Row to columnar conversion#

Fixed Schemas#

The following example converts an array of structs to a arrow::Table instance, and then converts it back to the original array of structs.

// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.

#include <arrow/api.h>
#include <arrow/result.h>

#include <cstdint>
#include <iomanip>
#include <iostream>
#include <vector>

using arrow::DoubleBuilder;
using arrow::Int64Builder;
using arrow::ListBuilder;

// While we want to use columnar data structures to build efficient operations, we
// often receive data in a row-wise fashion from other systems. In the following,
// we want give a brief introduction into the classes provided by Apache Arrow by
// showing how to transform row-wise data into a columnar table.
//
// The table contains an id for a product, the number of components in the product
// and the cost of each component.
//
// The data in this example is stored in the following struct:
struct data_row {
  int64_t id;
  int64_t components;
  std::vector<double> component_cost;
};

// Transforming a vector of structs into a columnar Table.
//
// The final representation should be an `arrow::Table` which in turn
// is made up of an `arrow::Schema` and a list of
// `arrow::ChunkedArray` instances. As the first step, we will iterate
// over the data and build up the arrays incrementally.  For this
// task, we provide `arrow::ArrayBuilder` classes that help in the
// construction of the final `arrow::Array` instances.
//
// For each type, Arrow has a specially typed builder class. For the primitive
// values `id` and `components` we can use the `arrow::Int64Builder`. For the
// `component_cost` vector, we need to have two builders, a top-level
// `arrow::ListBuilder` that builds the array of offsets and a nested
// `arrow::DoubleBuilder` that constructs the underlying values array that
// is referenced by the offsets in the former array.
arrow::Result<std::shared_ptr<arrow::Table>> VectorToColumnarTable(
    const std::vector<struct data_row>& rows) {
  // The builders are more efficient using
  // arrow::jemalloc::MemoryPool::default_pool() as this can increase the size of
  // the underlying memory regions in-place. At the moment, arrow::jemalloc is only
  // supported on Unix systems, not Windows.
  arrow::MemoryPool* pool = arrow::default_memory_pool();

  Int64Builder id_builder(pool);
  Int64Builder components_builder(pool);
  ListBuilder component_cost_builder(pool, std::make_shared<DoubleBuilder>(pool));
  // The following builder is owned by component_cost_builder.
  DoubleBuilder* component_item_cost_builder =
      (static_cast<DoubleBuilder*>(component_cost_builder.value_builder()));

  // Now we can loop over our existing data and insert it into the builders. The
  // `Append` calls here may fail (e.g. we cannot allocate enough additional memory).
  // Thus we need to check their return values. For more information on these values,
  // check the documentation about `arrow::Status`.
  for (const data_row& row : rows) {
    ARROW_RETURN_NOT_OK(id_builder.Append(row.id));
    ARROW_RETURN_NOT_OK(components_builder.Append(row.components));

    // Indicate the start of a new list row. This will memorise the current
    // offset in the values builder.
    ARROW_RETURN_NOT_OK(component_cost_builder.Append());
    // Store the actual values. The same memory layout is
    // used for the component cost data, in this case a vector of
    // type double, as for the memory that Arrow uses to hold this
    // data and will be created.
    ARROW_RETURN_NOT_OK(component_item_cost_builder->AppendValues(
        row.component_cost.data(), row.component_cost.size()));
  }

  // At the end, we finalise the arrays, declare the (type) schema and combine them
  // into a single `arrow::Table`:
  std::shared_ptr<arrow::Array> id_array;
  ARROW_RETURN_NOT_OK(id_builder.Finish(&id_array));
  std::shared_ptr<arrow::Array> components_array;
  ARROW_RETURN_NOT_OK(components_builder.Finish(&components_array));
  // No need to invoke component_item_cost_builder.Finish because it is implied by
  // the parent builder's Finish invocation.
  std::shared_ptr<arrow::Array> component_cost_array;
  ARROW_RETURN_NOT_OK(component_cost_builder.Finish(&component_cost_array));

  std::vector<std::shared_ptr<arrow::Field>> schema_vector = {
      arrow::field("id", arrow::int64()), arrow::field("components", arrow::int64()),
      arrow::field("component_cost", arrow::list(arrow::float64()))};

  auto schema = std::make_shared<arrow::Schema>(schema_vector);

  // The final `table` variable is the one we can then pass on to other functions
  // that can consume Apache Arrow memory structures. This object has ownership of
  // all referenced data, thus we don't have to care about undefined references once
  // we leave the scope of the function building the table and its underlying arrays.
  std::shared_ptr<arrow::Table> table =
      arrow::Table::Make(schema, {id_array, components_array, component_cost_array});

  return table;
}

arrow::Result<std::vector<data_row>> ColumnarTableToVector(
    const std::shared_ptr<arrow::Table>& table) {
  // To convert an Arrow table back into the same row-wise representation as in the
  // above section, we first will check that the table conforms to our expected
  // schema and then will build up the vector of rows incrementally.
  //
  // For the check if the table is as expected, we can utilise solely its schema.
  std::vector<std::shared_ptr<arrow::Field>> schema_vector = {
      arrow::field("id", arrow::int64()), arrow::field("components", arrow::int64()),
      arrow::field("component_cost", arrow::list(arrow::float64()))};
  auto expected_schema = std::make_shared<arrow::Schema>(schema_vector);

  if (!expected_schema->Equals(*table->schema())) {
    // The table doesn't have the expected schema thus we cannot directly
    // convert it to our target representation.
    return arrow::Status::Invalid("Schemas are not matching!");
  }

  // As we have ensured that the table has the expected structure, we can unpack the
  // underlying arrays. For the primitive columns `id` and `components` we can use the
  // high level functions to get the values whereas for the nested column
  // `component_costs` we need to access the C-pointer to the data to copy its
  // contents into the resulting `std::vector<double>`. Here we need to be careful to
  // also add the offset to the pointer. This offset is needed to enable zero-copy
  // slicing operations. While this could be adjusted automatically for double
  // arrays, this cannot be done for the accompanying bitmap as often the slicing
  // border would be inside a byte.

  auto ids = std::static_pointer_cast<arrow::Int64Array>(table->column(0)->chunk(0));
  auto components =
      std::static_pointer_cast<arrow::Int64Array>(table->column(1)->chunk(0));
  auto component_cost =
      std::static_pointer_cast<arrow::ListArray>(table->column(2)->chunk(0));
  auto component_cost_values =
      std::static_pointer_cast<arrow::DoubleArray>(component_cost->values());
  // To enable zero-copy slices, the native values pointer might need to account
  // for this slicing offset. This is not needed for the higher level functions
  // like Value(…) that already account for this offset internally.
  const double* ccv_ptr = component_cost_values->raw_values();
  std::vector<data_row> rows;
  for (int64_t i = 0; i < table->num_rows(); i++) {
    // Another simplification in this example is that we assume that there are
    // no null entries, e.g. each row is fill with valid values.
    int64_t id = ids->Value(i);
    int64_t component = components->Value(i);
    const double* first = ccv_ptr + component_cost->value_offset(i);
    const double* last = ccv_ptr + component_cost->value_offset(i + 1);
    std::vector<double> components_vec(first, last);
    rows.push_back({id, component, components_vec});
  }

  return rows;
}

arrow::Status RunRowConversion() {
  std::vector<data_row> original_rows = {
      {1, 1, {10.0}}, {2, 3, {11.0, 12.0, 13.0}}, {3, 2, {15.0, 25.0}}};
  std::shared_ptr<arrow::Table> table;
  std::vector<data_row> converted_rows;

  ARROW_ASSIGN_OR_RAISE(table, VectorToColumnarTable(original_rows));

  ARROW_ASSIGN_OR_RAISE(converted_rows, ColumnarTableToVector(table));

  assert(original_rows.size() == converted_rows.size());

  // Print out contents of table, should get
  // ID Components Component prices
  // 1  1          10
  // 2  3          11  12  13
  // 3  2          15  25
  std::cout << std::left << std::setw(3) << "ID " << std::left << std::setw(11)
            << "Components " << std::left << std::setw(15) << "Component prices "
            << std::endl;
  for (const auto& row : converted_rows) {
    std::cout << std::left << std::setw(3) << row.id << std::left << std::setw(11)
              << row.components;
    for (const auto& cost : row.component_cost) {
      std::cout << std::left << std::setw(4) << cost;
    }
    std::cout << std::endl;
  }
  return arrow::Status::OK();
}

int main(int argc, char** argv) {
  auto status = RunRowConversion();
  if (!status.ok()) {
    std::cerr << status.ToString() << std::endl;
    return EXIT_FAILURE;
  }
  return EXIT_SUCCESS;
}

Dynamic Schemas#

In many cases, we need to convert to and from row data that does not have a schema known at compile time. To help implement these conversions, this library provides several utilities:

The following example shows how to implement conversion between rapidjson::Document and Arrow objects. You can read the full code example at apache/arrow

Writing conversions to Arrow#

To convert rows to Arrow record batches, we’ll setup Array builders for all the columns and then for each field iterate through row values and append to the builders. We assume that we already know the target schema, which may have been provided by another system or was inferred in another function. Inferring the schema during conversion is a challenging proposition; many systems will check the first N rows to infer a schema if there is none already available.

At the top level, we define a function ConvertToRecordBatch:

495arrow::Result<std::shared_ptr<arrow::RecordBatch>> ConvertToRecordBatch(
496    const std::vector<rapidjson::Document>& rows, std::shared_ptr<arrow::Schema> schema) {
497  // RecordBatchBuilder will create array builders for us for each field in our
498  // schema. By passing the number of output rows (`rows.size()`) we can
499  // pre-allocate the correct size of arrays, except of course in the case of
500  // string, byte, and list arrays, which have dynamic lengths.
501  std::unique_ptr<arrow::RecordBatchBuilder> batch_builder;
502  ARROW_ASSIGN_OR_RAISE(
503      batch_builder,
504      arrow::RecordBatchBuilder::Make(schema, arrow::default_memory_pool(), rows.size()));
505
506  // Inner converter will take rows and be responsible for appending values
507  // to provided array builders.
508  JsonValueConverter converter(rows);
509  for (int i = 0; i < batch_builder->num_fields(); ++i) {
510    std::shared_ptr<arrow::Field> field = schema->field(i);
511    arrow::ArrayBuilder* builder = batch_builder->GetField(i);
512    ARROW_RETURN_NOT_OK(converter.Convert(*field.get(), builder));
513  }
514
515  std::shared_ptr<arrow::RecordBatch> batch;
516  ARROW_ASSIGN_OR_RAISE(batch, batch_builder->Flush());
517
518  // Use RecordBatch::ValidateFull() to make sure arrays were correctly constructed.
519  DCHECK_OK(batch->ValidateFull());
520  return batch;
521}  // ConvertToRecordBatch

First we use arrow::RecordBatchBuilder, which conveniently creates builders for each field in the schema. Then we iterate over the fields of the schema, get the builder, and call Convert() on our JsonValueConverter (to be discussed next). At the end, we call batch->ValidateFull(), which checks the integrity of our arrays to make sure the conversion was performed correctly, which is useful for debugging new conversion implementations.

One level down, the JsonValueConverter is responsible for appending row values for the provided field to a provided array builder. In order to specialize logic for each data type, it implements Visit methods and calls arrow::VisitTypeInline(). (See more about type visitors in Visitor Pattern.)

At the end of that class is the private method FieldValues(), which returns an iterator of the column values for the current field across the rows. In row-based structures that are flat (such as a vector of values) this may be trivial to implement. But if the schema is nested, as in the case of JSON documents, a special iterator is needed to navigate the levels of nesting. See the full example for the implementation details of DocValuesIterator.

323class JsonValueConverter {
324 public:
325  explicit JsonValueConverter(const std::vector<rapidjson::Document>& rows)
326      : rows_(rows), array_levels_(0) {}
327
328  JsonValueConverter(const std::vector<rapidjson::Document>& rows,
329                     const std::vector<std::string>& root_path, int64_t array_levels)
330      : rows_(rows), root_path_(root_path), array_levels_(array_levels) {}
331
332  /// \brief For field passed in, append corresponding values to builder
333  arrow::Status Convert(const arrow::Field& field, arrow::ArrayBuilder* builder) {
334    return Convert(field, field.name(), builder);
335  }
336
337  /// \brief For field passed in, append corresponding values to builder
338  arrow::Status Convert(const arrow::Field& field, const std::string& field_name,
339                        arrow::ArrayBuilder* builder) {
340    field_name_ = field_name;
341    builder_ = builder;
342    ARROW_RETURN_NOT_OK(arrow::VisitTypeInline(*field.type().get(), this));
343    return arrow::Status::OK();
344  }
345
346  // Default implementation
347  arrow::Status Visit(const arrow::DataType& type) {
348    return arrow::Status::NotImplemented(
349        "Cannot convert json value to Arrow array of type ", type.ToString());
350  }
351
352  arrow::Status Visit(const arrow::Int64Type& type) {
353    arrow::Int64Builder* builder = static_cast<arrow::Int64Builder*>(builder_);
354    for (const auto& maybe_value : FieldValues()) {
355      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
356      if (value->IsNull()) {
357        ARROW_RETURN_NOT_OK(builder->AppendNull());
358      } else {
359        if (value->IsUint()) {
360          ARROW_RETURN_NOT_OK(builder->Append(value->GetUint()));
361        } else if (value->IsInt()) {
362          ARROW_RETURN_NOT_OK(builder->Append(value->GetInt()));
363        } else if (value->IsUint64()) {
364          ARROW_RETURN_NOT_OK(builder->Append(value->GetUint64()));
365        } else if (value->IsInt64()) {
366          ARROW_RETURN_NOT_OK(builder->Append(value->GetInt64()));
367        } else {
368          return arrow::Status::Invalid("Value is not an integer");
369        }
370      }
371    }
372    return arrow::Status::OK();
373  }
374
375  arrow::Status Visit(const arrow::DoubleType& type) {
376    arrow::DoubleBuilder* builder = static_cast<arrow::DoubleBuilder*>(builder_);
377    for (const auto& maybe_value : FieldValues()) {
378      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
379      if (value->IsNull()) {
380        ARROW_RETURN_NOT_OK(builder->AppendNull());
381      } else {
382        ARROW_RETURN_NOT_OK(builder->Append(value->GetDouble()));
383      }
384    }
385    return arrow::Status::OK();
386  }
387
388  arrow::Status Visit(const arrow::StringType& type) {
389    arrow::StringBuilder* builder = static_cast<arrow::StringBuilder*>(builder_);
390    for (const auto& maybe_value : FieldValues()) {
391      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
392      if (value->IsNull()) {
393        ARROW_RETURN_NOT_OK(builder->AppendNull());
394      } else {
395        ARROW_RETURN_NOT_OK(builder->Append(value->GetString()));
396      }
397    }
398    return arrow::Status::OK();
399  }
400
401  arrow::Status Visit(const arrow::BooleanType& type) {
402    arrow::BooleanBuilder* builder = static_cast<arrow::BooleanBuilder*>(builder_);
403    for (const auto& maybe_value : FieldValues()) {
404      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
405      if (value->IsNull()) {
406        ARROW_RETURN_NOT_OK(builder->AppendNull());
407      } else {
408        ARROW_RETURN_NOT_OK(builder->Append(value->GetBool()));
409      }
410    }
411    return arrow::Status::OK();
412  }
413
414  arrow::Status Visit(const arrow::StructType& type) {
415    arrow::StructBuilder* builder = static_cast<arrow::StructBuilder*>(builder_);
416
417    std::vector<std::string> child_path(root_path_);
418    if (field_name_.size() > 0) {
419      child_path.push_back(field_name_);
420    }
421    auto child_converter = JsonValueConverter(rows_, child_path, array_levels_);
422
423    for (int i = 0; i < type.num_fields(); ++i) {
424      std::shared_ptr<arrow::Field> child_field = type.field(i);
425      std::shared_ptr<arrow::ArrayBuilder> child_builder = builder->child_builder(i);
426
427      ARROW_RETURN_NOT_OK(
428          child_converter.Convert(*child_field.get(), child_builder.get()));
429    }
430
431    // Make null bitmap
432    for (const auto& maybe_value : FieldValues()) {
433      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
434      ARROW_RETURN_NOT_OK(builder->Append(!value->IsNull()));
435    }
436
437    return arrow::Status::OK();
438  }
439
440  arrow::Status Visit(const arrow::ListType& type) {
441    arrow::ListBuilder* builder = static_cast<arrow::ListBuilder*>(builder_);
442
443    // Values and offsets needs to be interleaved in ListBuilder, so first collect the
444    // values
445    std::unique_ptr<arrow::ArrayBuilder> tmp_value_builder;
446    ARROW_ASSIGN_OR_RAISE(tmp_value_builder,
447                          arrow::MakeBuilder(builder->value_builder()->type()));
448    std::vector<std::string> child_path(root_path_);
449    child_path.push_back(field_name_);
450    auto child_converter = JsonValueConverter(rows_, child_path, array_levels_ + 1);
451    ARROW_RETURN_NOT_OK(
452        child_converter.Convert(*type.value_field().get(), "", tmp_value_builder.get()));
453
454    std::shared_ptr<arrow::Array> values_array;
455    ARROW_RETURN_NOT_OK(tmp_value_builder->Finish(&values_array));
456    std::shared_ptr<arrow::ArrayData> values_data = values_array->data();
457
458    arrow::ArrayBuilder* value_builder = builder->value_builder();
459    int64_t offset = 0;
460    for (const auto& maybe_value : FieldValues()) {
461      ARROW_ASSIGN_OR_RAISE(auto value, maybe_value);
462      ARROW_RETURN_NOT_OK(builder->Append(!value->IsNull()));
463      if (!value->IsNull() && value->Size() > 0) {
464        ARROW_RETURN_NOT_OK(
465            value_builder->AppendArraySlice(*values_data.get(), offset, value->Size()));
466        offset += value->Size();
467      }
468    }
469
470    return arrow::Status::OK();
471  }
472
473 private:
474  std::string field_name_;
475  arrow::ArrayBuilder* builder_;
476  const std::vector<rapidjson::Document>& rows_;
477  std::vector<std::string> root_path_;
478  int64_t array_levels_;
479
480  /// Return a flattened iterator over values at nested location
481  arrow::Iterator<const rapidjson::Value*> FieldValues() {
482    std::vector<std::string> path(root_path_);
483    if (field_name_.size() > 0) {
484      path.push_back(field_name_);
485    }
486    auto iter = DocValuesIterator(rows_, std::move(path), array_levels_);
487    auto fn = [iter]() mutable -> arrow::Result<const rapidjson::Value*> {
488      return iter.Next();
489    };
490
491    return arrow::MakeFunctionIterator(fn);
492  }
493};  // JsonValueConverter

Writing conversions from Arrow#

To convert into rows from Arrow record batches, we’ll process the table in smaller batches, visiting each field of the batch and filling the output rows column-by-column.

At the top-level, we define ArrowToDocumentConverter that provides the API for converting Arrow batches and tables to rows. In many cases, it’s more optimal to perform conversions to rows in smaller batches, rather than doing the entire table at once. So we define one ConvertToVector method to convert a single batch, then in the other conversion method we use arrow::TableBatchReader to iterate over slices of a table. This returns Arrow’s iterator type (arrow::Iterator) so rows could then be processed either one-at-a-time or be collected into a container.

179class ArrowToDocumentConverter {
180 public:
181  /// Convert a single batch of Arrow data into Documents
182  arrow::Result<std::vector<rapidjson::Document>> ConvertToVector(
183      std::shared_ptr<arrow::RecordBatch> batch) {
184    RowBatchBuilder builder{batch->num_rows()};
185
186    for (int i = 0; i < batch->num_columns(); ++i) {
187      builder.SetField(batch->schema()->field(i).get());
188      ARROW_RETURN_NOT_OK(arrow::VisitArrayInline(*batch->column(i).get(), &builder));
189    }
190
191    return std::move(builder).Rows();
192  }
193
194  /// Convert an Arrow table into an iterator of Documents
195  arrow::Iterator<rapidjson::Document> ConvertToIterator(
196      std::shared_ptr<arrow::Table> table, size_t batch_size) {
197    // Use TableBatchReader to divide table into smaller batches. The batches
198    // created are zero-copy slices with *at most* `batch_size` rows.
199    auto batch_reader = std::make_shared<arrow::TableBatchReader>(*table);
200    batch_reader->set_chunksize(batch_size);
201
202    auto read_batch = [this](const std::shared_ptr<arrow::RecordBatch>& batch)
203        -> arrow::Result<arrow::Iterator<rapidjson::Document>> {
204      ARROW_ASSIGN_OR_RAISE(auto rows, ConvertToVector(batch));
205      return arrow::MakeVectorIterator(std::move(rows));
206    };
207
208    auto nested_iter = arrow::MakeMaybeMapIterator(
209        read_batch, arrow::MakeIteratorFromReader(std::move(batch_reader)));
210
211    return arrow::MakeFlattenIterator(std::move(nested_iter));
212  }
213};  // ArrowToDocumentConverter

One level down, the output rows are filled in by RowBatchBuilder. The RowBatchBuilder implements Visit() methods, but to save on code we write a template method for array types that have primitive C equivalents (booleans, integers, and floats) using arrow::enable_if_primitive_ctype. See Type Traits for other type predicates.

 57class RowBatchBuilder {
 58 public:
 59  explicit RowBatchBuilder(int64_t num_rows) : field_(nullptr) {
 60    // Reserve all of the space required up-front to avoid unnecessary resizing
 61    rows_.reserve(num_rows);
 62
 63    for (int64_t i = 0; i < num_rows; ++i) {
 64      rows_.push_back(rapidjson::Document());
 65      rows_[i].SetObject();
 66    }
 67  }
 68
 69  /// \brief Set which field to convert.
 70  void SetField(const arrow::Field* field) { field_ = field; }
 71
 72  /// \brief Retrieve converted rows from builder.
 73  std::vector<rapidjson::Document> Rows() && { return std::move(rows_); }
 74
 75  // Default implementation
 76  arrow::Status Visit(const arrow::Array& array) {
 77    return arrow::Status::NotImplemented(
 78        "Cannot convert to json document for array of type ", array.type()->ToString());
 79  }
 80
 81  // Handles booleans, integers, floats
 82  template <typename ArrayType, typename DataClass = typename ArrayType::TypeClass>
 83  arrow::enable_if_primitive_ctype<DataClass, arrow::Status> Visit(
 84      const ArrayType& array) {
 85    assert(static_cast<int64_t>(rows_.size()) == array.length());
 86    for (int64_t i = 0; i < array.length(); ++i) {
 87      if (!array.IsNull(i)) {
 88        rapidjson::Value str_key(field_->name(), rows_[i].GetAllocator());
 89        rows_[i].AddMember(str_key, array.Value(i), rows_[i].GetAllocator());
 90      }
 91    }
 92    return arrow::Status::OK();
 93  }
 94
 95  arrow::Status Visit(const arrow::StringArray& array) {
 96    assert(static_cast<int64_t>(rows_.size()) == array.length());
 97    for (int64_t i = 0; i < array.length(); ++i) {
 98      if (!array.IsNull(i)) {
 99        rapidjson::Value str_key(field_->name(), rows_[i].GetAllocator());
100        std::string_view value_view = array.Value(i);
101        rapidjson::Value value;
102        value.SetString(value_view.data(),
103                        static_cast<rapidjson::SizeType>(value_view.size()),
104                        rows_[i].GetAllocator());
105        rows_[i].AddMember(str_key, value, rows_[i].GetAllocator());
106      }
107    }
108    return arrow::Status::OK();
109  }
110
111  arrow::Status Visit(const arrow::StructArray& array) {
112    const arrow::StructType* type = array.struct_type();
113
114    assert(static_cast<int64_t>(rows_.size()) == array.length());
115
116    RowBatchBuilder child_builder(rows_.size());
117    for (int i = 0; i < type->num_fields(); ++i) {
118      const arrow::Field* child_field = type->field(i).get();
119      child_builder.SetField(child_field);
120      ARROW_RETURN_NOT_OK(arrow::VisitArrayInline(*array.field(i).get(), &child_builder));
121    }
122    std::vector<rapidjson::Document> rows = std::move(child_builder).Rows();
123
124    for (int64_t i = 0; i < array.length(); ++i) {
125      if (!array.IsNull(i)) {
126        rapidjson::Value str_key(field_->name(), rows_[i].GetAllocator());
127        // Must copy value to new allocator
128        rapidjson::Value row_val;
129        row_val.CopyFrom(rows[i], rows_[i].GetAllocator());
130        rows_[i].AddMember(str_key, row_val, rows_[i].GetAllocator());
131      }
132    }
133    return arrow::Status::OK();
134  }
135
136  arrow::Status Visit(const arrow::ListArray& array) {
137    assert(static_cast<int64_t>(rows_.size()) == array.length());
138    // First create rows from values
139    std::shared_ptr<arrow::Array> values = array.values();
140    RowBatchBuilder child_builder(values->length());
141    const arrow::Field* value_field = array.list_type()->value_field().get();
142    std::string value_field_name = value_field->name();
143    child_builder.SetField(value_field);
144    ARROW_RETURN_NOT_OK(arrow::VisitArrayInline(*values.get(), &child_builder));
145
146    std::vector<rapidjson::Document> rows = std::move(child_builder).Rows();
147
148    int64_t values_i = 0;
149    for (int64_t i = 0; i < array.length(); ++i) {
150      if (array.IsNull(i)) continue;
151
152      rapidjson::Document::AllocatorType& allocator = rows_[i].GetAllocator();
153      auto array_len = array.value_length(i);
154
155      rapidjson::Value value;
156      value.SetArray();
157      value.Reserve(array_len, allocator);
158
159      for (int64_t j = 0; j < array_len; ++j) {
160        rapidjson::Value row_val;
161        // Must copy value to new allocator
162        row_val.CopyFrom(rows[values_i][value_field_name], allocator);
163        value.PushBack(row_val, allocator);
164        ++values_i;
165      }
166
167      rapidjson::Value str_key(field_->name(), allocator);
168      rows_[i].AddMember(str_key, value, allocator);
169    }
170
171    return arrow::Status::OK();
172  }
173
174 private:
175  const arrow::Field* field_;
176  std::vector<rapidjson::Document> rows_;
177};  // RowBatchBuilder