src/Gargantext/Utils/LangDetect_hs

   1
   2
   3 -- DEFINITIONS as SPECS
   4 -- (Engineering axioms for Gargantext)
   5
   6
   7 ------------------------------------------------------------------------
   8 -- From file to corpus
   9 ------------------------------------------------------------------------
  10
  11 -- > A Corpus is a list of Documents
  12 data Corpus   = [Document]
  13
  14 -- > A Document should have a date, some text and a maybe a language.
  15 -- > Remarks :
  16 --         > If no date then force one ?
  17 --         > Analyze either text or numbers
  18 --         > only one language per document
  19 data Document = Document { date     :: UTCTime
  20                          , uce      :: Map Text $ Either (Maybe Text) (Maybe Double)
  21                          , lang     :: Maybe Language
  22                          }
  23
  24 parseFiles :: Maybe ParserType -> [File] -> Corpus
  25 parseFiles = undefined
  26
  27 -- This function exists already (in Python)
  28 parseFile' :: ParserType -> File -> Maybe [Document]
  29 parseFile' = undefined
  30
  31 -- This function does not exist yet
  32 parseFile :: Maybe ParserType -> File -> Maybe [Document]
  33 parseFile parserType file = documents
  34     where
  35         documents = case parserType of
  36
  37                       Nothing           -> case guessParserType file of
  38                                             Nothing          -> askUser "Answer to the question with link to $doc"
  39                                             Just parserType' -> parseFile (Just parserType') file
  40
  41                       Just parserType'' -> case parserType''         of
  42                                             UnsupportedYet   -> askUser "Not supported yet, which priority ?"
  43                                             otherwise        -> parseFile' parserType'' file
  44
  45 data ParserType = RIS | ISI | XML | CSV | Europresse | Book | UnsupportedYet
  46 guessParserType :: File -> Maybe ParserType
  47 guessParserType = undefined
  48
  49
  50 ------------------------------------------------------------------------
  51 -- What kind of interactions with our users ?
  52 ------------------------------------------------------------------------
  53
  54 -- Question is Text only
  55 type Question = Text
  56
  57 -- Possible Answers:
  58 data Answer = ClosedAnswer | NumAnswer | OpenAnswer
  59 -- Definitions of the Answers
  60 type ClosedAnswer   = Bool
  61 type OpenAnswer     = Text
  62 type NumAnswer      = Int
  63 -- Un formulaire est un mapping entre question et peut-être une réponse
  64 -- Un formulaire vide a Nothing au champs (Maybe Answer)
  65 -- Une question répondue a la valeur (Just Response)
  66 type Formular = Map Question (Maybe Answer)
  67
  68 askUser :: Question -> ClosedAnswer
  69 askUser = undefined
  70
  71 data Advice = BugReport | WishList
  72 askUser' :: Question -> Advice
  73 askUser' question = case askUser question of
  74                       True  -> BugReport
  75                       False -> WishList
  76
  77
  78 ------------------------------------------------------------------------
  79 -- Specs for Lang Detection
  80 ------------------------------------------------------------------------
  81 data Language   = English | French
  82
  83 tagDoc :: Document -> Ngrams
  84 tagDoc doc = ngrams
  85     where
  86         ngrams = case lang doc of
  87                    Nothing -> case guessLang doc of
  88                                 Nothing -> tag
  89
  90
  91 ------------------------------------------------------------------------
  92 -- Specs for ngrams Worflow
  93 ------------------------------------------------------------------------