44 template <std::ranges::sized_range T,
typename CR,
typename Ext = empty_extension>
79 fixed_width_binary_traits::value_type,
80 fixed_width_binary_traits::const_reference>;
82 template <std::ranges::sized_range T,
typename CR,
typename Ext>
122 template <std::ranges::sized_range T,
class CR,
class Ext>
130 sizeof(std::ranges::range_value_t<T>) ==
sizeof(
byte_t),
131 "Only sequences of types with the same size as byte_t are supported"
197 template <
class... ARGS>
200 :
base_type(create_proxy(std::forward<ARGS>(args)...))
205 using base_type::get_arrow_proxy;
206 using base_type::size;
263 size_t element_count,
266 std::optional<std::string_view> name = std::nullopt,
267 std::optional<METADATA_RANGE> metadata = std::nullopt
285 template <input_metadata_container METADATA_RANGE = std::vector<metadata_pair>>
289 std::optional<std::string_view> name = std::nullopt,
290 std::optional<METADATA_RANGE> metadata = std::nullopt
314 std::ranges::input_range VALUES,
318 std::ranges::input_range<std::ranges::range_value_t<VALUES>>
324 std::optional<std::string_view> name = std::nullopt,
325 std::optional<METADATA_RANGE> metadata = std::nullopt
347 template <std::ranges::input_range VALUES, input_metadata_container METADATA_RANGE = std::vector<metadata_pair>>
349 std::ranges::input_range<std::ranges::range_value_t<VALUES>>
355 std::optional<std::string_view> name = std::nullopt,
356 std::optional<METADATA_RANGE> metadata = std::nullopt
376 template <std::ranges::input_range NULLABLE_VALUES, input_metadata_container METADATA_RANGE = std::vector<metadata_pair>>
378 && std::ranges::input_range<typename std::ranges::range_value_t<NULLABLE_VALUES>::value_type>
380 std::ranges::range_value_t<typename std::ranges::range_value_t<NULLABLE_VALUES>::value_type>,
384 std::optional<std::string_view> name = std::nullopt,
385 std::optional<METADATA_RANGE> metadata = std::nullopt
411 template <mpl::
char_like C, input_metadata_container METADATA_RANGE = std::vector<metadata_pair>>
412 [[nodiscard]]
static arrow_proxy create_proxy_impl(
414 size_t element_count,
416 std::optional<validity_bitmap>&& validity_input,
417 std::optional<std::string_view> name = std::nullopt,
418 std::optional<METADATA_RANGE> metadata = std::nullopt
421 static constexpr size_t DATA_BUFFER_INDEX = 1;
431 [[nodiscard]]
constexpr size_type byte_offset(size_type index)
const
433 return (index +
static_cast<size_type
>(this->get_arrow_proxy().offset())) * m_element_size;
443 [[nodiscard]]
constexpr auto& get_data_buffer()
445 return this->get_arrow_proxy().get_array_private_data()->buffers()[DATA_BUFFER_INDEX];
455 [[nodiscard]]
constexpr const auto& get_data_buffer()
const
457 return this->get_arrow_proxy().buffers()[DATA_BUFFER_INDEX];
540 template <std::ranges::sized_range U>
541 requires mpl::convertible_ranges<U, T>
563 template <std::ranges::sized_range U>
564 requires mpl::convertible_ranges<U, T>
592 template <
typename InputIt>
593 requires std::input_iterator<InputIt>
594 && mpl::convertible_ranges<typename std::iterator_traits<InputIt>::value_type, T>
633 template <std::ranges::sized_range U>
634 requires mpl::convertible_ranges<U, T>
635 constexpr void assign(U&& rhs,
size_type index);
637 size_t m_element_size = 0;
650 template <std::ranges::sized_range T,
typename CR,
typename Ext>
658 template <std::ranges::sized_range T,
typename CR,
typename Ext>
659 template <mpl::
char_like C, val
idity_bitmap_input VB, input_metadata_container METADATA_RANGE>
660 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy(
662 size_t element_count,
665 std::optional<std::string_view> name,
666 std::optional<METADATA_RANGE> metadata
670 return create_proxy_impl(
671 std::move(data_buffer),
680 template <std::ranges::sized_range T,
typename CR,
typename Ext>
681 template <input_metadata_container METADATA_RANGE>
682 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy(
685 std::optional<std::string_view> name,
686 std::optional<METADATA_RANGE> metadata
689 u8_buffer<char> data_buffer{};
690 std::optional<validity_bitmap> bitmap =
nullable ? std::make_optional<validity_bitmap>(
nullptr, 0)
692 return create_proxy_impl(
693 std::move(data_buffer),
702 template <std::ranges::sized_range T,
typename CR,
typename Ext>
703 template <std::ranges::input_range R, val
idity_bitmap_input VB, input_metadata_container METADATA_RANGE>
705 std::ranges::input_range<std::ranges::range_value_t<R>> &&
709 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy(
712 std::optional<std::string_view> name,
713 std::optional<METADATA_RANGE> metadata
716 using values_type = std::ranges::range_value_t<R>;
717 using values_inner_value_type = std::ranges::range_value_t<values_type>;
720 const size_t element_size = std::ranges::empty(values) ? 0 : std::ranges::size(*values.begin());
724 std::move(data_buffer),
727 std::forward<VB>(validity_input),
728 std::forward<std::optional<std::string_view>>(name),
729 std::forward<std::optional<METADATA_RANGE>>(metadata)
733 template <std::ranges::sized_range T,
typename CR,
typename Ext>
734 template <std::ranges::input_range R, input_metadata_container METADATA_RANGE>
736 std::ranges::input_range<std::ranges::range_value_t<R>> &&
740 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy(
743 std::optional<std::string_view> name,
744 std::optional<METADATA_RANGE> metadata
749 return create_proxy(std::forward<R>(values),
validity_bitmap{}, std::move(name), std::move(metadata));
753 using values_type = std::ranges::range_value_t<R>;
754 using values_inner_value_type = std::ranges::range_value_t<values_type>;
757 const size_t element_size = std::ranges::empty(values) ? 0 : std::ranges::size(*values.begin());
759 return create_proxy_impl(
760 std::move(data_buffer),
770 template <std::ranges::sized_range T,
typename CR,
typename Ext>
771 template <std::ranges::input_range NULLABLE_RANGE, input_metadata_container METADATA_RANGE>
773 && std::ranges::input_range<typename std::ranges::range_value_t<NULLABLE_RANGE>::value_type>
775 std::ranges::range_value_t<typename std::ranges::range_value_t<NULLABLE_RANGE>::value_type>,
777 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy(
778 NULLABLE_RANGE&& range,
779 std::optional<std::string_view> name,
780 std::optional<METADATA_RANGE> metadata
784 const auto values = range
785 | std::views::transform(
791 const auto is_non_null = range
792 | std::views::transform(
795 return v.has_value();
798 return self_type::create_proxy(values, is_non_null, std::move(name), std::move(metadata));
801 template <std::ranges::sized_range T,
typename CR,
typename Ext>
802 template <mpl::
char_like C, input_metadata_container METADATA_RANGE>
803 arrow_proxy fixed_width_binary_array_impl<T, CR, Ext>::create_proxy_impl(
805 size_t element_count,
807 std::optional<validity_bitmap>&& bitmap,
808 std::optional<std::string_view> name,
809 std::optional<METADATA_RANGE> metadata
813 element_size == 0 ? (data_buffer.size() == 0) : (data_buffer.size() % element_size == 0)
816 const auto null_count = bitmap.has_value() ? bitmap->null_count() : 0;
817 std::string format_str =
"w:" + std::to_string(element_size);
818 const std::optional<std::unordered_set<ArrowFlag>>
819 flags = bitmap.has_value()
824 std::move(format_str),
829 repeat_view<bool>(
true, 0),
834 std::vector<buffer<std::uint8_t>> arr_buffs = {
835 bitmap.has_value() ? std::move(*bitmap).extract_storage() : buffer<std::uint8_t>{
nullptr, 0},
836 std::move(data_buffer).extract_storage()
840 static_cast<std::int64_t
>(element_count),
841 static_cast<int64_t
>(null_count),
843 std::move(arr_buffs),
849 arrow_proxy proxy{std::move(arr), std::move(schema)};
854 template <std::ranges::sized_range T,
typename CR,
typename Ext>
855 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::data(size_type i) -> data_iterator
857 return const_cast<data_iterator
>(std::as_const(*this).data(i));
860 template <std::ranges::sized_range T,
typename CR,
typename Ext>
861 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::data(size_type i)
const -> const_data_iterator
863 const auto& data_buffer = get_data_buffer();
864 const size_t data_buffer_size = data_buffer.size();
865 const size_type index_offset = (
static_cast<size_type
>(this->get_arrow_proxy().offset())
869 return data_buffer.template data<const data_value_type>() + index_offset;
872 template <std::ranges::sized_range T,
typename CR,
typename Ext>
873 template <std::ranges::sized_range U>
875 constexpr void fixed_width_binary_array_impl<T, CR, Ext>::assign(U&& rhs, size_type index)
879 std::copy(std::ranges::begin(rhs), std::ranges::end(rhs), data(index * m_element_size));
882 template <std::ranges::sized_range T,
typename CR,
typename Ext>
889 template <std::ranges::sized_range T,
typename CR,
typename Ext>
893 const auto offset_begin = i * m_element_size;
894 const auto offset_end = offset_begin + m_element_size;
900 template <std::ranges::sized_range T,
typename CR,
typename Ext>
901 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::value_begin() -> value_iterator
903 return value_iterator{functor_type{&(this->derived_cast())}, 0};
906 template <std::ranges::sized_range T,
typename CR,
typename Ext>
907 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::value_end() -> value_iterator
912 template <std::ranges::sized_range T,
typename CR,
typename Ext>
913 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::value_cbegin() const -> const_value_iterator
915 return const_value_iterator{const_functor_type{&(this->derived_cast())}, 0};
918 template <std::ranges::sized_range T,
typename CR,
typename Ext>
919 constexpr auto fixed_width_binary_array_impl<T, CR, Ext>::value_cend() const -> const_value_iterator
924 template <std::ranges::sized_range T,
typename CR,
typename Ext>
925 template <std::ranges::sized_range U>
927 constexpr void fixed_width_binary_array_impl<T, CR, Ext>::resize_values(size_type new_length, U value)
930 if (new_length <
size())
932 const size_t new_size = new_length +
static_cast<size_t>(this->get_arrow_proxy().offset());
933 const auto offset = new_size * m_element_size;
934 auto& data_buffer = get_data_buffer();
935 data_buffer.resize(offset);
937 else if (new_length >
size())
939 insert_value(value_cend(), value, new_length -
size());
943 template <std::ranges::sized_range T,
typename CR,
typename Ext>
944 template <std::ranges::sized_range U>
951 const auto idx =
static_cast<size_t>(std::distance(value_cbegin(), pos));
953 const uint8_t* uint8_ptr =
reinterpret_cast<const uint8_t*
>(
value.data());
954 const std::vector<uint8_t> casted_value(uint8_ptr, uint8_ptr +
value.size());
956 const auto joined_repeated_value_range = std::ranges::views::join(my_repeat_view);
957 auto& data_buffer = get_data_buffer();
958 const auto offset_begin = byte_offset(idx);
959 const auto pos_to_insert =
sparrow::next(data_buffer.cbegin(), offset_begin);
960 data_buffer.insert(pos_to_insert, joined_repeated_value_range.begin(), joined_repeated_value_range.end());
964 template <std::ranges::sized_range T,
typename CR,
typename Ext>
965 template <
typename InputIt>
966 requires std::input_iterator<InputIt>
969 fixed_width_binary_array_impl<T, CR, Ext>::insert_values(
const_value_iterator pos, InputIt first, InputIt last)
978 auto values = std::ranges::subrange(first, last);
979 const size_t cumulative_sizes = values.size() * m_element_size;
980 auto& data_buffer = get_data_buffer();
981 data_buffer.resize(data_buffer.size() + cumulative_sizes);
982 const auto idx =
static_cast<size_t>(std::distance(value_cbegin(), pos));
984 const auto offset_begin = byte_offset(idx);
985 auto insert_pos =
sparrow::next(casted_values.begin(), offset_begin);
994 for (
const auto& val : values)
996 std::copy(val.begin(), val.end(), insert_pos);
997 std::advance(insert_pos, m_element_size);
1002 template <std::ranges::sized_range T,
typename CR,
typename Ext>
1004 fixed_width_binary_array_impl<T, CR, Ext>::erase_values(const_value_iterator pos, size_type count)
1009 const size_t index =
static_cast<size_t>(std::distance(value_cbegin(), pos));
1014 auto& data_buffer = get_data_buffer();
1015 const size_type byte_count = m_element_size * count;
1016 const auto offset_begin = byte_offset(index);
1017 const auto offset_end = offset_begin + byte_count;
1020 data_buffer.begin() +
static_cast<difference_type
>(offset_end),
1022 data_buffer.begin() +
static_cast<difference_type
>(offset_begin)
1024 data_buffer.resize(data_buffer.size() - byte_count);
typename base_type::const_bitmap_range const_bitmap_range
typename base_type::iterator_tag iterator_tag
std::conditional_t< is_mutable, mutable_array_base< D >, array_crtp_base< D > > base_type
typename base_type::bitmap_const_reference bitmap_const_reference
typename base_type::bitmap_type bitmap_type
typename base_type::difference_type difference_type
bitset_iterator< self_type, true > const_iterator
typename base_type::size_type size_type
typename base_type::iterator_tag iterator_tag
typename base_type::const_bitmap_range const_bitmap_range
typename inner_types::data_value_type data_value_type
fixed_width_binary_array_impl(arrow_proxy)
Constructs fixed-width binary array from Arrow proxy.
constexpr inner_reference value(size_type i)
Gets mutable reference to element at specified index.
nullable< inner_reference, bitmap_reference > reference
typename inner_types::inner_const_reference inner_const_reference
nullable< inner_value_type > value_type
typename inner_types::functor_type functor_type
mutable_array_bitmap_base< self_type > base_type
typename inner_types::const_value_iterator const_value_iterator
typename base_type::bitmap_const_reference bitmap_const_reference
typename base_type::bitmap_reference bitmap_reference
typename inner_types::inner_value_type inner_value_type
nullable< inner_const_reference, bitmap_const_reference > const_reference
typename inner_types::const_functor_type const_functor_type
array_inner_types< self_type > inner_types
typename inner_types::const_data_iterator const_data_iterator
typename inner_types::data_iterator data_iterator
typename inner_types::inner_reference inner_reference
typename inner_types::value_iterator value_iterator
fixed_width_binary_array_impl(ARGS &&... args)
Generic constructor for creating fixed-width binary array.
typename base_type::bitmap_type bitmap_type
typename base_type::difference_type difference_type
fixed_width_binary_array_impl< T, fixed_width_binary_traits::const_reference, Ext > self_type
constexpr inner_const_reference value(size_type i) const
Gets const reference to element at specified index.
A view that repeats a value a given number of times.
The class sequence_view describes an object that can refer to a constant contiguous sequence of T wit...
This buffer class is used as storage buffer for all sparrow arrays.
Concept for character-like types.
Concept for convertible range types.
#define SPARROW_ASSERT_TRUE(expr__)
constexpr std::size_t size(typelist< T... >={})
Gets the count of types contained in a typelist.
constexpr bool excludes_copy_and_move_ctor_v
Convenience variable template for excludes_copy_and_move_ctor.
constexpr bool is_type_instance_of_v
Variable template for convenient access to is_type_instance_of.
array_bitmap_base_impl< D, true > mutable_array_bitmap_base
Convenient alias for arrays with mutable validity bitmaps.
ArrowSchema make_arrow_schema(F format, N name, std::optional< M > metadata, std::optional< std::unordered_set< ArrowFlag > > flags, ArrowSchema **children, const CHILDREN_OWNERSHIP &children_ownership, ArrowSchema *dictionary, bool dictionary_ownership)
Creates an ArrowSchema owned by a unique_ptr and holding the provided data.
constexpr bool all_same_size(const Range &range)
arrow_traits< std::vector< byte_t > > fixed_width_binary_traits
constexpr InputIt next(InputIt it, Distance n)
fixed_width_binary_array_impl< fixed_width_binary_traits::value_type, fixed_width_binary_traits::const_reference > fixed_width_binary_array
ArrowArray make_arrow_array(int64_t length, int64_t null_count, int64_t offset, B buffers, ArrowArray **children, const CHILDREN_OWNERSHIP &children_ownership, ArrowArray *dictionary, bool dictionary_ownership)
Creates an ArrowArray.
dynamic_bitset< std::uint8_t > validity_bitmap
Type alias for a validity bitmap using 8-bit storage blocks.
SPARROW_API std::size_t num_bytes_for_fixed_sized_binary(std::string_view format)
Get the number of bytes for a fixed width binary layout from the ArrowArray format string.
validity_bitmap ensure_validity_bitmap(std::size_t size, R &&validity_input)
Ensures a validity bitmap of the specified size from various input types.
data_type
Runtime identifier of arrow data types, usually associated with raw bytes with the associated value.
const_iterator const_value_iterator
detail::layout_value_functor< array_type, inner_reference > functor_type
functor_index_iterator< functor_type > iterator
fixed_width_binary_reference< array_type > inner_reference
std::reverse_iterator< const_iterator > const_reverse_iterator
bitmap_type::const_iterator const_bitmap_iterator
fixed_width_binary_array_impl< T, CR, Ext > array_type
std::random_access_iterator_tag iterator_tag
data_value_type * data_iterator
functor_index_iterator< const_functor_type > const_iterator
detail::layout_value_functor< const array_type, inner_const_reference > const_functor_type
const data_value_type * const_data_iterator
typename T::value_type data_value_type
Base class for array_inner_types specializations.
Traits class that must be specialized by array implementations.
Provides compile-time information about Arrow data types.
static constexpr sparrow::data_type get()
Metafunction for retrieving the data_type of a typed array.