but create as and distributed by

Author: rekby

tools/greenplum-to-pg-tests/cmd/extrace_sessions_test.go

@@ -31,3 +31,45 @@ WHERE`,
 		})
 	}
 }
+
+func TestCutGreenplumSpecific(t *testing.T) {
+	table := []struct {
+		name string
+		from string
+		to   string
+	}{
+		{
+			name: "CreateAs",
+			from: `create table snb_eagle___hex_lavka_orders as
+        select *
+        from snb_geo_lavka___hex_lavka_orders;
+`,
+			to: `
+        select *
+        from snb_geo_lavka___hex_lavka_orders;
+`,
+		},
+		{
+			name: "CreateAndDistributedBy",
+			from: `        CREATE TEMPORARY TABLE updated_tickets
+        AS (
+            SELECT DISTINCT chatterbox_ticket_id AS ticket_id
+            FROM taxi_cdm_contactcenter___fct_chatterbox_ticket_event --fct_chatterbox_ticket_event
+            WHERE utc_event_dttm BETWEEN ''2024-04-10 00:00:00''::timestamp AND ''2024-04-30 23:59:59''::timestamp
+        )
+            DISTRIBUTED BY (ticket_id);
+`,
+			to: `        SELECT DISTINCT chatterbox_ticket_id AS ticket_id
+            FROM taxi_cdm_contactcenter___fct_chatterbox_ticket_event --fct_chatterbox_ticket_event
+            WHERE utc_event_dttm BETWEEN ''2024-04-10 00:00:00''::timestamp AND ''2024-04-30 23:59:59''::timestamp
+        ;
+`,
+		},
+	}
+	for _, test := range table {
+		t.Run(test.name, func(t *testing.T) {
+			res := cutGreenplumSpecific(test.from)
+			require.Equal(t, test.to, res)
+		})
+	}
+}

tools/greenplum-to-pg-tests/cmd/extraxtSessions.go

@@ -217,7 +217,6 @@ readLoop:
 
 func checkQueries(rules Rules, stats *QueryStats, pgSchema *internal.PgSchema, db *ydb.Driver, sessions []internal.Session) {
 	reasonFilter := regexp.MustCompile(extractSessionsConfig.filterReason)
-	//checked := map[string]bool{}
 
 	errorLimit := extractSessionsConfig.errorLimit
 	if errorLimit == 0 {
@@ -291,6 +290,7 @@ func checkQuery(stat *QueryStats, rules Rules, db *ydb.Driver, queryText string)
 	queryText = strings.TrimSpace(queryText)
 	queryText = fixSchemaNames(queryText)
 	queryText = fixCreateTable(queryText)
+	queryText = cutGreenplumSpecific(queryText)
 
 	ctx := context.Background()
 	res, err := db.Query().Execute(
@@ -349,6 +349,19 @@ func fixCreateTable(queryText string) string {
 	return queryText
 }
 
+func cutGreenplumSpecific(q string) string {
+	q = createAndDistributedByWithBrackets.ReplaceAllString(q, "$1;")
+	q = createTableAsSelect.ReplaceAllLiteralString(q, "")
+	q = distributedBy.ReplaceAllLiteralString(q, ";")
+	return q
+}
+
+var (
+	createAndDistributedByWithBrackets = regexp.MustCompile(`(?is)CREATE\s+.*\sTABLE\s+.*\s+AS\s+\(\s*(.*)\s*\)\s+DISTRIBUTED\s+BY\s\(.*\);`)
+	createTableAsSelect                = regexp.MustCompile(`(?i)create\s+(temporary\s+)?table .* as`)
+	distributedBy                      = regexp.MustCompile(`(?i)DISTRIBUTED BY ([^\)]+);`)
+)
+
 type QueryStats struct {
 	OkCount    int
 	TotalCount int