[Metrics] optim

src/Gargantext/API/Ngrams.hs

@@ -130,7 +130,7 @@ import GHC.Generics (Generic)
 import Gargantext.Core.Utils.Prefix (unPrefix, unPrefixSwagger)
 -- import Gargantext.Database.Schema.Ngrams (NgramsTypeId, ngramsTypeId, NgramsTableData(..))
 import Gargantext.Database.Config (userMaster)
-import Gargantext.Database.Metrics.NgramsByNode (getOccByNgramsOnlyFast)
+import Gargantext.Database.Metrics.NgramsByNode (getOccByNgramsOnlyFast')
 import Gargantext.Database.Schema.Ngrams (NgramsType)
 import Gargantext.Database.Types.Node (NodeType(..))
 import Gargantext.Database.Utils (fromField', HasConnection)
@@ -1019,7 +1019,8 @@ getTableNgrams _nType nId tabType listId limit_ offset
     setScores True  table = do
       let ngrams_terms = (table ^.. each . ne_ngrams)
       t1 <- getTime'
-      occurrences <- getOccByNgramsOnlyFast nId
+      occurrences <- getOccByNgramsOnlyFast' nId
+                                             listId
                                             ngramsType
                                             ngrams_terms
       t2 <- getTime'

src/Gargantext/Database/Metrics/NgramsByNode.hs

@@ -19,6 +19,7 @@ Ngrams by node enable contextual metrics.
 module Gargantext.Database.Metrics.NgramsByNode
   where
 
+import Debug.Trace (trace)
 import Data.Map.Strict (Map, fromListWith, elems, toList, fromList)
 import Data.Map.Strict.Patch (PatchMap, Replace, diff)
 import Data.Set (Set)
@@ -162,6 +163,46 @@ getOccByNgramsOnlyFast :: CorpusId
 getOccByNgramsOnlyFast cId nt ngs =
   fromListWith (+) <$> selectNgramsOccurrencesOnlyByNodeUser cId nt ngs
 
+
+getOccByNgramsOnlyFast' :: CorpusId
+                       -> ListId
+                       -> NgramsType
+                       -> [Text]
+                       -> Cmd err (Map Text Int)
+getOccByNgramsOnlyFast' cId lId nt tms = trace (show (cId, lId)) $ 
+  fromListWith (+) <$> map (second round) <$> run cId lId nt tms
+  
+    where
+      fields = [QualifiedIdentifier Nothing "text"]
+
+      run :: CorpusId
+           -> ListId
+           -> NgramsType
+           -> [Text]
+           -> Cmd err [(Text, Double)]
+      run cId' lId' _nt' tms' = runPGSQuery query
+                ( Values fields (DPS.Only <$> tms')
+                , cId'
+                , lId'
+                -- , ngramsTypeId nt'
+                )
+
+      query :: DPS.Query
+      query = [sql|
+        WITH input_rows(terms) AS (?)
+        SELECT ng.terms, nng.weight FROM node_node_ngrams nng
+          JOIN ngrams ng      ON nng.ngrams_id = ng.id
+          JOIN input_rows  ir ON ir.terms      = ng.terms
+          WHERE nng.node1_id     = ? -- CorpusId
+            AND nng.node2_id     = ?
+            -- AND nng.ngrams_type = ? -- NgramsTypeId
+            -- AND nn.category     > 0
+            GROUP BY ng.terms, nng.weight
+        |]
+
+
+
+
 -- just slower than getOccByNgramsOnlyFast
 getOccByNgramsOnlySlow :: NodeType
                        -> CorpusId
@@ -279,6 +320,38 @@ queryNgramsOnlyByNodeUser = [sql|
   |]
 
 
+
+
+selectNgramsOnlyByNodeUser' :: CorpusId -> [ListId] -> NgramsType -> [Text]
+                           -> Cmd err [(Text, Int)]
+selectNgramsOnlyByNodeUser' cId ls nt tms =
+  runPGSQuery queryNgramsOnlyByNodeUser
+                ( Values fields (DPS.Only <$> tms)
+                , Values [QualifiedIdentifier Nothing "int4"] (DPS.Only <$> (map (\(NodeId n) -> n) ls))
+                , cId
+                , nodeTypeId NodeDocument
+                , ngramsTypeId nt
+                )
+    where
+      fields = [QualifiedIdentifier Nothing "text"]
+
+queryNgramsOnlyByNodeUser' :: DPS.Query
+queryNgramsOnlyByNodeUser' = [sql|
+  WITH input_rows(terms) AS (?),
+       input_list(id)    AS (?)
+  SELECT ng.terms, nng.weight FROM node_node_ngrams nng
+    JOIN ngrams ng      ON nng.ngrams_id = ng.id
+    JOIN input_rows  ir ON ir.terms      = ng.terms
+    JOIN input_list  il ON il.id         = nng.node2_id
+    WHERE nng.node1_id     = ? -- CorpusId
+      AND nng.ngrams_type = ? -- NgramsTypeId
+      -- AND nn.category     > 0
+      GROUP BY ng.terms, nng.weight
+  |]
+
+
+
+
 getNgramsByDocOnlyUser :: NodeId -> [ListId] -> NgramsType -> [Text]
                          -> Cmd err (Map Text (Set NodeId))
 getNgramsByDocOnlyUser cId ls nt ngs =

src/Gargantext/Database/Triggers.hs

@@ -37,7 +37,7 @@ insertOccsUpdates cId lId = runPGSQuery query (cId, lId, nodeTypeId NodeList, no
     query :: DPS.Query
     query = [sql|
               INSERT INTO node_node_ngrams (node1_id, node2_id, ngrams_id, ngrams_type, weight)
-                SELECT nn.node1_id, lists.id, nnn.ngrams_id, count(*), 1  -- type of score
+                SELECT nn.node1_id, lists.id, nnn.ngrams_id, 1, count(*) as c  -- type of score
                    FROM node_node_ngrams nnn
                    INNER JOIN nodes_nodes nn    ON nn.node2_id = nnn.node2_id
                    INNER JOIN nodes       docs  ON docs.id     = nnn.node2_id
@@ -49,7 +49,9 @@ insertOccsUpdates cId lId = runPGSQuery query (cId, lId, nodeTypeId NodeList, no
                     AND docs.typename  = ?
                   GROUP BY nn.node1_id, lists.id, nnn.ngrams_id
                ON CONFLICT (node1_id, node2_id, ngrams_id, ngrams_type)
-                  DO UPDATE SET weight = excluded.weight +1; -- TOCHECK
+                  DO UPDATE SET weight = 3 -- c -- excluded.weight
+                  RETURNING 1
+                  -- TOCHECK
   |]