b6ce6a3 · streamly · git.iko.soy

git.iko.soy/streamly

Implement Data.Parser.ParserD.groupBy

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

commitb6ce6a3624932596e37016ffbadc79453e15959c authorAdithya Kumar <adithya@composewell.com> dateThu 10 Dec 2020, 12:20 (6 years ago) parentac3af87

6 files changed+82−17

compare with master

unifiedsplit

benchmark/Streamly/Benchmark/Data/Parser.hs+5 benchmark/Streamly/Benchmark/Data/Parser/ParserD.hs+5 src/Streamly/Internal/Data/Parser.hs+2−4 src/Streamly/Internal/Data/Parser/ParserD.hs+28−7 test/Streamly/Test/Data/Parser.hs+21−3 test/Streamly/Test/Data/Parser/ParserD.hs+21−3

benchmark/Streamly/Benchmark/Data/Parser.hs+5

@@ -72,6 +72,10 @@ drainWhile value = IP.parse (PR.drainWhile (<= value))

7272takeWhile :: MonadCatch m => Int -> SerialT m Int -> m ()

7373takeWhile value = IP.parse (PR.takeWhile (<= value) FL.drain)

7474

75+{-# INLINE groupBy #-}

76+groupBy :: MonadCatch m => SerialT m Int -> m ()

77+groupBy = IP.parse (PR.groupBy (<=) FL.drain)

78+

7579{-# INLINE many #-}

7680many :: MonadCatch m => SerialT m Int -> m Int

7781many = IP.parse (PR.many FL.length (PR.satisfy (> 0)))

@@ -258,6 +262,7 @@ o_1_space_serial value =

258262 [ benchIOSink value "takeEQ" $ takeEQ value

259263 , benchIOSink value "takeWhile" $ takeWhile value

260264 , benchIOSink value "drainWhile" $ drainWhile value

265+ , benchIOSink value "groupBy" $ groupBy

261266 , benchIOSink value "splitAp" $ splitAp value

262267 , benchIOSink value "splitApBefore" $ splitApBefore value

263268 , benchIOSink value "splitApAfter" $ splitApAfter value

benchmark/Streamly/Benchmark/Data/Parser/ParserD.hs+5

@@ -67,6 +67,10 @@ drainWhile p = PR.takeWhile p FL.drain

6767takeWhile :: MonadThrow m => Int -> SerialT m Int -> m ()

6868takeWhile value = IP.parseD (drainWhile (<= value))

6969

70+{-# INLINE groupBy #-}

71+groupBy :: MonadThrow m => SerialT m Int -> m ()

72+groupBy = IP.parseD (PR.groupBy (<=) FL.drain)

73+

7074{-# INLINE many #-}

7175many :: MonadCatch m => SerialT m Int -> m Int

7276many = IP.parseD (PR.many FL.length (PR.satisfy (> 0)))

@@ -195,6 +199,7 @@ moduleName = "Data.Parser.ParserD"

195199o_1_space_serial :: Int -> [Benchmark]

196200o_1_space_serial value =

197201 [ benchIOSink value "takeWhile" $ takeWhile value

202+ , benchIOSink value "groupBy" $ groupBy

198203 , benchIOSink value "split (all,any)" $ splitAllAny value

199204 , benchIOSink value "many" many

200205 , benchIOSink value "some" some

src/Streamly/Internal/Data/Parser.hs+2−4

@@ -573,10 +573,8 @@ wordBy = undefined

573573-- /Unimplemented/

574574--

575575{-# INLINABLE groupBy #-}

576−groupBy ::

577− -- Monad m =>

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

579−groupBy = undefined

576+groupBy :: MonadCatch m => (a -> a -> Bool) -> Fold m a b -> Parser m a b

577+groupBy cmp = K.toParserK . D.groupBy cmp

580578

581579-- | Match the given sequence of elements using the given comparison function.

582580--

src/Streamly/Internal/Data/Parser/ParserD.hs+28−7

@@ -457,15 +457,36 @@ wordBy ::

457457 (a -> Bool) -> Fold m a b -> Parser m a b

458458wordBy = undefined

459459

460+data GroupByState a s

461+ = GroupByInit !s

462+ | GroupByGrouping !a !s

463+

460464-- | See 'Streamly.Internal.Data.Parser.groupBy'.

461465--

462−-- /Unimplemented/

463−--

464−{-# INLINABLE groupBy #-}

465−groupBy ::

466− -- Monad m =>

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

468−groupBy = undefined

466+{-# INLINE groupBy #-}

467+groupBy :: Monad m => (a -> a -> Bool) -> Fold m a b -> Parser m a b

468+groupBy cmp (Fold fstep finitial fextract) = Parser step initial extract

469+

470+ where

471+

472+ {-# INLINE grouper #-}

473+ grouper s a0 a = do

474+ res <- fstep s a

475+ return

476+ $ case res of

477+ FL.Done b -> Done 0 b

478+ FL.Partial s1 -> Partial 0 (GroupByGrouping a0 s1)

479+

480+ initial = GroupByInit <$> finitial

481+

482+ step (GroupByInit s) a = grouper s a a

483+ step (GroupByGrouping a0 s) a =

484+ if cmp a0 a

485+ then grouper s a0 a

486+ else Done 1 <$> fextract s

487+

488+ extract (GroupByInit s) = fextract s

489+ extract (GroupByGrouping _ s) = fextract s

469490

470491-- XXX use an Unfold instead of a list?

471492-- XXX custom combinators for matching list, array and stream?

test/Streamly/Test/Data/Parser.hs+21−3

@@ -12,12 +12,13 @@ import Test.QuickCheck.Monadic (monadicIO, assert, run)

1212

1313import Prelude hiding (sequence)

1414

15+import qualified Data.List as List

16+import qualified Prelude

17+import qualified Streamly.Internal.Data.Array.Storable.Foreign as A

18+import qualified Streamly.Internal.Data.Fold as FL

1519import qualified Streamly.Internal.Data.Parser as P

1620import qualified Streamly.Internal.Data.Stream.IsStream as S

17−import qualified Streamly.Internal.Data.Fold as FL

18−import qualified Streamly.Internal.Data.Array.Storable.Foreign as A

1921import qualified Test.Hspec as H

20−import qualified Prelude

2122

2223#if MIN_VERSION_QuickCheck(2,14,0)

2324

@@ -300,6 +301,22 @@ takeWhile1 =

300301 where

301302 predicate = (== 0)

302303

304+groupBy :: Property

305+groupBy =

306+ forAll (listOf (chooseInt (0, 1)))

307+ $ \ls ->

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

309+ Right parsed -> checkListEqual parsed (groupByLF ls)

310+ Left _ -> property False

311+

312+ where

313+

314+ cmp = (==)

315+ parser = P.groupBy cmp FL.toList

316+ groupByLF lst

317+ | null lst = []

318+ | otherwise = head $ List.groupBy cmp lst

319+

303320-- splitWithPass :: Property

304321-- splitWithPass =

305322-- forAll (listOf (chooseInt (0, 1))) $ \ls ->

@@ -609,6 +626,7 @@ main =

609626 -- prop "lookAhead . take n >> lookAhead . take n = lookAhead . take n, else fail" lookAhead

610627 prop "P.takeWhile = Prelude.takeWhile" Main.takeWhile

611628 prop "P.takeWhile = Prelude.takeWhile if taken something, else check why failed" takeWhile1

629+ prop "P.groupBy = Prelude.head . Prelude.groupBy" groupBy

612630 -- prop "" splitWithPass

613631 -- prop "" splitWithFailLeft

614632 -- prop "" splitWithFailRight

test/Streamly/Test/Data/Parser/ParserD.hs+21−3

@@ -10,11 +10,12 @@ import Test.QuickCheck

1010 property, listOf, vectorOf, (.&&.), Gen, suchThat)

1111import Test.QuickCheck.Monadic (monadicIO, assert, run)

1212

13+import qualified Data.List as List

14+import qualified Prelude

15+import qualified Streamly.Internal.Data.Array.Storable.Foreign as A

16+import qualified Streamly.Internal.Data.Fold as FL

1317import qualified Streamly.Internal.Data.Parser.ParserD as P

1418import qualified Streamly.Internal.Data.Stream.IsStream as S

15−import qualified Streamly.Internal.Data.Fold as FL

16−import qualified Streamly.Internal.Data.Array.Storable.Foreign as A

17−import qualified Prelude

1819import qualified Test.Hspec as H

1920

2021import Prelude hiding (sequence)

@@ -287,6 +288,22 @@ takeWhile1 =

287288 where

288289 predicate = (== 0)

289290

291+groupBy :: Property

292+groupBy =

293+ forAll (listOf (chooseInt (0, 1)))

294+ $ \ls ->

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

296+ Right parsed -> checkListEqual parsed (groupByLF ls)

297+ Left _ -> property False

298+

299+ where

300+

301+ cmp = (==)

302+ parser = P.groupBy cmp FL.toList

303+ groupByLF lst

304+ | null lst = []

305+ | otherwise = head $ List.groupBy cmp lst

306+

290307sliceSepBy :: Property

291308sliceSepBy =

292309 forAll (listOf (chooseInt (0, 1))) $ \ls ->

@@ -637,6 +654,7 @@ main =

637654 prop "lookAhead . take n >> lookAhead . take n = lookAhead . take n, else fail" lookAhead

638655 prop "P.takeWhile = Prelude.takeWhile" Main.takeWhile

639656 prop "P.takeWhile1 = Prelude.takeWhile if taken something, else check why failed" takeWhile1

657+ prop "P.groupBy = Prelude.head . Prelude.groupBy" groupBy

640658 prop "P.sliceSepBy = Prelude.takeWhile (not . predicate)" sliceSepBy

641659 prop "P.sliceSepByMax = Prelude.take n (Prelude.takeWhile (not . predicate)" sliceSepByMax

642660 prop "parse 0, then 1, else fail" splitWith