pvary commented on a change in pull request #1407:
URL: https://github.com/apache/iceberg/pull/1407#discussion_r486489902
##########
File path:
mr/src/main/java/org/apache/iceberg/mr/hive/HiveIcebergOutputFormat.java
##########
@@ -19,21 +19,442 @@
package org.apache.iceberg.mr.hive;
+import java.io.IOException;
+import java.io.ObjectInputStream;
+import java.io.ObjectOutputStream;
+import java.io.Serializable;
+import java.nio.ByteBuffer;
+import java.util.HashMap;
+import java.util.HashSet;
+import java.util.Map;
+import java.util.Properties;
+import java.util.Set;
+import java.util.concurrent.ConcurrentHashMap;
+import java.util.concurrent.ExecutorService;
+import java.util.concurrent.Executors;
+import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FileSystem;
+import org.apache.hadoop.fs.Path;
+import org.apache.hadoop.hive.conf.HiveConf;
+import org.apache.hadoop.hive.ql.exec.FileSinkOperator;
+import org.apache.hadoop.hive.ql.io.HiveOutputFormat;
+import org.apache.hadoop.io.NullWritable;
+import org.apache.hadoop.io.Writable;
import org.apache.hadoop.mapred.JobConf;
+import org.apache.hadoop.mapred.JobContext;
+import org.apache.hadoop.mapred.OutputCommitter;
import org.apache.hadoop.mapred.OutputFormat;
-import org.apache.hadoop.mapred.RecordWriter;
+import org.apache.hadoop.mapred.Reporter;
+import org.apache.hadoop.mapred.TaskAttemptContext;
+import org.apache.hadoop.mapred.TaskAttemptID;
+import org.apache.hadoop.mapreduce.JobID;
+import org.apache.hadoop.mapreduce.TaskType;
import org.apache.hadoop.util.Progressable;
+import org.apache.iceberg.AppendFiles;
+import org.apache.iceberg.DataFile;
+import org.apache.iceberg.DataFiles;
+import org.apache.iceberg.FileFormat;
+import org.apache.iceberg.Metrics;
+import org.apache.iceberg.PartitionSpec;
+import org.apache.iceberg.Schema;
+import org.apache.iceberg.SchemaParser;
+import org.apache.iceberg.Table;
+import org.apache.iceberg.avro.Avro;
+import org.apache.iceberg.data.Record;
+import org.apache.iceberg.data.avro.DataWriter;
+import org.apache.iceberg.data.orc.GenericOrcWriter;
+import org.apache.iceberg.data.parquet.GenericParquetWriter;
+import org.apache.iceberg.exceptions.NotFoundException;
+import org.apache.iceberg.hadoop.HadoopFileIO;
+import org.apache.iceberg.io.FileAppender;
+import org.apache.iceberg.io.FileIO;
+import org.apache.iceberg.io.OutputFile;
+import org.apache.iceberg.mr.Catalogs;
+import org.apache.iceberg.mr.InputFormatConfig;
+import org.apache.iceberg.mr.mapreduce.IcebergWritable;
+import org.apache.iceberg.orc.ORC;
+import org.apache.iceberg.parquet.Parquet;
+import org.apache.iceberg.relocated.com.google.common.base.Preconditions;
+import
org.apache.iceberg.relocated.com.google.common.util.concurrent.ThreadFactoryBuilder;
+import org.apache.iceberg.util.Tasks;
+import org.slf4j.Logger;
+import org.slf4j.LoggerFactory;
-public class HiveIcebergOutputFormat<T> implements OutputFormat<Void, T> {
+public class HiveIcebergOutputFormat implements OutputFormat<NullWritable,
IcebergWritable>,
+ HiveOutputFormat<NullWritable, IcebergWritable> {
+
+ private static final Logger LOG =
LoggerFactory.getLogger(HiveIcebergOutputFormat.class);
+ private static final String TASK_ATTEMPT_ID_KEY = "mapred.task.id";
+ private static final String COMMITTED_EXTENSION = ".committed";
+
+ // <TaskAttemptId, ClosedFileData> map to store the data needed to create
DataFiles
+ // Stored in concurrent map, since some executor engines can share containers
+ private static final Map<TaskAttemptID, ClosedFileData> fileData = new
ConcurrentHashMap<>();
+
+ private Configuration overlayedConf = null;
+ private TaskAttemptID taskAttemptId = null;
+ private Schema schema = null;
+ private FileFormat fileFormat = null;
@Override
- public RecordWriter<Void, T> getRecordWriter(FileSystem ignored, JobConf
job, String name, Progressable progress) {
- throw new UnsupportedOperationException("Writing to an Iceberg table with
Hive is not supported");
+ @SuppressWarnings("rawtypes")
+ public FileSinkOperator.RecordWriter getHiveRecordWriter(JobConf jc, Path
finalOutPath, Class valueClass,
+ boolean isCompressed, Properties tableAndSerDeProperties, Progressable
progress) throws IOException {
+
+ this.overlayedConf = createOverlayedConf(jc, tableAndSerDeProperties);
+ this.taskAttemptId =
TaskAttemptID.forName(overlayedConf.get(TASK_ATTEMPT_ID_KEY));
+ this.schema =
SchemaParser.fromJson(overlayedConf.get(InputFormatConfig.TABLE_SCHEMA));
+ this.fileFormat =
FileFormat.valueOf(overlayedConf.get(InputFormatConfig.WRITE_FILE_FORMAT).toUpperCase());
+
+ fileData.remove(this.taskAttemptId);
+
+ return new IcebergRecordWriter(generateDataFileLocation(overlayedConf,
taskAttemptId));
+ }
+
+ /**
+ * Returns the union of the configuration and table properties with the
+ * table properties taking precedence.
+ */
+ private static Configuration createOverlayedConf(Configuration conf,
Properties tblProps) {
+ Configuration newConf = new Configuration(conf);
+ for (Map.Entry<Object, Object> prop : tblProps.entrySet()) {
+ newConf.set((String) prop.getKey(), (String) prop.getValue());
+ }
+ return newConf;
+ }
+
+ /**
+ * Generates query directory location based on the configuration.
+ * Currently it uses tableLocation/queryId
+ * @param conf The job's configuration
+ * @return The directory to store the query result files
+ */
+ private static String generateQueryLocation(Configuration conf) {
+ String tableLocation = conf.get(InputFormatConfig.TABLE_LOCATION);
+ String queryId = conf.get(HiveConf.ConfVars.HIVEQUERYID.varname);
+ return tableLocation + "/" + queryId;
+ }
+
+ /**
+ * Generates datafile location based on the task configuration.
+ * Currently it uses QUERY_LOCATION/jobId/taskAttemptId.
+ * @param conf The job's configuration
+ * @param taskAttemptId The TaskAttemptID for the task
+ * @return The file to store the results
+ */
+ private static String generateDataFileLocation(Configuration conf,
TaskAttemptID taskAttemptId) {
+ return generateQueryLocation(conf) + "/" + taskAttemptId.getJobID() + "/"
+ taskAttemptId.toString();
+ }
+
+ /**
+ * Generates commit file location based on the task configuration and a
specific task id.
+ * Currently it uses QUERY_LOCATION/jobId/task-[0..numTasks].committed.
+ * @param conf The job's configuration
+ * @param jobId The jobId for the task
+ * @param taskId The taskId for the commit file
+ * @return The file to store the results
+ */
+ private static String generateCommitFileLocation(Configuration conf, JobID
jobId, int taskId) {
+ return generateQueryLocation(conf) + "/" + jobId + "/task-" + taskId +
COMMITTED_EXTENSION;
+ }
+
+ /**
+ * Generates commit file location based on the task configuration.
+ * Currently it uses QUERY_LOCATION/jobId/task-[0..numTasks].committed.
+ * @param conf The job's configuration
+ * @param taskAttemptId The TaskAttemptID for the task
+ * @return The file to store the results
+ */
+ private static String generateCommitFileLocation(Configuration conf,
TaskAttemptID taskAttemptId) {
+ return generateCommitFileLocation(conf, taskAttemptId.getJobID(),
taskAttemptId.getTaskID().getId());
+ }
+
+ @Override
+ public org.apache.hadoop.mapred.RecordWriter<NullWritable, IcebergWritable>
getRecordWriter(FileSystem ignored,
+ JobConf job, String name, Progressable progress) {
+
+ throw new UnsupportedOperationException("Please implement if needed");
}
@Override
public void checkOutputSpecs(FileSystem ignored, JobConf job) {
- throw new UnsupportedOperationException("Writing to an Iceberg table with
Hive is not supported");
+ // Not doing any check.
+ }
+
+ protected class IcebergRecordWriter extends
org.apache.hadoop.mapreduce.RecordWriter<NullWritable, IcebergWritable>
+ implements FileSinkOperator.RecordWriter,
org.apache.hadoop.mapred.RecordWriter<NullWritable, IcebergWritable> {
+
+ private final String location;
+ private final FileAppender<Record> appender;
+ private final FileIO io;
+
+ IcebergRecordWriter(String location) throws IOException {
+ this.location = location;
+ io = new HadoopFileIO(overlayedConf);
+ OutputFile dataFile = io.newOutputFile(location);
+
+ switch (fileFormat) {
+ case AVRO:
+ this.appender = Avro.write(dataFile)
+ .schema(schema)
+ .createWriterFunc(DataWriter::create)
+ .named(fileFormat.name())
+ .build();
+ break;
+
+ case PARQUET:
+ this.appender = Parquet.write(dataFile)
+ .schema(schema)
+ .createWriterFunc(GenericParquetWriter::buildWriter)
+ .named(fileFormat.name())
+ .build();
+ break;
+
+ case ORC:
+ this.appender = ORC.write(dataFile)
+ .schema(schema)
+ .createWriterFunc(GenericOrcWriter::buildWriter)
+ .build();
+ break;
+
+ default:
+ throw new UnsupportedOperationException("Cannot write format: " +
fileFormat);
+ }
+ LOG.info("IcebergRecordWriter is created in {} with {}", location,
fileFormat);
+ }
+
+ @Override
+ public void write(Writable row) {
+ Preconditions.checkArgument(row instanceof IcebergWritable);
+
+ // TODO partition handling
+ appender.add(((IcebergWritable) row).record());
+ }
+
+ @Override
+ public void write(NullWritable key, IcebergWritable value) {
+ write(value);
+ }
+
+ @Override
+ public void close(boolean abort) throws IOException {
+ appender.close();
+ if (!abort) {
+ fileData.put(taskAttemptId, new ClosedFileData(location, fileFormat,
appender.length(), appender.metrics()));
Review comment:
Thanks for the confirmation @rdblue!
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
