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 qualified Data.List as List
  23 import qualified Data.Map as Map
  24 import qualified Data.Set as Set
  25 import qualified Data.HashMap.Strict as HashMap
  26
  27 import Gargantext.API.Node.Corpus.Export.Types
  28 import Gargantext.API.Ngrams.Types
  29 import Gargantext.API.Ngrams.Tools (filterListWithRoot, mapTermListRoot, getRepo')
  30 import Gargantext.API.Prelude (GargNoServer)
  31 import Gargantext.Prelude.Crypto.Hash (hash)
  32 import Gargantext.Core.Types
  33 import Gargantext.Core.NodeStory
  34 import Gargantext.Database.Action.Metrics.NgramsByNode (getNgramsByNodeOnlyUser)
  35 import Gargantext.Database.Admin.Config (userMaster)
  36 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  37 import Gargantext.Database.Prelude (Cmd)
  38 import Gargantext.Database.Query.Table.Node
  39 import Gargantext.Database.Query.Table.Node.Error (HasNodeError)
  40 import Gargantext.Database.Query.Table.Node.Select (selectNodesWithUsername)
  41 import Gargantext.Database.Query.Table.NodeNode (selectDocNodes)
  42 import Gargantext.Database.Schema.Ngrams (NgramsType(..))
  43 import Gargantext.Database.Schema.Node (_node_id, _node_hyperdata)
  44 import Gargantext.Prelude
  45
  46 --------------------------------------------------
  47 -- | Hashes are ordered by Set
  48 getCorpus :: CorpusId
  49           -> Maybe ListId
  50           -> Maybe NgramsType
  51           -> GargNoServer Corpus
  52 getCorpus cId lId nt' = do
  53
  54   let
  55     nt = case nt' of
  56       Nothing -> NgramsTerms
  57       Just  t -> t
  58
  59   ns   <- Map.fromList
  60        <$> map (\n -> (_node_id n, n))
  61        <$> selectDocNodes cId
  62
  63   repo <- getRepo' [fromMaybe (panic "[Gargantext.API.Node.Corpus.Export]") lId]
  64   ngs  <- getNodeNgrams cId lId nt repo
  65   let  -- uniqId is hash computed already for each document imported in database
  66     r = Map.intersectionWith
  67         (\a b -> Document { _d_document = a
  68                           , _d_ngrams = Ngrams (Set.toList b) (hash b)
  69                           , _d_hash = d_hash a b }
  70         ) ns (Map.map (Set.map unNgramsTerm) ngs)
  71           where
  72             d_hash  a b = hash [ fromMaybe "" (_hd_uniqId $ _node_hyperdata a)
  73                                , hash b
  74                                ]
  75   pure $ Corpus { _c_corpus = Map.elems r
  76                 , _c_hash = hash $ List.map _d_hash $ Map.elems r }
  77
  78 getNodeNgrams :: HasNodeError err
  79         => CorpusId
  80         -> Maybe ListId
  81         -> NgramsType
  82         -> NodeListStory
  83         -> Cmd err (Map NodeId (Set NgramsTerm))
  84 getNodeNgrams cId lId' nt repo = do
  85   lId <- case lId' of
  86     Nothing -> defaultList cId
  87     Just  l -> pure l
  88
  89   lIds <- selectNodesWithUsername NodeList userMaster
  90   let ngs = filterListWithRoot MapTerm $ mapTermListRoot [lId] nt repo
  91   -- TODO HashMap
  92   r <- getNgramsByNodeOnlyUser cId (lIds <> [lId]) nt (HashMap.keys ngs)
  93   pure r
  94
  95 -- TODO
  96 -- Exports List
  97 -- Version number of the list