Function ParseJSONFields

cpp/src/arrow/json/parser_benchmark.cc:193–219

}

static void ParseJSONFields(benchmark::State& state) {  // NOLINT non-const reference
  const bool ordered = !!state.range(0);
  const bool with_schema = !!state.range(1);
  const double sparsity = state.range(2) / 100.0;
  const auto num_fields = static_cast<int>(state.range(3));

  // This would generate approximately 400 kB of JSON data
  int32_t num_rows = static_cast<int32_t>(2e4 / (1.0 - sparsity) / num_fields);
  // ... however, we want enough rows to make setup/finish overhead negligible
  num_rows = std::max<int32_t>(num_rows, 200);
  // In the end, we will empirically generate between 400 kB and 4 MB of JSON data.

  auto fields = GenerateTestFields(num_fields, 10);

  auto parse_options = ParseOptions::Defaults();
  if (with_schema) {
    parse_options.explicit_schema = schema(fields);
    parse_options.unexpected_field_behavior = UnexpectedFieldBehavior::Error;
  }

  auto gen_options = GenerateOptions::Defaults();
  gen_options.field_probability = 1.0 - sparsity;
  gen_options.randomize_field_order = !ordered;

  auto json = GenerateTestData(fields, num_rows, gen_options);
  BenchmarkJSONParsing(state, std::make_shared<Buffer>(json), parse_options);
}

BENCHMARK(ChunkJSONPrettyPrinted);
BENCHMARK(ChunkJSONLineDelimited);