1
0

Refactoring HoodieTableFileSystemView using FileGroups/FileSlices

- Merged all filter* and get* methods
 - new constructor takes filestatus[]
 - All existing tests pass
 - FileGroup is all files that belong to a fileID within a partition
 - FileSlice is a generation of data and log files, starting at a base commit
This commit is contained in:
Vinoth Chandar
2017-06-19 00:34:37 -07:00
committed by prazanna
parent 23e7badd8a
commit c00f1a9ed9
42 changed files with 810 additions and 408 deletions

View File

@@ -22,10 +22,10 @@ import com.uber.hoodie.common.model.HoodieDeltaWriteStat;
import com.uber.hoodie.common.model.HoodieRecord;
import com.uber.hoodie.common.model.HoodieRecordLocation;
import com.uber.hoodie.common.model.HoodieRecordPayload;
import com.uber.hoodie.common.table.log.HoodieLogFile;
import com.uber.hoodie.common.table.log.HoodieLogFormat;
import com.uber.hoodie.common.table.log.HoodieLogFormat.Writer;
import com.uber.hoodie.common.table.log.block.HoodieAvroDataBlock;
import com.uber.hoodie.common.model.HoodieLogFile;
import com.uber.hoodie.common.util.FSUtils;
import com.uber.hoodie.common.util.HoodieAvroUtils;
import com.uber.hoodie.config.HoodieWriteConfig;
@@ -86,8 +86,9 @@ public class HoodieAppendHandle<T extends HoodieRecordPayload> extends HoodieIOH
partitionPath = record.getPartitionPath();
// HACK(vc) This also assumes a base file. It will break, if appending without one.
String latestValidFilePath =
fileSystemView.getLatestDataFilesForFileId(record.getPartitionPath(), fileId)
.findFirst().get().getFileName();
fileSystemView.getLatestDataFiles(record.getPartitionPath())
.filter(dataFile -> dataFile.getFileId().equals(fileId))
.findFirst().get().getFileName();
String baseCommitTime = FSUtils.getCommitTime(latestValidFilePath);
writeStatus.getStat().setPrevCommit(baseCommitTime);
writeStatus.setFileId(fileId);