Examples with FieldProjectionFilter org.apache.parquet.thrift.projection.FieldProjectionFilter used on opensource projects

Example 1 with FieldProjectionFilter

use of org.apache.parquet.thrift.projection.FieldProjectionFilter in project parquet-mr by apache.

the class ThriftReadSupport method init.

@Override
public org.apache.parquet.hadoop.api.ReadSupport.ReadContext init(InitContext context) {
    final Configuration configuration = context.getConfiguration();
    final MessageType fileMessageType = context.getFileSchema();
    MessageType requestedProjection = fileMessageType;
    String partialSchemaString = configuration.get(ReadSupport.PARQUET_READ_SCHEMA);
    FieldProjectionFilter projectionFilter = getFieldProjectionFilter(configuration);
    if (partialSchemaString != null && projectionFilter != null) {
        throw new ThriftProjectionException(String.format("You cannot provide both a partial schema and field projection filter." + "Only one of (%s, %s, %s) should be set.", PARQUET_READ_SCHEMA, STRICT_THRIFT_COLUMN_FILTER_KEY, THRIFT_COLUMN_FILTER_KEY));
    }
    // set requestedProjections only when it's specified
    if (partialSchemaString != null) {
        requestedProjection = getSchemaForRead(fileMessageType, partialSchemaString);
    } else if (projectionFilter != null) {
        try {
            initThriftClassFromMultipleFiles(context.getKeyValueMetadata(), configuration);
            requestedProjection = getProjectedSchema(projectionFilter);
        } catch (ClassNotFoundException e) {
            throw new ThriftProjectionException("can not find thriftClass from configuration", e);
        }
    }
    MessageType schemaForRead = getSchemaForRead(fileMessageType, requestedProjection);
    return new ReadContext(schemaForRead);
}