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 ByteString
  30 isidoreGet q = do
  31   let s = createSelectQuery $ simpleSelect q
  32   putStrLn s
  33   r <- selectQueryRaw' route s
  34   putStrLn $ show $ r ^. responseStatus
  35   pure $ r ^. responseBody
  36  -- res <- selectQuery route $ simpleSelect q
  37  -- pure res
  38
  39 simpleSelect :: Text -> Query SelectQuery
  40 simpleSelect 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   ore     <- prefix "ore"    (iriRef "http://www.openarchives.org/ore/terms/")
  48   bif     <- prefix "bif"    (iriRef "bif:")
  49
  50   link     <- var
  51   title    <- var
  52   date     <- var
  53   abstract <- var
  54   authors  <- var
  55   source   <- var
  56   lang     <- var
  57   publisher <- var
  58   agg       <- var
  59
  60   indexT  <- 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") lang
  67   triple_ link (dcterms .:. "source")   source
  68
  69   triple_ link (ore .:. "isAggregatedBy") agg
  70   --triple_ agg (dcterms .:. "title") title
  71
  72
  73   optional $ triple_ link (dcterms .:. "publisher") publisher
  74   optional $ triple_ link (dc      .:. "description") abstract
  75
  76   --filterExpr $ (.||.) (contains title q) (contains abstract q)
  77
  78   --triple_ indexT (bif .:. "contains") title
  79
  80   --filterExpr_ (contains indexT q) -- (contains abstract q)
  81   orderNextDesc date
  82   limit_ 10
  83   distinct_
  84   selectVars [link, date, authors, source, title, lang, publisher, abstract]