fdb303d · text · git.iko.soy

git.iko.soy/text

on master help sign in

implement strict Text to Builder encoder using BoundedEncodings

fdb303dSimon Meier15 years ago

Data/Text/Encoding.hs+32

compare with master

unifiedsplit

@@ -45,6 +45,10 @@ module Data.Text.Encoding

45 , encodeUtf16BE

46 , encodeUtf32LE

47 , encodeUtf32BE

48+

49+ -- * Generic encoding of Text

50+ , encodeStreamWithB

51+ , encodeTextWithB

48 ) where

52 ) where

50import Control.Exception (evaluate, try)

54import Control.Exception (evaluate, try)

@@ -56,6 +60,8 @@ import Control.Monad.ST (unsafeIOToST, unsafeSTToIO)

56import Data.Bits ((.&.))

60import Data.Bits ((.&.))

57import Data.ByteString as B

61import Data.ByteString as B

58import Data.ByteString.Internal as B

62import Data.ByteString.Internal as B

63+import Data.ByteString.Lazy.Builder.Internal as B

64+import Data.ByteString.Lazy.Builder.BasicEncoding.Internal as B

59import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

65import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

60import Data.Text.Internal (Text(..), textP)

66import Data.Text.Internal (Text(..), textP)

61import Data.Text.UnsafeChar (ord, unsafeWrite)

67import Data.Text.UnsafeChar (ord, unsafeWrite)

@@ -278,3 +284,29 @@ encodeUtf32BE txt = E.unstream (E.restreamUtf32BE (F.stream txt))

278foreign import ccall unsafe "_hs_text_decode_utf8" c_decode_utf8

284foreign import ccall unsafe "_hs_text_decode_utf8" c_decode_utf8

279 :: MutableByteArray# s -> Ptr CSize

285 :: MutableByteArray# s -> Ptr CSize

280 -> Ptr Word8 -> Ptr Word8 -> IO (Ptr Word8)

286 -> Ptr Word8 -> Ptr Word8 -> IO (Ptr Word8)

287+

288+-- | Encode all elements of a 'F.Stream' using a 'B.BoundedEncoding'.

289+{-# INLINE encodeStreamWithB #-}

290+encodeStreamWithB :: B.BoundedEncoding a -> F.Stream a -> B.Builder

291+encodeStreamWithB be =

292+ \(F.Stream next s0 _) -> B.builder $ step next s0

293+ where

294+ bound = B.sizeBound be

295+ step next s0 k (B.BufferRange op0 ope0) =

296+ go s0 op0

297+ where

298+ go s !op = case next s of

299+ F.Done -> k (B.BufferRange op ope0)

300+ F.Skip s' -> go s' op

301+ F.Yield x s'

302+ | op `plusPtr` bound <= ope0 -> B.runB be x op >>= go s'

303+ | otherwise ->

304+ return $ B.bufferFull bound op (step next s k)

305+

306+

307+-- |

308+-- | /Subject to fusion./

309+-- Encode all 'Char's of a 'T.Text' using a 'B.BoundedEncoding'.

310+{-# INLINE encodeTextWithB #-}

311+encodeTextWithB :: B.BoundedEncoding Char -> Text -> B.Builder

312+encodeTextWithB be = encodeStreamWithB be . F.stream

text.cabal+11−11

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

1name: text

2−version: 0.11.1.12

2+version: 0.11.2.0

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

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

5synopsis: An efficient packed Unicode text type.

@@ -111,7 +111,7 @@ library

111 build-depends:

112 array,

113 base < 5,

114− bytestring >= 0.9 && < 1.0

114+ bytestring >= 0.10 && < 1.0

115 if impl(ghc >= 6.10)

116 build-depends:

117 ghc-prim, base >= 4, deepseq >= 1.1.0.0

@@ -149,16 +149,16 @@ test-suite tests

149 -DASSERTS -DHAVE_DEEPSEQ

150

151 build-depends:

152− base >= 4 && < 5,

152+ base >= 4 && < 5,

153− bytestring >= 0.9 && < 0.10,

153+ bytestring >= 0.10 && < 1.0,

154 deepseq >= 1.1,

155− directory >= 1.0 && < 1.2,

155+ directory >= 1.0 && < 1.2,

156− random >= 1.0 && < 1.1,

156+ random >= 1.0 && < 1.1,

157− QuickCheck >= 2.4 && < 2.5,

157+ QuickCheck >= 2.4 && < 2.5,

158− HUnit >= 1.2 && < 1.3,

158+ HUnit >= 1.2 && < 1.3,

159− test-framework >= 0.4 && < 0.5,

159+ test-framework >= 0.4 && < 0.5,

160− test-framework-quickcheck2 >= 0.2 && < 0.3,

160+ test-framework-quickcheck2 >= 0.2 && < 0.3,

161− test-framework-hunit >= 0.2 && < 0.3

161+ test-framework-hunit >= 0.2 && < 0.3

162

163source-repository head

164 type: git