src/Gargantext/Text/Terms/WithList.hs

   1 {-|
   2 Module      : Gargantext.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.Text.Terms.WithList where
  17
  18 import Data.List (null, concatMap)
  19 import Data.Ord
  20 import Data.Text (Text, concat)
  21 import Gargantext.Prelude
  22 import Gargantext.Text.Context
  23 import Gargantext.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
  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 (label : alts)
  63       where
  64         f alt | "" `elem` alt = error "buildPatterns: ERR1"
  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 extractTermsWithList :: Patterns -> Text -> Corpus [Text]
  71 extractTermsWithList pats = map (replaceTerms pats) . monoTextsBySentence
  72
  73 -- | Extract terms
  74 -- >>> let termList = [(["chat blanc"], [["chat","blanc"]])] :: TermList
  75 -- extractTermsWithList' (buildPatterns termList) "Le chat blanc"["chat blanc"]
  76 -- ["chat blanc"]
  77 extractTermsWithList' :: Patterns -> Text -> [Text]
  78 extractTermsWithList' pats = map (concat . map concat . replaceTerms pats)
  79                            . monoTextsBySentence
  80
  81
  82 filterWith :: TermList
  83            -> (a -> Text)
  84            -> [a]
  85            -> [(a, [Text])]
  86 filterWith termList f xs = filterWith' termList f zip xs
  87
  88
  89 filterWith' :: TermList
  90            -> (a -> Text)
  91            -> ([a] -> [[Text]] -> [b])
  92            -> [a]
  93            -> [b]
  94 filterWith' termList f f' xs = f' xs
  95                             $ map (extractTermsWithList' pats)
  96                             $ map f xs
  97     where
  98       pats = buildPatterns termList
  99