diff --git a/benchmarks/benchmark_v0.2.hs b/benchmarks/benchmark_v0.2.hs --- a/benchmarks/benchmark_v0.2.hs +++ b/benchmarks/benchmark_v0.2.hs @@ -1,18 +1,20 @@ {-# LANGUAGE OverloadedStrings #-} import Bio.Sequence.Stockholm import Bio.Sequence.Stockholm.Stream -import Control.Monad.Exception.Synchronous (Exceptional(..)) import System.IO (stdin) import qualified Data.Conduit as C import qualified Data.Conduit.Binary as CB import qualified Data.Conduit.List as CL +main :: IO () main = main_event +main_doc :: IO () main_doc = C.runResourceT $ CB.sourceHandle stdin C.$$ parseStockholm C.=$ CL.mapM_ printName where printName (Stockholm file _ _) = print (findAnn AC file) +main_event :: IO () main_event = C.runResourceT $ CB.sourceHandle stdin C.$$ parseEvents C.=$ CL.mapM_ printName where printName (EvGF "AC" name) = print name diff --git a/biostockholm.cabal b/biostockholm.cabal --- a/biostockholm.cabal +++ b/biostockholm.cabal @@ -1,5 +1,5 @@ Name: biostockholm -Version: 0.2 +Version: 0.2.1 Synopsis: Parsing and rendering of Stockholm files (used by Pfam, Rfam and Infernal). License: BSD3 License-file: LICENSE @@ -40,11 +40,11 @@ , containers >= 0.2 && < 0.5 , bytestring == 0.9.* , deepseq >= 1.1 && < 1.3 - , conduit == 0.1.* + , conduit == 0.2.* , attoparsec == 0.10.* - , attoparsec-conduit == 0.0.* + , attoparsec-conduit == 0.2.* , blaze-builder == 0.3.* - , blaze-builder-conduit == 0.0.* + , blaze-builder-conduit == 0.2.* , biocore >= 0.1 && < 0.3 @@ -54,17 +54,15 @@ Main-is: runtests.hs Ghc-Options: -Wall Build-depends: - base >= 3 && < 5 - , containers >= 0.2 && < 0.5 - , bytestring == 0.9.* - , conduit == 0.1.* + base + , containers + , bytestring + , conduit , zlib-conduit - , transformers >= 0.2 - - , biocore >= 0.1 && < 0.3 + , biocore + , transformers >= 0.2 , hspec == 0.9.* , HUnit , QuickCheck - , biostockholm diff --git a/src/Bio/Sequence/Stockholm/Document.hs b/src/Bio/Sequence/Stockholm/Document.hs --- a/src/Bio/Sequence/Stockholm/Document.hs +++ b/src/Bio/Sequence/Stockholm/Document.hs @@ -333,7 +333,7 @@ close (InsideStockholm annots seqs) = return [makeStockholm annots seqs] push state (EvComment _) = - return (state, C.Producing []) + return (C.StateProducing state []) push LookingForHeader EvHeader = continue (emptyPA, M.empty) @@ -343,7 +343,7 @@ push (InsideStockholm _ _) EvHeader = fail "parseDoc: unexpected header" push (InsideStockholm annots seqs) EvEnd = - return (LookingForHeader, C.Producing [makeStockholm annots seqs]) + return (C.StateProducing LookingForHeader [makeStockholm annots seqs]) push (InsideStockholm annots seqs) (EvSeqData label data_) = continue (annots, insertDM (label, data_) seqs) push (InsideStockholm annots seqs) (EvGF feat data_) = @@ -355,7 +355,7 @@ push (InsideStockholm annots seqs) (EvGR sq feat data_) = continue (insertPA_GR sq (Ann (parseClmnFeature feat) data_) annots, seqs) - continue (annots, seqs) = return (InsideStockholm annots seqs, C.Producing []) + continue (annots, seqs) = return (C.StateProducing (InsideStockholm annots seqs) []) {-# INLINE continue #-} data ParseDoc = LookingForHeader diff --git a/src/Bio/Sequence/Stockholm/Stream.hs b/src/Bio/Sequence/Stockholm/Stream.hs --- a/src/Bio/Sequence/Stockholm/Stream.hs +++ b/src/Bio/Sequence/Stockholm/Stream.hs @@ -60,16 +60,17 @@ eventParser = hash *> (ann <|> comment) <|> end <|> seqdata + A. "Event" where word = A.takeTill A8.isHorizontalSpace <* spaces tillNextLine = A.takeLazyByteString hash = A8.char '#' - comment = EvComment <$> tillNextLine - end = EvEnd <$ A8.string "//" <* spaces - seqdata = EvSeqData <$> word <*> tillNextLine + comment = EvComment <$> tillNextLine A. "Comment" + end = EvEnd <$ A8.string "//" <* spaces A. "End of Stockholm file (//)" + seqdata = EvSeqData <$> word <*> tillNextLine A. "Sequence data" - ann = A8.string "=G" *> (gf <|> gc <|> gs <|> gr) + ann = A8.string "=G" *> (gf <|> gc <|> gs <|> gr) A. "Annotation" where gf = EvGF <$ A8.char 'F' <* spaces <*> word <*> tillNextLine gc = EvGC <$ A8.char 'C' <* spaces <*> word <*> tillNextLine @@ -79,6 +80,7 @@ -- | Parse 'EvHeader'. headerParser :: A.Parser Event headerParser = EvHeader <$ A8.char '#' <* spaces <* mystring "STOCKHOLM 1.0" <* spaces + A. "Stockholm Header" where mystring (x:xs) = A8.char x *> mystring xs mystring [] = pure () diff --git a/tests/runtests.hs b/tests/runtests.hs --- a/tests/runtests.hs +++ b/tests/runtests.hs @@ -85,7 +85,7 @@ liftIO (zipWithM_ (@?=) parsed2 parsed1) describe "renderEvents/parseEvents" $ do - prop "passes QuickCheck property" $ \(events :: [Event]) -> + prop "passes QuickCheck property" $ forAll eventList $ \(events :: [Event]) -> unsafePerformIO $ do rendered <- C.runResourceT $ CL.sourceList events C.$$ renderEvents C.=$ CL.consume again <- C.runResourceT $ CL.sourceList rendered C.$$ parseEvents C.=$ CL.consume @@ -192,13 +192,13 @@ , (2, EvGR <$> seqlabel <*> feature <*> arbitrary) ] where seqlabel = strict . unSL <$> arbitrary - seqdata = strict . unSD <$> arbitrary + seqdata = unSD <$> arbitrary feature = B.pack <$> listOf1 (elements alpha) strict = B.concat . L.toChunks eventList :: Gen [Event] eventList = sized $ \s -> frequency [ (100, single) - , (s, (++) <$> single <*> (resize (s `div` 2) eventList)) ] + , (s, (++) <$> single <*> (resize (s - 1) eventList)) ] where single = (\xs -> EvHeader : xs ++ [EvEnd]) <$> listOf arbitrary @@ -283,7 +283,7 @@ (before, rest) -> (EvHeader : before) : separate rest separate (x:xs) = [x] : separate xs separate [] = [] - (<>) = B.append + (<>) = L.append glue (EvSeqData sq1 data1 : EvSeqData sq2 data2 : xs) | sq1 == sq2 = glue (EvSeqData sq1 (data1 <> data2) : xs) glue (EvGF feat1 data1 : EvGF feat2 data2 : xs)