Method BuildArray

cpp/src/parquet/arrow/reader.cc:620–672

  }

  Status BuildArray(int64_t length_upper_bound,
                    std::shared_ptr<ChunkedArray>* out) override {
    const int16_t* def_levels;
    const int16_t* rep_levels;
    int64_t num_levels;
    RETURN_NOT_OK(item_reader_->GetDefLevels(&def_levels, &num_levels));
    RETURN_NOT_OK(item_reader_->GetRepLevels(&rep_levels, &num_levels));

    std::shared_ptr<ResizableBuffer> validity_buffer;
    ::parquet::internal::ValidityBitmapInputOutput validity_io;
    validity_io.values_read_upper_bound = length_upper_bound;
    if (field_->nullable()) {
      ARROW_ASSIGN_OR_RAISE(validity_buffer,
                            AllocateResizableBuffer(
                                bit_util::BytesForBits(length_upper_bound), ctx_->pool));
      validity_io.valid_bits = validity_buffer->mutable_data();
    }
    ARROW_ASSIGN_OR_RAISE(
        std::shared_ptr<ResizableBuffer> offsets_buffer,
        AllocateResizableBuffer(
            sizeof(IndexType) * std::max(int64_t{1}, length_upper_bound + 1),
            ctx_->pool));
    // Ensure zero initialization in case we have reached a zero length list (and
    // because first entry is always zero).
    IndexType* offset_data = reinterpret_cast<IndexType*>(offsets_buffer->mutable_data());
    offset_data[0] = 0;
    BEGIN_PARQUET_CATCH_EXCEPTIONS
    ::parquet::internal::DefRepLevelsToList(def_levels, rep_levels, num_levels,
                                            level_info_, &validity_io, offset_data);
    END_PARQUET_CATCH_EXCEPTIONS

    RETURN_NOT_OK(item_reader_->BuildArray(offset_data[validity_io.values_read], out));

    // Resize to actual number of elements returned.
    RETURN_NOT_OK(
        offsets_buffer->Resize((validity_io.values_read + 1) * sizeof(IndexType)));
    if (validity_buffer != nullptr) {
      RETURN_NOT_OK(
          validity_buffer->Resize(bit_util::BytesForBits(validity_io.values_read)));
      validity_buffer->ZeroPadding();
    }
    ARROW_ASSIGN_OR_RAISE(std::shared_ptr<ArrayData> item_chunk, ChunksToSingle(**out));

    std::vector<std::shared_ptr<Buffer>> buffers{
        validity_io.null_count > 0 ? validity_buffer : nullptr, offsets_buffer};
    auto data = std::make_shared<ArrayData>(
        field_->type(),
        /*length=*/validity_io.values_read, std::move(buffers),
        std::vector<std::shared_ptr<ArrayData>>{item_chunk}, validity_io.null_count);

    ARROW_ASSIGN_OR_RAISE(*out, AssembleArray(std::move(data)));
    return Status::OK();
  }

  const std::shared_ptr<Field> field() override { return field_; }