Snowflake: Improve support for reserved keywords for table factor

Author: yoavcloud

src/dialect/mod.rs

@@ -948,12 +948,6 @@ pub trait Dialect: Debug + Any {
         keywords::RESERVED_FOR_IDENTIFIER.contains(&kw)
     }
 
-    /// Returns reserved keywords when looking to parse a `TableFactor`.
-    /// See [Self::supports_from_trailing_commas]
-    fn get_reserved_keywords_for_table_factor(&self) -> &[Keyword] {
-        keywords::RESERVED_FOR_TABLE_FACTOR
-    }
-
     /// Returns reserved keywords that may prefix a select item expression
     /// e.g. `SELECT CONNECT_BY_ROOT name FROM Tbl2` (Snowflake)
     fn get_reserved_keywords_for_select_item_operator(&self) -> &[Keyword] {
@@ -1012,7 +1006,13 @@ pub trait Dialect: Debug + Any {
         explicit || self.is_column_alias(kw, parser)
     }
 
-    /// Returns true if the specified keyword should be parsed as a table identifier.
+    /// Returns true if the specified keyword should be parsed as a table factor identifier.
+    /// See [keywords::RESERVED_FOR_TABLE_FACTOR]
+    fn is_table_factor(&self, kw: &Keyword, _parser: &mut Parser) -> bool {
+        !keywords::RESERVED_FOR_TABLE_FACTOR.contains(kw)
+    }
+
+    /// Returns true if the specified keyword should be parsed as a table factor alias.
     /// See [keywords::RESERVED_FOR_TABLE_ALIAS]
     fn is_table_alias(&self, kw: &Keyword, _parser: &mut Parser) -> bool {
         !keywords::RESERVED_FOR_TABLE_ALIAS.contains(kw)

src/dialect/snowflake.rs

@@ -45,6 +45,82 @@ use super::keywords::RESERVED_FOR_IDENTIFIER;
 use sqlparser::ast::StorageSerializationPolicy;
 
 const RESERVED_KEYWORDS_FOR_SELECT_ITEM_OPERATOR: [Keyword; 1] = [Keyword::CONNECT_BY_ROOT];
+
+// See: <https://docs.snowflake.com/en/sql-reference/reserved-keywords>
+const RESERVED_KEYWORDS_FOR_TABLE_FACTOR: &[Keyword] = &[
+    Keyword::ALL,
+    Keyword::ALTER,
+    Keyword::AND,
+    Keyword::ANY,
+    Keyword::AS,
+    Keyword::BETWEEN,
+    Keyword::BY,
+    Keyword::CHECK,
+    Keyword::COLUMN,
+    Keyword::CONNECT,
+    Keyword::CREATE,
+    Keyword::CROSS,
+    Keyword::CURRENT,
+    Keyword::DELETE,
+    Keyword::DISTINCT,
+    Keyword::DROP,
+    Keyword::ELSE,
+    Keyword::EXISTS,
+    Keyword::FOLLOWING,
+    Keyword::FOR,
+    Keyword::FROM,
+    Keyword::FULL,
+    Keyword::GRANT,
+    Keyword::GROUP,
+    Keyword::HAVING,
+    Keyword::ILIKE,
+    Keyword::IN,
+    Keyword::INCREMENT,
+    Keyword::INNER,
+    Keyword::INSERT,
+    Keyword::INTERSECT,
+    Keyword::INTO,
+    Keyword::IS,
+    Keyword::JOIN,
+    Keyword::LEFT,
+    Keyword::LIKE,
+    Keyword::MINUS,
+    Keyword::NATURAL,
+    Keyword::NOT,
+    Keyword::NULL,
+    Keyword::OF,
+    Keyword::ON,
+    Keyword::OR,
+    Keyword::ORDER,
+    Keyword::QUALIFY,
+    Keyword::REGEXP,
+    Keyword::REVOKE,
+    Keyword::RIGHT,
+    Keyword::RLIKE,
+    Keyword::ROW,
+    Keyword::ROWS,
+    Keyword::SAMPLE,
+    Keyword::SELECT,
+    Keyword::SET,
+    Keyword::SOME,
+    Keyword::START,
+    Keyword::TABLE,
+    Keyword::TABLESAMPLE,
+    Keyword::THEN,
+    Keyword::TO,
+    Keyword::TRIGGER,
+    Keyword::UNION,
+    Keyword::UNIQUE,
+    Keyword::UPDATE,
+    Keyword::USING,
+    Keyword::VALUES,
+    Keyword::WHEN,
+    Keyword::WHENEVER,
+    Keyword::WHERE,
+    Keyword::WINDOW,
+    Keyword::WITH,
+];
+
 /// A [`Dialect`] for [Snowflake](https://www.snowflake.com/)
 #[derive(Debug, Default)]
 pub struct SnowflakeDialect;
@@ -427,6 +503,21 @@ impl Dialect for SnowflakeDialect {
         }
     }
 
+    fn is_table_factor(&self, kw: &Keyword, parser: &mut Parser) -> bool {
+        match kw {
+            Keyword::LIMIT
+                if matches!(
+                    parser.peek_token().token,
+                    Token::Number(_, _) | Token::Placeholder(_)
+                ) =>
+            {
+                false
+            }
+
+            _ => !RESERVED_KEYWORDS_FOR_TABLE_FACTOR.contains(kw),
+        }
+    }
+
     /// See: <https://docs.snowflake.com/en/sql-reference/constructs/at-before>
     fn supports_timestamp_versioning(&self) -> bool {
         true

src/parser/mod.rs

@@ -4369,11 +4369,7 @@ impl<'a> Parser<'a> {
         self.parse_comma_separated_with_trailing_commas(
             Parser::parse_table_and_joins,
             trailing_commas,
-            |kw, _parser| {
-                self.dialect
-                    .get_reserved_keywords_for_table_factor()
-                    .contains(kw)
-            },
+            |kw, parser| !self.dialect.is_table_factor(kw, parser),
         )
     }
 

tests/sqlparser_snowflake.rs

@@ -3543,6 +3543,18 @@ fn test_sql_keywords_as_table_aliases() {
     }
 }
 
+#[test]
+fn test_sql_keywords_as_table_factor() {
+    // LIMIT is a table factor, Snowflake does not reserve it
+    snowflake().one_statement_parses_to("SELECT * FROM tbl, LIMIT", "SELECT * FROM tbl, LIMIT");
+    // LIMIT is not a table factor
+    snowflake().one_statement_parses_to("SELECT * FROM tbl, LIMIT 1", "SELECT * FROM tbl LIMIT 1");
+    // ORDER is reserved
+    assert!(snowflake()
+        .parse_sql_statements("SELECT * FROM tbl, order")
+        .is_err());
+}
+
 #[test]
 fn test_timetravel_at_before() {
     snowflake().verified_only_select("SELECT * FROM tbl AT(TIMESTAMP => '2024-12-15 00:00:00')");
@@ -4438,3 +4450,16 @@ fn test_snowflake_identifier_function() {
         true
     );
 }
+
+#[test]
+fn test_x() {
+    println!(
+        "{:#?}",
+        snowflake()
+            .parse_sql_statements(
+                r#"
+SELECT email from customers, order by 1"#
+            )
+            .unwrap()
+    );
+}