Compare commits

..

4 Commits

Author SHA1 Message Date
537c0df198 Update tests 2021-10-29 22:39:27 +03:00
75aa615263 Clean up 2021-10-29 22:37:47 +03:00
a300c88cfb Try parallel 2021-10-29 22:09:43 +03:00
560ea23861 parallel 2021-10-29 20:47:04 +03:00
3 changed files with 29 additions and 49 deletions

View File

@ -29,7 +29,6 @@ import qualified Data.ByteString.Lazy.Char8 as LBC
import System.IO.Unsafe (unsafeInterleaveIO) import System.IO.Unsafe (unsafeInterleaveIO)
import Control.Parallel.Strategies (rseq, parMap) import Control.Parallel.Strategies (rseq, parMap)
import qualified Data.List as L import qualified Data.List as L
import Control.Monad (unless)
combine :: (MonadUnliftIO m, MonadResource m, MonadThrow m, MonadIO m) => ConduitM FilePath Header m () combine :: (MonadUnliftIO m, MonadResource m, MonadThrow m, MonadIO m) => ConduitM FilePath Header m ()
combine = await >>= \case combine = await >>= \case
@ -44,27 +43,23 @@ chunks n = L.unfoldr $ \case
run :: IO () run :: IO ()
run = do run = do
datDir <- fromMaybe "./" <$> lookupEnv "HOME" datDir <- fromMaybe "./" <$> lookupEnv "HOME"
let datFile = datDir </> ".addressbook.dat"
original <- Set.fromList . map LBS.toStrict . lbsLines <$> LBS.readFile datFile
xs <- LBS.getContents >>= stream xs <- LBS.getContents >>= stream
let set = original `Set.union` F.fold (parMap rseq F.fold (chunks 200 xs)) let set = F.fold (parMap rseq F.fold (chunks 20 xs))
unless (original == set) $
runResourceT $ runResourceT $
runConduit $ runConduit $
CL.sourceList (Set.elems set) CL.sourceList (Set.elems set)
.| C.map (<> "\n") .| C.map (<> "\n")
.| CB.sinkFileCautious datFile .| CB.sinkFileCautious (datDir </> ".addressbook.dat")
where where
separate = \case separate = \case
From x -> [x] From x -> [x]
To xs -> F.toList xs To xs -> F.toList xs
-- A set of (locally) unique addresses. Composes with parMap -- A set of (locally) unique addresses. Composes with parMap
lbsLines = LBS.split (fromIntegral $ ord '\n')
stream :: LBS.ByteString -> IO [Set ByteString] stream :: LBS.ByteString -> IO [Set ByteString]
stream = stream =
traverse (unsafeInterleaveIO . parse . LBC.unpack) traverse (unsafeInterleaveIO . parse . LBC.unpack)
. filter (not . LBS.null) . filter (not . LBS.null)
. lbsLines . LBS.split (fromIntegral $ ord '\n')
parse path = parse path =
runResourceT $ runResourceT $
runConduit $ runConduit $

View File

@ -1,5 +1,4 @@
{-# LANGUAGE OverloadedStrings #-} {-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE ApplicativeDo #-}
module Data.Email.Header where module Data.Email.Header where
import qualified Data.Foldable as F import qualified Data.Foldable as F
@ -22,39 +21,28 @@ data Header
decode :: ByteString -> Either String Header decode :: ByteString -> Either String Header
decode = parseOnly parseHeader decode = parseOnly parseHeader
{-# INLINE decode #-} where
parseHeader :: Parser Header parseHeader :: Parser Header
parseHeader = parseFrom <|> parseTo parseHeader = parseFrom <|> parseTo
{-# INLINE parseHeader #-}
parseFrom :: Parser Header parseFrom :: Parser Header
parseFrom = From <$> (string "From:" *> emptySpace *> (bracketEmail <|> email)) parseFrom = From <$> (string "From:" *> emptySpace *> (bracketEmail <|> email))
parseTo :: Parser Header parseTo :: Parser Header
parseTo = To <$> (string "To:" *> emptySpace *> emails) parseTo = To <$> (string "To:" *> emptySpace *> emails)
emptySpace = many' space
emptySpace :: Parser ()
emptySpace = () <$ many' space
emails :: Parser (Vector ByteString) emails :: Parser (Vector ByteString)
emails = V.fromList <$> (bracketEmail <|> email) `sepBy` char ',' emails = V.fromList <$> (bracketEmail <|> email) `sepBy` char ','
bracketEmail :: Parser ByteString bracketEmail :: Parser ByteString
bracketEmail = do bracketEmail = do
_ <- manyTill anyChar (char '<') _ <- manyTill anyChar (char '<')
email email
{-# INLINE bracketEmail #-}
email :: Parser ByteString email :: Parser ByteString
email = do email = do
_ <- many' space _ <- many' space
name <- BC.pack <$> many' (satisfy (\c -> not (isSpace c) && c /= '@')) name <- BC.pack <$> many' (satisfy (\c -> not (isSpace c) && c /= '@'))
_ <- char '@' _ <- char '@'
rest <- BC.pack <$> many' (satisfy (\c -> not (isSpace c) && c /= ',' && c /= '>' && c /= '<')) rest <- BC.pack <$> many' (satisfy (\c -> not (isSpace c) && c /= ',' && c /= '>'))
_ <- many' (notChar ',') _ <- many' (notChar ',')
pure (name <> "@" <> rest) pure (name <> "@" <> rest)
{-# INLINE email #-}
encode :: Header -> ByteString encode :: Header -> ByteString

View File

@ -31,7 +31,4 @@ tests = testGroup "Data.Email"
[ testCase "Can parse a sample email" $ do [ testCase "Can parse a sample email" $ do
got <- parseToList sample got <- parseToList sample
got @?= [ From "me@example.com", To ["you@example.com"]] got @?= [ From "me@example.com", To ["you@example.com"]]
, testCase "Combined mailto" $ do
let got = decode "To: James Doe<james.doe@example.com<mailto:james.doe@example.com>>"
got @?= Right (To ["james.doe@example.com"])
] ]