lanyuanxiaoyao/hudi
1
0
Fork 0
Code Activity

[HUDI-2161] Adding support to disable meta columns with bulk insert operation (#3247)

Browse Source
This commit is contained in:
Sivabalan Narayanan
2021-07-19 20:43:48 -04:00
committed by GitHub
parent 2099bf41db
commit d5026e9a24
53 changed files with 1063 additions and 269 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
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.common.table.HoodieTableConfig;
import org.apache.hudi.config.HoodieInternalConfig;
import org.apache.hudi.config.HoodieWriteConfig;
@@ -61,12 +62,14 @@ public class DefaultSource extends BaseDefaultSource implements DataSourceV2,
String instantTime = options.get(DataSourceInternalWriterHelper.INSTANT_TIME_OPT_KEY).get();
String path = options.get("path").get();
String tblName = options.get(HoodieWriteConfig.TABLE_NAME.key()).get();
boolean populateMetaFields = options.getBoolean(HoodieTableConfig.HOODIE_POPULATE_META_FIELDS.key(),
Boolean.parseBoolean(HoodieTableConfig.HOODIE_POPULATE_META_FIELDS.defaultValue()));
// 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(), options, arePartitionRecordsSorted));
getConfiguration(), options, populateMetaFields, arePartitionRecordsSorted));
}
}

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

@@ -36,10 +36,10 @@ public class HoodieBulkInsertDataInternalWriter implements DataWriter<InternalRo
private final BulkInsertDataInternalWriterHelper bulkInsertWriterHelper;
public HoodieBulkInsertDataInternalWriter(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, int taskPartitionId, long taskId, long taskEpochId,
StructType structType, boolean arePartitionRecordsSorted) {
String instantTime, int taskPartitionId, long taskId, long taskEpochId,
StructType structType, boolean populateMetaFields, boolean arePartitionRecordsSorted) {
this.bulkInsertWriterHelper = new BulkInsertDataInternalWriterHelper(hoodieTable,
writeConfig, instantTime, taskPartitionId, taskId, taskEpochId, structType, arePartitionRecordsSorted);
writeConfig, instantTime, taskPartitionId, taskId, taskEpochId, structType, populateMetaFields, arePartitionRecordsSorted);
}
@Override

7
hudi-spark-datasource/hudi-spark2/src/main/java/org/apache/hudi/internal/HoodieBulkInsertDataInternalWriterFactory.java
View File

@@ -36,19 +36,22 @@ public class HoodieBulkInsertDataInternalWriterFactory implements DataWriterFact
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final boolean arePartitionRecordsSorted;
private final boolean populateMetaFields;
public HoodieBulkInsertDataInternalWriterFactory(HoodieTable hoodieTable, HoodieWriteConfig writeConfig,
String instantTime, StructType structType, boolean arePartitionRecordsSorted) {
String instantTime, StructType structType, boolean populateMetaFields,
boolean arePartitionRecordsSorted) {
this.hoodieTable = hoodieTable;
this.writeConfig = writeConfig;
this.instantTime = instantTime;
this.structType = structType;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
}
@Override
public DataWriter<InternalRow> createDataWriter(int partitionId, long taskId, long epochId) {
return new HoodieBulkInsertDataInternalWriter(hoodieTable, writeConfig, instantTime, partitionId, taskId, epochId,
structType, arePartitionRecordsSorted);
structType, populateMetaFields, arePartitionRecordsSorted);
}
}

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

@@ -49,15 +49,17 @@ public class HoodieDataSourceInternalWriter implements DataSourceWriter {
private final HoodieWriteConfig writeConfig;
private final StructType structType;
private final DataSourceInternalWriterHelper dataSourceInternalWriterHelper;
private final boolean populateMetaFields;
private final Boolean arePartitionRecordsSorted;
private Map<String, String> extraMetadataMap = new HashMap<>();
public HoodieDataSourceInternalWriter(String instantTime, HoodieWriteConfig writeConfig, StructType structType,
SparkSession sparkSession, Configuration configuration, DataSourceOptions dataSourceOptions,
boolean arePartitionRecordsSorted) {
boolean populateMetaFields, boolean arePartitionRecordsSorted) {
this.instantTime = instantTime;
this.writeConfig = writeConfig;
this.structType = structType;
this.populateMetaFields = populateMetaFields;
this.arePartitionRecordsSorted = arePartitionRecordsSorted;
this.extraMetadataMap = DataSourceUtils.getExtraMetadata(dataSourceOptions.asMap());
this.dataSourceInternalWriterHelper = new DataSourceInternalWriterHelper(instantTime, writeConfig, structType,
@@ -69,7 +71,7 @@ public class HoodieDataSourceInternalWriter implements DataSourceWriter {
dataSourceInternalWriterHelper.createInflightCommit();
if (WriteOperationType.BULK_INSERT == dataSourceInternalWriterHelper.getWriteOperationType()) {
return new HoodieBulkInsertDataInternalWriterFactory(dataSourceInternalWriterHelper.getHoodieTable(),
writeConfig, instantTime, structType, arePartitionRecordsSorted);
writeConfig, instantTime, structType, populateMetaFields, arePartitionRecordsSorted);
} else {
throw new IllegalArgumentException("Write Operation Type + " + dataSourceInternalWriterHelper.getWriteOperationType() + " not supported ");
}