chunk.hpp

Go to the documentation of this file.

// -----------------------------------------------------------------------------------------------------
// Copyright (c) 2006-2022, Knut Reinert & Freie Universität Berlin
// Copyright (c) 2016-2022, Knut Reinert & MPI für molekulare Genetik
// This file may be used, modified and/or redistributed under the terms of the 3-clause BSD-License
// shipped with this file and also available at: https://github.com/seqan/seqan3/blob/master/LICENSE.md
// -----------------------------------------------------------------------------------------------------
 
#pragma once
 
#include <ranges>
 
#include <seqan3/alphabet/concept.hpp>
#include <seqan3/core/detail/all_view.hpp>
#include <seqan3/core/detail/iterator_traits.hpp>
#include <seqan3/core/platform.hpp>
#include <seqan3/core/range/detail/adaptor_from_functor.hpp>
#include <seqan3/core/range/type_traits.hpp>
#include <seqan3/utility/range/concept.hpp>
 
namespace seqan3::detail
{
// ---------------------------------------------------------------------------------------------------------------------
// chunk_view class
// ---------------------------------------------------------------------------------------------------------------------
 
template <std::ranges::input_range urng_t>
    requires std::ranges::view<urng_t>
class chunk_view : public std::ranges::view_interface<chunk_view<urng_t>>
{
private:
    urng_t urange;
 
    std::ranges::range_difference_t<urng_t> chunk_size;
 
    // The iterator type if `urng_t` is a pure input range. See class definition for details.
    template <bool const_range>
    class basic_input_iterator;
 
    // The iterator type if `urng_t` is at least a forward range. See class definition for details.
    template <bool const_range>
    class basic_iterator;
 
public:
    chunk_view()
        requires std::default_initializable<urng_t>
    = default;                                                
    chunk_view(chunk_view const & rhs) = default;             
    chunk_view(chunk_view && rhs) = default;                  
    chunk_view & operator=(chunk_view const & rhs) = default; 
    chunk_view & operator=(chunk_view && rhs) = default;      
    ~chunk_view() = default;                                  
 
    constexpr explicit chunk_view(urng_t urng, std::ranges::range_difference_t<urng_t> const size_of_chunk) :
        urange{std::move(urng)},
        chunk_size{size_of_chunk}
    {}
 
    auto begin() noexcept
    {
        if constexpr (std::ranges::forward_range<urng_t>)
            return basic_iterator<false>{std::ranges::begin(urange), std::ranges::end(urange), chunk_size};
        else
            return basic_input_iterator<false>{std::ranges::begin(urange), std::ranges::end(urange), chunk_size};
    }
 
    auto begin() const noexcept
        requires const_iterable_range<urng_t>
    {
        if constexpr (std::ranges::forward_range<urng_t>)
            return basic_iterator<true>{std::ranges::cbegin(urange), std::ranges::cend(urange), chunk_size};
        else
            return basic_input_iterator<true>{std::ranges::cbegin(urange), std::ranges::cend(urange), chunk_size};
    }
 
    auto end() noexcept
    {
        return std::ranges::end(urange);
    }
 
    auto end() const noexcept
        requires const_iterable_range<urng_t>
    {
        return std::ranges::cend(urange);
    }
 
    auto size()
        requires std::ranges::sized_range<urng_t>
    {
        using size_type = std::ranges::range_size_t<urng_t>;
        return static_cast<size_type>((std::ranges::size(urange) + chunk_size - 1) / chunk_size); // round up
    }
 
    auto size() const
        requires std::ranges::sized_range<urng_t const>
    {
        using size_type = std::ranges::range_size_t<urng_t const>;
        return static_cast<size_type>((std::ranges::size(urange) + chunk_size - 1) / chunk_size); // round up
    }
};
 
template <std::ranges::range rng_t>
chunk_view(rng_t &&, std::ranges::range_difference_t<rng_t> const &) -> chunk_view<seqan3::detail::all_t<rng_t>>;
 
// ---------------------------------------------------------------------------------------------------------------------
// chunk_view iterators (basic_input_iterator and basic_iterator)
// ---------------------------------------------------------------------------------------------------------------------
 
template <std::ranges::input_range urng_t>
    requires std::ranges::view<urng_t>
template <bool const_range>
class chunk_view<urng_t>::basic_input_iterator :
    public maybe_iterator_category<maybe_const_iterator_t<const_range, urng_t>>
{
private:
    using urng_it_t = maybe_const_iterator_t<const_range, urng_t>;
 
    using sentinel_t = maybe_const_sentinel_t<const_range, urng_t>;
 
    template <typename outer_it_type>
    class input_helper_iterator : public urng_it_t
    {
    public:
        constexpr input_helper_iterator() = default;                                          
        constexpr input_helper_iterator(input_helper_iterator const &) = default;             
        constexpr input_helper_iterator(input_helper_iterator &&) = default;                  
        constexpr input_helper_iterator & operator=(input_helper_iterator const &) = default; 
        constexpr input_helper_iterator & operator=(input_helper_iterator &&) = default;      
        ~input_helper_iterator() = default;                                                   
 
        constexpr explicit input_helper_iterator(outer_it_type & outer_iterator, urng_it_t urng_it) :
            urng_it_t(std::move(urng_it)),
            outer_it(std::addressof(outer_iterator))
        {}
 
        constexpr explicit input_helper_iterator(urng_it_t urng_it) : urng_it_t(std::move(urng_it))
        {}
 
        input_helper_iterator & operator++() noexcept
        {
            --(outer_it->remaining);
            urng_it_t::operator++();
            return *this;
        }
 
        input_helper_iterator operator++(int) noexcept
        {
            input_helper_iterator tmp{*this};
            ++(*this);
            return tmp;
        }
 
        bool operator==(sentinel_t const & /* rhs */) noexcept
        {
            return this->outer_it->remaining == 0u || this->outer_it->urng_begin == this->outer_it->urng_end;
        }
 
        outer_it_type * outer_it{nullptr};
    };
 
    using helper_it_t = input_helper_iterator<basic_input_iterator>;
 
    // befriend the iterator on a const range
    template <bool other_const_range>
    friend class basic_input_iterator;
 
    // befriend the inner iterator type
    template <typename outer_it_type>
    friend class input_helper_iterator;
 
public:
    using difference_type = typename std::iter_difference_t<urng_it_t>;
    using value_type = std::ranges::subrange<helper_it_t, sentinel_t>;
    using pointer = void;
    using reference = value_type;
    using iterator_concept = std::input_iterator_tag;
 
    constexpr basic_input_iterator() = default;                                         
    constexpr basic_input_iterator(basic_input_iterator const &) = default;             
    constexpr basic_input_iterator(basic_input_iterator &&) = default;                  
    constexpr basic_input_iterator & operator=(basic_input_iterator const &) = default; 
    constexpr basic_input_iterator & operator=(basic_input_iterator &&) = default;      
    ~basic_input_iterator() = default;                                                  
 
    constexpr explicit basic_input_iterator(basic_input_iterator<!const_range> it) noexcept
        requires const_range
    :
        chunk_size{std::move(it.chunk_size)},
        remaining{std::move(it.remaining)},
        urng_begin{std::move(it.urng_begin)},
        urng_end{std::move(it.urng_end)},
        current_chunk{std::move(it.current_chunk)}
    {}
 
    constexpr explicit basic_input_iterator(urng_it_t it_begin,
                                            sentinel_t it_end,
                                            std::ranges::range_difference_t<urng_t> const size_of_chunk) :
        chunk_size{size_of_chunk},
        remaining{size_of_chunk},
        urng_begin{std::move(it_begin)},
        urng_end{std::move(it_end)}
    {
        current_chunk = std::ranges::subrange<helper_it_t, sentinel_t>{helper_it_t{*this, it_begin}, it_end};
    }
 
 
    friend constexpr bool operator==(basic_input_iterator const & lhs, sentinel_t const & rhs) noexcept
    {
        return lhs.urng_begin == rhs;
    }
 
    friend constexpr bool operator==(basic_input_iterator const & lhs, basic_input_iterator const & rhs) noexcept
    {
        return (lhs.urng_begin == rhs.urng_begin) && (lhs.remaining == rhs.remaining);
    }
 
    constexpr basic_input_iterator & operator++() noexcept
    {
        while (remaining > 0u && urng_begin != urng_end) // if chunk was not fully consumed and range is not at end
        {
            ++urng_begin;
            --remaining;
        }
        current_chunk = std::ranges::subrange<helper_it_t, sentinel_t>{helper_it_t{*this, urng_begin}, urng_end};
        remaining = chunk_size;
        return *this;
    }
 
    constexpr basic_input_iterator operator++(int) noexcept
    {
        basic_input_iterator tmp{*this};
        ++(*this);
        return tmp;
    }
 
    constexpr value_type operator*() const noexcept
    {
        return current_chunk;
    }
 
private:
    std::ranges::range_difference_t<urng_t> chunk_size;
 
    std::ranges::range_difference_t<urng_t> remaining;
 
    urng_it_t urng_begin;
 
    sentinel_t urng_end;
 
    value_type current_chunk;
};
 
template <std::ranges::input_range urng_t>
    requires std::ranges::view<urng_t>
template <bool const_range>
class chunk_view<urng_t>::basic_iterator : public maybe_iterator_category<maybe_const_iterator_t<const_range, urng_t>>
{
private:
    using it_t = maybe_const_iterator_t<const_range, urng_t>;
    using sentinel_t = maybe_const_sentinel_t<const_range, urng_t>;
 
    // befriend the iterator on a const range
    template <bool other_const_range>
    friend class basic_iterator;
 
public:
    using difference_type = typename std::iter_difference_t<it_t>;
    using value_type = std::ranges::subrange<it_t, it_t>;
    using pointer = void;
    using reference = value_type;
    using iterator_concept = std::conditional_t<std::contiguous_iterator<it_t>,
                                                typename std::random_access_iterator_tag,
                                                detail::iterator_concept_tag_t<it_t>>;
 
    constexpr basic_iterator() = default;                                   
    constexpr basic_iterator(basic_iterator const &) = default;             
    constexpr basic_iterator(basic_iterator &&) = default;                  
    constexpr basic_iterator & operator=(basic_iterator const &) = default; 
    constexpr basic_iterator & operator=(basic_iterator &&) = default;      
    ~basic_iterator() = default;                                            
 
    constexpr basic_iterator(basic_iterator<!const_range> const & it) noexcept
        requires const_range
    :
        chunk_size{std::move(it.chunk_size)},
        urng_begin{std::move(it.urng_begin)},
        urng_end{std::move(it.urng_end)},
        current_chunk{std::move(it.current_chunk)}
    {}
 
    constexpr explicit basic_iterator(it_t it_start,
                                      sentinel_t it_end,
                                      std::ranges::range_difference_t<urng_t> const size_of_chunk) :
        chunk_size{size_of_chunk},
        urng_begin{std::move(it_start)},
        urng_end{std::move(it_end)}
    {
        current_chunk = value_type{it_start, get_next_end_of_chunk(it_start)};
    }
 
 
    friend constexpr bool operator==(basic_iterator const & lhs, sentinel_t const & rhs) noexcept
    {
        return lhs.current_chunk.begin() == rhs;
    }
 
    friend constexpr bool operator==(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return (lhs.current_chunk.begin() == rhs.current_chunk.begin()) && (lhs.chunk_size == rhs.chunk_size);
    }
 
    friend constexpr bool operator!=(basic_iterator const & lhs, sentinel_t const & rhs) noexcept
    {
        return !(lhs == rhs);
    }
 
    friend constexpr bool operator!=(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return !(lhs == rhs);
    }
 
    friend constexpr bool operator<(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return lhs.current_chunk.begin() < rhs.current_chunk.begin();
    }
 
    friend constexpr bool operator>(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return lhs.current_chunk.begin() > rhs.current_chunk.begin();
    }
 
    friend constexpr bool operator<=(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return lhs.current_chunk.begin() <= rhs.current_chunk.begin();
    }
 
    friend constexpr bool operator>=(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
    {
        return lhs.current_chunk.begin() >= rhs.current_chunk.begin();
    }
 
 
    constexpr basic_iterator & operator++() noexcept
    {
        current_chunk = value_type{current_chunk.end(), get_next_end_of_chunk(current_chunk.end())};
        return *this;
    }
 
    basic_iterator operator++(int) noexcept
    {
        basic_iterator tmp{*this};
        ++(*this);
        return tmp;
    }
 
    constexpr basic_iterator & operator--() noexcept
        requires std::bidirectional_iterator<it_t>
    {
        current_chunk = value_type{get_former_start_of_chunk(current_chunk.begin()), current_chunk.begin()};
        return *this;
    }
 
    constexpr basic_iterator operator--(int) noexcept
        requires std::bidirectional_iterator<it_t>
    {
        basic_iterator tmp{*this};
        --(*this);
        return tmp;
    }
 
    constexpr basic_iterator & operator+=(difference_type const skip) noexcept
        requires std::random_access_iterator<it_t>
    {
        auto new_start_it = current_chunk.begin() + (chunk_size * skip);
        current_chunk = value_type{new_start_it, get_next_end_of_chunk(new_start_it)};
        return *this;
    }
 
    constexpr basic_iterator operator+(difference_type const skip) const noexcept
        requires std::random_access_iterator<it_t>
    {
        basic_iterator tmp{*this};
        return tmp += skip;
    }
 
    friend constexpr basic_iterator operator+(difference_type const skip, basic_iterator const & it) noexcept
        requires std::random_access_iterator<it_t>
    {
        return it + skip;
    }
 
    constexpr basic_iterator & operator-=(difference_type const skip) noexcept
        requires std::random_access_iterator<it_t>
    {
        auto new_start_it = current_chunk.begin() - (chunk_size * skip);
        current_chunk = value_type{new_start_it, get_next_end_of_chunk(new_start_it)};
        return *this;
    }
 
    constexpr basic_iterator operator-(difference_type const skip) const noexcept
        requires std::random_access_iterator<it_t>
    {
        basic_iterator tmp{*this};
        return tmp -= skip;
    }
 
    friend constexpr basic_iterator operator-(difference_type const skip, basic_iterator const & it) noexcept
        requires std::random_access_iterator<it_t>
    {
        return it - skip;
    }
 
    friend constexpr difference_type operator-(basic_iterator const & lhs, basic_iterator const & rhs) noexcept
        requires std::sized_sentinel_for<it_t, it_t>
    {
        return static_cast<difference_type>((lhs.current_chunk.begin() - rhs.current_chunk.begin()) / lhs.chunk_size);
    }
 
    friend constexpr difference_type operator-(sentinel_t const & /* lhs */, basic_iterator const & rhs) noexcept
        requires std::sized_sentinel_for<sentinel_t, it_t>
    {
        return static_cast<difference_type>((rhs.urng_end - rhs.current_chunk.begin() + rhs.chunk_size - 1)
                                            / rhs.chunk_size);
    }
 
    friend constexpr difference_type operator-(basic_iterator const & lhs, sentinel_t const & rhs) noexcept
        requires std::sized_sentinel_for<sentinel_t, it_t>
    {
        return -(rhs - lhs);
    }
 
    constexpr reference operator[](difference_type const n) const
        requires std::random_access_iterator<it_t>
    {
        return *(*this + n);
    }
 
    constexpr value_type operator*() const noexcept
    {
        return current_chunk;
    }
 
private:
    std::ranges::range_difference_t<urng_t> chunk_size;
 
    it_t urng_begin;
 
    sentinel_t urng_end;
 
    value_type current_chunk;
 
    constexpr it_t get_next_end_of_chunk(it_t start_of_chunk) const
    {
        // Very similar to `return std::ranges::next(start_of_chunk, chunk_size, urng_end);`.
        // However, the STL checks that the direction of chunk_size and urng_end are the same for sized_sentinels when
        // -D_GLIBCXX_ASSERTIONS is enabled.
        // If start_of_chunk was moved past urng_end, we should constrain it.
        // =========X===========Y============
        //          ^           ^
        //        urng_end   new_start_it
        //          <-----------              // direction from iterator to bound
        //                      --------->    // direction from chunk_size
        // See https://eel.is/c++draft/range.iter.op.advance (next just takes and returns a copy of the iterator)
        // Note: n is chunk_size and always positive.
        if constexpr (std::sized_sentinel_for<sentinel_t, it_t>) // We can check whether we can jump.
        {
            if (chunk_size >= std::abs(urng_end - start_of_chunk))  // Remaining range smaller than chunk_size
                return std::ranges::next(start_of_chunk, urng_end); // Returns it_t which is equal to urng_end
            else                                                    // We can jump chunk_size many times
                return std::ranges::next(start_of_chunk, chunk_size);
        }
        else // We need to increment one by one to not cross urng_end.
        {
            for (std::ranges::range_difference_t<urng_t> increments{};
                 increments != chunk_size && start_of_chunk != urng_end;
                 ++increments)
            {
                ++start_of_chunk;
            }
 
            return start_of_chunk;
        }
    }
 
    constexpr it_t get_former_start_of_chunk(it_t end_of_chunk) const
    {
        // Very similar to `return std::ranges::prev(end_of_chunk, chunk_size, urng_begin);`.
        // However, the STL checks that the direction of chunk_size and urng_end are the same for sized_sentinels when
        // -D_GLIBCXX_ASSERTIONS is enabled.
        // If end_of_chunk was moved before urng_begin, we should constrain it.
        // =========X===========Y============
        //          ^           ^
        //      end_of_chunk  urng_begin
        // <---------                         // direction from chunk_size
        //           --------->               // direction from iterator to bound
        // See https://eel.is/c++draft/range.iter.op.advance (prev(i, n, bound) is equal to advance(i, -n, bound))
        // Note: n is chunk_size and always positive.
        if constexpr (std::sized_sentinel_for<sentinel_t, it_t>) // We can check whether we can jump.
        {
            if (chunk_size >= std::abs(urng_begin - end_of_chunk)) // Remaining range smaller than chunk_size
                return urng_begin;
            else // We can jump chunk_size many times
                return std::ranges::prev(end_of_chunk, chunk_size);
        }
        else // We need to decrement one by one to not cross urng_begin.
        {
            for (std::ranges::range_difference_t<urng_t> decrements{};
                 decrements != chunk_size && end_of_chunk != urng_begin;
                 ++decrements)
            {
                --end_of_chunk;
            }
 
            return end_of_chunk;
        }
    }
};
 
// ---------------------------------------------------------------------------------------------------------------------
// chunk_fn (adaptor definition)
// ---------------------------------------------------------------------------------------------------------------------
 
struct chunk_fn
{
    constexpr auto operator()(std::ptrdiff_t const chunk_size) const
    {
        return adaptor_from_functor{*this, chunk_size};
    }
 
    template <std::ranges::range urng_t>
    constexpr auto operator()(urng_t && urange, std::ranges::range_difference_t<urng_t> const chunk_size) const
    {
        static_assert(std::ranges::input_range<urng_t>,
                      "The range parameter to views::chunk must model std::ranges::input_range.");
 
        return chunk_view{std::forward<urng_t>(urange), chunk_size};
    }
};
 
} // namespace seqan3::detail
 
namespace seqan3::views
{
inline constexpr auto chunk = detail::chunk_fn{};
 
} // namespace seqan3::views