{-|
Module : Gargantext.Database.Metrics
Description : Get Metrics from Storage (Database like)
Copyright : (c) CNRS, 2017-Present
License : AGPL + CECILL v3
Maintainer : team@gargantext.org
Stability : experimental
Portability : POSIX
Node API
-}
{-# LANGUAGE QuasiQuotes #-}
module Gargantext.Database.Action.Metrics
where
import Database.PostgreSQL.Simple.SqlQQ (sql)
import Data.HashMap.Strict (HashMap)
import Data.Map.Strict (Map)
import Data.Set (Set)
import Database.PostgreSQL.Simple (Query, Only(..))
import Database.PostgreSQL.Simple.Types (Values(..), QualifiedIdentifier(..))
import Data.Vector (Vector)
import Gargantext.Core (HasDBid(toDBid))
import Gargantext.API.Ngrams.Tools (filterListWithRoot, groupNodesByNgrams, Diagonal(..), getCoocByNgrams, mapTermListRoot, RootTerm, getRepo)
import Gargantext.Database.Prelude (runPGSQuery{-, formatPGSQuery-})
import Gargantext.API.Ngrams.Types (TabType(..), ngramsTypeFromTabType, NgramsTerm(..))
import Gargantext.Core.Mail.Types (HasMail)
import Gargantext.Core.NodeStory hiding (runPGSQuery)
import Gargantext.Core.Text.Metrics (scored, Scored(..), {-localMetrics, toScored-})
import Database.PostgreSQL.Simple.ToField (toField, Action{-, ToField-})
import Gargantext.Core.Types (ListType(..), Limit, NodeType(..), ContextId)
import Gargantext.Database.Action.Flow.Types (FlowCmdM)
import Gargantext.Database.Action.Metrics.NgramsByContext (getContextsByNgramsOnlyUser{-, getTficfWith-})
import Gargantext.Database.Admin.Config (userMaster)
import Gargantext.Database.Admin.Types.Node (ListId, CorpusId)
import Gargantext.Database.Query.Table.Node (defaultList)
import Gargantext.Database.Query.Table.Node.Select
import Gargantext.Prelude
import qualified Data.HashMap.Strict as HM
import qualified Data.Map.Strict as Map
import qualified Data.Set as Set
import qualified Data.List as List
import qualified Data.Text as Text
getMetrics :: FlowCmdM env err m
=> CorpusId -> Maybe ListId -> TabType -> Maybe Limit
-> m (HashMap NgramsTerm (ListType, Maybe NgramsTerm), Vector (Scored NgramsTerm))
getMetrics cId maybeListId tabType maybeLimit = do
(ngs, _, myCooc) <- getNgramsCooc cId maybeListId tabType maybeLimit
-- TODO HashMap
pure (ngs, scored myCooc)
getNgramsCooc :: (FlowCmdM env err m)
=> CorpusId -> Maybe ListId -> TabType -> Maybe Limit
-> m ( HashMap NgramsTerm (ListType, Maybe NgramsTerm)
, HashMap NgramsTerm (Maybe RootTerm)
, HashMap (NgramsTerm, NgramsTerm) Int
)
getNgramsCooc cId maybeListId tabType maybeLimit = do
lId <- case maybeListId of
Nothing -> defaultList cId
Just lId' -> pure lId'
(ngs', ngs) <- getNgrams lId tabType
lIds <- selectNodesWithUsername NodeList userMaster
myCooc <- HM.filter (>1) <$> getCoocByNgrams (Diagonal True)
<$> groupNodesByNgrams ngs
<$> getContextsByNgramsOnlyUser cId
(lIds <> [lId])
(ngramsTypeFromTabType tabType)
(take' maybeLimit $ HM.keys ngs)
pure $ (ngs', ngs, myCooc)
------------------------------------------------------------------------
------------------------------------------------------------------------
updateNgramsOccurrences :: (FlowCmdM env err m)
=> CorpusId -> Maybe ListId
-> m ()
updateNgramsOccurrences cId mlId = do
_ <- mapM (updateNgramsOccurrences' cId mlId Nothing) [Terms, Sources, Authors, Institutes]
pure ()
updateNgramsOccurrences' :: (FlowCmdM env err m)
=> CorpusId -> Maybe ListId -> Maybe Limit -> TabType
-> m [Int]
updateNgramsOccurrences' cId maybeListId maybeLimit tabType = do
lId <- case maybeListId of
Nothing -> defaultList cId
Just lId' -> pure lId'
result <- getNgramsOccurrences cId lId tabType maybeLimit
let
toInsert :: [[Action]]
toInsert = map (\(ngramsTerm, score)
-> [ toField cId
, toField lId
, toField $ unNgramsTerm ngramsTerm
, toField $ toDBid $ ngramsTypeFromTabType tabType
, toField score
]
)
$ HM.toList result
queryInsert :: Query
queryInsert = [sql|
WITH input(corpus_id, list_id, terms, type_id, weight) AS (?)
INSERT into node_node_ngrams (node1_id, node2_id, ngrams_id, ngrams_type, weight)
SELECT input.corpus_id,input.list_id,ngrams.id,input.type_id,input.weight FROM input
JOIN ngrams on ngrams.terms = input.terms
ON CONFLICT (node1_id, node2_id, ngrams_id, ngrams_type)
DO UPDATE SET weight = excluded.weight
RETURNING 1
|]
let fields = map (\t-> QualifiedIdentifier Nothing t)
$ map Text.pack ["int4", "int4","text","int4","int4"]
map (\(Only a) -> a) <$> runPGSQuery queryInsert (Only $ Values fields toInsert)
------------------------------------------------------------------------
-- Used for scores in Ngrams Table
getNgramsOccurrences :: (FlowCmdM env err m)
=> CorpusId -> ListId -> TabType -> Maybe Limit
-> m (HashMap NgramsTerm Int)
getNgramsOccurrences c l t ml = HM.map Set.size <$> getNgramsContexts c l t ml
getNgramsContexts :: (FlowCmdM env err m)
=> CorpusId -> ListId -> TabType -> Maybe Limit
-> m (HashMap NgramsTerm (Set ContextId))
getNgramsContexts cId lId tabType maybeLimit = do
(_ngs', ngs) <- getNgrams lId tabType
lIds <- selectNodesWithUsername NodeList userMaster
-- TODO maybe add an option to group here
getContextsByNgramsOnlyUser cId
(lIds <> [lId])
(ngramsTypeFromTabType tabType)
(take' maybeLimit $ HM.keys ngs)
------------------------------------------------------------------------
updateContextScore :: (FlowCmdM env err m)
=> CorpusId -> Maybe ListId
-> m [Int]
updateContextScore cId maybeListId = do
lId <- case maybeListId of
Nothing -> defaultList cId
Just lId' -> pure lId'
result <- getContextsNgramsScore cId lId Terms MapTerm Nothing
let
toInsert :: [[Action]]
toInsert = map (\(contextId, score)
-> [ toField cId
, toField contextId
, toField score
]
)
$ Map.toList result
queryInsert :: Query
queryInsert = [sql|
WITH input(node_id, context_id, score) AS (?)
UPDATE nodes_contexts nc
SET score = input.score
FROM input
WHERE nc.node_id = input.node_id
AND nc.context_id = input.context_id
RETURNING 1
|]
let fields = map (\t-> QualifiedIdentifier Nothing t)
$ map Text.pack ["int4", "int4","int4"]
map (\(Only a) -> a) <$> runPGSQuery queryInsert (Only $ Values fields toInsert)
-- Used for scores in Doc Table
getContextsNgramsScore :: (FlowCmdM env err m)
=> CorpusId -> ListId -> TabType -> ListType -> Maybe Limit
-> m (Map ContextId Int)
getContextsNgramsScore cId lId tabType listType maybeLimit
= Map.map Set.size <$> getContextsNgrams cId lId tabType listType maybeLimit
getContextsNgrams :: (FlowCmdM env err m)
=> CorpusId -> ListId -> TabType -> ListType -> Maybe Limit
-> m (Map ContextId (Set NgramsTerm))
getContextsNgrams cId lId tabType listType maybeLimit = do
(ngs', ngs) <- getNgrams lId tabType
lIds <- selectNodesWithUsername NodeList userMaster
result <- groupNodesByNgrams ngs <$> getContextsByNgramsOnlyUser
cId
(lIds <> [lId])
(ngramsTypeFromTabType tabType)
( take' maybeLimit
$ HM.keys
$ HM.filter (\v -> fst v == listType) ngs'
)
-- printDebug "getCoocByNgrams" result
pure $ Map.fromListWith (<>)
$ List.concat
$ map (\(ng, contexts) -> List.zip (Set.toList contexts) (List.cycle [Set.singleton ng]))
$ HM.toList result
------------------------------------------------------------------------
------------------------------------------------------------------------
getNgrams :: (HasMail env, HasNodeStory env err m)
=> ListId -> TabType
-> m ( HashMap NgramsTerm (ListType, Maybe NgramsTerm)
, HashMap NgramsTerm (Maybe RootTerm)
)
getNgrams lId tabType = do
lists <- mapTermListRoot [lId] (ngramsTypeFromTabType tabType) <$> getRepo [lId]
-- TODO filterListWithRoot [MapTerm, StopTerm, CandidateTerm] lists
let maybeSyn = HM.unions $ map (\t -> filterListWithRoot t lists)
[[MapTerm], [StopTerm], [CandidateTerm]]
pure (lists, maybeSyn)
-- Some useful Tools
take' :: Maybe Int -> [a] -> [a]
take' Nothing xs = xs
take' (Just n) xs = take n xs