Make DataFileSet preserve insertion order for v3 row ID assignment

Change DataFileSet from std::unordered_set to a custom class that preserves
insertion order, similar to Java's DataFileSet which uses LinkedHashSet.
This is important for row ID assignment in v3 manifests, where row IDs
are assigned based on the order files are written.

The implementation uses both a vector (for insertion order) and an
unordered_set (for O(1) duplicate detection) to maintain the same API
while preserving order.

Author: zhjwpku

src/iceberg/util/content_file_util.h

@@ -22,12 +22,14 @@
 /// \file iceberg/util/content_file_util.h
 /// Utility functions for content files (data files and delete files).
 
+#include <algorithm>
 #include <functional>
 #include <memory>
 #include <optional>
 #include <span>
 #include <string>
 #include <unordered_set>
+#include <vector>
 
 #include "iceberg/iceberg_export.h"
 #include "iceberg/manifest/manifest_entry.h"
@@ -61,8 +63,87 @@ struct ICEBERG_EXPORT DataFilePtrEqual {
 };
 
 /// \brief A set of DataFile pointers, deduplicated by file path.
-using DataFileSet =
-    std::unordered_set<std::shared_ptr<DataFile>, DataFilePtrHash, DataFilePtrEqual>;
+///
+/// This preserves insertion order, which is important for row ID assignment in v3
+/// manifests. Similar to Java's DataFileSet which uses LinkedHashSet to maintain
+/// insertion order.
+class ICEBERG_EXPORT DataFileSet {
+ public:
+  using value_type = std::shared_ptr<DataFile>;
+  using iterator = typename std::vector<value_type>::iterator;
+  using const_iterator = typename std::vector<value_type>::const_iterator;
+
+  DataFileSet() = default;
+
+  /// \brief Insert a data file into the set.
+  /// \param file The data file to insert
+  /// \return A pair with an iterator to the inserted element (or the existing one) and
+  ///         a bool indicating whether insertion took place
+  std::pair<iterator, bool> insert(const value_type& file) {
+    if (!file) {
+      return {elements_.end(), false};
+    }
+    // Check if file already exists using the hash set for O(1) lookup
+    auto [hash_iter, hash_inserted] = hash_set_.insert(file);
+    if (!hash_inserted) {
+      // File already exists, find it in the vector using the element from hash_set_
+      const auto& existing_file = *hash_iter;
+      auto vec_iter = std::ranges::find_if(elements_, [&existing_file](const auto& elem) {
+        return DataFilePtrEqual{}(elem, existing_file);
+      });
+      return {vec_iter, false};
+    }
+    elements_.push_back(*hash_iter);
+    return {std::prev(elements_.end()), true};
+  }
+
+  /// \brief Insert a data file into the set (move version).
+  std::pair<iterator, bool> insert(value_type&& file) {
+    if (!file) {
+      return {elements_.end(), false};
+    }
+    // Check if file already exists
+    auto [hash_iter, hash_inserted] = hash_set_.insert(file);
+    if (!hash_inserted) {
+      // File already exists, find it in the vector using the element from hash_set_
+      const auto& existing_file = *hash_iter;
+      auto vec_iter = std::ranges::find_if(elements_, [&existing_file](const auto& elem) {
+        return DataFilePtrEqual{}(elem, existing_file);
+      });
+      return {vec_iter, false};
+    }
+    elements_.push_back(*hash_iter);
+    return {std::prev(elements_.end()), true};
+  }
+
+  /// \brief Get the number of elements in the set.
+  size_t size() const { return elements_.size(); }
+
+  /// \brief Check if the set is empty.
+  bool empty() const { return elements_.empty(); }
+
+  /// \brief Clear all elements from the set.
+  void clear() {
+    elements_.clear();
+    hash_set_.clear();
+  }
+
+  /// \brief Get iterator to the beginning.
+  iterator begin() { return elements_.begin(); }
+  const_iterator begin() const { return elements_.begin(); }
+  const_iterator cbegin() const { return elements_.cbegin(); }
+
+  /// \brief Get iterator to the end.
+  iterator end() { return elements_.end(); }
+  const_iterator end() const { return elements_.end(); }
+  const_iterator cend() const { return elements_.cend(); }
+
+ private:
+  // Vector to preserve insertion order
+  std::vector<value_type> elements_;
+  // Hash set for O(1) duplicate detection
+  std::unordered_set<value_type, DataFilePtrHash, DataFilePtrEqual> hash_set_;
+};
 
 /// \brief Utility functions for content files.
 struct ICEBERG_EXPORT ContentFileUtil {