src/Gargantext/Text/Metrics.hs

   1 {-|
   2 Module      : Gargantext.Text.Metrics
   3 Description : All parsers of Gargantext in one file.
   4 Copyright   : (c) CNRS, 2017 - present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Mainly reexport functions in @Data.Text.Metrics@
  11 -}
  12
  13 {-# LANGUAGE NoImplicitPrelude #-}
  14 {-# LANGUAGE OverloadedStrings #-}
  15
  16 module Gargantext.Text.Metrics where
  17
  18 import Data.Text (Text, pack)
  19 import qualified Data.Text as T
  20 import Data.List (concat)
  21
  22 --import GHC.Real (Ratio)
  23 --import qualified Data.Text.Metrics as DTM
  24
  25 import Gargantext.Prelude
  26
  27 import Gargantext.Text.Metrics.Count (occurrences, cooc)
  28 import Gargantext.Text.Terms (TermType(Multi), terms)
  29 import Gargantext.Core (Lang(EN))
  30 import Gargantext.Core.Types (Terms(..))
  31 import Gargantext.Text.Context (splitBy, SplitContext(Sentences))
  32
  33 --noApax :: Ord a => Map a Occ -> Map a Occ
  34 --noApax m = M.filter (>1) m
  35
  36
  37 metrics_text :: Text
  38 metrics_text = T.concat ["A table is an object."
  39                         ,"A glas is an object too."
  40                         ,"Using a glas to dring is a function."
  41                         ,"Using a spoon to eat is a function."
  42                         ,"The spoon is an object to eat."
  43                         ]
  44
  45 metrics_sentences' :: [Text]
  46 metrics_sentences' = splitBy (Sentences 0) metrics_text
  47
  48 -- | Sentences
  49 metrics_sentences :: [Text]
  50 metrics_sentences = ["A table is an object."
  51                     ,"A glas is an object too."
  52                     ,"The glas and the spoon are on the table."
  53                     ,"The spoon is an object to eat."
  54                     ,"The spoon is on the table and the plate and the glas."]
  55
  56
  57 metrics_sentences_Test = metrics_sentences == metrics_sentences'
  58
  59 -- | Terms reordered to visually check occurrences
  60 metrics_terms :: [[Text]]
  61 metrics_terms = undefined
  62
  63 metrics_terms' :: IO [[Terms]]
  64 metrics_terms' = mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text
  65
  66 --metrics_terms_Test =  metrics_terms == ((map _terms_label) <$> metrics_terms')
  67
  68 -- | Occurrences
  69 {-
  70 fromList [ (fromList ["table"] ,fromList [(["table"] , 3 )])]
  71          , (fromList ["object"],fromList [(["object"], 3 )])
  72          , (fromList ["glas"]  ,fromList [(["glas"]  , 2 )])
  73          , (fromList ["spoon"] ,fromList [(["spoon"] , 2 )])
  74 -}
  75 metrics_occ = occurrences <$> concat <$> (mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text)
  76
  77 {-
  78 -- fromList [((["glas"],["object"]),6)
  79             ,((["glas"],["spoon"]),4)
  80             ,((["glas"],["table"]),6),((["object"],["spoon"]),6),((["object"],["table"]),9),((["spoon"],["table"]),6)]
  81
  82 -}
  83 metrics_cooc = cooc <$> (mapM (terms Multi EN) $ splitBy (Sentences 0) metrics_text)
  84
  85 metrics_cooc' = (mapM (terms Multi EN) $ splitBy (Sentences 0) "The table object. The table object.")
  86
  87
  88