lanyuanxiaoyao/hudi
1
0
Fork 0
Code Activity

[HUDI-1104] Adding support for UserDefinedPartitioners and SortModes to BulkInsert with Rows (#3149)

Browse Source
This commit is contained in:
Sivabalan Narayanan
2021-07-07 11:15:25 -04:00
committed by GitHub
parent 55ecbc662e
commit ea9e5d0e8b
31 changed files with 618 additions and 82 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
hudi-spark-datasource/hudi-spark2/src/main/java/org/apache/hudi/internal/DefaultSource.java
View File

@@ -19,6 +19,7 @@
package org.apache.hudi.internal;
import org.apache.hudi.DataSourceUtils;
import org.apache.hudi.config.HoodieInternalConfig;
import org.apache.hudi.config.HoodieWriteConfig;
import org.apache.spark.sql.SaveMode;
@@ -62,7 +63,10 @@ public class DefaultSource extends BaseDefaultSource implements DataSourceV2,
String tblName = options.get(HoodieWriteConfig.TABLE_NAME.key()).get();
// 1st arg to createHooodieConfig is not really reuqired to be set. but passing it anyways.
HoodieWriteConfig config = DataSourceUtils.createHoodieConfig(options.get(HoodieWriteConfig.AVRO_SCHEMA.key()).get(), path, tblName, options.asMap());
boolean arePartitionRecordsSorted = HoodieInternalConfig.getBulkInsertIsPartitionRecordsSorted(
options.get(HoodieInternalConfig.BULKINSERT_ARE_PARTITIONER_RECORDS_SORTED).isPresent()
? options.get(HoodieInternalConfig.BULKINSERT_ARE_PARTITIONER_RECORDS_SORTED).get() : null);
return Optional.of(new HoodieDataSourceInternalWriter(instantTime, config, schema, getSparkSession(),
getConfiguration()));
getConfiguration(), arePartitionRecordsSorted));
}
}

4
hudi-spark-datasource/hudi-spark2/src/main/java/org/apache/hudi/internal/HoodieBulkInsertDataInternalWriter.java
View File

@@ -37,9 +37,9 @@ public class HoodieBulkInsertDataInternalWriter implements DataWriter<InternalRo
public HoodieBulkInsertDataInternalWriter(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, int taskPartitionId, long taskId, long taskEpochId,
StructType structType) {
StructType structType, boolean arePartitionRecordsSorted) {
this.bulkInsertWriterHelper = new BulkInsertDataInternalWriterHelper(hoodieTable,
writeConfig, instantTime, taskPartitionId, taskId, taskEpochId, structType);
writeConfig, instantTime, taskPartitionId, taskId, taskEpochId, structType, arePartitionRecordsSorted);
}
@Override

6
hudi-spark-datasource/hudi-spark2/src/main/java/org/apache/hudi/internal/HoodieBulkInsertDataInternalWriterFactory.java
View File

@@ -35,18 +35,20 @@ public class HoodieBulkInsertDataInternalWriterFactory implements DataWriterFact
private final HoodieTable hoodieTable;
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final boolean arePartitionRecordsSorted;
public HoodieBulkInsertDataInternalWriterFactory(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, StructType structType) {
String instantTime, StructType structType, boolean arePartitionRecordsSorted) {
this.hoodieTable = hoodieTable;
this.writeConfig = writeConfig;
this.instantTime = instantTime;
this.structType = structType;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
}
@Override
public DataWriter<InternalRow> createDataWriter(int partitionId, long taskId, long epochId) {
return new HoodieBulkInsertDataInternalWriter(hoodieTable, writeConfig, instantTime, partitionId, taskId, epochId,
structType);
structType, arePartitionRecordsSorted);
}
}

6
hudi-spark-datasource/hudi-spark2/src/main/java/org/apache/hudi/internal/HoodieDataSourceInternalWriter.java
View File

@@ -45,12 +45,14 @@ public class HoodieDataSourceInternalWriter implements DataSourceWriter {
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final DataSourceInternalWriterHelper dataSourceInternalWriterHelper;
private final Boolean arePartitionRecordsSorted;
public HoodieDataSourceInternalWriter(String instantTime, HoodieWriteConfig writeConfig, StructType structType,
SparkSession sparkSession, Configuration configuration) {
SparkSession sparkSession, Configuration configuration, boolean arePartitionRecordsSorted) {
this.instantTime = instantTime;
this.writeConfig = writeConfig;
this.structType = structType;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
this.dataSourceInternalWriterHelper = new DataSourceInternalWriterHelper(instantTime, writeConfig, structType,
sparkSession, configuration);
}
@@ -60,7 +62,7 @@ public class HoodieDataSourceInternalWriter implements DataSourceWriter {
dataSourceInternalWriterHelper.createInflightCommit();
if (WriteOperationType.BULK_INSERT == dataSourceInternalWriterHelper.getWriteOperationType()) {
return new HoodieBulkInsertDataInternalWriterFactory(dataSourceInternalWriterHelper.getHoodieTable(),
writeConfig, instantTime, structType);
writeConfig, instantTime, structType, arePartitionRecordsSorted);
} else {
throw new IllegalArgumentException("Write Operation Type + " + dataSourceInternalWriterHelper.getWriteOperationType() + " not supported ");
}