src/Gargantext/Ext/IMT.hs

   1 {-|
   2 Module      : Gargantext.API
   3 Description : Server API
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9 -}
  10
  11 {-# LANGUAGE NoImplicitPrelude    #-}
  12
  13 module Gargantext.Ext.IMT where
  14
  15 import Gargantext.Prelude
  16 import Data.Text (Text, pack, splitOn)
  17 import Data.Map (Map)
  18
  19 import qualified Data.Set    as S
  20 import qualified Data.List   as DL
  21 import qualified Data.Vector as DV
  22 import qualified Data.Map    as M
  23
  24 import Gargantext.Text.Metrics.Freq as F
  25 import Gargantext.Text.Parsers.CSV as CSV
  26
  27 data School = School { school_shortName :: Text
  28                      , school_longName  :: Text
  29                      , school_id        :: Text
  30 } deriving (Show, Read, Eq)
  31
  32 schools :: [School]
  33 schools = [ School
  34             (pack "Mines Albi-Carmaux")
  35             (pack "Mines Albi-Carmaux - École nationale supérieure des Mines d'Albi‐Carmaux")
  36             (pack "469216")
  37           , School
  38             (pack "Mines Alès")
  39             (pack "EMA - École des Mines d'Alès")
  40             (pack "6279")
  41
  42           , School
  43             (pack "Mines Douai")
  44             (pack "Mines Douai EMD - École des Mines de Douai")
  45             (pack "224096")
  46
  47           , School
  48             (pack "Mines Nantes")
  49             (pack "Mines Nantes - Mines Nantes")
  50             (pack "84538")
  51
  52 --          , School
  53 --            (pack "Mines ParisTech")
  54 --            (pack "MINES ParisTech - École nationale supérieure des mines de Paris")
  55 --            (pack "301492")
  56 --
  57           , School
  58             (pack "Mines Saint-Étienne")
  59             (pack "Mines Saint-Étienne MSE - École des Mines de Saint-Étienne")
  60             (pack "29212")
  61
  62           , School
  63             (pack "Télécom Bretagne")
  64             (pack "Télécom Bretagne")
  65             (pack "301262")
  66
  67           , School
  68             (pack "Télécom École de Management")
  69             (pack "TEM - Télécom Ecole de Management")
  70             (pack "301442")
  71
  72           , School
  73             (pack "Télécom ParisTech")
  74             (pack "Télécom ParisTech")
  75             (pack "300362")
  76
  77           , School
  78             (pack "Télécom SudParis")
  79             (pack "TSP - Télécom SudParis")
  80             (pack "352124")
  81
  82           , School
  83             (pack "IMT Atlantique")
  84             (pack "IMT Atlantique - IMT Atlantique Bretagne-Pays de la Loire")
  85             (pack "481355")
  86             ]
  87
  88 mapIdSchool :: Map Text Text
  89 mapIdSchool = M.fromList $ Gargantext.Prelude.map (\(School n _ i) -> (i,n)) schools
  90
  91 hal_data :: IO (DV.Vector CsvHal)
  92 hal_data = snd <$> CSV.readHal "doc/corpus_imt/Gargantext_Corpus.csv"
  93
  94 names :: S.Set Text
  95 names = S.fromList $ Gargantext.Prelude.map (\s -> school_id s) schools
  96
  97
  98 publisBySchool :: DV.Vector CsvHal -> [(Maybe Text, Int)]
  99 publisBySchool hal_data' = Gargantext.Prelude.map (\(i,n) -> (M.lookup i mapIdSchool, n))
 100                         $ DL.filter (\i -> S.member  (fst i) names)
 101                         $ DL.reverse
 102                         $ DL.sortOn snd
 103                         $ M.toList
 104                         $ F.freq
 105                         $ DL.concat
 106                         $ DV.toList
 107                         $ DV.map (\n -> splitOn (pack ", ") (csvHal_instStructId_i n) )
 108                         $ DV.filter (\n -> csvHal_publication_year n == 2017) hal_data'
 109