{-|
Module      : Main.hs
Description : Gargantext starter binary with Phylo
Copyright   : (c) CNRS, 2017-Present
License     : AGPL + CECILL v3
Maintainer  : team@gargantext.org
Stability   : experimental
Portability : POSIX

Phylo binaries

 -}

{-# LANGUAGE DataKinds         #-}
{-# LANGUAGE DeriveGeneric     #-}
{-# LANGUAGE FlexibleInstances #-}
{-# LANGUAGE NoImplicitPrelude #-}
{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE StandaloneDeriving #-}
{-# LANGUAGE TypeOperators     #-}
{-# LANGUAGE Strict             #-}

module Main where

import System.Directory (doesFileExist) 

import Data.Aeson
import Data.Text (Text, unwords)
import Data.List ((++))
import GHC.Generics
import GHC.IO (FilePath)
import Gargantext.Prelude
import Gargantext.Text.List.CSV (csvGraphTermList)
import Gargantext.Text.Parsers.CSV (csv_title, csv_abstract, csv_publication_year)
import qualified Gargantext.Text.Parsers.CSV as CSV
import Gargantext.Text.Parsers (FileFormat(..),parseFile)
import Gargantext.Text.Terms.WithList
import Gargantext.Text.Context (TermList)

import System.Environment

import Gargantext.Viz.Phylo
import Gargantext.Viz.Phylo.Tools
import Gargantext.Viz.Phylo.LevelMaker
import Gargantext.Viz.Phylo.View.Export
import Gargantext.Viz.Phylo.View.ViewMaker

import Gargantext.Database.Types.Node
import Data.Maybe

import qualified Data.Map    as DM
import qualified Data.Vector as DV
import qualified Data.List   as DL
import qualified Data.Text   as DT
import qualified Prelude     as P
import qualified Data.ByteString.Lazy as L


--------------
-- | Conf | --
--------------


type ListPath   = FilePath
type FisPath    = FilePath
type CorpusPath = FilePath
data CorpusType = Wos | Csv deriving (Show,Generic) 
type Limit = Int

data Conf = 
     Conf { corpusPath  :: CorpusPath
          , corpusType  :: CorpusType
          , listPath    :: ListPath
          , fisPath     :: FilePath
          , outputPath  :: FilePath
          , phyloName   :: Text
          , limit       :: Limit
          , timeGrain   :: Int
          , timeStep    :: Int
          , timeFrame   :: Int
          , timeTh      :: Double
          , timeSens    :: Double
          , clusterTh   :: Double
          , clusterSens :: Double
          , phyloLevel  :: Int
          , viewLevel   :: Int
          , fisSupport  :: Int
          , fisClique   :: Int
          , minSizeBranch :: Int 
     } deriving (Show,Generic)

instance FromJSON Conf
instance ToJSON Conf

instance FromJSON CorpusType
instance ToJSON CorpusType


decoder :: P.Either a b -> b
decoder (P.Left _) = P.error "Error"
decoder (P.Right x) = x

-- | Get the conf from a Json file
getJson :: FilePath -> IO L.ByteString
getJson path = L.readFile path


---------------
-- | Parse | --
---------------


-- | To filter the Ngrams of a document based on the termList
filterTerms :: Patterns -> (a, Text) -> (a, [Text])
filterTerms patterns (y,d) = (y,termsInText patterns d)
  where
    --------------------------------------
    termsInText :: Patterns -> Text -> [Text]
    termsInText pats txt = DL.nub $ DL.concat $ map (map unwords) $ extractTermsWithList pats txt
    --------------------------------------


-- | To transform a Csv nfile into a readable corpus
csvToCorpus :: Limit -> CorpusPath -> IO ([(Int,Text)])
csvToCorpus limit csv = DV.toList
                      -- . DV.reverse
                      . DV.take limit
                      -- . DV.reverse
                      . DV.map (\n -> (csv_publication_year n, (csv_title n) <> " " <> (csv_abstract n)))
                      . snd <$> CSV.readFile csv


-- | To transform a Wos nfile into a readable corpus
wosToCorpus :: Limit -> CorpusPath -> IO ([(Int,Text)])
wosToCorpus limit path = DL.take limit
                         . map (\d -> ((fromJust $_hyperdataDocument_publication_year d)
                                    ,(fromJust $_hyperdataDocument_title d) <> " " <> (fromJust $_hyperdataDocument_abstract d)))
                         . filter (\d -> (isJust $_hyperdataDocument_publication_year d)
                                      && (isJust $_hyperdataDocument_title d)
                                      && (isJust $_hyperdataDocument_abstract d))
                         <$> parseFile WOS path


-- | To use the correct parser given a CorpusType
fileToCorpus :: CorpusType -> Limit -> CorpusPath -> IO ([(Int,Text)])
fileToCorpus format limit path = case format of 
  Wos -> wosToCorpus limit path
  Csv -> csvToCorpus limit path


-- | To parse a file into a list of Document
parse :: CorpusType -> Limit -> CorpusPath -> TermList -> IO [Document]
parse format limit path l = do
  corpus <- fileToCorpus format limit path
  let patterns = buildPatterns l
  pure $ map ( (\(y,t) -> Document y t) . filterTerms patterns) corpus


-- | To parse an existing Fis file
parseFis :: FisPath -> Text -> Int -> Int -> Int -> Int -> IO [PhyloFis]
parseFis path name grain step support clique = do
  fisExists <- doesFileExist (path <> (DT.unpack name) <> "_" <> show(grain) <> "_" <> show(step) <> "_" <> show(support) <> "_" <> show(clique) <> ".json")
  if fisExists 
    then do 
      fisJson <- (eitherDecode <$> getJson (path <> (DT.unpack name) <> "_" <> show(grain) <> "_" <> show(step) <> "_" <> show(support) <> "_" <> show(clique) <> ".json")) :: IO (P.Either P.String [PhyloFis])
      case fisJson of 
        P.Left err -> do 
                       putStrLn err
                       pure []
        P.Right fis -> pure fis
    else pure []

writeFis :: FisPath -> Text -> Int -> Int -> Int -> Int -> DM.Map (Date,Date) [PhyloFis] -> IO ()
writeFis path name grain step support clique fis = do 
  let fisPath = path <> (DT.unpack name) <> "_" <> show(grain) <> "_" <> show(step) <> "_" <> show(support) <> "_" <> show(clique) <> ".json"
  L.writeFile fisPath $ encode (DL.concat $ DM.elems fis)

--------------
-- | Main | --
--------------


main :: IO ()
main = do 

  [jsonPath] <- getArgs

  confJson <- (eitherDecode <$> getJson jsonPath) :: IO (P.Either P.String Conf)

  case confJson of
    P.Left err -> putStrLn err
    P.Right conf -> do

      termList <- csvGraphTermList (listPath conf)

      corpus <- parse (corpusType conf) (limit conf) (corpusPath conf) termList

      putStrLn $ ("\n" <> show (length corpus) <> " parsed docs")

      let roots = DL.nub $ DL.concat $ map text corpus

      putStrLn $ ("\n" <> show (length roots) <> " parsed foundation roots")      

      fis <- parseFis (fisPath conf) (phyloName conf) (timeGrain conf) (timeStep conf) (fisSupport conf) (fisClique conf)

      putStrLn $ ("\n" <> show (length fis) <> " parsed fis")

      let mFis  = DM.fromListWith (++) $ DL.sortOn (fst . fst) $ map (\f -> (getFisPeriod f,[f])) fis    
      
      let query = PhyloQueryBuild (phyloName conf) "" (timeGrain conf) (timeStep conf) 
                  (Fis $ FisParams True (fisSupport conf) (fisClique conf)) [] [] (WeightedLogJaccard $ WLJParams (timeTh conf) (timeSens conf)) (timeFrame conf) (phyloLevel conf)
                  (RelatedComponents $ RCParams $ WeightedLogJaccard $ WLJParams (clusterTh conf) (clusterSens conf))

      let queryView = PhyloQueryView (viewLevel conf) Merge False 1 [BranchAge] [SizeBranch $ SBParams (minSizeBranch conf)] [BranchPeakFreq,GroupLabelCooc] (Just (ByBranchAge,Asc)) Json Flat True           

      let phylo = toPhylo query corpus roots termList mFis

      writeFis (fisPath conf) (phyloName conf) (timeGrain conf) (timeStep conf) (fisSupport conf) (fisClique conf) (getPhyloFis phylo)

      let view  = toPhyloView queryView phylo

      putStrLn $ ("phylo completed until level " <> show (phyloLevel conf) <> ", export at level " <> show (viewLevel conf)) 

      let outputFile = (outputPath conf) <> (DT.unpack $ phyloName conf)
                                         <> "_" <> show (limit conf) <> "_"
                                         <> "_" <> show (timeTh conf) <> "_"
                                         <> "_" <> show (timeSens conf) <> "_"
                                         <> "_" <> show (clusterTh conf) <> "_"
                                         <> "_" <> show (clusterSens conf) 
                                         <> ".dot"

      P.writeFile outputFile $ dotToString $ viewToDot view