4d3f9b4 · streamly · git.iko.soy

git.iko.soy/streamly

Implement Data.Parser.ParserD.wordBy

- Modify Data.Parser.wordBy accordingly - Add tests for both - Add benchmarks for both

commit4d3f9b41edac9e95861dc7a4ae7ad69f74cf1c68 authorAdithya Kumar <adithya@composewell.com> dateThu 10 Dec 2020, 19:01 (6 years ago) parent9fefed3

6 files changed+97−13

compare with master

unifiedsplit

benchmark/Streamly/Benchmark/Data/Parser.hs+10 benchmark/Streamly/Benchmark/Data/Parser/ParserD.hs+11 src/Streamly/Internal/Data/Parser.hs+3−7 src/Streamly/Internal/Data/Parser/ParserD.hs+37−6 test/Streamly/Test/Data/Parser.hs+18 test/Streamly/Test/Data/Parser/ParserD.hs+18

benchmark/Streamly/Benchmark/Data/Parser.hs+10

@@ -76,6 +76,14 @@ takeWhile value = IP.parse (PR.takeWhile (<= value) FL.drain)

7676groupBy :: MonadCatch m => SerialT m Int -> m ()

7777groupBy = IP.parse (PR.groupBy (<=) FL.drain)

7878

79+{-# INLINE wordBy #-}

80+wordBy :: MonadCatch m => Int -> SerialT m Int -> m ()

81+wordBy value = IP.parse (PR.wordBy (>= value) FL.drain)

82+

83+{-# INLINE manyWordByEven #-}

84+manyWordByEven :: MonadCatch m => SerialT m Int -> m ()

85+manyWordByEven = IP.parse (PR.many FL.drain (PR.wordBy (Prelude.even) FL.drain))

86+

7987{-# INLINE many #-}

8088many :: MonadCatch m => SerialT m Int -> m Int

8189many = IP.parse (PR.many FL.length (PR.satisfy (> 0)))

@@ -263,11 +271,13 @@ o_1_space_serial value =

263271 , benchIOSink value "takeWhile" $ takeWhile value

264272 , benchIOSink value "drainWhile" $ drainWhile value

265273 , benchIOSink value "groupBy" $ groupBy

274+ , benchIOSink value "wordBy" $ wordBy value

266275 , benchIOSink value "splitAp" $ splitAp value

267276 , benchIOSink value "splitApBefore" $ splitApBefore value

268277 , benchIOSink value "splitApAfter" $ splitApAfter value

269278 , benchIOSink value "splitWith" $ splitWith value

270279 , benchIOSink value "many" many

280+ , benchIOSink value "many (wordBy even)" $ manyWordByEven

271281 , benchIOSink value "some" some

272282 , benchIOSink value "manyTill" $ manyTill value

273283 , benchIOSink value "tee" $ teeAllAny value

benchmark/Streamly/Benchmark/Data/Parser/ParserD.hs+11

@@ -71,6 +71,15 @@ takeWhile value = IP.parseD (drainWhile (<= value))

7171groupBy :: MonadThrow m => SerialT m Int -> m ()

7272groupBy = IP.parseD (PR.groupBy (<=) FL.drain)

7373

74+{-# INLINE wordBy #-}

75+wordBy :: MonadThrow m => Int -> SerialT m Int -> m ()

76+wordBy value = IP.parseD (PR.wordBy (>= value) FL.drain)

77+

78+{-# INLINE manyWordByEven #-}

79+manyWordByEven :: MonadCatch m => SerialT m Int -> m ()

80+manyWordByEven =

81+ IP.parseD (PR.many FL.drain (PR.wordBy (Prelude.even) FL.drain))

82+

7483{-# INLINE many #-}

7584many :: MonadCatch m => SerialT m Int -> m Int

7685many = IP.parseD (PR.many FL.length (PR.satisfy (> 0)))

@@ -200,8 +209,10 @@ o_1_space_serial :: Int -> [Benchmark]

200209o_1_space_serial value =

201210 [ benchIOSink value "takeWhile" $ takeWhile value

202211 , benchIOSink value "groupBy" $ groupBy

212+ , benchIOSink value "wordBy" $ wordBy value

203213 , benchIOSink value "split (all,any)" $ splitAllAny value

204214 , benchIOSink value "many" many

215+ , benchIOSink value "many (wordBy even)" $ manyWordByEven

205216 , benchIOSink value "some" some

206217 , benchIOSink value "manyTill" $ manyTill value

207218 , benchIOSink value "tee (all,any)" $ teeAllAny value

src/Streamly/Internal/Data/Parser.hs+3−7

@@ -548,13 +548,9 @@ escapedFrameBy _begin _end _escape _p = undefined

548548-- S.wordsBy pred f = S.parseMany (PR.wordBy pred f)

549549-- @

550550--

551−-- /Unimplemented/

552−--

553−{-# INLINABLE wordBy #-}

554−wordBy ::

555− -- Monad m =>

556− (a -> Bool) -> Fold m a b -> Parser m a b

557−wordBy = undefined

551+{-# INLINE wordBy #-}

552+wordBy :: MonadCatch m => (a -> Bool) -> Fold m a b -> Parser m a b

553+wordBy f = K.toParserK . D.wordBy f

558554

559555-- | @groupBy cmp f $ S.fromList [a,b,c,...]@ assigns the element @a@ to the

560556-- first group, then if @a \`cmp` b@ is 'True' @b@ is also assigned to the same

src/Streamly/Internal/Data/Parser/ParserD.hs+37−6

@@ -447,15 +447,46 @@ sliceBeginWith ::

447447 (a -> Bool) -> Fold m a b -> Parser m a b

448448sliceBeginWith = undefined

449449

450+data WordByState s b = WBLeft !s | WBWord !s | WBRight !b

451+

450452-- | See 'Streamly.Internal.Data.Parser.wordBy'.

451453--

452−-- /Unimplemented/

453454--

454−{-# INLINABLE wordBy #-}

455−wordBy ::

456− -- Monad m =>

457− (a -> Bool) -> Fold m a b -> Parser m a b

458−wordBy = undefined

455+{-# INLINE wordBy #-}

456+wordBy :: Monad m => (a -> Bool) -> Fold m a b -> Parser m a b

457+wordBy predicate (Fold fstep finitial fextract) = Parser step initial extract

458+

459+ where

460+

461+ {-# INLINE worder #-}

462+ worder s a = do

463+ res <- fstep s a

464+ return

465+ $ case res of

466+ FL.Partial s1 -> Partial 0 $ WBWord s1

467+ FL.Done b -> Done 0 b

468+

469+ initial = WBLeft <$> finitial

470+

471+ step (WBLeft s) a =

472+ if not (predicate a)

473+ then worder s a

474+ else return $ Partial 0 $ WBLeft s

475+ step (WBWord s) a =

476+ if not (predicate a)

477+ then worder s a

478+ else do

479+ b <- fextract s

480+ return $ Partial 0 $ WBRight b

481+ step (WBRight b) a =

482+ return

483+ $ if not (predicate a)

484+ then Done 1 b

485+ else Partial 0 $ WBRight b

486+

487+ extract (WBLeft s) = fextract s

488+ extract (WBWord s) = fextract s

489+ extract (WBRight b) = return b

459490

460491data GroupByState a s

461492 = GroupByInit !s

test/Streamly/Test/Data/Parser.hs+18

@@ -317,6 +317,23 @@ groupBy =

317317 | null lst = []

318318 | otherwise = head $ List.groupBy cmp lst

319319

320+

321+wordBy :: Property

322+wordBy =

323+ forAll (listOf (elements [' ', 's']))

324+ $ \ls ->

325+ case S.parse parser (S.fromList ls) of

326+ Right parsed -> checkListEqual parsed (words' ls)

327+ Left _ -> property False

328+

329+ where

330+

331+ predicate = (== ' ')

332+ parser = P.many FL.toList (P.wordBy predicate FL.toList)

333+ words' lst =

334+ let wrds = words lst

335+ in if wrds == [] then [""] else wrds

336+

320337-- splitWithPass :: Property

321338-- splitWithPass =

322339-- forAll (listOf (chooseInt (0, 1))) $ \ls ->

@@ -627,6 +644,7 @@ main =

627644 prop "P.takeWhile = Prelude.takeWhile" Main.takeWhile

628645 prop "P.takeWhile1 = Prelude.takeWhile if taken something, else check why failed" takeWhile1

629646 prop "P.groupBy = Prelude.head . Prelude.groupBy" groupBy

647+ prop "many (P.wordBy ' ') = words'" wordBy

630648 -- prop "" splitWithPass

631649 -- prop "" splitWithFailLeft

632650 -- prop "" splitWithFailRight

test/Streamly/Test/Data/Parser/ParserD.hs+18

@@ -323,6 +323,23 @@ sliceSepByMax =

323323 where

324324 predicate = (== 1)

325325

326+wordBy :: Property

327+wordBy =

328+ forAll (listOf (elements [' ', 's']))

329+ $ \ls ->

330+ case S.parseD parser (S.fromList ls) of

331+ Right parsed -> checkListEqual parsed (words' ls)

332+ Left _ -> property False

333+

334+ where

335+

336+ predicate = (== ' ')

337+ parser = P.many FL.toList (P.wordBy predicate FL.toList)

338+ words' lst =

339+ let wrds = words lst

340+ in if wrds == [] then [""] else wrds

341+

342+

326343splitWith :: Property

327344splitWith =

328345 forAll (listOf (chooseInt (0, 1))) $ \ls ->

@@ -657,6 +674,7 @@ main =

657674 prop "P.groupBy = Prelude.head . Prelude.groupBy" groupBy

658675 prop "P.sliceSepBy = Prelude.takeWhile (not . predicate)" sliceSepBy

659676 prop "P.sliceSepByMax = Prelude.take n (Prelude.takeWhile (not . predicate)" sliceSepByMax

677+ prop "many (P.wordBy ' ') = words'" wordBy

660678 prop "parse 0, then 1, else fail" splitWith

661679 prop "fail due to die as left parser" splitWithFailLeft

662680 prop "fail due to die as right parser" splitWithFailRight