-- | Import CT type structures

module Biobase.DataSource.CT where

import qualified Data.Array.IArray as A
import Data.List.Split



-- | A CT file. Encodes one or more individual RNAs within one complex. The
-- 'nucs' array encodes (1,n) as '0' has a special meaning. It is used in
-- 'prv', 'nxt' do denote no next or previous nucleotide in the chain.

data CT = CT
  { comments :: [String]
  , lengthLine :: String
  , nucs :: A.Array Int Nuc
  }
  deriving (Show,Read)

-- | A nucleotide. Encodes character, previous and next nucleotide on the chain
-- and pairing partner.

data Nuc = Nuc
  { nuc :: Char
  , prv :: Int
  , nxt :: Int
  , partner :: Int
  }
  deriving (Show,Read)

-- | transforms a 'CT' object into our internal format.

fromCT :: CT -> ()
fromCT ct = res where
  res = ()
  chns = split (keepDelimsR $ whenElt ((==0) . nxt)) . A.elems $ nucs ct