Function SampleData

cpp/src/parquet/arrow/generate_encoding_fuzz_corpus.cc:90–134

};

Result<std::vector<DataWithEncodings>> SampleData() {
  auto rag = random::RandomArrayGenerator(/*seed=*/42);
  std::vector<DataWithEncodings> all_data;

  auto push_array = [&](std::shared_ptr<Array> arr, EncodingVector encodings = {}) {
    auto pq_type = FullParquetType::FromArrow(*arr->type());
    if (encodings.empty()) {
      encodings = ::parquet::SupportedEncodings(pq_type.type);
    }
    all_data.emplace_back(pq_type, arr, std::move(encodings));
  };

  for (const auto size : {100, 10000}) {
    push_array(rag.Int32(size, /*min=*/0, /*max=*/10));
    push_array(rag.Int32(size, /*min=*/std::numeric_limits<int32_t>::min(),
                         /*max=*/std::numeric_limits<int32_t>::max()));
    push_array(rag.Int64(size, /*min=*/0, /*max=*/0));
    push_array(rag.Int64(size, /*min=*/0, /*max=*/10'000));
    push_array(rag.Int64(size, /*min=*/std::numeric_limits<int64_t>::min(),
                         /*max=*/std::numeric_limits<int64_t>::max()));
    for (const double true_probability : {0.01, 0.5, 0.9}) {
      push_array(rag.Boolean(size, true_probability));
    }
    push_array(rag.Float32(size, /*min=*/-1.0f, /*max=*/1.0f));
    push_array(rag.Float32(size, /*min=*/std::numeric_limits<float>::lowest(),
                           /*max=*/std::numeric_limits<float>::max(),
                           /*null_probability=*/0.0, /*nan_probability=*/1e-2));
    push_array(rag.Float64(size, /*min=*/-1.0, /*max=*/1.0));
    push_array(rag.Float64(size, /*min=*/std::numeric_limits<double>::lowest(),
                           /*max=*/std::numeric_limits<double>::max(),
                           /*null_probability=*/0.0, /*nan_probability=*/1e-2));
    for (const int32_t byte_width : {1, 3, 16}) {
      const auto fsb_size = size / byte_width;
      push_array(rag.FixedSizeBinary(fsb_size, byte_width));
    }
    for (const int32_t max_binary_length : {1, 30}) {
      const auto binary_size = size / max_binary_length;
      push_array(rag.BinaryWithRepeats(binary_size, /*unique=*/binary_size / 2,
                                       /*min_length=*/0,
                                       /*max_length=*/max_binary_length));
    }
  }

  return all_data;
}

using Int96Vector = std::vector<Int96>;