use of org.apache.gobblin.source.extractor.hadoop.AvroFsHelper in project incubator-gobblin by apache.
the class PartitionedFileSourceBase method init.
/**
* Gobblin calls the {@link Source#getWorkunits(SourceState)} method after creating a {@link Source} object with a
* blank constructor, so any custom initialization of the object needs to be done here.
*/
protected void init(SourceState state) {
retriever.init(state);
try {
initFileSystemHelper(state);
} catch (FileBasedHelperException e) {
Throwables.propagate(e);
}
AvroFsHelper fsHelper = (AvroFsHelper) this.fsHelper;
this.fs = fsHelper.getFileSystem();
this.sourceState = state;
this.lowWaterMark = getLowWaterMark(state.getPreviousWorkUnitStates(), state.getProp(DATE_PARTITIONED_SOURCE_MIN_WATERMARK_VALUE, String.valueOf(DEFAULT_DATE_PARTITIONED_SOURCE_MIN_WATERMARK_VALUE)));
this.maxFilesPerJob = state.getPropAsInt(DATE_PARTITIONED_SOURCE_MAX_FILES_PER_JOB, DEFAULT_DATE_PARTITIONED_SOURCE_MAX_FILES_PER_JOB);
this.maxWorkUnitsPerJob = state.getPropAsInt(DATE_PARTITIONED_SOURCE_MAX_WORKUNITS_PER_JOB, DEFAULT_DATE_PARTITIONED_SOURCE_MAX_WORKUNITS_PER_JOB);
this.tableType = TableType.valueOf(state.getProp(ConfigurationKeys.EXTRACT_TABLE_TYPE_KEY).toUpperCase());
this.fileCount = 0;
this.sourceDir = new Path(state.getProp(ConfigurationKeys.SOURCE_FILEBASED_DATA_DIRECTORY));
}
use of org.apache.gobblin.source.extractor.hadoop.AvroFsHelper in project incubator-gobblin by apache.
the class PartitionedFileSourceBase method initFileSystemHelper.
@Override
public void initFileSystemHelper(State state) throws FileBasedHelperException {
this.fsHelper = new AvroFsHelper(state);
this.fsHelper.connect();
}
Aggregations