packed row group range read

Signed-off-by: shaoting-huang <[email protected]>
shaoting-huang · Dec 27, 2024 · 8b9f281 · 8b9f281
1 parent b968ad4
commit 8b9f281
Showing 1 changed file with 11 additions and 11 deletions.
diff --git a/cpp/include/milvus-storage/packed/reader.h b/cpp/include/milvus-storage/packed/reader.h
@@ -41,36 +41,28 @@ using RowOffsetMinHeap =
 
 class PackedRecordBatchReader : public arrow::RecordBatchReader {
   public:
-  // Test only
+  // Read a file with default buffer size
   PackedRecordBatchReader(arrow::fs::FileSystem& fs,
                           const std::string& path,
                           const std::shared_ptr<arrow::Schema> schema,
                           const int64_t buffer_size = DEFAULT_READ_BUFFER_SIZE);
 
-  // Test only
+  // Read a range of row groups from a file with default buffer size
   PackedRecordBatchReader(arrow::fs::FileSystem& fs,
                           const std::string& path,
                           const std::shared_ptr<arrow::Schema> schema,
                           const size_t start_row_group,
                           const size_t end_row_group,
                           const int64_t buffer_size = DEFAULT_READ_BUFFER_SIZE);
 
+  // Read all files with default buffer size
   PackedRecordBatchReader(arrow::fs::FileSystem& fs,
                           const std::vector<std::string>& paths,
                           const std::shared_ptr<arrow::Schema> schema,
                           const std::vector<ColumnOffset>& column_offsets,
                           const std::set<int>& needed_columns,
                           const int64_t buffer_size = DEFAULT_READ_BUFFER_SIZE);
 
-  PackedRecordBatchReader(arrow::fs::FileSystem& fs,
-                          const std::vector<std::string>& paths,
-                          const std::shared_ptr<arrow::Schema> schema,
-                          const std::vector<ColumnOffset>& column_offsets,
-                          const std::set<int>& needed_columns,
-                          const std::vector<size_t>& start_row_groups,
-                          const std::vector<size_t>& end_row_groups,
-                          const int64_t buffer_size = DEFAULT_READ_BUFFER_SIZE);
-
   std::shared_ptr<arrow::Schema> schema() const override;
 
   arrow::Status ReadNext(std::shared_ptr<arrow::RecordBatch>* batch) override;
@@ -80,6 +72,14 @@ class PackedRecordBatchReader : public arrow::RecordBatchReader {
   arrow::Status Close() override;
 
   private:
+  PackedRecordBatchReader(arrow::fs::FileSystem& fs,
+                          const std::vector<std::string>& paths,
+                          const std::shared_ptr<arrow::Schema> schema,
+                          const std::vector<ColumnOffset>& column_offsets,
+                          const std::set<int>& needed_columns,
+                          const std::vector<size_t>& start_row_groups,
+                          const std::vector<size_t>& end_row_groups,
+                          const int64_t buffer_size = DEFAULT_READ_BUFFER_SIZE);
   // Advance buffer to fill the expected buffer size
   arrow::Status advanceBuffer();
   std::vector<const arrow::Array*> collectChunks(int64_t chunksize) const;