[TOOLING] sql queries to remove unused documents and ngrams from database....

[TOOLING] sql queries to remove unused documents and ngrams from database. Usage: use a crontab to execute it every week for instance.

devops/postgres/tools/removeUnusedDocuments.sql

+
+WITH repeated AS
+  ( select nn.node2_id AS id, count(*) AS c
+    FROM nodes_nodes nn 
+    GROUP BY nn.node2_id
+  )
+
+DELETE FROM nodes n 
+USING repeated r
+WHERE
+n.id = r.id
+AND r.c <= 1
+; 
+

devops/postgres/tools/removeUnusedNgrams.sql

+
+WITH listed AS
+  ( select nn.ngrams_id AS id, count(*) AS c
+    FROM node_node_ngrams nn 
+    GROUP BY nn.ngrams_id
+  )
+
+--SELECT count(*) from listed l
+-- WHERE
+--l.c <= 1
+
+
+DELETE FROM ngrams n 
+  USING listed l
+  WHERE
+  n.id = l.id
+  AND l.c <= 1
+; 
+
+
+