org.apache.spark.sql.sources.v2.writer.DataWriter Java Examples

The following examples show how to use org.apache.spark.sql.sources.v2.writer.DataWriter. 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: ParallelRowReadWriteDataSource.java    From spark-data-sources with MIT License 5 votes vote down vote up
@Override
public DataWriter<Row> createDataWriter(int partitionId, int attemptNumber)
{
    try {
        log.info("creating a data writer for table " + _table +
                " partition " + partitionId + " attempt " + attemptNumber);
        return new TaskDataWriter(_host, _port, _table, _schema, partitionId, attemptNumber);
    } catch (UnknownTableException ute) {
        throw new RuntimeException(ute);
    }
}
 
Example #2
Source File: Writer.java    From iceberg with Apache License 2.0 5 votes vote down vote up
@Override
public DataWriter<InternalRow> createDataWriter(int partitionId, long taskId, long epochId) {
  OutputFileFactory fileFactory = new OutputFileFactory(
      spec, format, locations, io.value(), encryptionManager.value(), partitionId, taskId);
  SparkAppenderFactory appenderFactory = new SparkAppenderFactory(properties, writeSchema, dsSchema);

  if (spec.fields().isEmpty()) {
    return new Unpartitioned24Writer(spec, format, appenderFactory, fileFactory, io.value(), targetFileSize);
  } else {
    return new Partitioned24Writer(
        spec, format, appenderFactory, fileFactory, io.value(), targetFileSize, writeSchema);
  }
}
 
Example #3
Source File: Writer.java    From iceberg with Apache License 2.0 5 votes vote down vote up
@Override
public DataWriter<InternalRow> createDataWriter(int partitionId, int attemptNumber) {
  String filename = format.addExtension(String.format("%05d-%d-%s",
      partitionId, attemptNumber, uuid));
  AppenderFactory<InternalRow> factory = new SparkAppenderFactory();
  if (spec.fields().isEmpty()) {
    return new UnpartitionedWriter(lazyDataPath(), filename, format, conf.value(), factory);

  } else {
    Path baseDataPath = lazyDataPath(); // avoid calling this in the output path function
    Function<PartitionKey, Path> outputPathFunc = key ->
        new Path(new Path(baseDataPath, key.toPath()), filename);

    boolean useObjectStorage = (
        Boolean.parseBoolean(properties.get(OBJECT_STORE_ENABLED)) ||
        OBJECT_STORE_ENABLED_DEFAULT
    );

    if (useObjectStorage) {
      // try to get db and table portions of the path for context in the object store
      String context = pathContext(baseDataPath);
      String objectStore = properties.get(OBJECT_STORE_PATH);
      Preconditions.checkNotNull(objectStore,
          "Cannot use object storage, missing location: " + OBJECT_STORE_PATH);
      Path objectStorePath = new Path(objectStore);

      outputPathFunc = key -> {
        String partitionAndFilename = key.toPath() + "/" + filename;
        int hash = HASH_FUNC.apply(partitionAndFilename);
        return new Path(objectStorePath,
            String.format("%08x/%s/%s", hash, context, partitionAndFilename));
      };
    }

    return new PartitionedWriter(spec, format, conf.value(), factory, outputPathFunc);
  }
}
 
Example #4
Source File: HiveWarehouseDataWriterFactory.java    From spark-llap with Apache License 2.0 5 votes vote down vote up
@Override public DataWriter<InternalRow> createDataWriter(int partitionId, int attemptNumber) {
  Path filePath = new Path(this.path, String.format("%s_%s_%s", jobId, partitionId, attemptNumber));
  FileSystem fs = null;
  try {
    fs = filePath.getFileSystem(conf.get());
  } catch (IOException e) {
    e.printStackTrace();
  }
  return getDataWriter(conf.get(), jobId, schema, partitionId, attemptNumber,
      fs, filePath);
}
 
Example #5
Source File: HiveStreamingDataWriterFactory.java    From spark-llap with Apache License 2.0 5 votes vote down vote up
@Override
public DataWriter<InternalRow> createDataWriter(int partitionId, int attemptNumber) {
  ClassLoader restoredClassloader = Thread.currentThread().getContextClassLoader();
  ClassLoader isolatedClassloader = HiveIsolatedClassLoader.isolatedClassLoader();
  try {
    Thread.currentThread().setContextClassLoader(isolatedClassloader);
    return new HiveStreamingDataWriter(jobId, schema, commitIntervalRows, partitionId, attemptNumber, db,
      table, partition, metastoreUri, metastoreKrbPrincipal);
  } finally {
    Thread.currentThread().setContextClassLoader(restoredClassloader);
  }
}
 
Example #6
Source File: HiveWarehouseDataWriterFactory.java    From spark-llap with Apache License 2.0 4 votes vote down vote up
protected DataWriter<InternalRow> getDataWriter(Configuration conf, String jobId,
    StructType schema, int partitionId, int attemptNumber,
    FileSystem fs, Path filePath) {
  return new HiveWarehouseDataWriter(conf, jobId, schema, partitionId, attemptNumber, fs, filePath);
}
 
Example #7
Source File: MockWriteSupport.java    From spark-llap with Apache License 2.0 4 votes vote down vote up
protected DataWriter<InternalRow> getDataWriter(Configuration conf, String jobId,
    StructType schema, int partitionId, int attemptNumber,
    FileSystem fs, Path filePath) {
  return new MockHiveWarehouseDataWriter(conf, jobId, schema, partitionId, attemptNumber, fs, filePath);
}