: Addr# -> Text

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

47+{-# NOINLINE unpackCString# #-}

48+

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

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

51+ = unpackCString# a #-}

52+

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

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

55+ = unpackCString# a #-}

56+

57+{-# RULES "TEXT empty literal"

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

59+ = empty_ #-}

60+

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

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

63+ = singleton_ a #-}

64+

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

66+-- Performs replacement on invalid scalar values.

67+singleton :: Char -> Text

68+singleton = unstream . S.singleton . safe

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

70+

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

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

73+ = singleton_ a #-}

74+

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

76+singleton_ :: Char -> Text

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

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

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

80+ _ <- unsafeWrite arr 0 d

81+ return arr

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

83+ | otherwise = 2

84+ d = safe c

85+{-# NOINLINE singleton_ #-}

tests/text-tests.cabal+2

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

114 Data.Text.Internal.Lazy.Search

115 Data.Text.Internal.Private

116 Data.Text.Read

117+ Data.Text.Show

117 Data.Text.Internal.Read

118 Data.Text.Internal.Read

118 Data.Text.Internal.Search

119 Data.Text.Internal.Search

119 Data.Text.Unsafe

120 Data.Text.Unsafe

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

135 build-depends:

136 build-depends:

136 array,

137 array,

137 base == 4.*,

138 base == 4.*,

139+ binary,

138 bytestring,

140 bytestring,

139 deepseq,

141 deepseq,

140 ghc-prim,

142 ghc-prim,

text.cabal+5

@@ -121,9 +121,13 @@ library

121 Data.Text.Read

122 Data.Text.Unsafe

123

124+ other-modules:

125+ Data.Text.Show

126+

124 build-depends:

127 build-depends:

125 array >= 0.3,

128 array >= 0.3,

126 base >= 4.2 && < 5,

129 base >= 4.2 && < 5,

130+ binary,

127 deepseq >= 1.1.0.0,

131 deepseq >= 1.1.0.0,

128 ghc-prim >= 0.2

132 ghc-prim >= 0.2

129

133

@@ -164,6 +168,7 @@ test-suite tests

164 QuickCheck >= 2.7,

168 QuickCheck >= 2.7,

165 array,

169 array,

166 base,

170 base,

171+ binary,

167 bytestring,

172 bytestring,

168 deepseq,

173 deepseq,

169 directory,

174 directory,