-- Hoogle documentation, generated by Haddock
-- See Hoogle, http://www.haskell.org/hoogle/


-- | streaming FASTA parser
--   
--   Stream-based handling of FASTA files. The user selects a window size,
--   the library then handles the window. For each window, the previous
--   (past) window is available, in case some data sits on the boundary
--   between windows.
--   
--   FastaTool is a simple tool providing information on FASTA files, and
--   allowing to extract sequences and subsequences.
--   
--   Greg Schwartz' <a>http://hackage.haskell.org/package/fasta</a> package
--   is a lot more complete. This one is mostly tailored to my usage
--   requirements (and may at some point use his library).
@package BiobaseFasta
@version 0.2.0.0

module Biobase.Fasta.Types

data Fasta
Fasta :: ByteString -> ByteString -> Fasta
[fastaHeader] :: Fasta -> ByteString
[fastaSequence] :: Fasta -> ByteString
newtype RawFastaEntry
RawFastaEntry :: ByteString -> RawFastaEntry
[_rawFastaEntry] :: RawFastaEntry -> ByteString

-- | <a>StreamEvent</a>s are chunked pieces of data, where the raw data is
--   a strict <tt>ByteString</tt>. Each element also retains information on
--   the first and last line and column (via <a>streamLines</a>) that are
--   part of this chunk.
data StreamEvent

-- | A Header event, multiple header events signal that the header name was
--   longer than the chunk size.
StreamHeader :: !ByteString -> !LineInfo -> StreamEvent
[streamHeader] :: StreamEvent -> !ByteString
[streamLines] :: StreamEvent -> !LineInfo

-- | A data event. We keep a pointer to the previous chunk (which is useful
--   for some algorithms). The chunk is free of newlines!
StreamFasta :: !ByteString -> !ByteString -> !LineInfo -> !ByteString -> StreamEvent
[streamFasta] :: StreamEvent -> !ByteString
[prevStreamFasta] :: StreamEvent -> !ByteString
[streamLines] :: StreamEvent -> !LineInfo
[streamHeader] :: StreamEvent -> !ByteString

-- | Complete information on line and column start and end for a chunk.
--   
--   TODO This is a 1-based format? Lets use the BiobaseTypes facilities!
data LineInfo
LineInfo :: !Int -> !Int -> !Int -> !Int -> !Int -> LineInfo

-- | first line for this chunk <tt>(lines in complete file!)</tt>
[firstLine] :: LineInfo -> !Int

-- | first column in first line for this chunk
[firstCol] :: LineInfo -> !Int

-- | last line for this chunk <tt>(lines in complete file!)</tt>
[lastLine] :: LineInfo -> !Int

-- | last column in last line for this chunk
[lastCol] :: LineInfo -> !Int

-- | first index in this fasta block. Counts just the number of symbols in
--   the <tt>Fasta</tt> payload.
[firstIndex] :: LineInfo -> !Int
instance GHC.Generics.Generic Biobase.Fasta.Types.StreamEvent
instance GHC.Classes.Ord Biobase.Fasta.Types.StreamEvent
instance GHC.Classes.Eq Biobase.Fasta.Types.StreamEvent
instance GHC.Show.Show Biobase.Fasta.Types.StreamEvent
instance GHC.Generics.Generic Biobase.Fasta.Types.LineInfo
instance GHC.Classes.Ord Biobase.Fasta.Types.LineInfo
instance GHC.Classes.Eq Biobase.Fasta.Types.LineInfo
instance GHC.Show.Show Biobase.Fasta.Types.LineInfo
instance GHC.Classes.Ord Biobase.Fasta.Types.RawFastaEntry
instance GHC.Classes.Eq Biobase.Fasta.Types.RawFastaEntry
instance GHC.Show.Show Biobase.Fasta.Types.RawFastaEntry
instance GHC.Classes.Eq Biobase.Fasta.Types.Fasta
instance Control.DeepSeq.NFData Biobase.Fasta.Types.StreamEvent
instance Control.DeepSeq.NFData Biobase.Fasta.Types.LineInfo


-- | Streaming Fasta handling via the <tt>streaming</tt> library.
--   
--   The functions in here should be streaming in constant memory.
--   
--   TODO Check if this is actually true with some unit tests.
module Biobase.Fasta.Streaming

-- | Control structure for <a>streamingFasta</a>.
data FindHeader
FindHeader :: [ByteString] -> !Int -> FindHeader

-- | the collected header parts (in reverse order)
[headerParts] :: FindHeader -> [ByteString]

-- | accumulated header length
[headerLength] :: FindHeader -> !Int
HasHeader :: !ByteString -> !ByteString -> [ByteString] -> !Int -> !Int -> FindHeader

-- | the (size-truncated) header for this fasta file
[header] :: FindHeader -> !ByteString

-- | overlap (if any) from earlier parts of the fasta file
[dataOverlap] :: FindHeader -> !ByteString

-- | collection of dataParts, in reverse order!
[dataParts] :: FindHeader -> [ByteString]

-- | total length of data parts, simplifies checking if enough data was
--   collected
[dataLength] :: FindHeader -> !Int

-- | count how many entries we have seen
[entries] :: FindHeader -> !Int

-- | Current Fasta window, together with the start index (0-based).
data Current (which :: k)
Current :: ByteString -> Index 0 -> Current
[currentFasta] :: Current -> ByteString
[currentStart] :: Current -> Index 0
newtype Overlap (which :: k)
Overlap :: ByteString -> Overlap
[getOverlap] :: Overlap -> ByteString
newtype Header (which :: k)
Header :: ByteString -> Header
[getHeader] :: Header -> ByteString
newtype CurrentSize
CurrentSize :: Int -> CurrentSize
newtype OverlapSize
OverlapSize :: Int -> OverlapSize
newtype HeaderSize
HeaderSize :: Int -> HeaderSize

-- | Fully stream a fasta file, making sure to never exceed a constant
--   amount of memory. The <tt>go</tt> function yields values of type
--   <tt>a</tt> down the line for continued streaming.
--   
--   <pre>
--   r4 = toList . streamingFasta (HeaderSize 2) (OverlapSize 1) (CurrentSize 2) go . S8.fromStrict $ BS.pack t0
--    where go (Header h) (Overlap o) (Current c) = yield (h,o,c)
--   </pre>
streamingFasta :: forall m w r a. Monad m => HeaderSize -> OverlapSize -> CurrentSize -> (Header w -> Overlap w -> Current w -> Stream (Of a) m ()) -> ByteString m r -> Stream (Of a) m r
eachFasta :: Monad m => Header which1 -> Overlap which2 -> Current which3 -> Stream (Of (ByteString, ByteString, ByteString)) m ()
parseFastaFile :: FilePath -> IO [Fasta]
parseFasta :: ByteString -> [Fasta]
instance forall k (which :: k). GHC.Show.Show (Biobase.Fasta.Streaming.Current which)
instance forall k (which :: k). GHC.Classes.Ord (Biobase.Fasta.Streaming.Current which)
instance forall k (which :: k). GHC.Classes.Eq (Biobase.Fasta.Streaming.Current which)
instance forall k (which :: k). GHC.Show.Show (Biobase.Fasta.Streaming.Overlap which)
instance forall k (which :: k). GHC.Classes.Ord (Biobase.Fasta.Streaming.Overlap which)
instance forall k (which :: k). GHC.Classes.Eq (Biobase.Fasta.Streaming.Overlap which)
instance forall k (which :: k). GHC.Show.Show (Biobase.Fasta.Streaming.Header which)
instance forall k (which :: k). GHC.Classes.Ord (Biobase.Fasta.Streaming.Header which)
instance forall k (which :: k). GHC.Classes.Eq (Biobase.Fasta.Streaming.Header which)
instance GHC.Show.Show Biobase.Fasta.Streaming.CurrentSize
instance GHC.Classes.Ord Biobase.Fasta.Streaming.CurrentSize
instance GHC.Classes.Eq Biobase.Fasta.Streaming.CurrentSize
instance GHC.Show.Show Biobase.Fasta.Streaming.OverlapSize
instance GHC.Classes.Ord Biobase.Fasta.Streaming.OverlapSize
instance GHC.Classes.Eq Biobase.Fasta.Streaming.OverlapSize
instance GHC.Show.Show Biobase.Fasta.Streaming.HeaderSize
instance GHC.Classes.Ord Biobase.Fasta.Streaming.HeaderSize
instance GHC.Classes.Eq Biobase.Fasta.Streaming.HeaderSize


-- | Fasta export
module Biobase.Fasta.Export
prettyPrintFasta :: Int -> Fasta -> String
prettyByteStringFasta :: Int -> Fasta -> ByteString
breakByteString :: Int -> ByteString -> [ByteString]
fromIntToInt64 :: Int -> Int64
writeFastaFile :: FilePath -> [Fasta] -> IO ()
instance GHC.Show.Show Biobase.Fasta.Types.Fasta