Version 0.1.3.7. Generic fasta parser (#41)

ChangeLog.md

@@ -2,6 +2,10 @@
 
 ## [Unreleased]
 
+## [0.1.3.7] - 2020-10-14
+### Added
+- Generic fasta parser. 
+
 ## [0.1.3.6] - 2020-07-14
 ### Added
 - Convertation from `Model`s to `PDB`.

package.yaml

@@ -1,5 +1,5 @@
 name:                cobot-io
-version:             0.1.3.6
+version:             0.1.3.7
 github:              "biocad/cobot-io"
 license:             BSD3
 category:            Bio

src/Bio/FASTA.hs

@@ -5,6 +5,7 @@ module Bio.FASTA
   , fromFile
   , toFile
   , fastaP
+  , fastaPGeneric
   ) where
 
 import           Bio.FASTA.Parser

src/Bio/FASTA/Parser.hs

@@ -1,31 +1,35 @@
 module Bio.FASTA.Parser
   ( fastaP
+  , fastaPGeneric
   ) where
 
-import           Bio.FASTA.Type       (Fasta, FastaItem (..))
-import           Bio.Sequence         (BareSequence, bareSequence)
-import           Data.Attoparsec.Text (Parser, char, choice, endOfInput,
-                                       endOfLine, letter, many', many1',
-                                       takeWhile)
-import           Data.Text            (Text, strip)
-import           Prelude              hiding (takeWhile)
+import Bio.FASTA.Type       (Fasta, FastaItem (..))
+import Bio.Sequence         (BareSequence, bareSequence)
+import Data.Attoparsec.Text (Parser, char, choice, endOfInput, endOfLine, many', many1', satisfy,
+                             takeWhile)
+import Data.Char            (isLetter)
+import Data.Text            (Text, strip)
+import Prelude              hiding (takeWhile)
 
 -- | Parser of .fasta file.
 --
 fastaP :: Parser (Fasta Char)
-fastaP = many' item
+fastaP = fastaPGeneric isLetter
 
-item :: Parser (FastaItem Char)
-item = FastaItem <$> seqName <*> fastaSeq
+fastaPGeneric :: (Char -> Bool) -> Parser (Fasta Char)
+fastaPGeneric = many' . item
 
-seqName :: Parser (Text)
+item :: (Char -> Bool) -> Parser (FastaItem Char)
+item predicate = FastaItem <$> seqName <*> fastaSeq predicate
+
+seqName :: Parser Text
 seqName = strip <$> (char '>' *> tabs *> takeWhile (`notElem` ['\n', '\r']) <* tabs <* eol)
 
-fastaSeq :: Parser (BareSequence Char)
-fastaSeq = bareSequence . mconcat <$> many' line
+fastaSeq :: (Char -> Bool) -> Parser (BareSequence Char)
+fastaSeq predicate = bareSequence . mconcat <$> many' (line predicate)
 
-line :: Parser String
-line = concat <$> many1' (many1' letter <* many' (char ' ')) <* eol
+line :: (Char -> Bool) -> Parser String
+line predicate = concat <$> many1' (many1' (satisfy predicate) <* many' (char ' ')) <* eol
 
 eol :: Parser ()
 eol = tabs *> choice [slashN, endOfInput]

src/Bio/Sequence/Class.hs

@@ -14,6 +14,7 @@ module Bio.Sequence.Class
   , sequ
   , markings
   , weights
+  , bareSequ
 
   -- classes for weights and markings of sequence
   , IsMarking
@@ -87,7 +88,7 @@ instance Foldable (Sequence mk w) where
   length = V.length . _sequ
 
 instance Traversable (Sequence mk w) where
-  traverse f s@Sequence{..} = fmap (\newSeq -> s { _sequ = newSeq }) $ traverse f _sequ
+  traverse f s@Sequence{..} = (\newSeq -> s { _sequ = newSeq }) <$> traverse f _sequ
 
 -- | Exported constructor for 'Sequence'. Should be used ONLY in module Bio.Sequence.
 --
@@ -113,6 +114,7 @@ type BareSequence a = Sequence () () a
 --------------------------------------------------------------------------------
 -- Lenses for 'Sequence'.
 -- We create only getters, so user that couldn't ruin 'Sequence's invariant.
+-- But we can create a Lens for 'BareSequence', and it won't ruin any invariants.
 --------------------------------------------------------------------------------
 
 sequ :: Getter (Sequence mk w a) (Vector a)
@@ -124,6 +126,9 @@ markings = to _markings
 weights :: Getter (Sequence mk w a) (Vector w)
 weights  = to _weights
 
+bareSequ :: Lens' (BareSequence a) (Vector a)
+bareSequ = lens _sequ (\s v -> s { _sequ = v })
+
 
 --------------------------------------------------------------------------------
 -- IsMarking class.