fix cut unsupported constructions

Author: rekby

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

@@ -409,7 +409,7 @@ func checkQuery(stat *QueryStats, rules Rules, db *ydb.Driver, queryText string)
 	queryText = strings.TrimSpace(queryText)
 	queryText = fixSchemaNames(queryText)
 	queryText = fixCreateTable(queryText)
-	queryText = cutGreenplumSpecific(queryText)
+	queryText = cutUnsupportedConstructions(queryText)
 
 	ctx := context.Background()
 	res, err := db.Query().Execute(
@@ -468,17 +468,19 @@ func fixCreateTable(queryText string) string {
 	return queryText
 }
 
-func cutGreenplumSpecific(q string) string {
-	q = createAndDistributedByWithBrackets.ReplaceAllString(q, "$1")
-	q = createTableAsSelect.ReplaceAllLiteralString(q, "")
+func cutUnsupportedConstructions(q string) string {
+	q = createAS.ReplaceAllString(q, "$1")
+	q = createTableAsSelect.ReplaceAllLiteralString(q, "SELECT")
 	q = distributedBy.ReplaceAllLiteralString(q, "")
+	q = distributedWord.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(`(?is)create\s+(temporary\s+)?table .* as`)
-	distributedBy                      = regexp.MustCompile(`(?i)DISTRIBUTED BY \(.*\)`)
+	createAS            = regexp.MustCompile(`(?is)CREATE\s+.*\sTABLE\s+.*\s+AS\s+\(\s*(.*)\s*\)\s`)
+	createTableAsSelect = regexp.MustCompile(`(?is)CREATE\s+(TEMPORARY\s+)?TABLE .* AS\s+SELECT`)
+	distributedBy       = regexp.MustCompile(`(?is)DISTRIBUTED BY \(.*\)`)
+	distributedWord     = regexp.MustCompile(`(?is)DISTRIBUTED \w+`)
 )
 
 type QueryStats struct {

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

@@ -44,8 +44,7 @@ func TestCutGreenplumSpecific(t *testing.T) {
         select *
         from bbb;
 `,
-			to: `
-        select *
+			to: `SELECT *
         from bbb;
 `,
 		},
@@ -56,19 +55,35 @@ func TestCutGreenplumSpecific(t *testing.T) {
             SELECT DISTINCT b AS ticket_id
             FROM t2 --comment
             WHERE mytime BETWEEN ''2024-04-10 00:00:00''::timestamp AND ''2024-04-30 23:59:59''::timestamp
-        )
-            DISTRIBUTED BY (ticket_id);
+        ) DISTRIBUTED BY (ticket_id);
 `,
 			to: `        SELECT DISTINCT b AS ticket_id
             FROM t2 --comment
             WHERE mytime BETWEEN ''2024-04-10 00:00:00''::timestamp AND ''2024-04-30 23:59:59''::timestamp
         ;
+`,
+		},
+		{
+			name: "CreateAndDistributedSomething",
+			from: `        CREATE TEMPORARY TABLE result_table
+            ON COMMIT DROP AS
+        SELECT dt AS utc_dt
+            , diff_a
+            , diff_b
+        FROM t
+        DISTRIBUTED REPLICATED;
+`,
+			to: `        SELECT dt AS utc_dt
+            , diff_a
+            , diff_b
+        FROM t
+        ;
 `,
 		},
 	}
 	for _, test := range table {
 		t.Run(test.name, func(t *testing.T) {
-			res := cutGreenplumSpecific(test.from)
+			res := cutUnsupportedConstructions(test.from)
 			require.Equal(t, test.to, res)
 		})
 	}

tools/greenplum-to-pg-tests/issues.yaml

@@ -15,8 +15,6 @@ issues:
     query_regexp:
       - "(?i)^\\s*COMMIT\\s*;?\\s*$"
       - "(?i)^\\s*ROLLBACK\\s*;?\\s*$"
-  - name: Set variables
-    issue_regexp: "VariableSetStmt, not supported name"
   - name: drop external table
     count: 2
     tag: