fixup

facebookincubator · Feb 28, 2025 · 4f362da · 4f362da
1 parent eae4ff4
commit 4f362da
Show file tree

Hide file tree

Showing 5 changed files with 21 additions and 16 deletions.
diff --git a/velox/connectors/hive/HiveConnectorSplit.h b/velox/connectors/hive/HiveConnectorSplit.h
@@ -79,29 +79,30 @@ struct HiveConnectorSplit : public connector::ConnectorSplit {
       const std::unordered_map<std::string, std::optional<std::string>>&
           _partitionKeys = {},
       std::optional<int32_t> _tableBucketNumber = std::nullopt,
-      std::optional<HiveBucketConversion> _bucketConversion = std::nullopt,
       const std::unordered_map<std::string, std::string>& _customSplitInfo = {},
       const std::shared_ptr<std::string>& _extraFileInfo = {},
       const std::unordered_map<std::string, std::string>& _serdeParameters = {},
       int64_t splitWeight = 0,
       bool cacheable = true,
       const std::unordered_map<std::string, std::string>& _infoColumns = {},
       std::optional<FileProperties> _properties = std::nullopt,
-      std::optional<RowIdProperties> _rowIdProperties = std::nullopt)
+      std::optional<RowIdProperties> _rowIdProperties = std::nullopt,
+      const std::optional<HiveBucketConversion>& _bucketConversion =
+          std::nullopt)
       : ConnectorSplit(connectorId, splitWeight, cacheable),
         filePath(_filePath),
         fileFormat(_fileFormat),
         start(_start),
         length(_length),
         partitionKeys(_partitionKeys),
         tableBucketNumber(_tableBucketNumber),
-        bucketConversion(std::move(_bucketConversion)),
         customSplitInfo(_customSplitInfo),
         extraFileInfo(_extraFileInfo),
         serdeParameters(_serdeParameters),
         infoColumns(_infoColumns),
         properties(_properties),
-        rowIdProperties(_rowIdProperties) {}
+        rowIdProperties(_rowIdProperties),
+        bucketConversion(_bucketConversion) {}
 
   std::string toString() const override;
 
@@ -200,6 +201,12 @@ class HiveConnectorSplitBuilder {
     return *this;
   }
 
+  HiveConnectorSplitBuilder& rowIdProperties(
+      const RowIdProperties& rowIdProperties) {
+    rowIdProperties_ = rowIdProperties;
+    return *this;
+  }
+
   std::shared_ptr<connector::hive::HiveConnectorSplit> build() const {
     return std::make_shared<connector::hive::HiveConnectorSplit>(
         connectorId_,
@@ -209,14 +216,15 @@ class HiveConnectorSplitBuilder {
         length_,
         partitionKeys_,
         tableBucketNumber_,
-        bucketConversion_,
         customSplitInfo_,
         extraFileInfo_,
         serdeParameters_,
         splitWeight_,
         cacheable_,
         infoColumns_,
-        fileProperties_);
+        fileProperties_,
+        rowIdProperties_,
+        bucketConversion_);
   }
 
  private:
@@ -235,6 +243,7 @@ class HiveConnectorSplitBuilder {
   int64_t splitWeight_{0};
   bool cacheable_{true};
   std::optional<FileProperties> fileProperties_;
+  std::optional<RowIdProperties> rowIdProperties_ = std::nullopt;
 };
 
 } // namespace facebook::velox::connector::hive
diff --git a/velox/connectors/hive/iceberg/IcebergSplit.cpp b/velox/connectors/hive/iceberg/IcebergSplit.cpp
@@ -31,7 +31,6 @@ HiveIcebergSplit::HiveIcebergSplit(
     const std::unordered_map<std::string, std::optional<std::string>>&
         partitionKeys,
     std::optional<int32_t> tableBucketNumber,
-    std::optional<HiveBucketConversion> bucketConversion,
     const std::unordered_map<std::string, std::string>& customSplitInfo,
     const std::shared_ptr<std::string>& extraFileInfo,
     bool cacheable,
@@ -45,14 +44,15 @@ HiveIcebergSplit::HiveIcebergSplit(
           length,
           partitionKeys,
           tableBucketNumber,
-          std::move(bucketConversion),
           customSplitInfo,
           extraFileInfo,
           {},
           /*splitWeight=*/0,
           cacheable,
           infoColumns,
-          properties) {
+          properties,
+          std::nullopt,
+          std::nullopt) {
   // TODO: Deserialize _extraFileInfo to get deleteFiles;
 }
 
@@ -66,7 +66,6 @@ HiveIcebergSplit::HiveIcebergSplit(
     const std::unordered_map<std::string, std::optional<std::string>>&
         partitionKeys,
     std::optional<int32_t> tableBucketNumber,
-    std::optional<HiveBucketConversion> bucketConversion,
     const std::unordered_map<std::string, std::string>& customSplitInfo,
     const std::shared_ptr<std::string>& extraFileInfo,
     bool cacheable,
@@ -81,13 +80,14 @@ HiveIcebergSplit::HiveIcebergSplit(
           length,
           partitionKeys,
           tableBucketNumber,
-          std::move(bucketConversion),
           customSplitInfo,
           extraFileInfo,
           {},
           0,
           cacheable,
           infoColumns,
-          properties),
+          properties,
+          std::nullopt,
+          std::nullopt),
       deleteFiles(std::move(deletes)) {}
 } // namespace facebook::velox::connector::hive::iceberg
diff --git a/velox/connectors/hive/iceberg/IcebergSplit.h b/velox/connectors/hive/iceberg/IcebergSplit.h
@@ -35,7 +35,6 @@ struct HiveIcebergSplit : public connector::hive::HiveConnectorSplit {
       const std::unordered_map<std::string, std::optional<std::string>>&
           partitionKeys = {},
       std::optional<int32_t> tableBucketNumber = std::nullopt,
-      std::optional<HiveBucketConversion> bucketConversion = std::nullopt,
       const std::unordered_map<std::string, std::string>& customSplitInfo = {},
       const std::shared_ptr<std::string>& extraFileInfo = {},
       bool cacheable = true,
@@ -52,7 +51,6 @@ struct HiveIcebergSplit : public connector::hive::HiveConnectorSplit {
       const std::unordered_map<std::string, std::optional<std::string>>&
           partitionKeys = {},
       std::optional<int32_t> tableBucketNumber = std::nullopt,
-      std::optional<HiveBucketConversion> bucketConversion = std::nullopt,
       const std::unordered_map<std::string, std::string>& customSplitInfo = {},
       const std::shared_ptr<std::string>& extraFileInfo = {},
       bool cacheable = true,

diff --git a/velox/connectors/hive/iceberg/tests/IcebergReadTest.cpp b/velox/connectors/hive/iceberg/tests/IcebergReadTest.cpp
@@ -264,7 +264,6 @@ class HiveIcebergTest : public HiveConnectorTestBase {
           splitSize,
           partitionKeys,
           std::nullopt,
-          std::nullopt,
           customSplitInfo,
           nullptr,
           /*cacheable=*/true,

diff --git a/velox/connectors/hive/iceberg/tests/IcebergSplitReaderBenchmark.cpp b/velox/connectors/hive/iceberg/tests/IcebergSplitReaderBenchmark.cpp
@@ -113,7 +113,6 @@ IcebergSplitReaderBenchmark::makeIcebergSplit(
       fileSize,
       partitionKeys,
       std::nullopt,
-      std::nullopt,
       customSplitInfo,
       nullptr,
       /*cacheable=*/true,