src/Data/Gargantext/NLP/CoreNLP.hs

   1 {-# LANGUAGE DataKinds #-}
   2 {-# LANGUAGE DeriveGeneric #-}
   3 {-# LANGUAGE TypeOperators #-}
   4 {-# LANGUAGE TemplateHaskell #-}
   5
   6 module Data.Gargantext.NLP.CoreNLP where
   7
   8 import Data.Aeson
   9 import Data.Aeson.TH (deriveJSON)
  10 import Data.Proxy
  11 import GHC.Generics
  12 import Network.HTTP.Client (newManager, defaultManagerSettings)
  13 import Servant.API
  14 import Servant.Client
  15
  16 import Data.Gargantext.Prelude
  17 import Data.Gargantext.Utils.Prefix (unPrefix)
  18 import Data.Text (Text)
  19
  20 data Token = Token { _tokenIndex                :: Int
  21                    , _tokenWord                 :: Text
  22                    , _tokenOriginalText         :: Text
  23                    , _tokenLemma                :: Text
  24                    , _tokenCharacterOffsetBegin :: Int
  25                    , _tokenCharacterOffsetEnd   :: Int
  26                    , _tokenPos                  :: Text
  27                    , _tokenNer                  :: Text
  28                    , _tokenBefore               :: Text
  29                    , _tokenAfter                :: Text
  30                    } deriving (Show, Generic)
  31 $(deriveJSON (unPrefix "_token") ''Token)
  32
  33 data Sentence  = Sentence { _sentenceIndex :: Int
  34                           , _sentenceToken :: [Token]
  35                           } deriving (Show, Generic)
  36
  37 $(deriveJSON (unPrefix "_sentence") ''Sentence)
  38
  39 data Properties = Properties { _propertiesAnnotators  :: Text
  40                              , _propertiesOutputFormat :: Text
  41                              } deriving (Show, Generic)
  42
  43 $(deriveJSON (unPrefix "_properties") ''Properties)
  44
  45 data Sentences = Sentences { sentences :: [Sentence]}
  46   deriving (Show, Generic)
  47 instance ToJSON Sentences
  48
  49 -- API Client configuration
  50
  51 -- Example of Client Request :
  52 -- wget --post-data 'Alexandre Grothendieck is a mathematician who lived in France which is a european country. There is another sentence  here.' 'localhost:9000/?properties={"annotators": "tokenize,ssplit,pos,ner", "outputFormat": "json"}' -O
  53
  54 -- the result is Sentence as a JSON
  55 -- {"sentences":[{"index":0,"tokens":[{"index":1,"word":"Alexandre","originalText":"Alexandre","lemma":"Alexandre","characterOffsetBegin":0,"characterOffsetEnd":9,"pos":"NNP","ner":"PERSON","before":"","after":" "},{"index":2,"word":"Grothendieck","originalText":"Grothendieck","lemma":"Grothendieck","characterOffsetBegin":10,"characterOffsetEnd":22,"pos":"NNP","ner":"PERSON","before":" ","after":" "},{"index":3,"word":"is","originalText":"is","lemma":"be","characterOffsetBegin":23,"characterOffsetEnd":25,"pos":"VBZ","ner":"O","before":" ","after":" "},{"index":4,"word":"a","originalText":"a","lemma":"a","characterOffsetBegin":26,"characterOffsetEnd":27,"pos":"DT","ner":"O","before":" ","after":" "},{"index":5,"word":"mathematician","originalText":"mathematician","lemma":"mathematician","characterOffsetBegin":28,"characterOffsetEnd":41,"pos":"NN","ner":"O","before":" ","after":" "},{"index":6,"word":"who","originalText":"who","lemma":"who","characterOffsetBegin":42,"characterOffsetEnd":45,"pos":"WP","ner":"O","before":" ","after":" "},{"index":7,"word":"lived","originalText":"lived","lemma":"live","characterOffsetBegin":46,"characterOffsetEnd":51,"pos":"VBD","ner":"O","before":" ","after":" "},{"index":8,"word":"in","originalText":"in","lemma":"in","characterOffsetBegin":52,"characterOffsetEnd":54,"pos":"IN","ner":"O","before":" ","after":" "},{"index":9,"word":"France","originalText":"France","lemma":"France","characterOffsetBegin":55,"characterOffsetEnd":61,"pos":"NNP","ner":"LOCATION","before":" ","after":" "},{"index":10,"word":"which","originalText":"which","lemma":"which","characterOffsetBegin":62,"characterOffsetEnd":67,"pos":"WDT","ner":"O","before":" ","after":" "},{"index":11,"word":"is","originalText":"is","lemma":"be","characterOffsetBegin":68,"characterOffsetEnd":70,"pos":"VBZ","ner":"O","before":" ","after":" "},{"index":12,"word":"a","originalText":"a","lemma":"a","characterOffsetBegin":71,"characterOffsetEnd":72,"pos":"DT","ner":"O","before":" ","after":" "},{"index":13,"word":"european","originalText":"european","lemma":"european","characterOffsetBegin":73,"characterOffsetEnd":81,"pos":"JJ","ner":"O","before":" ","after":" "},{"index":14,"word":"country","originalText":"country","lemma":"country","characterOffsetBegin":82,"characterOffsetEnd":89,"pos":"NN","ner":"O","before":" ","after":""},{"index":15,"word":".","originalText":".","lemma":".","characterOffsetBegin":89,"characterOffsetEnd":90,"pos":".","ner":"O","before":"","after":" "}]},{"index":1,"tokens":[{"index":1,"word":"There","originalText":"There","lemma":"there","characterOffsetBegin":91,"characterOffsetEnd":96,"pos":"EX","ner":"O","before":" ","after":" "},{"index":2,"word":"is","originalText":"is","lemma":"be","characterOffsetBegin":97,"characterOffsetEnd":99,"pos":"VBZ","ner":"O","before":" ","after":" "},{"index":3,"word":"another","originalText":"another","lemma":"another","characterOffsetBegin":100,"characterOffsetEnd":107,"pos":"DT","ner":"O","before":" ","after":" "},{"index":4,"word":"sentence","originalText":"sentence","lemma":"sentence","characterOffsetBegin":108,"characterOffsetEnd":116,"pos":"NN","ner":"O","before":" ","after":"  "},{"index":5,"wo
  56
  57
  58 type API = "" :> QueryParam "properties" Properties :> ReqBody '[JSON] String :> Post '[JSON] String
  59
  60 corenlp :: Maybe Properties -> Text -> ClientM Sentence
  61 corenlp p t = client api
  62
  63 -- text2nlp :: Text -> ClientM
  64
  65 api :: Proxy API
  66 api = Proxy
  67
  68
  69 -- corenlp t = client api
  70
  71 -- | URI scheme to use
  72 --data Scheme =
  73 --    Http  -- ^ http://
  74 --  | Https -- ^ https://
  75 --
  76 ---- | Simple data type to represent the target of HTTP requests
  77 ----   for servant's automatically-generated clients.
  78 --data BaseUrl = BaseUrl
  79 --  { baseUrlScheme :: Scheme -- ^ URI scheme to use
  80 --  , baseUrlHost :: String   -- ^ host (eg "haskell.org")
  81 --  , baseUrlPort :: Int      -- ^ port (eg 80)
  82 --  , baseUrlPath :: String   -- ^ path (eg "/a/b/c")
  83 --  }
  84 --
  85
  86 queries :: ClientM (Text, Properties)
  87 queries = do
  88     let text = "Alexandre Grothendieck is free even in a sentence."
  89     let prop = Properties "tokenize,ssplit,pos,ner" "json"
  90     return (text, prop)
  91
  92 run :: IO ()
  93 run = do
  94   manager <- newManager defaultManagerSettings
  95   res <- runClientM queries (ClientEnv manager (BaseUrl Http "localhost" 9000 ""))
  96   case res of
  97     Left err -> putStrLn $ "Error: " ++ show err
  98     Right x -> do
  99       print x
 100
 101
 102
 103
 104
 105
 106