Function TEST

cpp/src/parquet/page_index_test.cc:33–78

namespace parquet {

TEST(PageIndex, ReadOffsetIndex) {
  std::string dir_string(parquet::test::get_data_dir());
  std::string path = dir_string + "/alltypes_tiny_pages.parquet";
  auto reader = ParquetFileReader::OpenFile(path, false);
  auto file_metadata = reader->metadata();

  // Get offset index location to column 0 of row group 0.
  const int row_group_id = 0;
  const int column_id = 0;
  ASSERT_LT(row_group_id, file_metadata->num_row_groups());
  ASSERT_LT(column_id, file_metadata->num_columns());
  auto index_location = file_metadata->RowGroup(row_group_id)
                            ->ColumnChunk(column_id)
                            ->GetOffsetIndexLocation();
  ASSERT_TRUE(index_location.has_value());

  // Read serialized offset index from the file.
  std::shared_ptr<::arrow::io::RandomAccessFile> source;
  PARQUET_ASSIGN_OR_THROW(source, ::arrow::io::ReadableFile::Open(path));
  PARQUET_ASSIGN_OR_THROW(auto buffer,
                          source->ReadAt(index_location->offset, index_location->length));
  PARQUET_THROW_NOT_OK(source->Close());

  // Deserialize offset index.
  auto properties = default_reader_properties();
  std::unique_ptr<OffsetIndex> offset_index = OffsetIndex::Make(
      buffer->data(), static_cast<uint32_t>(buffer->size()), properties);

  // Verify only partial data as it contains 325 pages in total.
  const size_t num_pages = 325;
  const std::vector<size_t> page_indices = {0, 100, 200, 300};
  const std::vector<PageLocation> page_locations = {
      PageLocation{4, 109, 0}, PageLocation{11480, 133, 2244},
      PageLocation{22980, 133, 4494}, PageLocation{34480, 133, 6744}};

  ASSERT_EQ(num_pages, offset_index->page_locations().size());
  for (size_t i = 0; i < page_indices.size(); ++i) {
    size_t page_id = page_indices.at(i);
    const auto& read_page_location = offset_index->page_locations().at(page_id);
    const auto& expected_page_location = page_locations.at(i);
    ASSERT_EQ(expected_page_location.offset, read_page_location.offset);
    ASSERT_EQ(expected_page_location.compressed_page_size,
              read_page_location.compressed_page_size);
    ASSERT_EQ(expected_page_location.first_row_index, read_page_location.first_row_index);
  }
}

template <typename DType, typename T = typename DType::c_type>
void TestReadTypedColumnIndex(const std::string& file_name, int column_id,