[Pick-2183][AMORO-2168] Support hive commit protocol in mixed-hive fo…

…rmat. (apache#2220)

* [Pick-2183][AMORO-2168] Support hive commit protocol in mixed-hive format. (apache#2183)

pick from master branch and PR-2183
---------

Co-authored-by: ZhouJinsong <[email protected]>

* consistent-write.enable docs

* FileNameGenerator pattern.

* move to HiveTableProperties

* fix checkstyle

---------

Co-authored-by: ZhouJinsong <[email protected]>

core/src/main/java/com/netease/arctic/data/file/FileNameGenerator.java

@@ -52,7 +52,7 @@
  */
 public class FileNameGenerator {
 
-  private static final String KEYED_FILE_NAME_PATTERN_STRING = "(\\d+)-(\\w+)-(\\d+)-(\\d+)-(\\d+)-.*";
+  private static final String KEYED_FILE_NAME_PATTERN_STRING = "\\.?(\\d+)-(\\w+)-(\\d+)-(\\d+)-(\\d+)-.*";
   private static final Pattern KEYED_FILE_NAME_PATTERN = Pattern.compile(KEYED_FILE_NAME_PATTERN_STRING);
 
   private static final String FORMAT = "%d-%s-%d-%05d-%d-%s-%05d";

core/src/main/java/com/netease/arctic/table/TableProperties.java

@@ -298,7 +298,7 @@ private TableProperties() {
 
   public static final String LOG_STORE_DATA_VERSION = "log-store.data-version";
   public static final String LOG_STORE_DATA_VERSION_DEFAULT = "v1";
-  
+
   public static final String LOG_STORE_PROPERTIES_PREFIX = "properties.";
 
   public static final String OWNER = "owner";

core/src/test/java/com/netease/arctic/io/DataTestHelpers.java

@@ -45,6 +45,7 @@
 import org.apache.iceberg.io.CloseableIterable;
 import org.apache.iceberg.io.CloseableIterator;
 import org.apache.iceberg.io.InputFile;
+import org.apache.iceberg.io.TaskWriter;
 import org.apache.iceberg.io.WriteResult;
 import org.apache.iceberg.orc.ORC;
 import org.apache.iceberg.parquet.Parquet;
@@ -122,6 +123,24 @@ public static List<DataFile> writeBaseStore(KeyedTable keyedTable, long txId, Li
     }
   }
 
+  public static List<DataFile> writeRecords(TaskWriter<Record> taskWriter, List<Record> records) {
+    try {
+      records.forEach(
+          d -> {
+            try {
+              taskWriter.write(d);
+            } catch (IOException e) {
+              throw new UncheckedIOException(e);
+            }
+          });
+
+      WriteResult result = taskWriter.complete();
+      return Lists.newArrayList(Arrays.asList(result.dataFiles()));
+    } catch (IOException e) {
+      throw new UncheckedIOException(e);
+    }
+  }
+
   public static List<DataFile> writeAndCommitChangeStore(
       KeyedTable keyedTable, long txId, ChangeAction action,
       List<Record> records) {

flink/v1.12/flink/src/main/java/com/netease/arctic/flink/write/FlinkTaskWriterBuilder.java

@@ -18,6 +18,7 @@
 
 package com.netease.arctic.flink.write;
 
+import com.netease.arctic.hive.HiveTableProperties;
 import com.netease.arctic.hive.io.writer.AdaptHiveOperateToTableRelation;
 import com.netease.arctic.hive.io.writer.AdaptHiveOutputFileFactory;
 import com.netease.arctic.hive.table.HiveLocationKind;
@@ -137,10 +138,14 @@ private FlinkBaseTaskWriter buildBaseWriter(LocationKind locationKind) {
 
     Schema selectSchema = TypeUtil.reassignIds(
         FlinkSchemaUtil.convert(FlinkSchemaUtil.toSchema(flinkSchema)), schema);
+    boolean hiveConsistentWriteEnabled = PropertyUtil.propertyAsBoolean(
+        table.properties(),
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED,
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED_DEFAULT);
 
     OutputFileFactory outputFileFactory = locationKind == HiveLocationKind.INSTANT ?
         new AdaptHiveOutputFileFactory(((SupportHive) table).hiveLocation(), table.spec(), fileFormat, table.io(),
-            encryptionManager, partitionId, taskId, transactionId) :
+            encryptionManager, partitionId, taskId, transactionId, hiveConsistentWriteEnabled) :
         new CommonOutputFileFactory(baseLocation, table.spec(), fileFormat, table.io(),
             encryptionManager, partitionId, taskId, transactionId);
     FileAppenderFactory<RowData> appenderFactory = TableTypeUtil.isHive(table) ?

flink/v1.14/flink/src/main/java/com/netease/arctic/flink/write/FlinkTaskWriterBuilder.java

@@ -18,6 +18,7 @@
 
 package com.netease.arctic.flink.write;
 
+import com.netease.arctic.hive.HiveTableProperties;
 import com.netease.arctic.hive.io.writer.AdaptHiveOperateToTableRelation;
 import com.netease.arctic.hive.io.writer.AdaptHiveOutputFileFactory;
 import com.netease.arctic.hive.table.HiveLocationKind;
@@ -135,12 +136,17 @@ private FlinkBaseTaskWriter buildBaseWriter(LocationKind locationKind) {
       schema = table.schema();
     }
 
-    Schema selectSchema = TypeUtil.reassignIds(
-        FlinkSchemaUtil.convert(FlinkSchemaUtil.toSchema(flinkSchema)), schema);
+    Schema selectSchema =
+        TypeUtil.reassignIds(
+            FlinkSchemaUtil.convert(FlinkSchemaUtil.toSchema(flinkSchema)), schema);
+    boolean hiveConsistentWriteEnabled = PropertyUtil.propertyAsBoolean(
+        table.properties(),
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED,
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED_DEFAULT);
 
     OutputFileFactory outputFileFactory = locationKind == HiveLocationKind.INSTANT ?
         new AdaptHiveOutputFileFactory(((SupportHive) table).hiveLocation(), table.spec(), fileFormat, table.io(),
-            encryptionManager, partitionId, taskId, transactionId) :
+            encryptionManager, partitionId, taskId, transactionId, hiveConsistentWriteEnabled) :
         new CommonOutputFileFactory(baseLocation, table.spec(), fileFormat, table.io(),
             encryptionManager, partitionId, taskId, transactionId);
     FileAppenderFactory<RowData> appenderFactory = TableTypeUtil.isHive(table) ?

flink/v1.15/flink/src/main/java/com/netease/arctic/flink/write/FlinkTaskWriterBuilder.java

@@ -18,6 +18,7 @@
 
 package com.netease.arctic.flink.write;
 
+import com.netease.arctic.hive.HiveTableProperties;
 import com.netease.arctic.hive.io.writer.AdaptHiveOperateToTableRelation;
 import com.netease.arctic.hive.io.writer.AdaptHiveOutputFileFactory;
 import com.netease.arctic.hive.table.HiveLocationKind;
@@ -135,12 +136,17 @@ private FlinkBaseTaskWriter buildBaseWriter(LocationKind locationKind) {
       schema = table.schema();
     }
 
-    Schema selectSchema = TypeUtil.reassignIds(
-        FlinkSchemaUtil.convert(FlinkSchemaUtil.toSchema(flinkSchema)), schema);
+    Schema selectSchema =
+        TypeUtil.reassignIds(
+            FlinkSchemaUtil.convert(FlinkSchemaUtil.toSchema(flinkSchema)), schema);
+    boolean hiveConsistentWriteEnabled = PropertyUtil.propertyAsBoolean(
+        table.properties(),
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED,
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED_DEFAULT);
 
     OutputFileFactory outputFileFactory = locationKind == HiveLocationKind.INSTANT ?
         new AdaptHiveOutputFileFactory(((SupportHive) table).hiveLocation(), table.spec(), fileFormat, table.io(),
-            encryptionManager, partitionId, taskId, transactionId) :
+            encryptionManager, partitionId, taskId, transactionId, hiveConsistentWriteEnabled) :
         new CommonOutputFileFactory(baseLocation, table.spec(), fileFormat, table.io(),
             encryptionManager, partitionId, taskId, transactionId);
     FileAppenderFactory<RowData> appenderFactory = TableTypeUtil.isHive(table) ?

hive/src/main/java/com/netease/arctic/hive/HiveTableProperties.java

@@ -51,6 +51,11 @@ public class HiveTableProperties {
   public static final String AUTO_SYNC_HIVE_DATA_WRITE = "base.hive.auto-sync-data-write";
   public static final boolean AUTO_SYNC_HIVE_DATA_WRITE_DEFAULT = false;
 
+  /** enable consistent write for hive store */
+  public static final String HIVE_CONSISTENT_WRITE_ENABLED = "base.hive.consistent-write.enabled";
+
+  public static final boolean HIVE_CONSISTENT_WRITE_ENABLED_DEFAULT = true;
+
   public static final String ALLOW_HIVE_TABLE_EXISTED = "allow-hive-table-existed";
 
   public static final String WATERMARK_HIVE = "watermark.hive";

hive/src/main/java/com/netease/arctic/hive/io/writer/AdaptHiveGenericTaskWriterBuilder.java

@@ -19,6 +19,7 @@
 package com.netease.arctic.hive.io.writer;
 
 import com.netease.arctic.data.ChangeAction;
+import com.netease.arctic.hive.HiveTableProperties;
 import com.netease.arctic.hive.table.HiveLocationKind;
 import com.netease.arctic.hive.table.SupportHive;
 import com.netease.arctic.hive.utils.TableTypeUtil;
@@ -68,9 +69,14 @@ public class AdaptHiveGenericTaskWriterBuilder implements TaskWriterBuilder<Reco
   private String customHiveSubdirectory;
   private Long targetFileSize;
   private boolean orderedWriter = false;
+  private Boolean hiveConsistentWrite;
 
   private AdaptHiveGenericTaskWriterBuilder(ArcticTable table) {
     this.table = table;
+    this.hiveConsistentWrite = PropertyUtil.propertyAsBoolean(
+        table.properties(),
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED,
+        HiveTableProperties.HIVE_CONSISTENT_WRITE_ENABLED_DEFAULT);
   }
 
   public AdaptHiveGenericTaskWriterBuilder withTransactionId(Long transactionId) {
@@ -97,7 +103,7 @@ public AdaptHiveGenericTaskWriterBuilder withCustomHiveSubdirectory(String custo
     this.customHiveSubdirectory = customHiveSubdirectory;
     return this;
   }
-  
+
   public AdaptHiveGenericTaskWriterBuilder withTargetFileSize(long targetFileSize) {
     this.targetFileSize = targetFileSize;
     return this;
@@ -108,6 +114,11 @@ public AdaptHiveGenericTaskWriterBuilder withOrdered() {
     return this;
   }
 
+  public AdaptHiveGenericTaskWriterBuilder hiveConsistentWrite(boolean enabled) {
+    this.hiveConsistentWrite = enabled;
+    return this;
+  }
+
   @Override
   public TaskWriter<Record> buildWriter(WriteOperationKind writeOperationKind) {
     LocationKind locationKind = AdaptHiveOperateToTableRelation.INSTANT.getLocationKindsFromOperateKind(
@@ -179,17 +190,41 @@ private GenericBaseTaskWriter buildBaseWriter(LocationKind locationKind) {
       schema = table.schema();
     }
 
-    OutputFileFactory outputFileFactory = locationKind == HiveLocationKind.INSTANT ?
-        new AdaptHiveOutputFileFactory(((SupportHive) table).hiveLocation(), table.spec(), fileFormat,
-            table.io(), encryptionManager, partitionId, taskId, transactionId, customHiveSubdirectory) :
-        new CommonOutputFileFactory(baseLocation, table.spec(), fileFormat, table.io(),
-            encryptionManager, partitionId, taskId, transactionId);
-    FileAppenderFactory<Record> appenderFactory = TableTypeUtil.isHive(table) ?
-        new AdaptHiveGenericAppenderFactory(schema, table.spec()) :
-        new GenericAppenderFactory(schema, table.spec());
-    return new GenericBaseTaskWriter(fileFormat, appenderFactory,
+    OutputFileFactory outputFileFactory =
+        locationKind == HiveLocationKind.INSTANT ? new AdaptHiveOutputFileFactory(
+                ((SupportHive) table).hiveLocation(),
+                table.spec(),
+                fileFormat,
+                table.io(),
+                encryptionManager,
+                partitionId,
+                taskId,
+                transactionId,
+                customHiveSubdirectory,
+                hiveConsistentWrite)
+            : new CommonOutputFileFactory(
+                baseLocation,
+                table.spec(),
+                fileFormat,
+                table.io(),
+                encryptionManager,
+                partitionId,
+                taskId,
+                transactionId);
+    FileAppenderFactory<Record> appenderFactory =
+        TableTypeUtil.isHive(table) ? new AdaptHiveGenericAppenderFactory(schema, table.spec())
+            : new GenericAppenderFactory(schema, table.spec());
+    return new GenericBaseTaskWriter(
+        fileFormat,
+        appenderFactory,
         outputFileFactory,
-        table.io(), fileSizeBytes, mask, schema, table.spec(), primaryKeySpec, orderedWriter);
+        table.io(),
+        fileSizeBytes,
+        mask,
+        schema,
+        table.spec(),
+        primaryKeySpec,
+        orderedWriter);
   }
 
   private GenericChangeTaskWriter buildChangeWriter() {

hive/src/main/java/com/netease/arctic/hive/io/writer/AdaptHiveOutputFileFactory.java

@@ -50,36 +50,49 @@
  */
 public class AdaptHiveOutputFileFactory implements OutputFileFactory {
 
-  private final String baseLocation;
+  private final String hiveLocation;
   private final String hiveSubDirectory;
   private final PartitionSpec partitionSpec;
   private final ArcticFileIO io;
   private final EncryptionManager encryptionManager;
   private final FileNameGenerator fileNameGenerator;
+  private final boolean hiveConsistentWrite;
 
   public AdaptHiveOutputFileFactory(
-      String baseLocation,
+      String hiveLocation,
       PartitionSpec partitionSpec,
       FileFormat format,
       ArcticFileIO io,
       EncryptionManager encryptionManager,
       int partitionId,
       long taskId,
-      Long transactionId) {
-    this(baseLocation, partitionSpec, format, io, encryptionManager, partitionId, taskId, transactionId, null);
+      Long transactionId,
+      boolean hiveConsistentWrite) {
+    this(
+        hiveLocation,
+        partitionSpec,
+        format,
+        io,
+        encryptionManager,
+        partitionId,
+        taskId,
+        transactionId,
+        null,
+        hiveConsistentWrite);
   }
 
   public AdaptHiveOutputFileFactory(
-      String baseLocation,
+      String hiveLocation,
       PartitionSpec partitionSpec,
       FileFormat format,
       ArcticFileIO io,
       EncryptionManager encryptionManager,
       int partitionId,
       long taskId,
       Long transactionId,
-      String hiveSubDirectory) {
-    this.baseLocation = baseLocation;
+      String hiveSubDirectory,
+      boolean hiveConsistentWrite) {
+    this.hiveLocation = hiveLocation;
     this.partitionSpec = partitionSpec;
     this.io = io;
     this.encryptionManager = encryptionManager;
@@ -90,15 +103,23 @@ public AdaptHiveOutputFileFactory(
       this.hiveSubDirectory = hiveSubDirectory;
     }
     this.fileNameGenerator = new FileNameGenerator(format, partitionId, taskId, transactionId);
+    this.hiveConsistentWrite = hiveConsistentWrite;
   }
 
   private String generateFilename(TaskWriterKey key) {
-    return fileNameGenerator.fileName(key);
+    String filename = fileNameGenerator.fileName(key);
+    if (hiveConsistentWrite) {
+      filename = "." + filename;
+    }
+    return filename;
   }
 
   private String fileLocation(StructLike partitionData, String fileName) {
-    return String.format("%s/%s",
-        HiveTableUtil.newHiveDataLocation(baseLocation, partitionSpec, partitionData, hiveSubDirectory), fileName);
+    return String.format(
+        "%s/%s",
+        HiveTableUtil.newHiveDataLocation(
+            hiveLocation, partitionSpec, partitionData, hiveSubDirectory),
+        fileName);
   }
 
   public EncryptedOutputFile newOutputFile(TaskWriterKey key) {

hive/src/main/java/com/netease/arctic/hive/op/OverwriteHiveFiles.java

@@ -29,6 +29,7 @@
 import org.apache.iceberg.expressions.Expressions;
 import org.apache.iceberg.relocated.com.google.common.base.Preconditions;
 
+import java.util.List;
 import java.util.function.Consumer;
 
 import static com.netease.arctic.op.OverwriteBaseFiles.PROPERTIES_TRANSACTION_ID;
@@ -59,11 +60,10 @@ public OverwriteFiles overwriteByRowFilter(Expression expr) {
 
   @Override
   public OverwriteFiles addFile(DataFile file) {
-    delegate.addFile(file);
-    String hiveLocationRoot = table.hiveLocation();
-    String dataFileLocation = file.path().toString();
-    if (dataFileLocation.toLowerCase().contains(hiveLocationRoot.toLowerCase())) {
-      // only handle file in hive location
+    if (!isHiveDataFile(file)) {
+      delegate.addFile(file);
+    } else {
+      // handle file in hive location when commit
       this.addFiles.add(file);
     }
     return this;
@@ -72,9 +72,7 @@ public OverwriteFiles addFile(DataFile file) {
   @Override
   public OverwriteFiles deleteFile(DataFile file) {
     delegate.deleteFile(file);
-    String hiveLocation = table.hiveLocation();
-    String dataFileLocation = file.path().toString();
-    if (dataFileLocation.toLowerCase().contains(hiveLocation.toLowerCase())) {
+    if (isHiveDataFile(file)) {
       // only handle file in hive location
       this.deleteFiles.add(file);
     }
@@ -123,6 +121,11 @@ public OverwriteFiles validateNoConflictingDeletes() {
     return this;
   }
 
+  @Override
+  protected void postHiveDataCommitted(List<DataFile> committedDataFile) {
+    committedDataFile.forEach(delegate::addFile);
+  }
+
   @Override
   public OverwriteFiles set(String property, String value) {
     if (PROPERTIES_TRANSACTION_ID.equals(property)) {