bigquery identifier escaping

sqlconnect/internal/bigquery/dialect.go

@@ -1,6 +1,7 @@
 package bigquery
 
 import (
+	"regexp"
 	"strings"
 
 	"github.com/rudderlabs/sqlconnect-go/sqlconnect"
@@ -9,6 +10,11 @@ import (
 
 type dialect struct{}
 
+var (
+	escape   = regexp.MustCompile("('|\"|`)")
+	unescape = regexp.MustCompile("\\\\('|\")")
+)
+
 // QuoteTable quotes a table name
 func (d dialect) QuoteTable(table sqlconnect.RelationRef) string {
 	if table.Schema != "" {
@@ -19,7 +25,7 @@ func (d dialect) QuoteTable(table sqlconnect.RelationRef) string {
 
 // QuoteIdentifier quotes an identifier, e.g. a column name
 func (d dialect) QuoteIdentifier(name string) string {
-	return "`" + strings.ReplaceAll(name, "`", "``") + "`"
+	return "`" + escape.ReplaceAllString(name, "\\$1") + "`"
 }
 
 // FormatTableName formats a table name, typically by lower or upper casing it, depending on the database
@@ -31,11 +37,16 @@ var identityFn = func(s string) string { return s }
 
 // NormaliseIdentifier normalises identifier parts that are unquoted, typically by lower or upper casing them, depending on the database
 func (d dialect) NormaliseIdentifier(identifier string) string {
-	return base.NormaliseIdentifier(identifier, '`', identityFn)
+	return base.NormaliseIdentifier(normalise(identifier), '`', identityFn)
 }
 
 // ParseRelationRef parses a string into a RelationRef after normalising the identifier and stripping out surrounding quotes.
 // The result is a RelationRef with case-sensitive fields, i.e. it can be safely quoted (see [QuoteTable] and, for instance, used for matching against the database's information schema.
 func (d dialect) ParseRelationRef(identifier string) (sqlconnect.RelationRef, error) {
-	return base.ParseRelationRef(identifier, '`', identityFn)
+	return base.ParseRelationRef(normalise(identifier), '`', identityFn)
+}
+
+func normalise(identifier string) string {
+	identifier = strings.ReplaceAll(identifier, "\\`", "``")
+	return unescape.ReplaceAllString(identifier, "$1")
 }

sqlconnect/internal/bigquery/dialect_test.go

@@ -18,6 +18,8 @@ func TestDialect(t *testing.T) {
 	t.Run("quote identifier", func(t *testing.T) {
 		quoted := d.QuoteIdentifier("column")
 		require.Equal(t, "`column`", quoted, "column name should be quoted with backticks")
+
+		require.Equal(t, "`col\\`umn`", d.QuoteIdentifier("col`umn"), "column name with backtick should be escaped")
 	})
 
 	t.Run("quote table", func(t *testing.T) {
@@ -62,8 +64,8 @@ func TestDialect(t *testing.T) {
 		require.NoError(t, err)
 		require.Equal(t, sqlconnect.RelationRef{Schema: "ScHeMA", Name: "TaBle"}, parsed)
 
-		parsed, err = d.ParseRelationRef("`CaTa``LoG`.ScHeMA.`TaBle`")
+		parsed, err = d.ParseRelationRef("`CaTa``LoG`.ScHeMA.`TaB\\`\\\"\\'le`")
 		require.NoError(t, err)
-		require.Equal(t, sqlconnect.RelationRef{Catalog: "CaTa`LoG", Schema: "ScHeMA", Name: "TaBle"}, parsed)
+		require.Equal(t, sqlconnect.RelationRef{Catalog: "CaTa`LoG", Schema: "ScHeMA", Name: "TaB`\"'le"}, parsed)
 	})
 }

sqlconnect/internal/integration_test/db_integration_test_scenario.go

@@ -163,7 +163,7 @@ func TestDatabaseScenarios(t *testing.T, warehouse string, configJSON json.RawMe
 		t.Run("with quoted table", func(t *testing.T) {
 			identifier := db.QuoteIdentifier(schema.Name) + "." + db.QuoteIdentifier("Quoted_TablE")
 			_, err := db.Exec("CREATE TABLE " + identifier + " (c1 int)")
-			require.NoError(t, err, "it should be able to create a quoted table")
+			require.NoErrorf(t, err, "it should be able to create a quoted table: %s", identifier)
 
 			table, err := db.ParseRelationRef(identifier)
 			require.NoError(t, err, "it should be able to parse a quoted table")