src/Gargantext/Text/Parsers/Isidore.hs

   1 {-# LANGUAGE NoImplicitPrelude   #-}
   2 {-# LANGUAGE ScopedTypeVariables #-}
   3 {-# LANGUAGE OverloadedStrings   #-}
   4
   5 module Gargantext.Text.Parsers.Isidore where
   6
   7 import Gargantext.Prelude
   8 import Database.HSparql.Connection
   9 import Database.HSparql.QueryGenerator
  10
  11 -- import Data.RDF hiding (triple)
  12 import Data.Text hiding (groupBy)
  13
  14 import Control.Lens hiding (contains)
  15 import Data.ByteString.Lazy (ByteString)
  16 import Prelude (String)
  17 import Network.Wreq
  18
  19 route :: EndPoint
  20 route = "https://isidore.science/sparql/"
  21
  22 selectQueryRaw' :: String -> String -> IO (Response ByteString)
  23 selectQueryRaw' uri q = getWith opts uri
  24   where
  25     opts = defaults & header "Accept"     .~ ["application/sparql-results+xml"]
  26                     & header "User-Agent" .~ ["gargantext-hsparql-client"]
  27                     & param  "query"      .~ [Data.Text.pack q]
  28
  29 isidoreGet :: Text -> IO (Maybe [[BindingValue]])
  30 isidoreGet q = do
  31   let s = createSelectQuery $ isidoreSelect q
  32   putStrLn s
  33   r <- selectQueryRaw' route s
  34   putStrLn $ show $ r ^. responseStatus
  35   pure $ structureContent $ r ^. responseBody
  36  -- res <- selectQuery route $ simpleSelect q
  37  -- pure res
  38
  39 isidoreSelect :: Text -> Query SelectQuery
  40 isidoreSelect q = do
  41   -- See Predefined Namespace Prefixes:
  42   -- https://isidore.science/sparql?nsdecl
  43   isidore <- prefix "isidore" (iriRef "http://www.rechercheisidore.fr/class/")
  44   rdf     <- prefix "rdf"     (iriRef "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
  45   dcterms <- prefix "dcterms" (iriRef "http://purl.org/dc/terms/")
  46   dc      <- prefix "dc"      (iriRef "http://purl.org/dc/elements/1.1/")
  47   --iso     <- prefix "fra"        (iriRef "http://lexvo.org/id/iso639-3/")
  48   --ore     <- prefix "ore"    (iriRef "http://www.openarchives.org/ore/terms/")
  49   --bif     <- prefix "bif"    (iriRef "bif:")
  50
  51   link     <- var
  52   title    <- var
  53   date     <- var
  54   abstract <- var
  55   authors  <- var
  56   source   <- var
  57   langDoc  <- var
  58   publisher <- var
  59   --langFr   <- var
  60   --agg       <- var
  61
  62   triple_ link (rdf     .:. "type")     (isidore .:. "BibliographicalResource")
  63   triple_ link (dcterms .:. "title")    title
  64   triple_ link (dcterms .:. "date")     date
  65   triple_ link (dcterms .:. "creator")  authors
  66   triple_ link (dcterms .:. "language") langDoc
  67   triple_ link (dc      .:. "description") abstract
  68   --triple_ link (ore .:. "isAggregatedBy") agg
  69   --triple_ agg (dcterms .:. "title") title
  70
  71   optional_ $ triple_ link (dcterms .:. "source")      source
  72   optional_ $ triple_ link (dcterms .:. "publisher")   publisher
  73
  74   -- TODO FIX BUG with (.||.) operator
  75   --filterExpr $ (.||.) (contains title q) (contains title q)
  76   filterExpr_ (containsWith title q) -- (contains abstract q)
  77   --filterExpr (containsWith abstract q) -- (contains abstract q)
  78
  79   -- TODO FIX filter with lang
  80   --filterExpr $ langMatches title (str ("fra" :: Text))
  81   --filterExpr $ (.==.) lang (str ("http://lexvo.org/id/iso639-3/fra" :: Text))
  82
  83   orderNextDesc date
  84   limit_ 10
  85   distinct_
  86   selectVars [link, date, langDoc, authors, source, publisher, title, abstract]