Java Code Examples for org.apache.iceberg.PartitionSpec#identitySourceIds()

The following examples show how to use org.apache.iceberg.PartitionSpec#identitySourceIds() . You can vote up the ones you like or vote down the ones you don't like, and go to the original project or source file by following the links above each example. You may check out the related API usage on the sidebar.
Example 1
Source File: RowDataReader.java    From iceberg with Apache License 2.0 6 votes vote down vote up 
@Override
CloseableIterator<InternalRow> open(FileScanTask task) {
  DataFile file = task.file();

  // update the current file for Spark's filename() function
  InputFileBlockHolder.set(file.path().toString(), task.start(), task.length());

  // schema or rows returned by readers
  PartitionSpec spec = task.spec();
  Set<Integer> idColumns = spec.identitySourceIds();
  Schema partitionSchema = TypeUtil.select(expectedSchema, idColumns);
  boolean projectsIdentityPartitionColumns = !partitionSchema.columns().isEmpty();

  if (projectsIdentityPartitionColumns) {
    return open(task, expectedSchema, PartitionUtil.constantsMap(task, RowDataReader::convertConstant))
        .iterator();
  }
  // return the base iterator
  return open(task, expectedSchema, ImmutableMap.of()).iterator();
}
Example 2
Source File: IcebergInputFormat.java    From iceberg with Apache License 2.0 5 votes vote down vote up 
private Map<Integer, ?> constantsMap(FileScanTask task, BiFunction<Type, Object, Object> converter) {
  PartitionSpec spec = task.spec();
  Set<Integer> idColumns = spec.identitySourceIds();
  Schema partitionSchema = TypeUtil.select(expectedSchema, idColumns);
  boolean projectsIdentityPartitionColumns = !partitionSchema.columns().isEmpty();
  if (projectsIdentityPartitionColumns) {
    return PartitionUtil.constantsMap(task, converter);
  } else {
    return Collections.emptyMap();
  }
}
Example 3
Source File: IcebergPigInputFormat.java    From iceberg with Apache License 2.0 4 votes vote down vote up 
@SuppressWarnings("unchecked")
private boolean advance() throws IOException {
  if (reader != null) {
    reader.close();
  }

  if (!tasks.hasNext()) {
    return false;
  }

  FileScanTask currentTask = tasks.next();

  Schema tableSchema = (Schema) ObjectSerializer.deserialize(context.getConfiguration().get(scope(ICEBERG_SCHEMA)));
  LOG.debug("[{}]: Task table schema: {}", signature, tableSchema);

  List<String> projectedFields =
      (List<String>) ObjectSerializer.deserialize(context.getConfiguration().get(scope(ICEBERG_PROJECTED_FIELDS)));
  LOG.debug("[{}]: Task projected fields: {}", signature, projectedFields);

  Schema projectedSchema = projectedFields != null ? SchemaUtil.project(tableSchema, projectedFields) : tableSchema;

  PartitionSpec spec = currentTask.asFileScanTask().spec();
  DataFile file = currentTask.file();
  InputFile inputFile = HadoopInputFile.fromLocation(file.path(), context.getConfiguration());

  Set<Integer> idColumns = spec.identitySourceIds();

  // schema needed for the projection and filtering
  boolean hasJoinedPartitionColumns = !idColumns.isEmpty();

  switch (file.format()) {
    case PARQUET:
      Map<Integer, Object> partitionValueMap = Maps.newHashMap();

      if (hasJoinedPartitionColumns) {

        Schema readSchema = TypeUtil.selectNot(projectedSchema, idColumns);
        Schema projectedPartitionSchema = TypeUtil.select(projectedSchema, idColumns);

        Map<String, Integer> partitionSpecFieldIndexMap = Maps.newHashMap();
        for (int i = 0; i < spec.fields().size(); i++) {
          partitionSpecFieldIndexMap.put(spec.fields().get(i).name(), i);
        }

        for (Types.NestedField field : projectedPartitionSchema.columns()) {
          int partitionIndex = partitionSpecFieldIndexMap.get(field.name());

          Object partitionValue = file.partition().get(partitionIndex, Object.class);
          partitionValueMap.put(field.fieldId(), convertPartitionValue(field.type(), partitionValue));
        }

        reader = Parquet.read(inputFile)
            .project(readSchema)
            .split(currentTask.start(), currentTask.length())
            .filter(currentTask.residual())
            .createReaderFunc(
                fileSchema -> PigParquetReader.buildReader(fileSchema, projectedSchema, partitionValueMap))
            .build();
      } else {
        reader = Parquet.read(inputFile)
            .project(projectedSchema)
            .split(currentTask.start(), currentTask.length())
            .filter(currentTask.residual())
            .createReaderFunc(
                fileSchema -> PigParquetReader.buildReader(fileSchema, projectedSchema, partitionValueMap))
            .build();
      }

      recordIterator = reader.iterator();

      break;
    default:
      throw new UnsupportedOperationException("Unsupported file format: " + file.format());
  }

  return true;
}