vinishjail97 commented on code in PR #801: URL: https://github.com/apache/incubator-xtable/pull/801#discussion_r3331833922
########## xtable-core/src/test/java/org/apache/xtable/kernel/TestDeltaKernelDataFileUpdatesExtractor.java: ########## @@ -0,0 +1,385 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.xtable.kernel; + +import static org.junit.jupiter.api.Assertions.assertFalse; +import static org.junit.jupiter.api.Assertions.assertNotNull; +import static org.junit.jupiter.api.Assertions.assertTrue; + +import java.io.IOException; +import java.nio.charset.StandardCharsets; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import java.time.Instant; +import java.util.Arrays; +import java.util.Collections; +import java.util.List; + +import org.apache.hadoop.conf.Configuration; +import org.junit.jupiter.api.BeforeEach; +import org.junit.jupiter.api.Test; +import org.junit.jupiter.api.io.TempDir; + +import scala.collection.JavaConverters; + +import io.delta.kernel.Table; +import io.delta.kernel.defaults.engine.DefaultEngine; +import io.delta.kernel.engine.Engine; +import io.delta.kernel.internal.actions.AddFile; +import io.delta.kernel.internal.actions.RemoveFile; +import io.delta.kernel.internal.actions.RowBackedAction; +import io.delta.kernel.types.IntegerType; +import io.delta.kernel.types.StringType; +import io.delta.kernel.types.StructField; +import io.delta.kernel.types.StructType; + +import org.apache.xtable.model.schema.InternalField; +import org.apache.xtable.model.schema.InternalPartitionField; +import org.apache.xtable.model.schema.InternalSchema; +import org.apache.xtable.model.schema.InternalType; +import org.apache.xtable.model.schema.PartitionTransformType; +import org.apache.xtable.model.stat.PartitionValue; +import org.apache.xtable.model.stat.Range; +import org.apache.xtable.model.storage.InternalDataFile; +import org.apache.xtable.model.storage.PartitionFileGroup; + +public class TestDeltaKernelDataFileUpdatesExtractor { + + @TempDir private Path tempDir; + + private Engine engine; + private DeltaKernelDataFileUpdatesExtractor extractor; + private InternalSchema testSchema; + private StructType physicalSchema; + + @BeforeEach + public void setup() { + Configuration hadoopConf = new Configuration(); + engine = DefaultEngine.create(hadoopConf); + + // Create test schema + testSchema = + InternalSchema.builder() + .name("record") + .dataType(InternalType.RECORD) + .fields( + Arrays.asList( + InternalField.builder() + .name("id") + .schema( + InternalSchema.builder() + .name("integer") + .dataType(InternalType.INT) + .isNullable(false) + .build()) + .build(), + InternalField.builder() + .name("name") + .schema( + InternalSchema.builder() + .name("string") + .dataType(InternalType.STRING) + .isNullable(true) + .build()) + .build())) + .build(); + + // Create physical schema + physicalSchema = + new StructType() + .add(new StructField("id", IntegerType.INTEGER, false)) + .add(new StructField("name", StringType.STRING, true)); + + // Initialize extractor + extractor = + DeltaKernelDataFileUpdatesExtractor.builder() + .engine(engine) + .basePath(tempDir.toString()) + .includeColumnStats(false) + .build(); + } + + @Test + public void testCreateAddFileAction() throws IOException { + // Create a test data file + String testFilePath = tempDir.resolve("test_data.parquet").toString(); + Files.createFile(Paths.get(testFilePath)); + + InternalDataFile dataFile = + InternalDataFile.builder() + .physicalPath(testFilePath) + .fileSizeBytes(1024L) + .lastModified(Instant.now().toEpochMilli()) + .recordCount(100L) + .partitionValues(Collections.emptyList()) + .columnStats(Collections.emptyList()) + .build(); + + // Create a simple Delta table for testing + Table table = createSimpleDeltaTable(); + + List<PartitionFileGroup> partitionedDataFiles = + Collections.singletonList( + PartitionFileGroup.builder() + .files(Collections.singletonList(dataFile)) + .partitionValues(Collections.emptyList()) + .build()); + + // Execute applySnapshot + scala.collection.Seq<RowBackedAction> actions = + extractor.applySnapshot(table, partitionedDataFiles, testSchema); + + // Verify actions are created + assertNotNull(actions); + List<RowBackedAction> actionList = JavaConverters.seqAsJavaList(actions); + assertFalse(actionList.isEmpty(), "Should have at least one action"); + + // Verify we have AddFile actions + boolean hasAddFile = actionList.stream().anyMatch(action -> action instanceof AddFile); + assertTrue(hasAddFile, "Should contain AddFile actions"); + } + + @Test + public void testApplySnapshotWithPartitionedData() throws IOException { + // Create test data files with partitions + String testFilePath1 = tempDir.resolve("partition1/test_data1.parquet").toString(); + String testFilePath2 = tempDir.resolve("partition2/test_data2.parquet").toString(); + Files.createDirectories(Paths.get(testFilePath1).getParent()); + Files.createDirectories(Paths.get(testFilePath2).getParent()); + Files.createFile(Paths.get(testFilePath1)); + Files.createFile(Paths.get(testFilePath2)); + + InternalPartitionField partitionField = + InternalPartitionField.builder() + .sourceField( + InternalField.builder() + .name("partition_col") + .schema( + InternalSchema.builder() + .name("string") + .dataType(InternalType.STRING) + .build()) + .build()) + .transformType(PartitionTransformType.VALUE) + .build(); + + PartitionValue partitionValue1 = + PartitionValue.builder() + .partitionField(partitionField) + .range(Range.scalar("partition1")) + .build(); + + PartitionValue partitionValue2 = + PartitionValue.builder() + .partitionField(partitionField) + .range(Range.scalar("partition2")) + .build(); + + InternalDataFile dataFile1 = + InternalDataFile.builder() + .physicalPath(testFilePath1) + .fileSizeBytes(1024L) + .lastModified(Instant.now().toEpochMilli()) + .recordCount(50L) + .partitionValues(Collections.singletonList(partitionValue1)) + .columnStats(Collections.emptyList()) + .build(); + + InternalDataFile dataFile2 = + InternalDataFile.builder() + .physicalPath(testFilePath2) + .fileSizeBytes(2048L) + .lastModified(Instant.now().toEpochMilli()) + .recordCount(75L) + .partitionValues(Collections.singletonList(partitionValue2)) + .columnStats(Collections.emptyList()) + .build(); + + Table table = createSimpleDeltaTable(); + + List<PartitionFileGroup> partitionedDataFiles = + Arrays.asList( + PartitionFileGroup.builder() + .files(Collections.singletonList(dataFile1)) + .partitionValues(Collections.singletonList(partitionValue1)) + .build(), + PartitionFileGroup.builder() + .files(Collections.singletonList(dataFile2)) + .partitionValues(Collections.singletonList(partitionValue2)) + .build()); + + // Execute applySnapshot + scala.collection.Seq<RowBackedAction> actions = + extractor.applySnapshot(table, partitionedDataFiles, testSchema); + + // Verify + assertNotNull(actions); + List<RowBackedAction> actionList = JavaConverters.seqAsJavaList(actions); + assertFalse(actionList.isEmpty(), "Should have actions for partitioned data"); + + // Should have AddFile actions for new files + long addFileCount = actionList.stream().filter(action -> action instanceof AddFile).count(); + assertTrue(addFileCount >= 2, "Should have at least 2 AddFile actions"); + } + + @Test + public void testDifferentialSyncWithExistingData() throws IOException { + // This test simulates a real differential sync scenario: + // 1. Delta table has existing files: file1.parquet, file2.parquet + // 2. New sync brings: file2.parquet (unchanged), file3.parquet (new) + // 3. Expected result: AddFile for file3, RemoveFile for file1 + + // Step 1: Create a Delta table with existing data + Path tablePath = tempDir.resolve("delta_table_with_data"); + Files.createDirectories(tablePath); + Path deltaLogPath = tablePath.resolve("_delta_log"); + Files.createDirectories(deltaLogPath); + + // Create existing data files + Path existingFile1 = tablePath.resolve("file1.parquet"); + Path existingFile2 = tablePath.resolve("file2.parquet"); + Files.createFile(existingFile1); + Files.createFile(existingFile2); + + // Create initial commit with file1 and file2 + Path initialCommit = deltaLogPath.resolve("00000000000000000000.json"); + String initialCommitJson = + "{\"protocol\":{\"minReaderVersion\":1,\"minWriterVersion\":2}}\n" + + "{\"metaData\":{\"id\":\"test-id\",\"format\":{\"provider\":\"parquet\",\"options\":{}},\"schemaString\":\"" + + physicalSchema.toJson().replace("\"", "\\\"") + + "\",\"partitionColumns\":[],\"configuration\":{},\"createdTime\":" + + System.currentTimeMillis() + + "}}\n" + + "{\"add\":{\"path\":\"file1.parquet\",\"partitionValues\":{},\"size\":1024,\"modificationTime\":" + + Instant.now().toEpochMilli() + + ",\"dataChange\":true,\"stats\":\"{}\"}}\n" + + "{\"add\":{\"path\":\"file2.parquet\",\"partitionValues\":{},\"size\":2048,\"modificationTime\":" + + Instant.now().toEpochMilli() + + ",\"dataChange\":true,\"stats\":\"{}\"}}\n"; + Files.write(initialCommit, initialCommitJson.getBytes(StandardCharsets.UTF_8)); + + // Create the table + Table table = Table.forPath(engine, tablePath.toString()); + assertNotNull(table); + + // Step 2: Prepare new sync data - file2 (unchanged) + file3 (new) + Path newFile3 = tablePath.resolve("file3.parquet"); + Files.createFile(newFile3); + + InternalDataFile dataFile2 = + InternalDataFile.builder() + .physicalPath(existingFile2.toString()) + .fileSizeBytes(2048L) + .lastModified(Instant.now().toEpochMilli()) + .recordCount(100L) + .partitionValues(Collections.emptyList()) + .columnStats(Collections.emptyList()) + .build(); + + InternalDataFile dataFile3 = + InternalDataFile.builder() + .physicalPath(newFile3.toString()) + .fileSizeBytes(3072L) + .lastModified(Instant.now().toEpochMilli()) + .recordCount(150L) + .partitionValues(Collections.emptyList()) + .columnStats(Collections.emptyList()) + .build(); + + List<PartitionFileGroup> newPartitionedDataFiles = + Collections.singletonList( + PartitionFileGroup.builder() + .files(Arrays.asList(dataFile2, dataFile3)) + .partitionValues(Collections.emptyList()) + .build()); + + // Step 3: Apply snapshot (differential sync) + DeltaKernelDataFileUpdatesExtractor syncExtractor = + DeltaKernelDataFileUpdatesExtractor.builder() + .engine(engine) + .basePath(tablePath.toString()) + .includeColumnStats(false) + .build(); + + scala.collection.Seq<RowBackedAction> actions = + syncExtractor.applySnapshot(table, newPartitionedDataFiles, testSchema); + + // Step 4: Verify the differential sync results + assertNotNull(actions, "Actions should not be null"); + List<RowBackedAction> actionList = JavaConverters.seqAsJavaList(actions); + assertFalse(actionList.isEmpty(), "Should have actions for differential sync"); + + // Count AddFile and RemoveFile actions + long addFileCount = actionList.stream().filter(action -> action instanceof AddFile).count(); + long removeFileCount = + actionList.stream().filter(action -> action instanceof RemoveFile).count(); + + // Verify: Should have AddFile for file3 (new file) + assertTrue(addFileCount >= 1, "Should have at least 1 AddFile action for new file (file3)"); Review Comment: Addressed — the test now asserts exact counts, e.g. `assertEquals(1, addFileCount, ...)` for the added file. ########## xtable-core/src/main/java/org/apache/xtable/kernel/DeltaKernelConversionTarget.java: ########## @@ -0,0 +1,483 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + */ + +package org.apache.xtable.kernel; + +import static io.delta.kernel.utils.CloseableIterable.inMemoryIterable; + +import java.io.File; +import java.util.ArrayList; +import java.util.HashMap; +import java.util.List; +import java.util.Map; +import java.util.Optional; + +import lombok.Getter; +import lombok.extern.log4j.Log4j2; + +import org.apache.hadoop.conf.Configuration; + +import scala.collection.JavaConverters; +import scala.collection.Seq; + +import com.google.common.annotations.VisibleForTesting; + +import io.delta.kernel.Operation; +import io.delta.kernel.Snapshot; +import io.delta.kernel.Table; +import io.delta.kernel.Transaction; +import io.delta.kernel.TransactionBuilder; +import io.delta.kernel.TransactionCommitResult; +import io.delta.kernel.data.Row; +import io.delta.kernel.defaults.engine.DefaultEngine; +import io.delta.kernel.engine.Engine; +import io.delta.kernel.exceptions.TableNotFoundException; +import io.delta.kernel.hook.PostCommitHook; +import io.delta.kernel.internal.SnapshotImpl; +import io.delta.kernel.internal.actions.AddFile; +import io.delta.kernel.internal.actions.Metadata; +import io.delta.kernel.internal.actions.RemoveFile; +import io.delta.kernel.internal.actions.RowBackedAction; +import io.delta.kernel.internal.actions.SingleAction; +import io.delta.kernel.types.StructField; +import io.delta.kernel.types.StructType; +import io.delta.kernel.utils.CloseableIterable; +import io.delta.kernel.utils.CloseableIterator; + +import org.apache.xtable.conversion.TargetTable; +import org.apache.xtable.exception.NotSupportedException; +import org.apache.xtable.exception.UpdateException; +import org.apache.xtable.model.InternalTable; +import org.apache.xtable.model.metadata.TableSyncMetadata; +import org.apache.xtable.model.schema.InternalPartitionField; +import org.apache.xtable.model.schema.InternalSchema; +import org.apache.xtable.model.storage.InternalFilesDiff; +import org.apache.xtable.model.storage.PartitionFileGroup; +import org.apache.xtable.model.storage.TableFormat; +import org.apache.xtable.spi.sync.ConversionTarget; + +/** + * Implementation of {@link ConversionTarget} for Delta Lake using the Delta Kernel API. + * + * <p>This implementation uses Delta Kernel (io.delta.kernel) instead of Delta Standalone for write + * operations, providing better compatibility with cloud storage (S3, GCS, Azure Blob Storage, HDFS) + * and improved support for Delta Lake 3.x features. + * + * <p><strong>Initialization:</strong> This class supports two initialization patterns: + * + * <ul> + * <li><strong>Factory/ServiceLoader Pattern:</strong> Use the no-arg constructor followed by + * {@link #init(TargetTable, Configuration)}. This is used by {@link + * org.apache.xtable.conversion.ConversionTargetFactory}. + * <li><strong>Direct Testing Pattern:</strong> Use the constructor with {@link TargetTable} and + * {@link Engine} parameters for direct instantiation with custom dependencies in tests. + * </ul> + * + * <p><strong>Important:</strong> Do not mix initialization patterns. If you use the parameterized + * constructor, do not call {@link #init(TargetTable, Configuration)} afterward, as it will + * overwrite the custom Engine. + * + * <p><strong>Exception Handling:</strong> This implementation only catches {@link + * io.delta.kernel.exceptions.TableNotFoundException} when checking for table existence, allowing + * other exceptions (network errors, permission issues, corrupted metadata) to propagate rather than + * being silently masked. This ensures real errors are visible and fail fast. + * + * <p><strong>Known Limitations:</strong> + * + * <ul> + * <li><strong>Commit Tags:</strong> Delta Kernel 4.0.0 does not support commit tags in commitInfo + * (e.g., XTABLE_METADATA tags). This affects source-to-target commit identifier mapping. + * Tracked in: https://github.com/apache/incubator-xtable/issues/819 + * <li><strong>Schema Evolution:</strong> Schema changes are handled through Delta Kernel's + * transaction API, which may have different semantics compared to Delta Standalone. + * <li><strong>Internal API Usage:</strong> This implementation casts to internal classes + * (SnapshotImpl, TableImpl) to access metadata and commit history, as Delta Kernel 4.0.0 + * lacks public APIs for these operations. These casts are brittle and may break on version + * upgrades. Public API alternatives should be used when available. + * </ul> + * + * <p><strong>Implementation Choice:</strong> Delta Kernel API was chosen over Delta Standalone to: + * + * <ul> + * <li>Support newer Delta Lake features and protocol versions + * <li>Align with the Delta Lake community's direction (Kernel is the recommended API) + * <li>Reduce dependency on Spark-specific implementations + * </ul> + * + * @see ConversionTarget + * @see io.delta.kernel.Table + * @see io.delta.kernel.Transaction + */ +@Log4j2 +public class DeltaKernelConversionTarget implements ConversionTarget { + private static final String DELTA_LOG_RETENTION_DURATION = "delta.logRetentionDuration"; + + private DeltaKernelSchemaExtractor schemaExtractor; + private DeltaKernelPartitionExtractor partitionExtractor; + private DeltaKernelDataFileUpdatesExtractor dataKernelFileUpdatesExtractor; + + private String basePath; + private long logRetentionInHours; + private DeltaKernelConversionTarget.TransactionState transactionState; + private Engine engine; + + /** + * No-arg constructor for ServiceLoader instantiation. Must call {@link #init(TargetTable, + * Configuration)} before use. + */ + public DeltaKernelConversionTarget() {} + + /** + * Creates a fully initialized DeltaKernelConversionTarget with custom Engine. Typically used in + * tests. Do not call {@link #init(TargetTable, Configuration)} after this. + * + * @param targetTable the target table configuration + * @param engine custom Delta Kernel engine instance + */ + public DeltaKernelConversionTarget(TargetTable targetTable, Engine engine) { + this( + targetTable.getBasePath(), + targetTable.getMetadataRetention().toHours(), + engine, + DeltaKernelSchemaExtractor.getInstance(), + DeltaKernelPartitionExtractor.getInstance(), + DeltaKernelDataFileUpdatesExtractor.builder() + .engine(engine) + .basePath(targetTable.getBasePath()) + // Column statistics are not needed for conversion operations + .includeColumnStats(false) + .build()); + } + + @VisibleForTesting + DeltaKernelConversionTarget( + String tableDataPath, + long logRetentionInHours, + Engine engine, + DeltaKernelSchemaExtractor schemaExtractor, + DeltaKernelPartitionExtractor partitionExtractor, + DeltaKernelDataFileUpdatesExtractor dataKernelFileUpdatesExtractor) { + _init( + tableDataPath, + logRetentionInHours, + engine, + schemaExtractor, + partitionExtractor, + dataKernelFileUpdatesExtractor); + } + + /** + * Private initialization helper to avoid code duplication between constructor and init() paths. + */ + private void _init( + String tableDataPath, + long logRetentionInHours, + Engine engine, + DeltaKernelSchemaExtractor schemaExtractor, + DeltaKernelPartitionExtractor partitionExtractor, + DeltaKernelDataFileUpdatesExtractor dataKernelFileUpdatesExtractor) { + this.basePath = tableDataPath; + this.schemaExtractor = schemaExtractor; + this.partitionExtractor = partitionExtractor; + this.dataKernelFileUpdatesExtractor = dataKernelFileUpdatesExtractor; + this.engine = engine; + this.logRetentionInHours = logRetentionInHours; + } + + @Override + public void init(TargetTable targetTable, Configuration configuration) { + Engine engine = DefaultEngine.create(configuration); + + _init( + targetTable.getBasePath(), + targetTable.getMetadataRetention().toHours(), + engine, + DeltaKernelSchemaExtractor.getInstance(), + DeltaKernelPartitionExtractor.getInstance(), + DeltaKernelDataFileUpdatesExtractor.builder() + .engine(engine) + .basePath(targetTable.getBasePath()) + .includeColumnStats(true) + .build()); + } + + @Override + public void beginSync(InternalTable table) { + this.transactionState = + new DeltaKernelConversionTarget.TransactionState(engine, logRetentionInHours); + } + + @Override + public void syncSchema(InternalSchema schema) { + transactionState.setLatestSchema(schema); + } + + @Override + public void syncPartitionSpec(List<InternalPartitionField> partitionSpec) { + if (partitionSpec != null) { + Map<String, StructField> spec = + partitionExtractor.convertToDeltaPartitionFormat(partitionSpec); + for (Map.Entry<String, StructField> partitionEntry : spec.entrySet()) { + String partitionColumnName = partitionEntry.getKey(); + StructField partitionField = partitionEntry.getValue(); + + transactionState.addPartitionColumn(partitionColumnName); + if (partitionField != null + && transactionState.getLatestSchema().fields().stream() + .noneMatch(field -> field.getName().equals(partitionField.getName()))) { + // add generated columns to schema. + transactionState.addColumn(partitionField); + } + } + } + } + + @Override + public void syncMetadata(TableSyncMetadata metadata) { + transactionState.setMetadata(metadata); + } + + @Override + public void syncFilesForSnapshot(List<PartitionFileGroup> partitionedDataFiles) { + Table table = Table.forPath(engine, basePath); + transactionState.setActions( + dataKernelFileUpdatesExtractor.applySnapshot( + table, partitionedDataFiles, transactionState.getLatestSchemaInternal())); + } + + @Override + public void syncFilesForDiff(InternalFilesDiff internalFilesDiff) { + Table table = Table.forPath(engine, basePath); + transactionState.setActions( + dataKernelFileUpdatesExtractor.applyDiff( + internalFilesDiff, basePath, table.getLatestSnapshot(engine).getSchema())); + } + + @Override + public void completeSync() { + transactionState.commitTransaction(); + transactionState = null; + } + + @Override + public Optional<TableSyncMetadata> getTableMetadata() { + Table table = Table.forPath(engine, basePath); + Snapshot snapshot = table.getLatestSnapshot(engine); + + // WORKAROUND: Cast to SnapshotImpl (internal class) to access metadata configuration. + // Delta Kernel 4.0.0 does not provide a public API to access table metadata/configuration. + // This cast is brittle and may break on Kernel version upgrades. + Metadata metadata = ((SnapshotImpl) snapshot).getMetadata(); + + // Get configuration from metadata + Map<String, String> configuration = metadata.getConfiguration(); + String metadataJson = configuration.get(TableSyncMetadata.XTABLE_METADATA); + + return TableSyncMetadata.fromJson(metadataJson); + } + + @Override + public String getTableFormat() { + return TableFormat.DELTA; + } + + @Override + public Optional<String> getTargetCommitIdentifier(String sourceIdentifier) { + // Delta Kernel 4.0.0 does not support commit tags in commitInfo, which are required for + // source-to-target commit identifier mapping. This limitation is documented in: + // https://github.com/delta-io/delta/issues/6167 + // XTable tracking issue: https://github.com/apache/incubator-xtable/issues/819 + // + // Unlike DeltaConversionTarget (which uses Delta Standalone with commit tag support), + // DeltaKernelConversionTarget cannot retrieve commit tags from Delta Kernel's API. + // Rather than silently scanning all commits (O(n) performance cost) and always returning + // empty, we explicitly throw an exception to indicate this feature is unsupported. + // + // When Delta Kernel adds commit tag support, this method can be reimplemented to: + // 1. Scan commit history using tableImpl.getChanges(engine, 0, currentVersion, actionSet) + // 2. Extract tags from CommitInfo.tags MapValue + // 3. Parse XTABLE_METADATA from tags and match sourceIdentifier + throw new NotSupportedException( + "Source-to-target commit identifier mapping is not supported in DeltaKernelConversionTarget. " + + "Delta Kernel 4.0.0 does not support commit tags in commitInfo. " + + "See: https://github.com/delta-io/delta/issues/6167"); + } + + private class TransactionState { + private final Engine engine; + private final long retentionInHours; + private final List<String> partitionColumns; + @Getter private StructType latestSchema; + @Getter private InternalSchema latestSchemaInternal; + private TableSyncMetadata metadata; + private List<RowBackedAction> actions; + + private TransactionState(Engine engine, long retentionInHours) { + this.engine = engine; + this.partitionColumns = new ArrayList<>(); + this.retentionInHours = retentionInHours; + + try { + Table table = Table.forPath(engine, basePath); + this.latestSchema = table.getLatestSnapshot(engine).getSchema(); + } catch (TableNotFoundException e) { + // Expected: table doesn't exist yet on first sync + this.latestSchema = null; + } + // Let other exceptions propagate (network issues, permissions, corrupted metadata, etc.) + } + + /** + * Adds a partition column name to the list. Package-private to allow access from outer class. + */ + void addPartitionColumn(String columnName) { + partitionColumns.add(columnName); + } + + void setMetadata(TableSyncMetadata metadata) { + this.metadata = metadata; + } + + /** + * Sets the actions to be committed. Converts from Scala Seq to Java List for internal storage. + */ + void setActions(Seq<RowBackedAction> scalaActions) { + this.actions = JavaConverters.seqAsJavaList(scalaActions); + } + + private void addColumn(StructField field) { + latestSchema = latestSchema.add(field); + latestSchemaInternal = schemaExtractor.toInternalSchema(latestSchema); + } + + private void setLatestSchema(InternalSchema schema) { + this.latestSchemaInternal = schema; + this.latestSchema = schemaExtractor.fromInternalSchema(schema); + } + + private void commitTransaction() { + boolean tableExists = checkTableExists(); + + Operation operation = tableExists ? Operation.WRITE : Operation.CREATE_TABLE; + + if (!tableExists) { + File tableDir = new File(basePath); + if (!tableDir.exists()) { + tableDir.mkdirs(); + } + } + + Table table = Table.forPath(engine, basePath); + TransactionBuilder txnBuilder = + table.createTransactionBuilder(engine, "XTable Delta Sync", operation); + + // LIMITATION: Schema evolution for existing tables is NOT supported in Delta Kernel 4.0.0. + // The withSchema() method only works during CREATE_TABLE operations. For existing tables: + // - AddFile/RemoveFile actions are created using the old schema from existing snapshot + // - If source schema has evolved (columns added/removed/type changed), the Delta table + // will have mismatched metadata and data, causing query failures or incorrect results + // This is a known Delta Kernel limitation: https://github.com/delta-io/delta/issues/4305 + if (!tableExists) { + txnBuilder = txnBuilder.withSchema(engine, latestSchema); + + if (!partitionColumns.isEmpty()) { + txnBuilder = txnBuilder.withPartitionColumns(engine, partitionColumns); + } + } + + Map<String, String> tableProperties = getConfigurationsForDeltaSync(); + txnBuilder = txnBuilder.withTableProperties(engine, tableProperties); + + Transaction txn = txnBuilder.build(engine); + List<Row> allActionRows = new ArrayList<>(); + + // Iterate through actions (Java List) and convert to Row format + for (RowBackedAction action : actions) { + + if (action instanceof AddFile) { + AddFile addFile = (AddFile) action; + Row wrappedRow = SingleAction.createAddFileSingleAction(addFile.toRow()); + allActionRows.add(wrappedRow); + } else if (action instanceof RemoveFile) { + RemoveFile removeFile = (RemoveFile) action; + Row wrappedRow = SingleAction.createRemoveFileSingleAction(removeFile.toRow()); + allActionRows.add(wrappedRow); + } + } + + CloseableIterator<Row> allActionsIterator = + new CloseableIterator<Row>() { + private int currentIndex = 0; + + @Override + public boolean hasNext() { + return currentIndex < allActionRows.size(); + } + + @Override + public Row next() { + return allActionRows.get(currentIndex++); + } + + @Override + public void close() {} + }; + + CloseableIterable<Row> dataActions = inMemoryIterable(allActionsIterator); + + try { + TransactionCommitResult result = txn.commit(engine, dataActions); + + // Execute PostCommitHooks to create checkpoints and _last_checkpoint metadata file + List<PostCommitHook> hooks = result.getPostCommitHooks(); + if (hooks != null && !hooks.isEmpty()) { + for (PostCommitHook hook : hooks) { + try { + hook.threadSafeInvoke(engine); + } catch (Exception hookEx) { + // Post-commit hooks are optimizations; log but don't fail the transaction + log.warn("Post-commit hook failed but transaction succeeded", hookEx); + } + } + } + } catch (Exception e) { + throw new UpdateException("Failed to commit Delta Kernel transaction", e); + } + + // NOTE: Delta Kernel API limitations compared to Delta Standalone: + // - Commit tags (like XTABLE_METADATA in commitInfo.tags) are not yet supported + // - Operation type metadata (like DeltaOperations.Update) is simplified to + // Operation.WRITE/CREATE_TABLE + // - The commit timestamp is managed by Delta Kernel automatically + } + + private boolean checkTableExists() { + return DeltaKernelUtils.tableExists(engine, basePath); + } + + private Map<String, String> getConfigurationsForDeltaSync() { Review Comment: On Delta Kernel 4.0.0, setting `minReaderVersion`/`minWriterVersion` explicitly via table properties throws `UnknownConfiguration`. Kernel instead auto-upgrades the protocol when a feature requiring a higher version is written, so the floors are managed by Kernel rather than set here. This is documented inline in `getConfigurationsForDeltaSync()`. Resolving as a documented Kernel limitation. ########## xtable-core/src/main/java/org/apache/xtable/kernel/DeltaKernelSchemaExtractor.java: ########## @@ -229,4 +230,118 @@ private InternalSchema toInternalSchema( .fields(fields) .build(); } + + /** + * Converts an InternalSchema to Delta Kernel StructType. + * + * @param internalSchema the internal schema representation + * @return Delta Kernel StructType + */ + public StructType fromInternalSchema(InternalSchema internalSchema) { + List<StructField> fields = + internalSchema.getFields().stream() + .map( + field -> + new StructField( + field.getName(), + convertFieldType(field), + field.getSchema().isNullable(), + getFieldMetadata(field.getSchema()))) + .collect(CustomCollectors.toList(internalSchema.getFields().size())); + return new StructType(fields); + } + + /** + * Converts an InternalField to Delta Kernel DataType. + * + * @param field the internal field + * @return Delta Kernel DataType + */ + private DataType convertFieldType(InternalField field) { + switch (field.getSchema().getDataType()) { + case STRING: + case ENUM: + return StringType.STRING; + case INT: + return IntegerType.INTEGER; + case LONG: + return LongType.LONG; + case BYTES: + case FIXED: + case UUID: + return BinaryType.BINARY; + case BOOLEAN: + return BooleanType.BOOLEAN; + case FLOAT: + return FloatType.FLOAT; + case DATE: + return DateType.DATE; + case TIMESTAMP: + return TimestampType.TIMESTAMP; + case TIMESTAMP_NTZ: + return TimestampNTZType.TIMESTAMP_NTZ; + case DOUBLE: + return DoubleType.DOUBLE; + case DECIMAL: + int precision = + (int) field.getSchema().getMetadata().get(InternalSchema.MetadataKey.DECIMAL_PRECISION); Review Comment: Addressed — instead of a raw cast, it now throws `IllegalStateException` with the field name when DECIMAL metadata or precision/scale is missing or non-numeric. -- 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. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
