diff --git a/src/datatypes/src/value.rs b/src/datatypes/src/value.rs index 56a551484f2..2a72516d5db 100644 --- a/src/datatypes/src/value.rs +++ b/src/datatypes/src/value.rs @@ -1152,12 +1152,13 @@ impl TryFrom for Value { ScalarValue::UInt16(u) => Value::from(u), ScalarValue::UInt32(u) => Value::from(u), ScalarValue::UInt64(u) => Value::from(u), - ScalarValue::Utf8(s) | ScalarValue::LargeUtf8(s) => { + ScalarValue::Utf8(s) | ScalarValue::LargeUtf8(s) | ScalarValue::Utf8View(s) => { Value::from(s.map(StringBytes::from)) } ScalarValue::Binary(b) | ScalarValue::LargeBinary(b) - | ScalarValue::FixedSizeBinary(_, b) => Value::from(b.map(Bytes::from)), + | ScalarValue::FixedSizeBinary(_, b) + | ScalarValue::BinaryView(b) => Value::from(b.map(Bytes::from)), ScalarValue::List(array) => { // this is for item type let datatype = ConcreteDataType::try_from(&array.value_type())?; @@ -1221,6 +1222,12 @@ impl TryFrom for Value { .map(|v| Value::Decimal128(Decimal128::new(v, p, s))) .unwrap_or(Value::Null), ScalarValue::Struct(struct_array) => { + // A struct scalar carries a single element; a null (or empty) + // element means the struct itself is null, for example a null + // struct inside a list. + if struct_array.is_empty() || struct_array.is_null(0) { + return Ok(Value::Null); + } let struct_type = StructType::from(struct_array.fields()); let items = struct_array .columns() @@ -1244,8 +1251,6 @@ impl TryFrom for Value { | ScalarValue::LargeListView(_) | ScalarValue::Union(_, _, _) | ScalarValue::Float16(_) - | ScalarValue::Utf8View(_) - | ScalarValue::BinaryView(_) | ScalarValue::Map(_) | ScalarValue::Date64(_) | ScalarValue::RunEndEncoded(_, _, _) => { @@ -2157,6 +2162,29 @@ pub(crate) mod tests { Value::Struct(struct_value), scalar_struct_value.try_into().unwrap() ); + + // view-typed strings and binaries convert like their non-view forms + assert_eq!( + Value::String("abc".into()), + ScalarValue::Utf8View(Some("abc".into())) + .try_into() + .unwrap() + ); + assert_eq!(Value::Null, ScalarValue::Utf8View(None).try_into().unwrap()); + assert_eq!( + Value::Binary(Bytes::from(vec![1, 2])), + ScalarValue::BinaryView(Some(vec![1, 2])) + .try_into() + .unwrap() + ); + assert_eq!( + Value::Null, + ScalarValue::BinaryView(None).try_into().unwrap() + ); + + // a null struct scalar is a null value, not a struct of null fields + let null_struct = ScalarStructBuilder::new_null(build_struct_type().as_arrow_fields()); + assert_eq!(Value::Null, null_struct.try_into().unwrap()); } #[test] diff --git a/tests-integration/tests/sql.rs b/tests-integration/tests/sql.rs index d6be6c647ef..13b51c6cab9 100644 --- a/tests-integration/tests/sql.rs +++ b/tests-integration/tests/sql.rs @@ -96,6 +96,7 @@ macro_rules! sql_tests { test_mysql_prepare_tql_and_show, test_postgres_extended_query_row_returning_statements, test_postgres_native_histogram, + test_postgres_struct_types, test_declare_fetch_close_cursor, test_alter_update_on, ); @@ -1041,6 +1042,55 @@ pub async fn test_postgres_native_histogram(store_type: StorageType) { guard.remove_all().await; } +pub async fn test_postgres_struct_types(store_type: StorageType) { + let (mut guard, fe_pg_server) = setup_pg_server(store_type, "test_postgres_struct_types").await; + let addr = fe_pg_server.bind_addr().unwrap().to_string(); + + let (client, connection) = tokio_postgres::connect(&format!("postgres://{addr}/public"), NoTls) + .await + .unwrap(); + let (tx, rx) = tokio::sync::oneshot::channel(); + tokio::spawn(async move { + connection.await.unwrap(); + tx.send(()).unwrap(); + }); + + async fn query_one(client: &Client, sql: &str) -> String { + let messages = client.simple_query(sql).await.unwrap(); + messages + .iter() + .filter_map(|message| match message { + SimpleQueryMessage::Row(row) => row.get(0).map(|v| v.to_string()), + _ => None, + }) + .next() + .unwrap() + } + + // Struct and list-of-struct columns are serialized as JSON. View-typed + // struct fields must not fail row extraction, and a null struct inside a + // list stays null instead of becoming a struct of null fields. + let row = query_one(&client, "SELECT struct(arrow_cast('abc', 'Utf8View'))").await; + assert_eq!(row, "{\"c0\":\"abc\"}"); + + // '0102' casts to its UTF-8 bytes when interpreted as BinaryView. + let row = query_one(&client, "SELECT struct(arrow_cast('0102', 'BinaryView'))").await; + assert_eq!(row, "{\"c0\":[48,49,48,50]}"); + + let row = query_one(&client, "SELECT struct([struct(1), NULL])").await; + assert_eq!(row, "{\"c0\":[{\"c0\":1},null]}"); + + // Top-level control: the same null element in a bare list stays null too. + let row = query_one(&client, "SELECT [struct(1), NULL]").await; + assert_eq!(row, "[{\"c0\":1},null]"); + + drop(client); + rx.await.unwrap(); + + let _ = fe_pg_server.shutdown().await; + guard.remove_all().await; +} + pub async fn test_postgres_slow_query(store_type: StorageType) { let (mut guard, fe_pg_server) = setup_pg_server_with_slow_query_threshold( store_type,