{-|
Module : Main.hs
-Description : Gargantext starter binary with Phylo
+Description : Gargantext starter binary with Adaptative Phylo
Copyright : (c) CNRS, 2017-Present
License : AGPL + CECILL v3
Maintainer : team@gargantext.org
Stability : experimental
Portability : POSIX
-Phylo binaries
-
+Adaptative Phylo binaries
-}
-{-# LANGUAGE DataKinds #-}
-{-# LANGUAGE DeriveGeneric #-}
-{-# LANGUAGE FlexibleInstances #-}
-{-# LANGUAGE NoImplicitPrelude #-}
-{-# LANGUAGE OverloadedStrings #-}
{-# LANGUAGE StandaloneDeriving #-}
-{-# LANGUAGE TypeOperators #-}
+{-# LANGUAGE TypeOperators #-}
{-# LANGUAGE Strict #-}
module Main where
+-- import Debug.Trace (trace)
+import Control.Concurrent.Async (mapConcurrently)
+import Crypto.Hash.SHA256 (hash)
import Data.Aeson
-import Data.Text (Text, unwords)
-import GHC.Generics
+import Data.Either (Either(..), fromRight)
+import Data.List (concat, nub, isSuffixOf)
+import Data.List.Split
+import Data.Maybe (fromMaybe)
+import Data.String (String)
+import Data.Text (Text, unwords, unpack, replace, pack)
import GHC.IO (FilePath)
+import Gargantext.API.Ngrams.Prelude (toTermList)
+import Gargantext.API.Ngrams.Types
+import Gargantext.Core.Text.Context (TermList)
+import Gargantext.Core.Text.Corpus.Parsers (FileFormat(..), FileType(..), parseFile)
+import Gargantext.Core.Text.Corpus.Parsers.CSV (csv_title, csv_abstract, csv_publication_year, csv_publication_month, csv_publication_day, csv'_source, csv'_title, csv'_abstract, csv'_publication_year, csv'_publication_month, csv'_publication_day, csv'_weight)
+import Gargantext.Core.Text.List.Formats.CSV (csvMapTermList)
+import Gargantext.Core.Text.Terms.WithList (Patterns, buildPatterns, extractTermsWithList)
+import Gargantext.Core.Types.Main (ListType(..))
+import Gargantext.Core.Viz.Phylo
+import Gargantext.Core.Viz.Phylo.API.Tools
+import Gargantext.Core.Viz.Phylo.PhyloExport (toPhyloExport, dotToFile)
+import Gargantext.Core.Viz.Phylo.PhyloMaker (toPhylo, toPhyloStep)
+import Gargantext.Core.Viz.Phylo.PhyloTools (printIOMsg, printIOComment, setConfig)
+import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
+import Gargantext.Database.Schema.Ngrams (NgramsType(..))
import Gargantext.Prelude
-import Gargantext.Text.List.CSV (csvGraphTermList)
-import Gargantext.Text.Parsers.CSV (readCsv, csv_title, csv_abstract, csv_publication_year)
-import Gargantext.Text.Parsers (FileFormat(..),parseDocs)
-import Gargantext.Text.Terms.WithList
-import Gargantext.Text.Context (TermList)
-
+import System.Directory (listDirectory,doesFileExist)
import System.Environment
+import qualified Data.ByteString.Char8 as C8
+import qualified Data.Text as T
+import qualified Data.Vector as Vector
+import qualified Gargantext.Core.Text.Corpus.Parsers.CSV as Csv
-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
+data PhyloStage = PhyloWithCliques | PhyloWithLinks deriving (Show)
-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
+---------------
+-- | Tools | --
+---------------
+-- | To get all the files in a directory or just a file
+getFilesFromPath :: FilePath -> IO [FilePath]
+getFilesFromPath path = do
+ if (isSuffixOf "/" path)
+ then (listDirectory path)
+ else return [path]
+
+----------------
+-- | Parser | --
+----------------
+
+-- | To filter the Ngrams of a document based on the termList
+termsInText :: Patterns -> Text -> [Text]
+termsInText pats txt = nub $ concat $ map (map unwords) $ extractTermsWithList pats txt
+
+
+-- | To transform a Wos file (or [file]) into a list of Docs
+wosToDocs :: Int -> Patterns -> TimeUnit -> FilePath -> IO [Document]
+wosToDocs limit patterns time path = do
+ files <- getFilesFromPath path
+ take limit
+ <$> map (\d -> let title = fromJust $ _hd_title d
+ abstr = if (isJust $ _hd_abstract d)
+ then fromJust $ _hd_abstract d
+ else ""
+ in Document (toPhyloDate
+ (fromIntegral $ fromJust $ _hd_publication_year d)
+ (fromJust $ _hd_publication_month d)
+ (fromJust $ _hd_publication_day d) time)
+ (toPhyloDate'
+ (fromIntegral $ fromJust $ _hd_publication_year d)
+ (fromJust $ _hd_publication_month d)
+ (fromJust $ _hd_publication_day d) time)
+ (termsInText patterns $ title <> " " <> abstr) Nothing [])
+ <$> concat
+ <$> mapConcurrently (\file ->
+ filter (\d -> (isJust $ _hd_publication_year d)
+ && (isJust $ _hd_title d))
+ <$> fromRight [] <$> parseFile WOS Plain (path <> file) ) files
+
+
+-- To transform a Csv file into a list of Document
+csvToDocs :: CorpusParser -> Patterns -> TimeUnit -> FilePath -> IO [Document]
+csvToDocs parser patterns time path =
+ case parser of
+ Wos _ -> undefined
+ Csv limit -> Vector.toList
+ <$> Vector.take limit
+ <$> Vector.map (\row -> Document (toPhyloDate (Csv.fromMIntOrDec Csv.defaultYear $ csv_publication_year row) (fromMaybe Csv.defaultMonth $ csv_publication_month row) (fromMaybe Csv.defaultDay $ csv_publication_day row) time)
+ (toPhyloDate' (Csv.fromMIntOrDec Csv.defaultYear $ csv_publication_year row) (fromMaybe Csv.defaultMonth $ csv_publication_month row) (fromMaybe Csv.defaultDay $ csv_publication_day row) time)
+ (termsInText patterns $ (csv_title row) <> " " <> (csv_abstract row))
+ Nothing
+ []
+ ) <$> snd <$> either (\err -> panic $ cs $ "CSV error" <> (show err)) identity <$> Csv.readCSVFile path
+ Csv' limit -> Vector.toList
+ <$> Vector.take limit
+ <$> Vector.map (\row -> Document (toPhyloDate (csv'_publication_year row) (csv'_publication_month row) (csv'_publication_day row) time)
+ (toPhyloDate' (csv'_publication_year row) (csv'_publication_month row) (csv'_publication_day row) time)
+ (termsInText patterns $ (csv'_title row) <> " " <> (csv'_abstract row))
+ (Just $ csv'_weight row)
+ (map (T.strip . pack) $ splitOn ";" (unpack $ (csv'_source row)))
+ ) <$> snd <$> Csv.readWeightedCsv path
+
+
+-- To parse a file into a list of Document
+fileToDocs' :: CorpusParser -> FilePath -> TimeUnit -> TermList -> IO [Document]
+fileToDocs' parser path time lst = do
+ let patterns = buildPatterns lst
+ case parser of
+ Wos limit -> wosToDocs limit patterns time path
+ Csv _ -> csvToDocs parser patterns time path
+ Csv' _ -> csvToDocs parser patterns time path
---------------
--- | Conf | --
---------------
+---------------
+-- | Label | --
+---------------
-type ListPath = FilePath
-type CorpusPath = FilePath
-data CorpusType = Wos | Csv deriving (Show,Generic)
-type Limit = Int
-data Conf =
- Conf { corpusPath :: CorpusPath
- , corpusType :: CorpusType
- , listPath :: ListPath
- , outputPath :: FilePath
- , phyloName :: Text
- , limit :: Limit
- } deriving (Show,Generic)
+-- Config time parameters to label
+timeToLabel :: PhyloConfig -> [Char]
+timeToLabel config = case (timeUnit config) of
+ Epoch p s f -> ("time_epochs" <> "_" <> (show p) <> "_" <> (show s) <> "_" <> (show f))
+ Year p s f -> ("time_years" <> "_" <> (show p) <> "_" <> (show s) <> "_" <> (show f))
+ Month p s f -> ("time_months" <> "_" <> (show p) <> "_" <> (show s) <> "_" <> (show f))
+ Week p s f -> ("time_weeks" <> "_" <> (show p) <> "_" <> (show s) <> "_" <> (show f))
+ Day p s f -> ("time_days" <> "_" <> (show p) <> "_" <> (show s) <> "_" <> (show f))
-instance FromJSON Conf
-instance ToJSON Conf
-instance FromJSON CorpusType
-instance ToJSON CorpusType
+seaToLabel :: PhyloConfig -> [Char]
+seaToLabel config = case (seaElevation config) of
+ Constante start step -> ("sea_cst_" <> (show start) <> "_" <> (show step))
+ Adaptative granularity -> ("sea_adapt" <> (show granularity))
--- | Get the conf from a Json file
-getJson :: FilePath -> IO L.ByteString
-getJson path = L.readFile path
+sensToLabel :: PhyloConfig -> [Char]
+sensToLabel config = case (phyloProximity config) of
+ Hamming _ -> undefined
+ WeightedLogJaccard s -> ("WeightedLogJaccard_" <> show s)
+ WeightedLogSim s -> ( "WeightedLogSim-sens_" <> show s)
----------------
--- | Parse | --
----------------
+cliqueToLabel :: PhyloConfig -> [Char]
+cliqueToLabel config = case (clique config) of
+ Fis s s' -> "fis_" <> (show s) <> "_" <> (show s')
+ MaxClique s t f -> "clique_" <> (show s)<> "_" <> (show f)<> "_" <> (show t)
-filterTerms :: Patterns -> (a, Text) -> (a, [Text])
-filterTerms patterns (year', doc) = (year',termsInText patterns doc)
- where
- termsInText :: Patterns -> Text -> [Text]
- termsInText pats txt = DL.nub $ DL.concat $ map (map unwords) $ extractTermsWithList pats txt
-
-
-csvToCorpus :: Int -> CorpusPath -> IO ([(Int,Text)])
-csvToCorpus limit csv = DV.toList
- . DV.take limit
- . DV.map (\n -> (csv_publication_year n, (csv_title n) <> " " <> (csv_abstract n)))
- . snd <$> readCsv csv
+syncToLabel :: PhyloConfig -> [Char]
+syncToLabel config = case (phyloSynchrony config) of
+ ByProximityThreshold scl sync_sens scope _ -> ("scale_" <> (show scope) <> "_" <> (show sync_sens) <> "_" <> (show scl))
+ ByProximityDistribution _ _ -> undefined
-wosToCorpus :: Int -> CorpusPath -> IO ([(Int,Text)])
-wosToCorpus limit path = undefined
+qualToConfig :: PhyloConfig -> [Char]
+qualToConfig config = case (phyloQuality config) of
+ Quality g m -> "quality_" <> (show g) <> "_" <> (show m)
-fileToCorpus :: CorpusType -> Int -> CorpusPath -> IO ([(Int,Text)])
-fileToCorpus format limit path = case format of
- Wos -> wosToCorpus limit path
- Csv -> csvToCorpus limit path
+-- To set up the export file's label from the configuration
+configToLabel :: PhyloConfig -> [Char]
+configToLabel config = outputPath config
+ <> (unpack $ phyloName config)
+ <> "-" <> (timeToLabel config)
+ <> "-scale_" <> (show (phyloLevel config))
+ <> "-" <> (seaToLabel config)
+ <> "-" <> (sensToLabel config)
+ <> "-" <> (cliqueToLabel config)
+ <> "-level_" <> (show (_qua_granularity $ phyloQuality config))
+ <> "-" <> (syncToLabel config)
+ <> ".dot"
-parse :: Limit -> CorpusPath -> TermList -> IO [Document]
-parse limit corpus lst = do
- corpus' <- csvToCorpus limit corpus
- let patterns = buildPatterns lst
- pure $ map ( (\(y,t) -> Document y t) . filterTerms patterns) corpus'
+-- To write a sha256 from a set of config's parameters
+configToSha :: PhyloStage -> PhyloConfig -> [Char]
+configToSha stage config = unpack
+ $ replace "/" "-"
+ $ T.pack (show (hash $ C8.pack label))
+ where
+ label :: [Char]
+ label = case stage of
+ PhyloWithCliques -> (corpusPath config)
+ <> (listPath config)
+ <> (timeToLabel config)
+ <> (cliqueToLabel config)
+ PhyloWithLinks -> (corpusPath config)
+ <> (listPath config)
+ <> (timeToLabel config)
+ <> (cliqueToLabel config)
+ <> (sensToLabel config)
+ <> (seaToLabel config)
+ <> (syncToLabel config)
+ <> (qualToConfig config)
+ <> (show (phyloLevel config))
+
+
+readListV4 :: [Char] -> IO NgramsList
+readListV4 path = do
+ listJson <- (eitherDecode <$> readJson path) :: IO (Either String NgramsList)
+ case listJson of
+ Left err -> do
+ putStrLn err
+ undefined
+ Right listV4 -> pure listV4
+
+
+fileToList :: ListParser -> FilePath -> IO TermList
+fileToList parser path =
+ case parser of
+ V3 -> csvMapTermList path
+ V4 -> fromJust
+ <$> toTermList MapTerm NgramsTerms
+ <$> readListV4 path
--------------
main :: IO ()
-main = do
+main = do
+
+ printIOMsg "Starting the reconstruction"
+
+ printIOMsg "Read the configuration file"
+ [args] <- getArgs
+ jsonArgs <- (eitherDecode <$> readJson args) :: IO (Either String PhyloConfig)
+
+ case jsonArgs of
+ Left err -> putStrLn err
+ Right config -> do
+
+ printIOMsg "Parse the corpus"
+ mapList <- fileToList (listParser config) (listPath config)
+ corpus <- fileToDocs' (corpusParser config) (corpusPath config) (timeUnit config) mapList
+ printIOComment (show (length corpus) <> " parsed docs from the corpus")
+
+ printIOMsg "Reconstruct the phylo"
- putStrLn $ show "--| Read the conf |--"
+ let phyloWithCliquesFile = (outputPath config) <> "phyloWithCliques_" <> (configToSha PhyloWithCliques config) <> ".json"
+ let phyloWithLinksFile = (outputPath config) <> "phyloWithLinks_" <> (configToSha PhyloWithLinks config) <> ".json"
- [jsonPath] <- getArgs
+ phyloWithCliquesExists <- doesFileExist phyloWithCliquesFile
+ phyloWithLinksExists <- doesFileExist phyloWithLinksFile
- confJson <- (eitherDecode <$> getJson jsonPath) :: IO (P.Either P.String Conf)
+ -- phyloStep <- if phyloWithCliquesExists
+ -- then do
+ -- printIOMsg "Reconstruct the phylo step from an existing file"
+ -- readPhylo phyloWithCliquesFile
+ -- else do
+ -- printIOMsg "Reconstruct the phylo step from scratch"
+ -- pure $ toPhyloStep corpus mapList config
- case confJson of
- P.Left err -> putStrLn err
- P.Right conf -> do
+ -- writePhylo phyloWithCliquesFile phyloStep
- putStrLn $ show "--| Parse the corpus |--"
+ -- let phylo = toPhylo (setConfig config phyloStep)
- termList <- csvGraphTermList (listPath conf)
+ phyloWithLinks <- if phyloWithLinksExists
+ then do
+ printIOMsg "Reconstruct the phylo from an existing file with intertemporal links"
+ readPhylo phyloWithLinksFile
+ else do
+ if phyloWithCliquesExists
+ then do
+ printIOMsg "Reconstruct the phylo from an existing file with cliques"
+ phyloWithCliques <- readPhylo phyloWithCliquesFile
+ writePhylo phyloWithCliquesFile phyloWithCliques
+ pure $ toPhylo (setConfig config phyloWithCliques)
+ else do
+ printIOMsg "Reconstruct the phylo from scratch"
+ phyloWithCliques <- pure $ toPhyloStep corpus mapList config
+ writePhylo phyloWithCliquesFile phyloWithCliques
+ pure $ toPhylo (setConfig config phyloWithCliques)
- corpus <- parse (limit conf) (corpusPath conf) termList
+ writePhylo phyloWithLinksFile phyloWithLinks
- let roots = DL.nub $ DL.concat $ map text corpus
- putStrLn $ show "--| Build the phylo |--"
-
- let query = PhyloQueryBuild (phyloName conf) "" 5 3 defaultFis [] [] (WeightedLogJaccard $ WLJParams 0.00001 10) 2 (RelatedComponents $ RCParams $ WeightedLogJaccard $ WLJParams 0.5 10)
+ -- probes
- let queryView = PhyloQueryView 2 Merge False 1 [BranchAge] [defaultSmallBranch] [BranchPeakFreq,GroupLabelCooc] (Just (ByBranchAge,Asc)) Json Flat True
+ -- writeFile ((outputPath config) <> (unpack $ phyloName config) <> "_synchronic_distance_cumu_jaccard.txt")
+ -- $ synchronicDistance' phylo 1
- let phylo = toPhylo query corpus roots termList
+ -- writeFile ((outputPath config) <> (unpack $ phyloName config) <> "_inflexion_points.txt")
+ -- $ inflexionPoints phylo 1
- let view = toPhyloView queryView phylo
+ printIOMsg "End of reconstruction, start the export"
- putStrLn $ show "--| Export the phylo as a dot graph |--"
+ let dot = toPhyloExport (setConfig config phyloWithLinks)
- let outputFile = (outputPath conf) P.++ (DT.unpack $ phyloName conf) P.++ ".dot"
+ let output = configToLabel config
- P.writeFile outputFile $ dotToString $ viewToDot view
+ dotToFile output dot