[Kernel] [#5] Introduce deletion vector support (delta-io#4869)

<!--
Thanks for sending a pull request!  Here are some tips for you:
1. If this is your first time, please read our contributor guidelines:
https://github.com/delta-io/delta/blob/master/CONTRIBUTING.md
2. If the PR is unfinished, add '[WIP]' in your PR title, e.g., '[WIP]
Your PR title ...'.
  3. Be sure to keep the PR description updated to reflect all changes.
  4. Please write your PR title to summarize what this PR proposes.
5. If possible, provide a concise example to reproduce the issue for a
faster review.
6. If applicable, include the corresponding issue number in the PR title
and link it in the body.
-->

#### Which Delta project/connector is this regarding?
<!--
Please add the component selected below to the beginning of the pull
request title
For example: [Spark] Title of my pull request
-->

- [ ] Spark
- [ ] Standalone
- [ ] Flink
- [x] Kernel
- [ ] Other (fill in here)

## Description

<!--
- Describe what this PR changes.
- Describe why we need the change.
 
If this PR resolves an issue be sure to include "Resolves #XXX" to
correctly link and close the issue upon merge.
-->
Exposes deletion vector descriptor field as a parameter of AddFile for
generating v3 add actions.

## How was this patch tested?

<!--
If tests were added, say they were added here. Please make sure to test
the changes thoroughly including negative and positive cases if
possible.
If the changes were tested in any way other than unit tests, please
clarify how you tested step by step (ideally copy and paste-able, so
that other reviewers can test and check, and descendants can verify in
the future).
If the changes were not tested, please explain why.
-->
Unit tests. 

## Does this PR introduce _any_ user-facing changes?

<!--
If yes, please clarify the previous behavior and the change this PR
proposes - provide the console output, description and/or an example to
show the behavior difference if possible.
If possible, please also clarify if this is a user-facing change
compared to the released Delta Lake versions or within the unreleased
branches such as master.
If no, write 'No'.
-->
No.

Author: murali-db

kernel/kernel-api/src/main/java/io/delta/kernel/Transaction.java

@@ -270,7 +270,8 @@ static CloseableIterator<Row> generateAppendActions(
                   // TODO: populate tags in generateAppendActions
                   Collections.emptyMap() /* tags */,
                   Optional.empty() /* baseRowId */,
-                  Optional.empty() /* defaultRowCommitVersion */);
+                  Optional.empty() /* defaultRowCommitVersion */,
+                  Optional.empty() /* deletionVectorDescriptor */);
           return SingleAction.createAddFileSingleAction(addFileRow.toRow());
         });
   }

kernel/kernel-api/src/main/java/io/delta/kernel/internal/actions/AddFile.java

@@ -82,7 +82,8 @@ public static AddFile convertDataFileStatus(
       boolean dataChange,
       Map<String, String> tags,
       Optional<Long> baseRowId,
-      Optional<Long> defaultRowCommitVersion) {
+      Optional<Long> defaultRowCommitVersion,
+      Optional<DeletionVectorDescriptor> deletionVectorDescriptor) {
     Optional<MapValue> tagMapValue =
         !tags.isEmpty() ? Optional.of(VectorUtils.stringStringMapValue(tags)) : Optional.empty();
     Row row =
@@ -93,7 +94,7 @@ public static AddFile convertDataFileStatus(
             dataFileStatus.getSize(),
             dataFileStatus.getModificationTime(),
             dataChange,
-            Optional.empty(), // deletionVector
+            deletionVectorDescriptor,
             tagMapValue, // tags
             baseRowId,
             defaultRowCommitVersion,
@@ -118,8 +119,6 @@ public static Row createAddFileRow(
 
     checkArgument(path != null, "path is not nullable");
     checkArgument(partitionValues != null, "partitionValues is not nullable");
-    // TODO - Add support for DeletionVectorDescriptor
-    checkArgument(!deletionVector.isPresent(), "DeletionVectorDescriptor is unsupported");
 
     Map<Integer, Object> fieldMap = new HashMap<>();
     fieldMap.put(FULL_SCHEMA.indexOf("path"), path);
@@ -133,7 +132,11 @@ public static Row createAddFileRow(
         version -> fieldMap.put(FULL_SCHEMA.indexOf("defaultRowCommitVersion"), version));
     stats.ifPresent(
         stat -> fieldMap.put(FULL_SCHEMA.indexOf("stats"), stat.serializeAsJson(physicalSchema)));
-
+    deletionVector.ifPresent(
+        dv -> {
+          Row dvRow = dv.toRow();
+          fieldMap.put(FULL_SCHEMA.indexOf("deletionVector"), dvRow);
+        });
     return new GenericRow(FULL_SCHEMA, fieldMap);
   }
 

kernel/kernel-api/src/main/java/io/delta/kernel/internal/actions/DeletionVectorDescriptor.java

@@ -22,6 +22,7 @@
 
 import io.delta.kernel.data.ColumnVector;
 import io.delta.kernel.data.Row;
+import io.delta.kernel.internal.data.GenericRow;
 import io.delta.kernel.internal.deletionvectors.Base85Codec;
 import io.delta.kernel.internal.fs.Path;
 import io.delta.kernel.types.IntegerType;
@@ -30,10 +31,7 @@
 import io.delta.kernel.types.StructType;
 import java.net.URI;
 import java.net.URISyntaxException;
-import java.util.Map;
-import java.util.Objects;
-import java.util.Optional;
-import java.util.UUID;
+import java.util.*;
 import java.util.stream.IntStream;
 
 /** Information about a deletion vector attached to a file action. */
@@ -237,6 +235,25 @@ public String toString() {
         storageType, pathOrInlineDv, offset, sizeInBytes, cardinality);
   }
 
+  /** @return Row representation of this deletion vector descriptor */
+  public Row toRow() {
+    Map<Integer, Object> fieldMap = new HashMap<>();
+
+    fieldMap.put(COL_NAME_TO_ORDINAL.get("storageType"), storageType);
+    fieldMap.put(COL_NAME_TO_ORDINAL.get("pathOrInlineDv"), pathOrInlineDv);
+
+    // Only add offset if it's present
+    if (offset.isPresent()) {
+      fieldMap.put(COL_NAME_TO_ORDINAL.get("offset"), offset.get());
+    }
+    // If offset is not present, the field remains null in the map
+
+    fieldMap.put(COL_NAME_TO_ORDINAL.get("sizeInBytes"), sizeInBytes);
+    fieldMap.put(COL_NAME_TO_ORDINAL.get("cardinality"), cardinality);
+
+    return new GenericRow(READ_SCHEMA, fieldMap);
+  }
+
   @Override
   public boolean equals(Object o) {
     if (o == this) {

kernel/kernel-api/src/main/java/io/delta/kernel/internal/actions/GenerateIcebergCompatActionUtils.java

@@ -97,7 +97,8 @@ public static Row generateIcebergCompatWriterV1AddAction(
             dataChange,
             tags,
             Optional.empty() /* baseRowId */,
-            Optional.empty() /* defaultRowCommitVersion */);
+            Optional.empty() /* defaultRowCommitVersion */,
+            Optional.empty() /* deletionVectorDescriptor */);
     return SingleAction.createAddFileSingleAction(addFile.toRow());
   }
 
@@ -119,6 +120,7 @@ public static Row generateIcebergCompatWriterV1AddAction(
    * @param partitionValues the partition values for the add
    * @param dataChange whether or not the add constitutes a dataChange (i.e. append vs. compaction)
    * @param tags key-value metadata to be attached to the add action
+   * @param deletionVectorDescriptor optional deletion vector descriptor for the add action
    * @return add action row that can be included in the transaction
    * @throws UnsupportedOperationException if icebergWriterCompatV3 is not enabled
    * @throws UnsupportedOperationException if maxRetries != 0 in the transaction
@@ -132,13 +134,17 @@ public static Row generateIcebergCompatWriterV3AddAction(
       boolean dataChange,
       Map<String, String> tags,
       Optional<Long> baseRowId,
-      Optional<Long> defaultRowCommitVersion) {
+      Optional<Long> defaultRowCommitVersion,
+      Optional<DeletionVectorDescriptor> deletionVectorDescriptor) {
     Map<String, String> configuration = TransactionStateRow.getConfiguration(transactionState);
 
     /* ----- Validate that this is a valid usage of this API ----- */
     validateIcebergWriterCompatV3Enabled(configuration);
     validateMaxRetriesSetToZero(transactionState);
 
+    /* ----- Validate that deletion vector is passed in only when the table supports it ----- */
+    deletionVectorDescriptor.ifPresent(dv -> validateIcebergDeletionVectorsEnabled(configuration));
+
     /* ----- Validate this is valid write given the table's protocol & configurations ----- */
     checkState(
         TableConfig.ICEBERG_COMPAT_V3_ENABLED.fromMetadata(configuration),
@@ -161,7 +167,8 @@ public static Row generateIcebergCompatWriterV3AddAction(
             dataChange,
             tags,
             baseRowId,
-            defaultRowCommitVersion);
+            defaultRowCommitVersion,
+            deletionVectorDescriptor);
     return SingleAction.createAddFileSingleAction(addFile.toRow());
   }
 
@@ -210,7 +217,8 @@ public static Row generateIcebergCompatWriterV1RemoveAction(
             partitionValues,
             dataChange,
             Optional.empty() /* baseRowId */,
-            Optional.empty() /* defaultRowCommitVersion */);
+            Optional.empty() /* defaultRowCommitVersion */,
+            Optional.empty() /* deletionVectorDescriptor */);
     return SingleAction.createRemoveFileSingleAction(removeFileRow);
   }
 
@@ -224,6 +232,7 @@ public static Row generateIcebergCompatWriterV1RemoveAction(
    * @param partitionValues the partition values for the remove
    * @param dataChange whether or not the remove constitutes a dataChange (i.e. delete vs.
    *     compaction)
+   * @param deletionVectorDescriptor optional deletion vector descriptor for the add action
    * @return remove action row that can be committed to the transaction
    * @throws UnsupportedOperationException if icebergWriterCompatV3 is not enabled
    * @throws UnsupportedOperationException if maxRetries != 0 in the transaction
@@ -236,13 +245,17 @@ public static Row generateIcebergCompatWriterV3RemoveAction(
       Map<String, Literal> partitionValues,
       boolean dataChange,
       Optional<Long> baseRowId,
-      Optional<Long> defaultRowCommitVersion) {
+      Optional<Long> defaultRowCommitVersion,
+      Optional<DeletionVectorDescriptor> deletionVectorDescriptor) {
     Map<String, String> config = TransactionStateRow.getConfiguration(transactionState);
 
     /* ----- Validate that this is a valid usage of this API ----- */
     validateIcebergWriterCompatV3Enabled(config);
     validateMaxRetriesSetToZero(transactionState);
 
+    /* ----- Validate that deletion vector is passed in only when the table supports it ----- */
+    deletionVectorDescriptor.ifPresent(dv -> validateIcebergDeletionVectorsEnabled(config));
+
     /* ----- Validate this is valid write given the table's protocol & configurations ----- */
     // We only allow removes with dataChange=false when appendOnly=true
     if (dataChange && TableConfig.APPEND_ONLY_ENABLED.fromMetadata(config)) {
@@ -268,7 +281,8 @@ public static Row generateIcebergCompatWriterV3RemoveAction(
             partitionValues,
             dataChange,
             Optional.empty() /* baseRowId */,
-            Optional.empty() /* defaultRowCommitVersion */);
+            Optional.empty() /* defaultRowCommitVersion */,
+            deletionVectorDescriptor);
     return SingleAction.createRemoveFileSingleAction(removeFileRow);
   }
 
@@ -307,6 +321,20 @@ private static void validateIcebergWriterCompatV3Enabled(Map<String, String> con
     }
   }
 
+  /**
+   * Validates that table feature `deletion vectors` is enabled. Checked when a deletion vector
+   * descriptor is passed to generateIcebergCompatWriterV3AddAction.
+   */
+  private static void validateIcebergDeletionVectorsEnabled(Map<String, String> config) {
+    if (!TableConfig.DELETION_VECTORS_CREATION_ENABLED.fromMetadata(config)) {
+      throw new UnsupportedOperationException(
+          String.format(
+              "APIs within GenerateIcebergCompatActionUtils are only supported on tables with"
+                  + " '%s' set to true",
+              TableConfig.DELETION_VECTORS_CREATION_ENABLED.getKey()));
+    }
+  }
+
   /**
    * Throws an exception if `maxRetries` was not set to 0 in the transaction. We restrict these APIs
    * to require `maxRetries = 0` since conflict resolution is not supported for operations other
@@ -359,7 +387,8 @@ public static Row convertRemoveDataFileStatus(
       Map<String, Literal> partitionValues,
       boolean dataChange,
       Optional<Long> baseRowId,
-      Optional<Long> defaultRowCommitVersion) {
+      Optional<Long> defaultRowCommitVersion,
+      Optional<DeletionVectorDescriptor> deletionVectorDescriptor) {
     return createRemoveFileRowWithExtendedFileMetadata(
         relativizePath(new Path(dataFileStatus.getPath()), tableRoot).toUri().toString(),
         dataFileStatus.getModificationTime(),
@@ -369,7 +398,8 @@ public static Row convertRemoveDataFileStatus(
         dataFileStatus.getStatistics(),
         physicalSchema,
         baseRowId,
-        defaultRowCommitVersion);
+        defaultRowCommitVersion,
+        deletionVectorDescriptor);
   }
 
   @VisibleForTesting
@@ -382,7 +412,8 @@ public static Row createRemoveFileRowWithExtendedFileMetadata(
       Optional<DataFileStatistics> stats,
       StructType physicalSchema,
       Optional<Long> baseRowId,
-      Optional<Long> defaultRowCommitVersion) {
+      Optional<Long> defaultRowCommitVersion,
+      Optional<DeletionVectorDescriptor> deletionVector) {
     Map<Integer, Object> fieldMap = new HashMap<>();
     fieldMap.put(RemoveFile.FULL_SCHEMA.indexOf("path"), requireNonNull(path));
     fieldMap.put(RemoveFile.FULL_SCHEMA.indexOf("deletionTimestamp"), deletionTimestamp);
@@ -399,6 +430,11 @@ public static Row createRemoveFileRowWithExtendedFileMetadata(
     defaultRowCommitVersion.ifPresent(
         version ->
             fieldMap.put(RemoveFile.FULL_SCHEMA.indexOf("defaultRowCommitVersion"), version));
+    deletionVector.ifPresent(
+        dv -> {
+          Row dvRow = dv.toRow();
+          fieldMap.put(RemoveFile.FULL_SCHEMA.indexOf("deletionVector"), dvRow);
+        });
     return new GenericRow(RemoveFile.FULL_SCHEMA, fieldMap);
   }
 }