8dec5e3 · text · git.iko.soy

git.iko.soy/text

Speed up lazy text equality

--HG-- extra : convert_revision : cbc12c191e1a5e94932e3b1581ca9ee1c7884d5b

commit8dec5e3ce2864447bf69c056496ab360e9d2b18d authorBryan O'Sullivan <bos@serpentine.com> dateTue 19 Oct 2010, 05:42 (16 years ago) parentfe46b27

2 files changed+27−1

compare with master

unifiedsplit

Data/Text/Lazy.hs+15−1 Data/Text/Unsafe.hs+12

Data/Text/Lazy.hs+15−1

@@ -204,8 +204,22 @@ import Data.Text.Internal (textP)

204204import qualified Data.Text.Util as U

205205import Data.Text.Lazy.Search (indices)

206206

207+equal :: Text -> Text -> Bool

208+equal Empty Empty = True

209+equal Empty _ = False

210+equal _ Empty = False

211+equal (Chunk a as) (Chunk b bs) =

212+ case compare lenA lenB of

213+ LT -> a == (T.takeWord16 lenA b) &&

214+ as `equal` Chunk (T.dropWord16 lenA b) bs

215+ EQ -> a == b && as `equal` bs

216+ GT -> T.takeWord16 lenB a == b &&

217+ Chunk (T.dropWord16 lenB a) as `equal` bs

218+ where lenA = T.lengthWord16 a

219+ lenB = T.lengthWord16 b

220+

207221instance Eq Text where

208− t1 == t2 = stream t1 == stream t2

222+ (==) = equal

209223 {-# INLINE (==) #-}

210224

211225instance Ord Text where

Data/Text/Unsafe.hs+12

@@ -21,6 +21,8 @@ module Data.Text.Unsafe

2121 , unsafeHead

2222 , unsafeTail

2323 , lengthWord16

24+ , takeWord16

25+ , dropWord16

2426 ) where

2527

2628#if defined(ASSERTS)

@@ -132,3 +134,13 @@ inlineInterleaveST (ST m) = ST $ \ s ->

132134lengthWord16 :: Text -> Int

133135lengthWord16 (Text _arr _off len) = len

134136{-# INLINE lengthWord16 #-}

137+

138+-- | /O(1)/ Unchecked take of 'k' 'Word16's from the front of a 'Text'.

139+takeWord16 :: Int -> Text -> Text

140+takeWord16 k (Text arr off _len) = Text arr off k

141+{-# INLINE takeWord16 #-}

142+

143+-- | /O(1)/ Unchecked drop of 'k' 'Word16's from the front of a 'Text'.

144+dropWord16 :: Int -> Text -> Text

145+dropWord16 k (Text arr off len) = Text arr (off+k) (len-k)

146+{-# INLINE dropWord16 #-}