src/Gargantext/API/Ngrams/Tools.hs

   1 {-|
   2 Module      : Gargantext.API.Ngrams.Tools
   3 Description : Tools to manage Ngrams Elements (from the 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
  12 {-# LANGUAGE NoImplicitPrelude #-}
  13 {-# LANGUAGE OverloadedStrings #-}
  14 {-# LANGUAGE RankNTypes        #-}
  15
  16 module Gargantext.API.Ngrams.Tools
  17   where
  18
  19 import Control.Concurrent
  20 import Control.Lens (_Just, (^.), at, view)
  21 import Control.Monad.Reader
  22 import Data.Map.Strict (Map)
  23 import Data.Set (Set)
  24 import Data.Text (Text)
  25 import Data.Validity
  26 import Gargantext.API.Ngrams
  27 import Gargantext.Core.Types (ListType(..), NodeId, ListId)
  28 import Gargantext.Database.Schema.Ngrams (NgramsType)
  29 import Gargantext.Prelude
  30 import qualified Data.Map.Strict as Map
  31 import qualified Data.Set as Set
  32
  33
  34 type RootTerm = Text
  35
  36
  37 getListNgrams :: RepoCmdM env err m
  38                => [ListId] -> NgramsType
  39                -> m (Map Text NgramsRepoElement)
  40 getListNgrams nodeIds ngramsType = do
  41   v    <- view repoVar
  42   repo <- liftIO $ readMVar v
  43
  44   let
  45     ngramsMap = repo ^. r_state . at ngramsType . _Just
  46
  47     ngrams    = Map.unionsWith mergeNgramsElement
  48               [ ngramsMap ^. at nodeId . _Just | nodeId <- nodeIds ]
  49
  50   pure ngrams
  51
  52 mapTermListRoot :: RepoCmdM env err m
  53                => [ListId] -> NgramsType
  54                -> m (Map Text (ListType, (Maybe Text)))
  55 mapTermListRoot nodeIds ngramsType = do
  56   ngrams <- getListNgrams nodeIds ngramsType
  57   pure $ Map.fromList [(t, (_nre_list nre, _nre_root nre))
  58                       | (t, nre) <- Map.toList ngrams
  59                       ]
  60
  61 filterListWithRoot :: ListType -> Map Text (ListType, Maybe Text)
  62                       -> Map Text (Maybe RootTerm)
  63 filterListWithRoot lt m = Map.fromList
  64                     $ map (\(t,(_,r)) -> (t,r))
  65                     $ filter isGraphTerm (Map.toList m)
  66   where
  67     isGraphTerm (_t,(l, maybeRoot)) = case maybeRoot of
  68       Nothing -> l == lt
  69       Just  r -> case Map.lookup r m of
  70         Nothing -> panic $ "Garg.API.Ngrams.Tools: filterWithRoot, unknown key: " <> r
  71         Just  (l',_) -> l' == lt
  72
  73 groupNodesByNgrams :: Map Text (Maybe RootTerm)
  74                    -> Map Text (Set NodeId)
  75                    -> Map Text (Set NodeId)
  76 groupNodesByNgrams syn occs = Map.fromListWith (<>) occs'
  77   where
  78     occs' = map toSyn (Map.toList occs)
  79     toSyn (t,ns) = case Map.lookup t syn of
  80       Nothing -> panic $ "[Garg.API.Ngrams.Tools.groupNodesByNgrams] unknown key: " <> t
  81       Just  r -> case r of
  82         Nothing  -> (t, ns)
  83         Just  r' -> (r',ns)
  84
  85 data Diagonal = Diagonal Bool
  86
  87 getCoocByNgrams :: Diagonal -> Map Text (Set NodeId) -> Map (Text, Text) Int
  88 getCoocByNgrams (Diagonal diag) m =
  89   Map.fromList [((t1,t2)
  90                 ,maybe 0 Set.size $ Set.intersection
  91                                  <$> Map.lookup t1 m
  92                                  <*> Map.lookup t2 m
  93                 ) | (t1,t2) <- case diag of
  94                                  True   -> [ (x,y) | x <- Map.keys m, y <- Map.keys m, x <= y]
  95                                  False  -> listToCombi identity (Map.keys m)
  96                ]
  97