src/Gargantext/Core/Text/Terms/WithList.hs

   1 {-|
   2 Module      : Gargantext.Core.Text.Terms.WithList
   3 Description :
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Here is a longer description of this module, containing some
  11 commentary with @some markup@.
  12
  13 -}
  14 {-# LANGUAGE BangPatterns      #-}
  15
  16 module Gargantext.Core.Text.Terms.WithList where
  17
  18 import Data.List (null)
  19 import Data.Ord
  20 import Data.Text (Text, concat, unwords)
  21 import Gargantext.Prelude
  22 import Gargantext.Core.Text.Context
  23 import Gargantext.Core.Text.Terms.Mono (monoTextsBySentence)
  24 import Prelude (error)
  25 import qualified Data.Algorithms.KMP as KMP
  26 import qualified Data.IntMap.Strict  as IntMap
  27 import qualified Data.List           as List
  28 ------------------------------------------------------------------------
  29
  30 data Pattern = Pattern
  31   { _pat_table  :: !(KMP.Table Text)
  32   , _pat_length :: !Int
  33   , _pat_terms  :: ![Text]
  34   }
  35 type Patterns = [Pattern]
  36
  37 ------------------------------------------------------------------------
  38 replaceTerms :: Patterns -> [Text] -> [[Text]]
  39 replaceTerms pats terms = go 0
  40   where
  41     terms_len = length terms
  42
  43     go ix | ix >= terms_len = []
  44           | otherwise =
  45       case IntMap.lookup ix m of
  46         Nothing -> go (ix + 1)
  47         Just (len, term) ->
  48           term : go (ix + len)
  49
  50
  51     merge (len1, lab1) (len2, lab2) =
  52       if len2 < len1 then (len1, lab1) else (len2, lab2)
  53
  54     m =
  55       IntMap.fromListWith merge
  56         [ (ix, (len, term))
  57         | Pattern pat len term <- pats, ix <- KMP.match pat terms ]
  58
  59 buildPatterns :: TermList -> Patterns
  60 buildPatterns = sortWith (Down . _pat_length) . concatMap buildPattern
  61   where
  62     buildPattern (label, alts) = map f $ map (\alt -> filter (/= "") alt) (label : alts)
  63       where
  64         f alt | "" `elem` alt = error ("buildPatterns: ERR1" <> show(label))
  65               | null alt      = error "buildPatterns: ERR2"
  66               | otherwise     =
  67                 Pattern (KMP.build alt) (length alt) label
  68                         --(Terms label $ Set.empty) -- TODO check stems
  69
  70
  71 --------------------------------------------------------------------------
  72 -- Utils
  73 type MatchedText = Text
  74 termsInText :: Patterns -> Text -> [MatchedText]
  75 termsInText pats txt = List.nub
  76                      $ List.concat
  77                      $ map (map unwords)
  78                      $ extractTermsWithList pats txt
  79
  80 --------------------------------------------------------------------------
  81
  82 extractTermsWithList :: Patterns -> Text -> Corpus [Text]
  83 extractTermsWithList pats = map (replaceTerms pats) . monoTextsBySentence
  84
  85 -- | Extract terms
  86 -- >>> let termList = [(["chat blanc"], [["chat","blanc"]])] :: TermList
  87 -- extractTermsWithList' (buildPatterns termList) "Le chat blanc"["chat blanc"]
  88 -- ["chat blanc"]
  89 extractTermsWithList' :: Patterns -> Text -> [Text]
  90 extractTermsWithList' pats = map (concat . map concat . replaceTerms pats)
  91                            . monoTextsBySentence
  92
  93 --------------------------------------------------------------------------
  94
  95 {- | Not used
  96 filterWith :: TermList
  97            -> (a -> Text)
  98            -> [a]
  99            -> [(a, [Text])]
 100 filterWith termList f xs = filterWith' termList f zip xs
 101
 102
 103 filterWith' :: TermList
 104            -> (a -> Text)
 105            -> ([a] -> [[Text]] -> [b])
 106            -> [a]
 107            -> [b]
 108 filterWith' termList f f' xs = f' xs
 109                             $ map (extractTermsWithList' pats)
 110                             $ map f xs
 111     where
 112       pats = buildPatterns termList
 113 -}