Add CDB_EstimateRowCount function

See #295

scripts-available/CDB_EstimateRowCount.sql

@@ -0,0 +1,29 @@
+-- Internal function to generate stats for a table if they don't exist
+CREATE OR REPLACE FUNCTION _CDB_GenerateStats(reloid REGCLASS)
+RETURNS VOID
+AS $$
+DECLARE
+  num_cols INTEGER;
+BEGIN
+  SELECT COUNT(*) FROM pg_catalog.pg_statistic WHERE starelid = reloid INTO num_cols;
+  IF num_cols = 0 THEN
+    EXECUTE Format('ANALYZE %s;', reloid);
+  END IF;
+END
+$$ LANGUAGE 'plpgsql' VOLATILE STRICT SECURITY DEFINER;
+
+-- Return a row count estimate of the result of a query using statistics
+CREATE OR REPLACE FUNCTION CDB_EstimateRowCount(query text)
+RETURNS Numeric
+AS $$
+DECLARE
+  plan JSON;
+BEGIN
+  -- Make sure statistics exist for all the tables of the query
+  PERFORM _CDB_GenerateStats(tabname) FROM  unnest(CDB_QueryTablesText(query)) AS tabname;
+
+  -- Use the query planner to obtain an estimate of the number of result rows
+  EXECUTE 'EXPLAIN (FORMAT JSON) ' || query INTO STRICT plan;
+  RETURN plan->0->'Plan'->'Plan Rows';
+END
+$$ LANGUAGE 'plpgsql' VOLATILE STRICT;