Fix parsing error when having fields after nested struct in BigQuery (#1897)
Some checks are pending
license / Release Audit Tool (RAT) (push) Waiting to run
Rust / codestyle (push) Waiting to run
Rust / lint (push) Waiting to run
Rust / benchmark-lint (push) Waiting to run
Rust / compile (push) Waiting to run
Rust / docs (push) Waiting to run
Rust / compile-no-std (push) Waiting to run
Rust / test (beta) (push) Waiting to run
Rust / test (nightly) (push) Waiting to run
Rust / test (stable) (push) Waiting to run

This commit is contained in:
hulk 2025-06-20 22:56:26 +08:00 committed by GitHub
parent b1b379e570
commit 185a490218
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
2 changed files with 77 additions and 38 deletions

View file

@ -3034,7 +3034,6 @@ impl<'a> Parser<'a> {
where
F: FnMut(&mut Parser<'a>) -> Result<(StructField, MatchedTrailingBracket), ParserError>,
{
let start_token = self.peek_token();
self.expect_keyword_is(Keyword::STRUCT)?;
// Nothing to do if we have no type information.
@ -3047,16 +3046,10 @@ impl<'a> Parser<'a> {
let trailing_bracket = loop {
let (def, trailing_bracket) = elem_parser(self)?;
field_defs.push(def);
if !self.consume_token(&Token::Comma) {
// The struct field definition is finished if it occurs `>>` or comma.
if trailing_bracket.0 || !self.consume_token(&Token::Comma) {
break trailing_bracket;
}
// Angle brackets are balanced so we only expect the trailing `>>` after
// we've matched all field types for the current struct.
// e.g. this is invalid syntax `STRUCT<STRUCT<INT>>>, INT>(NULL)`
if trailing_bracket.0 {
return parser_err!("unmatched > in STRUCT definition", start_token.span.start);
}
};
Ok((

View file

@ -635,35 +635,6 @@ fn parse_nested_data_types() {
}
}
#[test]
fn parse_invalid_brackets() {
let sql = "SELECT STRUCT<INT64>>(NULL)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError("unmatched > in STRUCT literal".to_string())
);
let sql = "SELECT STRUCT<STRUCT<INT64>>>(NULL)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError("Expected: (, found: >".to_string())
);
let sql = "CREATE TABLE table (x STRUCT<STRUCT<INT64>>>)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError(
"Expected: ',' or ')' after column definition, found: >".to_string()
)
);
}
#[test]
fn parse_tuple_struct_literal() {
// tuple syntax: https://cloud.google.com/bigquery/docs/reference/standard-sql/data-types#tuple_syntax
@ -2472,3 +2443,78 @@ fn test_struct_field_options() {
")",
));
}
#[test]
fn test_struct_trailing_and_nested_bracket() {
bigquery().verified_stmt(concat!(
"CREATE TABLE my_table (",
"f0 STRING, ",
"f1 STRUCT<a STRING, b STRUCT<c INT64, d STRING>>, ",
"f2 STRING",
")",
));
// More complex nested structs
bigquery().verified_stmt(concat!(
"CREATE TABLE my_table (",
"f0 STRING, ",
"f1 STRUCT<a STRING, b STRUCT<c INT64, d STRUCT<e STRING>>>, ",
"f2 STRUCT<h STRING, i STRUCT<j INT64, k STRUCT<l STRUCT<m STRING>>>>, ",
"f3 STRUCT<e STRING, f STRUCT<c INT64>>",
")",
));
// Bad case with missing closing bracket
assert_eq!(
ParserError::ParserError("Expected: >, found: )".to_owned()),
bigquery()
.parse_sql_statements("CREATE TABLE my_table(f1 STRUCT<a STRING, b INT64)")
.unwrap_err()
);
// Bad case with redundant closing bracket
assert_eq!(
ParserError::ParserError(
"unmatched > after parsing data type STRUCT<a STRING, b INT64>)".to_owned()
),
bigquery()
.parse_sql_statements("CREATE TABLE my_table(f1 STRUCT<a STRING, b INT64>>)")
.unwrap_err()
);
// Base case with redundant closing bracket in nested struct
assert_eq!(
ParserError::ParserError(
"Expected: ',' or ')' after column definition, found: >".to_owned()
),
bigquery()
.parse_sql_statements("CREATE TABLE my_table(f1 STRUCT<a STRUCT<b INT>>>, c INT64)")
.unwrap_err()
);
let sql = "SELECT STRUCT<INT64>>(NULL)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError("unmatched > in STRUCT literal".to_string())
);
let sql = "SELECT STRUCT<STRUCT<INT64>>>(NULL)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError("Expected: (, found: >".to_string())
);
let sql = "CREATE TABLE table (x STRUCT<STRUCT<INT64>>>)";
assert_eq!(
bigquery_and_generic()
.parse_sql_statements(sql)
.unwrap_err(),
ParserError::ParserError(
"Expected: ',' or ')' after column definition, found: >".to_string()
)
);
}