Remove duplicate entries in pg_namespace

BemiHQ · Jan 9, 2025 · 8612ba4 · 8612ba4
1 parent dce453e
commit 8612ba4
Show file tree

Hide file tree

Showing 5 changed files with 27 additions and 10 deletions.
diff --git a/src/duckdb.go b/src/duckdb.go
@@ -14,6 +14,7 @@ import (
 var DEFAULT_BOOT_QUERIES = []string{
 	"INSTALL iceberg",
 	"LOAD iceberg",
+	"SELECT oid FROM pg_catalog.pg_namespace",
 	"CREATE SCHEMA public",
 	"USE public",
 }

diff --git a/src/parser_where.go b/src/parser_where.go
@@ -17,13 +17,23 @@ func (parser *ParserWhere) FunctionCall(whereNode *pgQuery.Node) *pgQuery.FuncCa
 	return whereNode.GetFuncCall()
 }
 
-// WHERE column != 'value'
-func (parser *ParserWhere) MakeExpressionNode(column string, operation string, value string) *pgQuery.Node {
+// WHERE column NOT IN (values)
+func (parser *ParserWhere) MakeNotInExpressionNode(column string, values []int64, alias string) *pgQuery.Node {
+	columnRefNodes := []*pgQuery.Node{pgQuery.MakeStrNode(column)}
+	if alias != "" {
+		columnRefNodes = []*pgQuery.Node{pgQuery.MakeStrNode(alias), pgQuery.MakeStrNode(column)}
+	}
+
+	valuesNodes := make([]*pgQuery.Node, len(values))
+	for i, value := range values {
+		valuesNodes[i] = pgQuery.MakeAConstIntNode(value, 0)
+	}
+
 	return pgQuery.MakeAExprNode(
-		pgQuery.A_Expr_Kind_AEXPR_OP,
-		[]*pgQuery.Node{pgQuery.MakeStrNode(operation)},
-		pgQuery.MakeColumnRefNode([]*pgQuery.Node{pgQuery.MakeStrNode(column)}, 0),
-		pgQuery.MakeAConstStrNode(value, 0),
+		pgQuery.A_Expr_Kind_AEXPR_IN,
+		[]*pgQuery.Node{pgQuery.MakeStrNode("<>")},
+		pgQuery.MakeColumnRefNode(columnRefNodes, 0),
+		pgQuery.MakeListNode(valuesNodes),
 		0,
 	)
 }

diff --git a/src/query_handler.go b/src/query_handler.go
@@ -395,6 +395,10 @@ func (queryHandler *QueryHandler) remapQuery(query string) (string, error) {
 		return "", err
 	}
 
+	if strings.HasSuffix(query, " --INSPECT") {
+		LogDebug(queryHandler.config, queryTree.Stmts[0].Stmt)
+	}
+
 	queryTree.Stmts, err = queryHandler.queryRemapper.RemapStatements(queryTree.Stmts)
 	if err != nil {
 		return "", err

diff --git a/src/query_handler_test.go b/src/query_handler_test.go
@@ -173,7 +173,7 @@ func TestHandleQuery(t *testing.T) {
 			"types":       {Uint32ToString(pgtype.TextOID)},
 			"values":      {},
 		},
-		"SELECT n.nspname FROM pg_catalog.pg_namespace n LEFT OUTER JOIN pg_catalog.pg_description d ON d.objoid = n.oid ORDER BY n.oid LIMIT 1": {
+		"SELECT n.nspname FROM pg_catalog.pg_namespace n LEFT OUTER JOIN pg_catalog.pg_description d ON d.objoid = n.oid ORDER BY n.oid DESC LIMIT 1": {
 			"description": {"nspname"},
 			"types":       {Uint32ToString(pgtype.TextOID)},
 			"values":      {"public"},

diff --git a/src/query_remapper_table.go b/src/query_remapper_table.go
@@ -6,6 +6,8 @@ import (
 	pgQuery "github.com/pganalyze/pg_query_go/v5"
 )
 
+var REDUNDANT_PG_NAMESPACE_OIDS = []int64{0, 1148, 1253, 1264, 1265, 1266, 1267}
+
 type QueryRemapperTable struct {
 	parserTable         *ParserTable
 	parserWhere         *ParserWhere
@@ -184,10 +186,10 @@ func (remapper *QueryRemapperTable) RemapWhereClauseForTable(qSchemaTable QueryS
 	if remapper.isTableFromPgCatalog(qSchemaTable) {
 		switch qSchemaTable.Table {
 
-		// FROM pg_catalog.pg_namespace -> FROM pg_catalog.pg_namespace WHERE nspname != 'main'
+		// FROM pg_catalog.pg_namespace -> FROM pg_catalog.pg_namespace WHERE oid NOT IN (3 'main' schema oids, 2 'pg_catalog' and 2 'information_schema' duplicate oids)
 		case PG_TABLE_PG_NAMESPACE:
-			withoutMainSchemaWhereCondition := remapper.parserWhere.MakeExpressionNode("nspname", "!=", "main")
-			return remapper.parserWhere.AppendWhereCondition(selectStatement, withoutMainSchemaWhereCondition)
+			withoutDuckdbOidsWhereCondition := remapper.parserWhere.MakeNotInExpressionNode("oid", REDUNDANT_PG_NAMESPACE_OIDS, qSchemaTable.Alias)
+			remapper.parserWhere.AppendWhereCondition(selectStatement, withoutDuckdbOidsWhereCondition)
 
 		// FROM pg_catalog.pg_statio_user_tables -> FROM pg_catalog.pg_statio_user_tables WHERE false
 		case PG_TABLE_PG_STATIO_USER_TABLES: