used std::map instead of std::unordered_map to store num_data_pages

Author: omega-bigstream

cpp/src/parquet/column_writer.cc

@@ -20,9 +20,9 @@
 #include <algorithm>
 #include <cstdint>
 #include <cstring>
+#include <map>
 #include <memory>
 #include <string>
-#include <unordered_map>
 #include <utility>
 #include <vector>
 
@@ -406,7 +406,7 @@ class SerializedPageWriter : public PageWriter {
   std::shared_ptr<ResizableBuffer> encryption_buffer_;
 
   int32_t num_dict_pages_;
-  std::unordered_map<Encoding::type, int32_t> num_data_pages_;
+  std::map<Encoding::type, int32_t> num_data_pages_;
 };
 
 // This implementation of the PageWriter writes to the final sink on Close .

cpp/src/parquet/metadata.cc

@@ -983,7 +983,7 @@ class ColumnChunkMetaDataBuilder::ColumnChunkMetaDataBuilderImpl {
               int64_t index_page_offset, int64_t data_page_offset,
               int64_t compressed_size, int64_t uncompressed_size, bool has_dictionary,
               bool dictionary_fallback, int32_t num_dict_pages,
-              std::unordered_map<Encoding::type, int32_t>& num_data_pages,
+              std::map<Encoding::type, int32_t>& num_data_pages,
               const std::shared_ptr<Encryptor>& encryptor) {
     if (dictionary_page_offset > 0) {
       column_chunk_->meta_data.__set_dictionary_page_offset(dictionary_page_offset);
@@ -1174,12 +1174,14 @@ void ColumnChunkMetaDataBuilder::set_file_path(const std::string& path) {
   impl_->set_file_path(path);
 }
 
-void ColumnChunkMetaDataBuilder::Finish(
-    int64_t num_values, int64_t dictionary_page_offset, int64_t index_page_offset,
-    int64_t data_page_offset, int64_t compressed_size, int64_t uncompressed_size,
-    bool has_dictionary, bool dictionary_fallback, int32_t num_dict_pages,
-    std::unordered_map<Encoding::type, int32_t> num_data_pages,
-    const std::shared_ptr<Encryptor>& encryptor) {
+void ColumnChunkMetaDataBuilder::Finish(int64_t num_values,
+                                        int64_t dictionary_page_offset,
+                                        int64_t index_page_offset,
+                                        int64_t data_page_offset, int64_t compressed_size,
+                                        int64_t uncompressed_size, bool has_dictionary,
+                                        bool dictionary_fallback, int32_t num_dict_pages,
+                                        std::map<Encoding::type, int32_t> num_data_pages,
+                                        const std::shared_ptr<Encryptor>& encryptor) {
   impl_->Finish(num_values, dictionary_page_offset, index_page_offset, data_page_offset,
                 compressed_size, uncompressed_size, has_dictionary, dictionary_fallback,
                 num_dict_pages, num_data_pages, encryptor);

cpp/src/parquet/metadata.h

@@ -22,7 +22,6 @@
 #include <map>
 #include <memory>
 #include <string>
-#include <unordered_map>
 #include <utility>
 #include <vector>
 
@@ -324,7 +323,7 @@ class PARQUET_EXPORT ColumnChunkMetaDataBuilder {
               int64_t index_page_offset, int64_t data_page_offset,
               int64_t compressed_size, int64_t uncompressed_size, bool has_dictionary,
               bool dictionary_fallback, int32_t num_dict_pages,
-              std::unordered_map<Encoding::type, int32_t>,
+              std::map<Encoding::type, int32_t>,
               const std::shared_ptr<Encryptor>& encryptor = NULLPTR);
 
   // The metadata contents, suitable for passing to ColumnChunkMetaData::Make

cpp/src/parquet/metadata_test.cc

@@ -40,8 +40,8 @@ std::unique_ptr<parquet::FileMetaData> GenerateTableMetaData(
   auto col1_builder = rg1_builder->NextColumnChunk();
   auto col2_builder = rg1_builder->NextColumnChunk();
   // column metadata
-  std::unordered_map<Encoding::type, int32_t> encoding_stats = {
-      {Encoding::RLE_DICTIONARY, 1}, {Encoding::PLAIN, 1}, {Encoding::RLE, 1}};
+  std::map<Encoding::type, int32_t> encoding_stats(
+      {{Encoding::RLE_DICTIONARY, 1}, {Encoding::PLAIN, 1}, {Encoding::RLE, 1}});
   stats_int.set_is_signed(true);
   col1_builder->SetStatistics(stats_int);
   stats_float.set_is_signed(true);