Revert "feat: Add Iceberg partition name generator (facebookincubator#15461)"

This reverts commit 7576f4e.

Alchemy-item: (ID = 885) Iceberg staging hub commit 1/15 - 9582b8b

Author: PingLiuPing

velox/connectors/hive/CMakeLists.txt

@@ -27,7 +27,7 @@ velox_add_library(
   HiveConnectorSplit.cpp
   HiveDataSink.cpp
   HiveDataSource.cpp
-  HivePartitionName.cpp
+  HivePartitionUtil.cpp
   PartitionIdGenerator.cpp
   SplitReader.cpp
   TableHandle.cpp

velox/connectors/hive/HiveDataSink.cpp

@@ -200,7 +200,6 @@ FOLLY_ALWAYS_INLINE int32_t
 getBucketCount(const HiveBucketProperty* bucketProperty) {
   return bucketProperty == nullptr ? 0 : bucketProperty->bucketCount();
 }
-
 } // namespace
 
 const HiveWriterId& HiveWriterId::unpartitionedId() {
@@ -383,9 +382,7 @@ HiveDataSink::HiveDataSink(
               ? createBucketFunction(
                     *insertTableHandle->bucketProperty(),
                     inputType)
-              : nullptr,
-          getPartitionChannels(insertTableHandle),
-          nullptr) {}
+              : nullptr) {}
 
 HiveDataSink::HiveDataSink(
     RowTypePtr inputType,
@@ -395,27 +392,6 @@ HiveDataSink::HiveDataSink(
     const std::shared_ptr<const HiveConfig>& hiveConfig,
     uint32_t bucketCount,
     std::unique_ptr<core::PartitionFunction> bucketFunction)
-    : HiveDataSink(
-          std::move(inputType),
-          insertTableHandle,
-          connectorQueryCtx,
-          commitStrategy,
-          hiveConfig,
-          bucketCount,
-          std::move(bucketFunction),
-          getPartitionChannels(insertTableHandle),
-          nullptr) {}
-
-HiveDataSink::HiveDataSink(
-    RowTypePtr inputType,
-    std::shared_ptr<const HiveInsertTableHandle> insertTableHandle,
-    const ConnectorQueryCtx* connectorQueryCtx,
-    CommitStrategy commitStrategy,
-    const std::shared_ptr<const HiveConfig>& hiveConfig,
-    uint32_t bucketCount,
-    std::unique_ptr<core::PartitionFunction> bucketFunction,
-    const std::vector<column_index_t>& partitionChannels,
-    std::unique_ptr<PartitionIdGenerator> partitionIdGenerator)
     : inputType_(std::move(inputType)),
       insertTableHandle_(std::move(insertTableHandle)),
       connectorQueryCtx_(connectorQueryCtx),
@@ -424,15 +400,16 @@ HiveDataSink::HiveDataSink(
       updateMode_(getUpdateMode()),
       maxOpenWriters_(hiveConfig_->maxPartitionsPerWriters(
           connectorQueryCtx->sessionProperties())),
-      partitionChannels_(partitionChannels),
+      partitionChannels_(getPartitionChannels(insertTableHandle_)),
       partitionIdGenerator_(
-          partitionIdGenerator ? std::move(partitionIdGenerator)
-              : !partitionChannels_.empty()
+          !partitionChannels_.empty()
               ? std::make_unique<PartitionIdGenerator>(
                     inputType_,
                     partitionChannels_,
                     maxOpenWriters_,
-                    connectorQueryCtx_->memoryPool())
+                    connectorQueryCtx_->memoryPool(),
+                    hiveConfig_->isPartitionPathAsLowerCase(
+                        connectorQueryCtx->sessionProperties()))
               : nullptr),
       dataChannels_(
           getNonPartitionChannels(partitionChannels_, inputType_->size())),
@@ -444,8 +421,6 @@ HiveDataSink::HiveDataSink(
       sortWriterFinishTimeSliceLimitMs_(getFinishTimeSliceLimitMsFromHiveConfig(
           hiveConfig_,
           connectorQueryCtx->sessionProperties())),
-      partitionKeyAsLowerCase_(hiveConfig_->isPartitionPathAsLowerCase(
-          connectorQueryCtx_->sessionProperties())),
       fileNameGenerator_(insertTableHandle_->fileNameGenerator()) {
   fileSystemStats_ = std::make_unique<filesystems::File::IoStats>();
   if (isBucketed()) {
@@ -782,7 +757,8 @@ uint32_t HiveDataSink::appendWriter(const HiveWriterId& id) {
 
   std::optional<std::string> partitionName;
   if (isPartitioned()) {
-    partitionName = getPartitionName(id.partitionId.value());
+    partitionName =
+        partitionIdGenerator_->partitionName(id.partitionId.value());
   }
 
   // Without explicitly setting flush policy, the default memory based flush
@@ -888,15 +864,6 @@ uint32_t HiveDataSink::appendWriter(const HiveWriterId& id) {
   return writerIndexMap_[id];
 }
 
-std::string HiveDataSink::getPartitionName(uint32_t partitionId) const {
-  VELOX_CHECK_NOT_NULL(partitionIdGenerator_);
-
-  return HivePartitionName::partitionName(
-      partitionId,
-      partitionIdGenerator_->partitionValues(),
-      partitionKeyAsLowerCase_);
-}
-
 std::unique_ptr<facebook::velox::dwio::common::Writer>
 HiveDataSink::maybeCreateBucketSortWriter(
     std::unique_ptr<facebook::velox::dwio::common::Writer> writer) {

velox/connectors/hive/HiveDataSink.h

@@ -18,7 +18,6 @@
 #include "velox/common/compression/Compression.h"
 #include "velox/connectors/Connector.h"
 #include "velox/connectors/hive/HiveConfig.h"
-#include "velox/connectors/hive/HivePartitionName.h"
 #include "velox/connectors/hive/PartitionIdGenerator.h"
 #include "velox/connectors/hive/TableHandle.h"
 #include "velox/dwio/common/Options.h"
@@ -517,17 +516,6 @@ class HiveDataSink : public DataSink {
   };
   static std::string stateString(State state);
 
-  /// Creates a HiveDataSink for writing data to Hive table files.
-  ///
-  /// @param inputType The schema of input data rows to be written.
-  /// @param insertTableHandle Metadata about the table write operation,
-  /// including storage format, compression, bucketing, and partitioning
-  /// configuration.
-  /// @param connectorQueryCtx Query context with session properties, memory
-  /// pools, and spill configuration.
-  /// @param commitStrategy Strategy for committing written data (kNoCommit or
-  /// kTaskCommit).
-  /// @param hiveConfig Hive connector configuration.
   HiveDataSink(
       RowTypePtr inputType,
       std::shared_ptr<const HiveInsertTableHandle> insertTableHandle,
@@ -544,40 +532,6 @@ class HiveDataSink : public DataSink {
       uint32_t bucketCount,
       std::unique_ptr<core::PartitionFunction> bucketFunction);
 
-  /// Constructor with explicit bucketing and partitioning parameters.
-  ///
-  /// @param inputType The schema of input data rows to be written.
-  /// @param insertTableHandle Metadata about the table write operation,
-  /// including storage format, compression, location, and serialization
-  /// parameters.
-  /// @param connectorQueryCtx Query context with session properties, memory
-  /// pools, and spill configuration.
-  /// @param commitStrategy Strategy for committing written data (kNoCommit or
-  /// kTaskCommit). Determines whether temporary files need to be renamed on
-  /// commit.
-  /// @param hiveConfig Hive connector configuration with settings for max
-  /// partitions, bucketing limits etc.
-  /// @param bucketCount Number of buckets for bucketed tables (0 if not
-  /// bucketed). Must be less than the configured max bucket count.
-  /// @param bucketFunction Function to compute bucket IDs from row data
-  /// (nullptr if not bucketed). Used to distribute rows across buckets.
-  /// @param partitionChannels Column indices used for partitioning (empty if
-  /// not partitioned). These columns are extracted to determine partition
-  /// directories.
-  /// @param partitionIdGenerator Generates partition IDs from partition column
-  /// values (nullptr if not partitioned). Compute partition key combinations to
-  /// unique IDs.
-  HiveDataSink(
-      RowTypePtr inputType,
-      std::shared_ptr<const HiveInsertTableHandle> insertTableHandle,
-      const ConnectorQueryCtx* connectorQueryCtx,
-      CommitStrategy commitStrategy,
-      const std::shared_ptr<const HiveConfig>& hiveConfig,
-      uint32_t bucketCount,
-      std::unique_ptr<core::PartitionFunction> bucketFunction,
-      const std::vector<column_index_t>& partitionChannels,
-      std::unique_ptr<PartitionIdGenerator> partitionIdGenerator);
-
   void appendData(RowVectorPtr input) override;
 
   bool finish() override;
@@ -667,7 +621,7 @@ class HiveDataSink : public DataSink {
       io::IoStatistics* ioStats);
 
   // Compute the partition id and bucket id for each row in 'input'.
-  virtual void computePartitionAndBucketIds(const RowVectorPtr& input);
+  void computePartitionAndBucketIds(const RowVectorPtr& input);
 
   // Get the HiveWriter corresponding to the row
   // from partitionIds and bucketIds.
@@ -687,12 +641,6 @@ class HiveDataSink : public DataSink {
   // the newly created writer in 'writers_'.
   uint32_t appendWriter(const HiveWriterId& id);
 
-  // Returns the Hive partition directory name for the given partition ID.
-  // Converts the partition values associated with the partition ID into a
-  // Hive-formatted directory path. Returns std::nullopt if the table is
-  // unpartitioned. Should be called only when writing to a partitioned table.
-  virtual std::string getPartitionName(uint32_t partitionId) const;
-
   std::unique_ptr<facebook::velox::dwio::common::Writer>
   maybeCreateBucketSortWriter(
       std::unique_ptr<facebook::velox::dwio::common::Writer> writer);
@@ -747,7 +695,6 @@ class HiveDataSink : public DataSink {
   const std::shared_ptr<dwio::common::WriterFactory> writerFactory_;
   const common::SpillConfig* const spillConfig_;
   const uint64_t sortWriterFinishTimeSliceLimitMs_{0};
-  const bool partitionKeyAsLowerCase_;
 
   std::vector<column_index_t> sortColumnIndices_;
   std::vector<CompareFlags> sortCompareFlags_;