Function GetReadRecordBatchReadRanges

cpp/src/arrow/ipc/read_write_test.cc:3022–3095

}

void GetReadRecordBatchReadRanges(
    uint32_t num_rows, const std::vector<int>& included_fields,
    const std::vector<int64_t>& expected_body_read_lengths) {
  auto buffer = MakeBooleanInt32Int64File(num_rows, /*num_batches=*/1);

  io::BufferReader buffer_reader(buffer);
  std::unique_ptr<io::TrackedRandomAccessFile> tracked =
      io::TrackedRandomAccessFile::Make(&buffer_reader);

  auto read_options = IpcReadOptions::Defaults();
  // if empty, return all fields
  read_options.included_fields = included_fields;
  ASSERT_OK_AND_ASSIGN(auto reader,
                       RecordBatchFileReader::Open(tracked.get(), read_options));
  ASSERT_OK_AND_ASSIGN(auto out_batch, reader->ReadRecordBatch(0));

  ASSERT_EQ(out_batch->num_rows(), num_rows);
  ASSERT_EQ(out_batch->num_columns(),
            included_fields.empty() ? 3 : included_fields.size());

  auto read_ranges = tracked->get_read_ranges();

  const int32_t magic_size = static_cast<int>(ipc::internal::kArrowMagicBytes.size());
  // read magic and footer length IO
  auto file_end_size = magic_size + sizeof(int32_t);
  auto footer_length_offset = buffer->size() - file_end_size;
  auto footer_length = bit_util::FromLittleEndian(
      util::SafeLoadAs<int32_t>(buffer->data() + footer_length_offset));

  // there are at least 2 read IOs before reading body:
  // 1) read magic and footer length IO
  // 2) footer IO
  EXPECT_GE(read_ranges.size(), 2);

  // read magic and footer length IO
  EXPECT_EQ(read_ranges[0].length, file_end_size);
  // read footer IO
  EXPECT_EQ(read_ranges[1].length, footer_length);

  if (included_fields.empty()) {
    // When no fields are explicitly included, the reader optimizes by
    // reading metadata and the entire body in a single IO.
    // Thus, there are exactly 3 read IOs in total:
    // 1) magic and footer length
    // 2) footer
    // 3) record batch metadata + body
    EXPECT_EQ(read_ranges.size(), 3);

    int64_t total_body = 0;
    for (auto len : expected_body_read_lengths) total_body += len;

    // In the optimized path (included_fields is empty), the 3rd read operation
    // fetches both the message metadata (flatbuffer) and the entire message body
    // in one contiguous block. Therefore, its length must at least exceed the
    // total body length by the size of the metadata.
    EXPECT_GT(read_ranges[2].length, total_body);
    EXPECT_LE(read_ranges[2].length, total_body + footer_length);
  } else {