org.apache.parquet.filter.UnboundRecordFilter Java Examples

private static UnboundRecordFilter getUnboundRecordFilterInstance(Configuration configuration) {
  Class<?> clazz = ConfigurationUtil.getClassFromConfig(configuration, UNBOUND_RECORD_FILTER, UnboundRecordFilter.class);
  if (clazz == null) { return null; }

  try {
    UnboundRecordFilter unboundRecordFilter = (UnboundRecordFilter) clazz.newInstance();

    if (unboundRecordFilter instanceof Configurable) {
      ((Configurable)unboundRecordFilter).setConf(configuration);
    }

    return unboundRecordFilter;
  } catch (InstantiationException | IllegalAccessException e) {
    throw new BadConfigurationException(
        "could not instantiate unbound record filter class", e);
  }
}
 

/**
 * Given either a FilterPredicate or the class of an UnboundRecordFilter, or neither (but not both)
 * return a Filter that wraps whichever was provided.
 * <p>
 * Either filterPredicate or unboundRecordFilterClass must be null, or an exception is thrown.
 * <p>
 * If both are null, the no op filter will be returned.
 *
 * @param filterPredicate a filter predicate, or null
 * @param unboundRecordFilter an unbound record filter, or null
 * @return a Filter wrapping either the predicate or the unbound record filter (from the old API)
 */
public static Filter get(FilterPredicate filterPredicate, UnboundRecordFilter unboundRecordFilter) {
  checkArgument(filterPredicate == null || unboundRecordFilter == null,
      "Cannot provide both a FilterPredicate and an UnboundRecordFilter");

  if (filterPredicate != null) {
    return get(filterPredicate);
  }

  if (unboundRecordFilter != null) {
    return get(unboundRecordFilter);
  }

  return NOOP;
}
 

/**
 * @param root          the root of the schema
 * @param validating
 * @param columnStore
 * @param unboundFilter Filter records, pass in NULL_FILTER to leave unfiltered.
 */
public FilteredRecordReader(MessageColumnIO root, RecordMaterializer<T> recordMaterializer, boolean validating,
                            ColumnReadStoreImpl columnStore, UnboundRecordFilter unboundFilter, long recordCount) {
  super(root, recordMaterializer, validating, columnStore);
  this.recordCount = recordCount;
  if ( unboundFilter != null ) {
    recordFilter = unboundFilter.bind(getColumnReaders());
  } else {
    recordFilter = null;
  }
}
 

public static void setUnboundRecordFilter(Job job, Class<? extends UnboundRecordFilter> filterClass) {
  Configuration conf = ContextUtil.getConfiguration(job);
  checkArgument(getFilterPredicate(conf) == null,
      "You cannot provide an UnboundRecordFilter after providing a FilterPredicate");

  conf.set(UNBOUND_RECORD_FILTER, filterClass.getName());
}
 

private UnboundRecordFilterCompat(UnboundRecordFilter unboundRecordFilter) {
  this.unboundRecordFilter = Objects.requireNonNull(unboundRecordFilter, "unboundRecordFilter cannot be null");
}
 

public UnboundRecordFilter getUnboundRecordFilter() {
  return unboundRecordFilter;
}
 

/**
 * Creates a new TajoParquetReader.
 *
 * @param conf the configuration
 * @param file The file to read from.
 * @param readSchema Tajo schema of the table.
 * @param requestedSchema Tajo schema of the projection.
 * @param recordFilter Record filter.
 */
public TajoParquetReader(Configuration conf, Path file, Schema readSchema,
                         Schema requestedSchema,
                         UnboundRecordFilter recordFilter)
    throws IOException {
  super(conf, file, new TajoReadSupport(readSchema, requestedSchema),
        recordFilter);
}
 

/**
 * @param columns a page read store with the column data
 * @param recordMaterializer a record materializer
 * @param filter a record filter
 * @param <T> the type of records returned by the reader
 * @return a record reader
 * @deprecated use getRecordReader(PageReadStore, RecordMaterializer, Filter)
 */
@Deprecated
public <T> RecordReader<T> getRecordReader(PageReadStore columns,
                                           RecordMaterializer<T> recordMaterializer,
                                           UnboundRecordFilter filter) {
  return getRecordReader(columns, recordMaterializer, FilterCompat.get(filter));
}
 

/**
 * @param readSupport Object which helps reads files of the given type, e.g. Thrift, Avro.
 * @param filter Optional filter for only returning matching records.
 * @deprecated use {@link #InternalParquetRecordReader(ReadSupport, Filter)}
 */
@Deprecated
public InternalParquetRecordReader(ReadSupport<T> readSupport, UnboundRecordFilter filter) {
  this(readSupport, FilterCompat.get(filter));
}
 

/**
 * @param conf a configuration
 * @param file a file path
 * @param unboundRecordFilter an unbound record filter (from the old filter API)
 * @throws IOException if there is an error while reading
 * @deprecated will be removed in 2.0.0; use {@link #builder(InputFile)} instead.
 */
@Deprecated
public AvroParquetReader(Configuration conf, Path file, UnboundRecordFilter unboundRecordFilter) throws IOException {
  super(conf, file, new AvroReadSupport<T>(), unboundRecordFilter);
}
 

/**
 * @param file a file path
 * @param unboundRecordFilter an unbound record filter (from the old filter API)
 * @throws IOException if there is an error while reading
 * @deprecated will be removed in 2.0.0; use {@link #builder(InputFile)} instead.
 */
@Deprecated
public AvroParquetReader(Path file, UnboundRecordFilter unboundRecordFilter) throws IOException {
  super(file, new AvroReadSupport<T>(), unboundRecordFilter);
}
 

/**
 * @param file a file path
 * @param recordFilter an unbound record filter
 * @throws IOException if there is an error while reading
 * @deprecated use {@link #builder(Path)}
 */
@Deprecated
@SuppressWarnings("unchecked")
public ProtoParquetReader(Path file, UnboundRecordFilter recordFilter) throws IOException {
  super(file, new ProtoReadSupport(), recordFilter);
}
 

/**
 * @param readSupport Object which helps reads files of the given type, e.g. Thrift, Avro.
 * @param filter for filtering individual records
 * @deprecated will be removed in 2.0.0.
 */
@Deprecated
public ParquetRecordReader(ReadSupport<T> readSupport, UnboundRecordFilter filter) {
  this(readSupport, FilterCompat.get(filter));
}
 

/**
 * @param configuration a configuration
 * @return an unbound record filter class
 * @deprecated use {@link #getFilter(Configuration)}
 */
@Deprecated
public static Class<?> getUnboundRecordFilter(Configuration configuration) {
  return ConfigurationUtil.getClassFromConfig(configuration, UNBOUND_RECORD_FILTER, UnboundRecordFilter.class);
}
 

/**
 * @param conf the configuration
 * @param file the file to read
 * @param readSupport to materialize records
 * @param unboundRecordFilter the filter to use to filter records
 * @throws IOException if there is an error while reading
 * @deprecated use {@link #builder(ReadSupport, Path)}
 */
@Deprecated
public ParquetReader(Configuration conf, Path file, ReadSupport<T> readSupport, UnboundRecordFilter unboundRecordFilter) throws IOException {
  this(conf, file, readSupport, FilterCompat.get(unboundRecordFilter));
}
 

/**
 * @param file the file to read
 * @param readSupport to materialize records
 * @param unboundRecordFilter the filter to use to filter records
 * @throws IOException if there is an error while reading
 * @deprecated use {@link #builder(ReadSupport, Path)}
 */
@Deprecated
public ParquetReader(Path file, ReadSupport<T> readSupport, UnboundRecordFilter unboundRecordFilter) throws IOException {
  this(new Configuration(), file, readSupport, FilterCompat.get(unboundRecordFilter));
}
 

/**
 * @param readSupport Object which helps reads files of the given type, e.g. Thrift, Avro.
 * @param filter Optional filter for only returning matching records.
 * @deprecated use {@link #InternalParquetRecordReader(ReadSupport, Filter)}
 */
@Deprecated
public InternalParquetRecordReader(ReadSupport<T> readSupport, UnboundRecordFilter filter) {
  this(readSupport, FilterCompat.get(filter));
}
 

/**
 * Given an UnboundRecordFilter, return a Filter that wraps it.
 *
 * @param unboundRecordFilter an unbound record filter
 * @return a Filter for the given record filter (from the old API)
 */
public static Filter get(UnboundRecordFilter unboundRecordFilter) {
  return new UnboundRecordFilterCompat(unboundRecordFilter);
}
 

/**
 * @param conf the configuration
 * @param file the file to read
 * @param readSupport to materialize records
 * @param unboundRecordFilter the filter to use to filter records
 * @throws IOException
 * @deprecated use {@link #builder(ReadSupport, Path)}
 */
@Deprecated
public ParquetReader(Configuration conf, Path file, ReadSupport<T> readSupport, UnboundRecordFilter unboundRecordFilter) throws IOException {
  this(conf, file, readSupport, FilterCompat.get(unboundRecordFilter));
}
 

/**
 * @param file the file to read
 * @param readSupport to materialize records
 * @param unboundRecordFilter the filter to use to filter records
 * @throws IOException
 * @deprecated use {@link #builder(ReadSupport, Path)}
 */
@Deprecated
public ParquetReader(Path file, ReadSupport<T> readSupport, UnboundRecordFilter unboundRecordFilter) throws IOException {
  this(new Configuration(), file, readSupport, FilterCompat.get(unboundRecordFilter));
}