src-test/Ngrams/Query/PaginationCorpus.hs

   1 {-# LANGUAGE ScopedTypeVariables #-}
   2 module Ngrams.Query.PaginationCorpus where
   3
   4 import           Prelude
   5
   6 import           Data.Aeson
   7 import           Data.Map.Strict (Map)
   8 import           Gargantext.API.Ngrams
   9 import           Gargantext.API.Ngrams.Types
  10 import           Gargantext.Core.Types.Main
  11 import           Gargantext.Database.Admin.Types.Node
  12 import           System.IO.Unsafe
  13 import qualified Data.ByteString as B
  14 import qualified Data.Map.Strict as Map
  15 import qualified Data.Set as Set
  16
  17 import Paths_gargantext
  18
  19
  20 implementationElem :: NgramsElement
  21 implementationElem = NgramsElement {
  22     _ne_ngrams = "implementation"
  23   , _ne_size = 1
  24   , _ne_list = MapTerm
  25   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3, NodeId 4, NodeId 5 ]
  26   , _ne_root = Nothing
  27   , _ne_parent = Nothing
  28   , _ne_children = mSetFromList [ "code", "functions", "language", "programs" ]
  29   }
  30
  31 languagesElem :: NgramsElement
  32 languagesElem = NgramsElement {
  33     _ne_ngrams = NgramsTerm {unNgramsTerm = "languages"}
  34   , _ne_size = 1
  35   , _ne_list = MapTerm
  36   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2 , NodeId 3 , NodeId 4 ]
  37   , _ne_root = Nothing
  38   , _ne_parent = Nothing
  39   , _ne_children = mSetFromList [ "approach", "use" ]
  40   }
  41
  42 termsElem :: NgramsElement
  43 termsElem = NgramsElement {
  44     _ne_ngrams = NgramsTerm {unNgramsTerm = "terms"}
  45   , _ne_size = 1
  46   , _ne_list = MapTerm
  47   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2 , NodeId 3 ]
  48   , _ne_root = Nothing
  49   , _ne_parent = Nothing
  50   , _ne_children = mSetFromList [ "algorithm", "evaluation", "monad", "programmers" ]
  51   }
  52
  53 proofElem :: NgramsElement
  54 proofElem = NgramsElement {
  55     _ne_ngrams = NgramsTerm {unNgramsTerm = "proof"}
  56   , _ne_size = 1
  57   , _ne_list = MapTerm
  58   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2 ]
  59   , _ne_root = Nothing
  60   , _ne_parent = Nothing
  61   , _ne_children = mSetFromList [ "proofs" ]
  62   }
  63
  64 sideEffectsElem :: NgramsElement
  65 sideEffectsElem = NgramsElement {
  66     _ne_ngrams = NgramsTerm {unNgramsTerm = "side effects"}
  67   , _ne_size = 1
  68   , _ne_list = StopTerm
  69   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3, NodeId 4, NodeId 5, NodeId 6 ]
  70   , _ne_root = Nothing
  71   , _ne_parent = Nothing
  72   , _ne_children = mSetFromList [ ]
  73   }
  74
  75 ooElem :: NgramsElement
  76 ooElem = NgramsElement {
  77     _ne_ngrams = NgramsTerm {unNgramsTerm = "object oriented"}
  78   , _ne_size = 1
  79   , _ne_list = StopTerm
  80   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3, NodeId 4, NodeId 5 ]
  81   , _ne_root = Nothing
  82   , _ne_parent = Nothing
  83   , _ne_children = mSetFromList [ "null pointer exception" ]
  84   }
  85
  86 javaElem :: NgramsElement
  87 javaElem = NgramsElement {
  88     _ne_ngrams = NgramsTerm {unNgramsTerm = "java"}
  89   , _ne_size = 1
  90   , _ne_list = StopTerm
  91   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3 ]
  92   , _ne_root = Nothing
  93   , _ne_parent = Nothing
  94   , _ne_children = mSetFromList [ "JVM" ]
  95   }
  96
  97 pascalElem :: NgramsElement
  98 pascalElem = NgramsElement {
  99     _ne_ngrams = NgramsTerm {unNgramsTerm = "pascal"}
 100   , _ne_size = 1
 101   , _ne_list = StopTerm
 102   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2 ]
 103   , _ne_root = Nothing
 104   , _ne_parent = Nothing
 105   , _ne_children = mSetFromList [ "turbo", "borland" ]
 106   }
 107
 108 haskellElem :: NgramsElement
 109 haskellElem = NgramsElement {
 110     _ne_ngrams = NgramsTerm {unNgramsTerm = "haskell"}
 111   , _ne_size = 1
 112   , _ne_list = CandidateTerm
 113   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3, NodeId 4, NodeId 5, NodeId 6, NodeId 7, NodeId 8 ]
 114   , _ne_root = Nothing
 115   , _ne_parent = Nothing
 116   , _ne_children = mSetFromList [ ]
 117   }
 118
 119 concHaskellElem :: NgramsElement
 120 concHaskellElem = NgramsElement {
 121     _ne_ngrams = NgramsTerm {unNgramsTerm = "concurrent haskell"}
 122   , _ne_size = 1
 123   , _ne_list = CandidateTerm
 124   , _ne_occurrences = Set.fromList [ NodeId 1, NodeId 2, NodeId 3, NodeId 4, NodeId 5 ]
 125   , _ne_root = Nothing
 126   , _ne_parent = Nothing
 127   , _ne_children = mSetFromList [ "Simon Marlow" ]
 128   }
 129
 130 -- | A big (for the sake of the tests anyway) corpus which has
 131 -- * 4 @MapTerm@s
 132 -- * 4 @StopTerm@s
 133 -- * 2 @CandidateTerm@s
 134 paginationCorpus :: Versioned (Map NgramsTerm NgramsElement)
 135 paginationCorpus = Versioned 0 $ Map.fromList [
 136     -- Map terms
 137     ( "implementation", implementationElem)
 138   , ( "languages", languagesElem)
 139   , ( "terms", termsElem)
 140   , ("proof", proofElem)
 141
 142     -- Stop terms
 143   , ("side effects", sideEffectsElem)
 144   , ("object oriented", ooElem)
 145   , ("java", javaElem)
 146   , ("pascal", pascalElem)
 147
 148     -- Candidate terms
 149   , ("haskell", haskellElem)
 150   , ("concurrent haskell", concHaskellElem)
 151   ]
 152
 153 quantumComputingCorpus :: Versioned (Map NgramsTerm NgramsElement)
 154 quantumComputingCorpus = unsafePerformIO $ do
 155   pth      <- getDataFileName "test-data/ngrams/GarganText_NgramsTerms-QuantumComputing.json"
 156   jsonBlob <- B.readFile pth
 157   case eitherDecodeStrict' jsonBlob of
 158     Left err    -> error err
 159     Right (Versioned ver (mp :: Map NgramsTerm NgramsRepoElement)) ->
 160       pure $ Versioned ver (Map.mapWithKey (\k -> ngramsElementFromRepo k) mp)
 161 {-# NOINLINE quantumComputingCorpus #-}