apache/shardingsphere · error · SQLFeatureNotSupportedException

getUnicodeStream

Error message

getUnicodeStream

What it means

java.sql.SQLFeatureNotSupportedException thrown by the Apache ShardingSphere JDBC driver when ResultSet.getUnicodeStream(int columnIndex) is called on a DatabaseMetaData result set. Metadata calls such as Connection.getMetaData().getTables()/getColumns()/getIndexInfo()/getPrimaryKeys() on a jdbc:shardingsphere: URL return a synthesized DatabaseMetaDataResultSet that buffers every row as plain Java objects, so only scalar getters (getString, getInt, getBytes, single-arg getDate/getTime/getTimestamp, ...) are implemented. getUnicodeStream(int columnIndex) is declared final in AbstractUnsupportedDatabaseMetaDataResultSet and always throws, so this is a permanent, intentional limitation, not a version bug. getUnicodeStream has itself been deprecated since JDBC 2.0, so drivers are not expected to support it.

Source

Thrown at jdbc/src/main/java/org/apache/shardingsphere/driver/jdbc/unsupported/AbstractUnsupportedDatabaseMetaDataResultSet.java:51

/**
 * Unsupported Database meta data result set.
 */
public abstract class AbstractUnsupportedDatabaseMetaDataResultSet extends AbstractUnsupportedOperationResultSet {
    
    @Override
    public final InputStream getAsciiStream(final int columnIndex) throws SQLException {
        throw new SQLFeatureNotSupportedException("getAsciiStream");
    }
    
    @Override
    public final InputStream getAsciiStream(final String columnLabel) throws SQLException {
        throw new SQLFeatureNotSupportedException("getAsciiStream");
    }
    
    @Override
    public final InputStream getUnicodeStream(final int columnIndex) throws SQLException {
        throw new SQLFeatureNotSupportedException("getUnicodeStream");
    }
    
    @Override
    public final InputStream getUnicodeStream(final String columnLabel) throws SQLException {
        throw new SQLFeatureNotSupportedException("getUnicodeStream");
    }
    
    @Override
    public final InputStream getBinaryStream(final int columnIndex) throws SQLException {
        throw new SQLFeatureNotSupportedException("getBinaryStream");
    }
    
    @Override
    public final InputStream getBinaryStream(final String columnLabel) throws SQLException {
        throw new SQLFeatureNotSupportedException("getBinaryStream");
    }
    
    @Override

View on GitHub (pinned to e952770a21)

Solutions

1. Replace rs.getUnicodeStream(int columnIndex) with rs.getString(...), which DatabaseMetaDataResultSet implements, then build the stream/reader yourself (new ByteArrayInputStream(s.getBytes(StandardCharsets.US_ASCII)) or new StringReader(s)).
2. If binary data was intended, use rs.getBytes(...) instead of a character stream.
3. If you do not control the calling code (third-party tool), point its metadata introspection at the underlying physical datasource instead of the ShardingSphere datasource.
4. As a last resort wrap the metadata ResultSet in a delegating adapter that implements the stream getters on top of getString.
5. Do not wait for a driver upgrade: the throw is final and by design in every ShardingSphere 5.x release.

Example fix

// before - throws SQLFeatureNotSupportedException on ShardingSphere metadata result sets
InputStream ascii = metaDataRs.getAsciiStream("REMARKS");
Reader reader = metaDataRs.getCharacterStream("REMARKS");

// after - getString is implemented; build the stream/reader yourself
String remarks = metaDataRs.getString("REMARKS");
InputStream ascii = remarks == null ? null : new ByteArrayInputStream(remarks.getBytes(StandardCharsets.US_ASCII));
Reader reader = remarks == null ? null : new StringReader(remarks);

// Run BEFORE using exotic getters: detect a ShardingSphere-synthesized metadata ResultSet
static boolean isShardingSphereMetaDataResultSet(final ResultSet rs) {
    return rs != null && "org.apache.shardingsphere.driver.jdbc.core.resultset.DatabaseMetaDataResultSet"
            .equals(rs.getClass().getName());
}

static boolean supportsMetaDataGetter(final Connection c) throws SQLException {
    // jdbc:shardingsphere: URLs return the synthesized metadata result set
    return c != null && !c.getMetaData().getURL().startsWith("jdbc:shardingsphere:");
}

private static boolean isUnsupportedMetaDataResultSet(final ResultSet rs) {
    return rs != null && rs.getClass().getName().startsWith("org.apache.shardingsphere.driver.jdbc.core.resultset.DatabaseMetaDataResultSet");
}

try {
    return rs.getUnicodeStream(int columnIndex);
} catch (final SQLFeatureNotSupportedException e) {
    // ShardingSphere metadata result sets: fall back to the implemented scalar getter
    final String s = rs.getString(col);
}

When it happens

Trigger: Calling getUnicodeStream(int columnIndex) on the ResultSet returned by ShardingSphereDatabaseMetaData, e.g. connection.getMetaData().getTables(null, null, "%", new String[]{"TABLE"}") followed by rs.getUnicodeStream(1)) where connection was opened with the org.apache.shardingsphere.driver JDBC URL (jdbc:shardingsphere:classpath:sharding.yaml or jdbc:shardingsphere:...). Both the int-columnIndex and String-columnLabel overloads of every listed getter throw; the methods are final so no subclass can override them.

Common situations: Code that ran unchanged against the native MySQL/PostgreSQL/OpenGauss driver and is repointed to the ShardingSphere driver URL, so the same metadata loop now hits the stubbed getter; generic schema tooling that enumerates JDBC metadata with feature-rich getters (DBeaver, Squirrel SQL, Liquibase/Hibernate/Flyway schema validators, code generators, DBDiff tools); hand-written metadata loops that defensively call getWarnings()/clearWarnings() or use the Calendar overloads to pin a timezone.

Related errors

• getBinaryStream
• getCharacterStream
• getWarnings
• clearWarnings
• getArray

AI-assisted analysis of apache/shardingsphere@e952770a21 (2026-08-14). Data as JSON: /api/errors/a631a7c17353d22c. Report an issue: GitHub.