Function Batches

cpp/src/arrow/csv/generate_fuzz_corpus.cc:78–153

}

Result<RecordBatchVector> Batches() {
  ::arrow::random::RandomArrayGenerator gen(/*seed=*/42);
  RecordBatchVector batches;

  auto append_batch = [&](auto array_factory, int64_t length) -> Status {
    ARROW_ASSIGN_OR_RAISE(auto batch, MakeBatch(array_factory, length));
    batches.push_back(batch);
    return Status::OK();
  };

  // Ideally, we should exercise all possible inference kinds (see inference_internal.h)
  auto make_nulls = [&](int64_t length, double null_probability) {
    return MakeArrayOfNull(null(), length);
  };
  auto make_ints = [&](int64_t length, double null_probability) {
    return gen.Int64(length, /*min=*/-1'000'000, /*max=*/1'000'000, null_probability);
  };
  auto make_floats = [&](int64_t length, double null_probability) {
    return gen.Float64(length, /*min=*/-100.0, /*max=*/100.0, null_probability);
  };
  auto make_booleans = [&](int64_t length, double null_probability) {
    return gen.Boolean(length, /*true_probability=*/0.8, null_probability);
  };
  auto make_dates = [&](int64_t length, double null_probability) {
    return gen.Date64(length, /*min=*/1, /*max=*/365 * 60, null_probability);
  };
  auto make_times = [&](int64_t length, double null_probability) {
    return gen.Int32(length, /*min=*/0, /*max=*/86399, null_probability)
        ->View(time32(TimeUnit::SECOND));
  };

  std::string timezone;
  auto make_timestamps = [&](int64_t length, double null_probability) {
    return gen.Int64(length, /*min=*/1, /*max=*/1764079190, null_probability)
        ->View(timestamp(TimeUnit::SECOND, timezone));
  };
  auto make_timestamps_ns = [&](int64_t length, double null_probability) {
    return gen
        .Int64(length, /*min=*/1, /*max=*/1764079190LL * 1'000'000'000, null_probability)
        ->View(timestamp(TimeUnit::NANO, timezone));
  };

  auto make_strings = [&](int64_t length, double null_probability) {
    return gen.String(length, /*min_length=*/3, /*max_length=*/15, null_probability);
  };
  auto make_string_with_repeats = [&](int64_t length, double null_probability) {
    // `unique` should be less than `auto_dict_max_cardinality` in fuzz target
    return gen.StringWithRepeats(length, /*unique=*/10, /*min_length=*/3,
                                 /*max_length=*/15, null_probability);
  };

  RETURN_NOT_OK(append_batch(make_nulls, /*length=*/2000));
  RETURN_NOT_OK(append_batch(make_ints, /*length=*/500));
  RETURN_NOT_OK(append_batch(make_floats, /*length=*/150));
  RETURN_NOT_OK(append_batch(make_booleans, /*length=*/500));

  RETURN_NOT_OK(append_batch(make_dates, /*length=*/200));
  RETURN_NOT_OK(append_batch(make_times, /*length=*/400));