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

   1 {-|
   2 Module      : Gargantext.API.Node.Corpus.Export
   3 Description : Corpus export
   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 Data.Text (Text)
  23 import qualified Data.List as List
  24 import qualified Data.Map as Map
  25 import qualified Data.Set as Set
  26 import qualified Data.HashMap.Strict as HashMap
  27
  28 import Gargantext.API.Node.Corpus.Export.Types
  29 import qualified Gargantext.API.Node.Document.Export.Types as DocumentExport
  30 import Gargantext.API.Ngrams.Types
  31 import Gargantext.API.Ngrams.Tools (filterListWithRoot, mapTermListRoot, getRepo')
  32 import Gargantext.API.Prelude (GargNoServer)
  33 import Gargantext.Prelude.Crypto.Hash (hash)
  34 import Gargantext.Core.Types
  35 import Gargantext.Core.NodeStory
  36 import Gargantext.Database.Action.Metrics.NgramsByContext (getNgramsByContextOnlyUser)
  37 import Gargantext.Database.Admin.Config (userMaster)
  38 import Gargantext.Database.Admin.Types.Hyperdata (HyperdataDocument(..))
  39 import Gargantext.Database.Prelude (Cmd)
  40 import Gargantext.Database.Query.Table.Node
  41 import Gargantext.Database.Query.Table.Node.Error (HasNodeError)
  42 import Gargantext.Database.Query.Table.Node.Select (selectNodesWithUsername)
  43 import Gargantext.Database.Query.Table.NodeContext (selectDocNodes)
  44 import Gargantext.Database.Schema.Ngrams (NgramsType(..))
  45 import Gargantext.Database.Schema.Context (_context_id, _context_hyperdata)
  46 import Gargantext.Prelude
  47
  48 --------------------------------------------------
  49 -- | Hashes are ordered by Set
  50 getCorpus :: CorpusId
  51           -> Maybe ListId
  52           -> Maybe NgramsType
  53           -> GargNoServer Corpus
  54 getCorpus cId lId nt' = do
  55
  56   let
  57     nt = case nt' of
  58       Nothing -> NgramsTerms
  59       Just  t -> t
  60
  61   listId <- case lId of
  62     Nothing -> defaultList cId
  63     Just l  -> pure l
  64
  65   ns   <- Map.fromList
  66        <$> map (\n -> (_context_id n, n))
  67        <$> selectDocNodes cId
  68
  69   repo <- getRepo' [listId]
  70   ngs  <- getContextNgrams cId listId MapTerm nt repo
  71   let  -- uniqId is hash computed already for each document imported in database
  72     r = Map.intersectionWith
  73         (\a b -> DocumentExport.Document { _d_document = context2node a
  74                                          , _d_ngrams = DocumentExport.Ngrams (Set.toList b) (hash b)
  75                                          , _d_hash = d_hash a b }
  76         ) ns (Map.map (Set.map unNgramsTerm) ngs)
  77           where
  78             d_hash :: Context HyperdataDocument -> Set Text -> Text
  79             d_hash  a b = hash [ fromMaybe "" (_hd_uniqId $ _context_hyperdata a)
  80                                , hash b
  81                                ]
  82   pure $ Corpus { _c_corpus = Map.elems r
  83                 , _c_hash = hash $ List.map DocumentExport._d_hash $ Map.elems r }
  84
  85 getContextNgrams :: HasNodeError err
  86         => CorpusId
  87         -> ListId
  88         -> ListType
  89         -> NgramsType
  90         -> NodeListStory
  91         -> Cmd err (Map ContextId (Set NgramsTerm))
  92 getContextNgrams cId lId listType nt repo = do
  93 --  lId <- case lId' of
  94 --    Nothing -> defaultList cId
  95 --    Just  l -> pure l
  96
  97   lIds <- selectNodesWithUsername NodeList userMaster
  98   let ngs = filterListWithRoot listType $ mapTermListRoot [lId] nt repo
  99   -- TODO HashMap
 100   r <- getNgramsByContextOnlyUser cId (lIds <> [lId]) nt (HashMap.keys ngs)
 101   pure r
 102
 103 -- TODO
 104 -- Exports List
 105 -- Version number of the list