Unable to Stream Parquet File using JDBC

[rickysaltzer]

I've seemed to hit a roadblock when trying to stream a Parquet file into ClickHouse using the JDBC library. Although the library doesn't contain a ClickHouseFormat.PARQUET I attempted to just specify the format via the sql parameter.

Curious to know if this is a known limitation, or if I'm just doing something wrong.

Query:

        connection.createStatement()
            .write()
            .sql("INSERT INTO `${studio.getDBName()}`.`${table}` FORMAT Parquet")
            .data(File("/tmp/pq/dump.parquet"))
            .send()

Error:

DB::Exception: Error while reading Parquet data: IOError: Couldn't deserialize thrift: TProtocolException: Invalid data
Deserializing page header failed.
 (version 20.4.4.18 (official build))

I've verified the Parquet file's correctness by using the clickhouse-client to import it, which went without issue.

cat /tmp/pq/dump.parquet | clickhouse-client ...

[den-crane]

DB::Exception: Error while reading Parquet data: IOError:
Couldn't deserialize thrift: TProtocolException: Invalid data
Deserializing page header failed.
(version 20.4.4.18 (official build))

BTW, this error is not related to JDBC.
ClickHouse server was unable to parse this specific dump.parquet.
This code should work with other (normal) parquet files.