cobot-io 0.1.3.14 → 0.1.3.15

raw patch · 4 files changed

+28/−15 lines, 4 files

Files

• ChangeLog.md +4/−0
• cobot-io.cabal +2/−2
• src/Bio/GB/Parser.hs +15/−13
• test/GBParserSpec.hs +7/−0

@@ -2,6 +2,10 @@  ## [Unreleased] +## [0.1.3.15] - 2021-02-16+### Fixed+- Unknown fields in GB.+ ## [0.1.3.14] - 2021-02-10 ### Changed - Added ability to parse modifications in Fasta.

@@ -4,10 +4,10 @@ -- -- see: https://github.com/sol/hpack ----- hash: b4805dfc259e39022dd3313837d29888036016ead10be1b3b52215e16a76f7a7+-- hash: 41de9b4b3a1a9916fe78ee831e4e9811e27725529c468f9b43d3378b25a5bc5b  name:           cobot-io-version:        0.1.3.14+version:        0.1.3.15 synopsis:       Biological data file formats and IO description:    Please see the README on GitHub at <https://github.com/biocad/cobot-io#readme> category:       Bio

@@ -4,17 +4,18 @@   ( genBankP   ) where -import Bio.GB.Type          (Feature (..), Form (..), GenBankSequence (..), Locus (..), Meta (..),-                             Reference (..), Source (..), Version (..))-import Bio.Sequence         (MarkedSequence, Range, markedSequence)-import Control.Applicative  ((<|>))-import Data.Attoparsec.Text (Parser, char, decimal, digit, endOfInput, endOfLine, letter, many',-                             many1', satisfy, string, takeWhile, takeWhile1)-import Data.Bifunctor       (bimap)-import Data.Char            (isAlphaNum, isSpace, isUpper)-import Data.Functor         (($>))-import Data.Text            (Text, intercalate, pack, splitOn, unpack)-import Prelude              hiding (takeWhile)+import Bio.GB.Type                (Feature (..), Form (..), GenBankSequence (..), Locus (..),+                                   Meta (..), Reference (..), Source (..), Version (..))+import Bio.Sequence               (MarkedSequence, Range, markedSequence)+import Control.Applicative        ((<|>))+import Data.Attoparsec.Combinator (manyTill)+import Data.Attoparsec.Text       (Parser, char, decimal, digit, endOfInput, endOfLine, letter,+                                   many', many1', satisfy, string, takeWhile, takeWhile1)+import Data.Bifunctor             (bimap)+import Data.Char                  (isAlphaNum, isSpace, isUpper)+import Data.Functor               (($>))+import Data.Text                  (Text, intercalate, pack, splitOn, unpack)+import Prelude                    hiding (takeWhile)  -- | Parser of .gb file. --@@ -55,7 +56,7 @@     textP = takeWhile1 $ not . isSpace      formP :: Parser Form-    formP = (string "linear" $> Linear) <|> (string "circular" *> pure Circular)+    formP = (string "linear" $> Linear) <|> (string "circular" $> Circular)  definitionP :: Parser Text definitionP =  string "DEFINITION" *> space *> (emptyP <|> someLinesP)@@ -104,7 +105,8 @@ --------------------------------------------------------------------------------  featuresP :: Parser [(Feature, Range)]-featuresP =  string "FEATURES" *> space+featuresP =  manyTill (textWithSpacesP <* eolSpaceP) (string "FEATURES") *> space+          -- ^ skip unknown fields and stop on line with "FEATURES"           *> textWithSpacesP <* eolSpaceP           *> many1' featureP 

@@ -13,6 +13,7 @@     pAAVGFPSpecP "test/GB/pAAV-GFP-CellBioLab.gb"     pAAVCMVSpecP "test/GB/pAAV_CMV_RPE65_PolyA_linkers.gb"     dottedMetaSpecP "test/GB/pAAV-GFP-CellBioLab-dots.gb"+    unknownFieldsSpecP "test/GB/pIntA-TRBV.gb"  pAAVGFPSpecP :: FilePath -> Spec pAAVGFPSpecP path = describe "pAAVGFP" $ do@@ -39,6 +40,12 @@     it "correctly parses meta information with all info set to dots" $ do         mt <- meta <$> fromFile path         mt `shouldBe` dottedMeta++unknownFieldsSpecP :: FilePath -> Spec+unknownFieldsSpecP path = describe "Unknown fields" $ do+    it "correctly parses and skip unknown fields before FEATURES" $ do+        mt <- meta <$> fromFile path+        name (locus mt) `shouldBe` "P2-32_pIntA-TRBV5-1_J1-1-Fc-lama-knob-EPEA"   pAAVGFPMeta :: Meta