feat: replace sort order and refactor transaction / pending update

Author: HeartLinked

src/iceberg/CMakeLists.txt

@@ -77,6 +77,7 @@ set(ICEBERG_SOURCES
     transform_function.cc
     type.cc
     update/pending_update.cc
+    update/replace-sort-order.cc
     update/update_properties.cc
     util/bucket_util.cc
     util/conversions.cc

src/iceberg/expression/term.h

@@ -151,6 +151,8 @@ class ICEBERG_EXPORT BoundReference
 
   std::shared_ptr<Type> type() const override { return field_.type(); }
 
+  int32_t field_id() const { return field_.field_id(); }
+
   bool MayProduceNull() const override { return field_.optional(); }
 
   bool Equals(const BoundTerm& other) const override;

src/iceberg/meson.build

@@ -99,6 +99,7 @@ iceberg_sources = files(
     'transform_function.cc',
     'type.cc',
     'update/pending_update.cc',
+    'update/replace_sort_order.cc',
     'update/update_properties.cc',
     'util/bucket_util.cc',
     'util/conversions.cc',

src/iceberg/sort_order.cc

@@ -111,7 +111,9 @@ Result<std::unique_ptr<SortOrder>> SortOrder::Make(const Schema& schema, int32_t
   }
 
   if (fields.empty() && sort_id != kUnsortedOrderId) [[unlikely]] {
-    return InvalidArgument("Sort order must have at least one sort field");
+    return InvalidArgument(
+        "Sorted order must have at least one sort field. Use SortOrder::Unsorted() to "
+        "create an unsorted order");
   }
 
   auto sort_order = std::unique_ptr<SortOrder>(new SortOrder(sort_id, std::move(fields)));

src/iceberg/table.cc

@@ -154,6 +154,13 @@ Result<std::shared_ptr<UpdateProperties>> Table::NewUpdateProperties() {
   return transaction->NewUpdateProperties();
 }
 
+Result<std::shared_ptr<ReplaceSortOrder>> Table::NewReplaceSortOrder() {
+  ICEBERG_ASSIGN_OR_RAISE(
+      auto transaction, Transaction::Make(shared_from_this(), Transaction::Kind::kUpdate,
+                                          /*auto_commit=*/true));
+  return transaction->NewReplaceSortOrder();
+}
+
 Result<std::shared_ptr<StagedTable>> StagedTable::Make(
     TableIdentifier identifier, std::shared_ptr<TableMetadata> metadata,
     std::string metadata_location, std::shared_ptr<FileIO> io,

src/iceberg/table.h

@@ -132,6 +132,10 @@ class ICEBERG_EXPORT Table : public std::enable_shared_from_this<Table> {
   /// changes.
   virtual Result<std::shared_ptr<UpdateProperties>> NewUpdateProperties();
 
+  /// \brief Create a new ReplaceSortOrder to replace the table sort order and commit the
+  /// changes.
+  virtual Result<std::shared_ptr<ReplaceSortOrder>> NewReplaceSortOrder();
+
  protected:
   Table(TableIdentifier identifier, std::shared_ptr<TableMetadata> metadata,
         std::string metadata_location, std::shared_ptr<FileIO> io,

src/iceberg/table_metadata.cc

@@ -721,7 +721,7 @@ TableMetadataBuilder& TableMetadataBuilder::SetProperties(
 }
 
 TableMetadataBuilder& TableMetadataBuilder::RemoveProperties(
-    const std::vector<std::string>& removed) {
+    const std::unordered_set<std::string>& removed) {
   // If removed is empty, return early (no-op)
   if (removed.empty()) {
     return *this;

src/iceberg/table_metadata.h

@@ -394,7 +394,7 @@ class ICEBERG_EXPORT TableMetadataBuilder : public ErrorCollector {
   ///
   /// \param removed Set of property keys to remove
   /// \return Reference to this builder for method chaining
-  TableMetadataBuilder& RemoveProperties(const std::vector<std::string>& removed);
+  TableMetadataBuilder& RemoveProperties(const std::unordered_set<std::string>& removed);
 
   /// \brief Set the table location
   ///

src/iceberg/table_update.h

@@ -26,6 +26,7 @@
 #include <optional>
 #include <string>
 #include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 #include "iceberg/iceberg_export.h"
@@ -379,10 +380,10 @@ class ICEBERG_EXPORT SetProperties : public TableUpdate {
 /// \brief Represents removing table properties
 class ICEBERG_EXPORT RemoveProperties : public TableUpdate {
  public:
-  explicit RemoveProperties(std::vector<std::string> removed)
+  explicit RemoveProperties(std::unordered_set<std::string> removed)
       : removed_(std::move(removed)) {}
 
-  const std::vector<std::string>& removed() const { return removed_; }
+  const std::unordered_set<std::string>& removed() const { return removed_; }
 
   void ApplyTo(TableMetadataBuilder& builder) const override;
 
@@ -391,7 +392,7 @@ class ICEBERG_EXPORT RemoveProperties : public TableUpdate {
   Kind kind() const override { return Kind::kRemoveProperties; }
 
  private:
-  std::vector<std::string> removed_;
+  std::unordered_set<std::string> removed_;
 };
 
 /// \brief Represents setting the table location

src/iceberg/test/CMakeLists.txt

@@ -153,7 +153,8 @@ if(ICEBERG_BUILD_BUNDLE)
                    USE_BUNDLE
                    SOURCES
                    transaction_test.cc
-                   update_properties_test.cc)
+                   update_properties_test.cc
+                   replace_sort_order_test.cc)
 
 endif()