src/Gargantext/API/Node/Corpus/Export.hs

   1 {-|
   2 Module      : Gargantext.API.Node.Corpus.Export
   3 Description : Get Metrics from Storage (Database like)
   4 Copyright   : (c) CNRS, 2017-Present
   5 License     : AGPL + CECILL v3
   6 Maintainer  : team@gargantext.org
   7 Stability   : experimental
   8 Portability : POSIX
   9
  10 Main exports of Gargantext:
  11 - corpus
  12 - document and ngrams
  13 - lists
  14 -}
  15
  16 module Gargantext.API.Node.Corpus.Export
  17   where
  18
  19 import Data.Map (Map)
  20 import Data.Maybe (fromMaybe)
  21 import Data.Set (Set)
  22 import Gargantext.API.Node.Corpus.Export.Types
  23 import Gargantext.API.Ngrams.Types
  24 import Gargantext.API.Ngrams.Tools (filterListWithRoot, mapTermListRoot, getRepo')
  25 import Gargantext.API.Prelude (GargNoServer)
  26 import Gargantext.Prelude.Crypto.Hash (hash)
  27 import Gargantext.Core.Types
  28 import Gargantext.Core.NodeStory
  29 import Gargantext.Database.Action.Metrics.NgramsByNode (getNgramsByNodeOnlyUser)
  30 import Gargantext.Database.Admin.Config (userMaster)
  31 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  32 import Gargantext.Database.Prelude (Cmd)
  33 import Gargantext.Database.Query.Table.Node
  34 import Gargantext.Database.Query.Table.Node.Error (HasNodeError)
  35 import Gargantext.Database.Query.Table.Node.Select (selectNodesWithUsername)
  36 import Gargantext.Database.Query.Table.NodeNode (selectDocNodes)
  37 import Gargantext.Database.Schema.Ngrams (NgramsType(..))
  38 import Gargantext.Database.Schema.Node (_node_id, _node_hyperdata)
  39 import Gargantext.Prelude
  40 import qualified Data.List as List
  41 import qualified Data.Map as Map
  42 import qualified Data.Set as Set
  43 import qualified Data.HashMap.Strict as HashMap
  44
  45 --------------------------------------------------
  46 -- | Hashes are ordered by Set
  47 getCorpus :: CorpusId
  48           -> Maybe ListId
  49           -> Maybe NgramsType
  50           -> GargNoServer Corpus
  51 getCorpus cId lId nt' = do
  52
  53   let
  54     nt = case nt' of
  55       Nothing -> NgramsTerms
  56       Just  t -> t
  57
  58   ns   <- Map.fromList
  59        <$> map (\n -> (_node_id n, n))
  60        <$> selectDocNodes cId
  61
  62   repo <- getRepo' [fromMaybe (panic "[Gargantext.API.Node.Corpus.Export]") lId]
  63   ngs  <- getNodeNgrams cId lId nt repo
  64   let  -- uniqId is hash computed already for each document imported in database
  65     r = Map.intersectionWith (\a b -> Document a (Ngrams (Set.toList b) (hash b)) (d_hash a b)
  66                              ) ns (Map.map (Set.map unNgramsTerm) ngs)
  67           where
  68             d_hash  a b = hash [ fromMaybe "" (_hd_uniqId $ _node_hyperdata a)
  69                                , hash b
  70                                ]
  71   pure $ Corpus (Map.elems r) (hash $ List.map _d_hash
  72                                     $ Map.elems r
  73                               )
  74
  75 getNodeNgrams :: HasNodeError err
  76         => CorpusId
  77         -> Maybe ListId
  78         -> NgramsType
  79         -> NodeListStory
  80         -> Cmd err (Map NodeId (Set NgramsTerm))
  81 getNodeNgrams cId lId' nt repo = do
  82   lId <- case lId' of
  83     Nothing -> defaultList cId
  84     Just  l -> pure l
  85
  86   lIds <- selectNodesWithUsername NodeList userMaster
  87   let ngs = filterListWithRoot MapTerm $ mapTermListRoot [lId] nt repo
  88   -- TODO HashMap
  89   r <- getNgramsByNodeOnlyUser cId (lIds <> [lId]) nt (HashMap.keys ngs)
  90   pure r
  91
  92 -- TODO
  93 -- Exports List
  94 -- Version number of the list