fdb303d · text · git.iko.soy

git.iko.soy/text

implement strict Text to Builder encoder using BoundedEncodings

A first test of the infrastructure can be found in my 'aeson' branch.

commitfdb303d72b825028b14361d25f842f2c312666d7 authorSimon Meier <iridcode@gmail.com> dateSat 28 Jan 2012, 00:12 (15 years ago) parent073662c

2 files changed+43−11

compare with master

unifiedsplit

Data/Text/Encoding.hs+32 text.cabal+11−11

Data/Text/Encoding.hs+32

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

4545 , encodeUtf16BE

4646 , encodeUtf32LE

4747 , encodeUtf32BE

48+

49+ -- * Generic encoding of Text

50+ , encodeStreamWithB

51+ , encodeTextWithB

4852 ) where

4953

5054import Control.Exception (evaluate, try)

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

5660import Data.Bits ((.&.))

5761import Data.ByteString as B

5862import Data.ByteString.Internal as B

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

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

5965import Data.Text.Encoding.Error (OnDecodeError, UnicodeException, strictDecode)

6066import Data.Text.Internal (Text(..), textP)

6167import Data.Text.UnsafeChar (ord, unsafeWrite)

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

278284foreign import ccall unsafe "_hs_text_decode_utf8" c_decode_utf8

279285 :: MutableByteArray# s -> Ptr CSize

280286 -> 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 @@

11name: text

2−version: 0.11.1.12

2+version: 0.11.2.0

33homepage: https://github.com/bos/text

44bug-reports: https://github.com/bos/text/issues

55synopsis: An efficient packed Unicode text type.

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

111111 build-depends:

112112 array,

113113 base < 5,

114− bytestring >= 0.9 && < 1.0

114+ bytestring >= 0.10 && < 1.0

115115 if impl(ghc >= 6.10)

116116 build-depends:

117117 ghc-prim, base >= 4, deepseq >= 1.1.0.0

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

149149 -DASSERTS -DHAVE_DEEPSEQ

150150

151151 build-depends:

152− base >= 4 && < 5,

153− bytestring >= 0.9 && < 0.10,

152+ base >= 4 && < 5,

153+ bytestring >= 0.10 && < 1.0,

154154 deepseq >= 1.1,

155− directory >= 1.0 && < 1.2,

156− random >= 1.0 && < 1.1,

157− QuickCheck >= 2.4 && < 2.5,

158− HUnit >= 1.2 && < 1.3,

159− test-framework >= 0.4 && < 0.5,

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

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

155+ directory >= 1.0 && < 1.2,

156+ random >= 1.0 && < 1.1,

157+ QuickCheck >= 2.4 && < 2.5,

158+ HUnit >= 1.2 && < 1.3,

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

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

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

162162

163163source-repository head

164164 type: git