Split range adaptors

include/beman/str_split/CMakeLists.txt

@@ -10,6 +10,7 @@ if(BEMAN_STR_SPLIT_USE_MODULES)
                     config.hpp
                     str_split.hpp
                     todo.hpp
+                    split_when.hpp
                     "${PROJECT_BINARY_DIR}/include/beman/str_split/config_generated.hpp"
     )
 else()
@@ -21,6 +22,7 @@ else()
                     config.hpp
                     str_split.hpp
                     todo.hpp
+                    split_when.hpp
                     "${PROJECT_BINARY_DIR}/include/beman/str_split/config_generated.hpp"
     )
 endif()

include/beman/str_split/detail/non_propagating_cache.hpp

@@ -0,0 +1,43 @@
+#ifndef BEMAN_STR_SPLIT_DETAIL_NON_PROPAGATING_CACHE_HPP
+#define BEMAN_STR_SPLIT_DETAIL_NON_PROPAGATING_CACHE_HPP
+
+#include <optional>
+#include <type_traits>
+
+namespace beman::str_split::detail {
+
+template <class T>
+    requires std::is_object_v<T>
+struct non_propagating_cache : std::optional<T> {
+    non_propagating_cache() = default;
+
+    constexpr non_propagating_cache(const non_propagating_cache&) noexcept {}
+
+    constexpr non_propagating_cache(non_propagating_cache&& from) noexcept { this->reset(); }
+
+    ~non_propagating_cache() = default;
+
+    constexpr non_propagating_cache& operator=(const non_propagating_cache& from) noexcept {
+        if (this != std::addressof(from)) {
+            this->reset();
+        }
+        return *this;
+    }
+
+    constexpr non_propagating_cache& operator=(non_propagating_cache&& from) noexcept {
+        this->reset();
+        from.reset();
+        return *this;
+    }
+
+    template <class It>
+        requires std::is_constructible_v<T, std::iter_reference_t<const It&>>
+    constexpr T&
+    emplace_deref(const It& it) noexcept(std::is_nothrow_constructible_v<T, std::iter_reference_t<const It&>>) {
+        return this->emplace(*it);
+    }
+};
+
+} // namespace beman::str_split::detail
+
+#endif

include/beman/str_split/detail/split_iterator.hpp

@@ -0,0 +1,87 @@
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#ifndef BEMAN_STR_SPLIT_SPLIT_ITERATOR_HPP
+#define BEMAN_STR_SPLIT_SPLIT_ITERATOR_HPP
+
+#include <iterator>
+#include <ranges>
+
+namespace beman::str_split::detail {
+
+template <class Parent>
+class split_sentinel;
+
+template <class Parent>
+class split_iterator {
+    friend split_sentinel<Parent>;
+
+    using base_iterator = Parent::base_iterator;
+
+  public:
+    using iterator_concept  = std::forward_iterator_tag;
+    using iterator_category = std::input_iterator_tag;
+    using value_type        = Parent::base_subrange;
+    using difference_type   = std::iter_difference_t<base_iterator>;
+
+    split_iterator() = default;
+
+    constexpr split_iterator(Parent& parent, base_iterator current, value_type next)
+        : parent_(std::addressof(parent)), current_(std::move(current)), next_(std::move(next)) {}
+
+    constexpr base_iterator base() const { return current_; }
+
+    constexpr value_type operator*() const { return {current_, next_.begin()}; }
+
+    constexpr split_iterator& operator++() {
+        current_ = next_.begin();
+
+        if (current_ == std::ranges::end(parent_->base_)) {
+            trailing_empty_ = false;
+        } else {
+            current_ = next_.end();
+            if (current_ == std::ranges::end(parent_->base_)) {
+                trailing_empty_ = true;
+                next_           = {current_, current_};
+            } else {
+                next_ = parent_->find_next(current_);
+            }
+        }
+
+        return *this;
+    }
+
+    constexpr split_iterator operator++(int) {
+        auto prev = *this;
+        ++(*this);
+        return prev;
+    }
+
+    friend constexpr bool operator==(const split_iterator& lhs, const split_iterator& rhs) {
+        return lhs.current_ == rhs.current_ && lhs.trailing_empty_ == rhs.trailing_empty_;
+    }
+
+  private:
+    Parent*       parent_ = nullptr;
+    base_iterator current_{};
+    value_type    next_{};
+    bool          trailing_empty_ = false;
+};
+
+template <class Parent>
+class split_sentinel {
+  public:
+    split_sentinel() = default;
+
+    constexpr split_sentinel(Parent& parent) : end_(std::ranges::end(parent.base_)) {}
+
+    friend constexpr bool operator==(const split_iterator<Parent>& lhs, const split_sentinel& rhs) {
+        return lhs.current_ == rhs.end_ && !lhs.trailing_empty_;
+    }
+
+  private:
+    std::ranges::sentinel_t<decltype(std::declval<Parent>().base())> end_{};
+};
+
+} // namespace beman::str_split::detail
+
+#endif

include/beman/str_split/split.hpp

@@ -0,0 +1,160 @@
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+
+#ifndef BEMAN_STR_SPLIT_SPLIT_HPP
+#define BEMAN_STR_SPLIT_SPLIT_HPP
+
+#include <beman/str_split/detail/non_propagating_cache.hpp>
+#include <beman/str_split/detail/split_iterator.hpp>
+
+#include <ranges>
+#include <functional>
+
+namespace beman::str_split {
+
+namespace detail {
+
+template <class V, class Pattern>
+concept splittable_view =
+    std::ranges::view<V> && std::ranges::view<Pattern> &&
+    std::indirectly_comparable<std::ranges::iterator_t<V>, std::ranges::iterator_t<Pattern>, std::ranges::equal_to>;
+
+template <class Range>
+using single_value_view_t = std::ranges::single_view<std::ranges::range_value_t<Range>>;
+
+} // namespace detail
+
+template <std::ranges::forward_range V, std::ranges::forward_range Pattern>
+    requires detail::splittable_view<V, Pattern>
+class split_view : public std::ranges::view_interface<split_view<V, Pattern>> {
+  public:
+    using iterator = detail::split_iterator<split_view>;
+    friend iterator;
+
+    using sentinel = detail::split_sentinel<split_view>;
+    friend sentinel;
+
+    split_view()
+        requires std::default_initializable<V> && std::default_initializable<Pattern>
+    = default;
+
+    constexpr explicit split_view(V base, Pattern pattern) : base_(std::move(base)), pattern_(std::move(pattern)) {}
+
+    template <std::ranges::forward_range R>
+        requires std::constructible_from<V, std::ranges::views::all_t<R>> &&
+                     std::constructible_from<Pattern, detail::single_value_view_t<R>>
+    constexpr explicit split_view(R&& range, std::ranges::range_value_t<R> value)
+        : base_(std::ranges::views::all(std::forward<R>(range))),
+          pattern_(std::ranges::views::single(std::move(value))) {}
+
+    constexpr V base() const&
+        requires std::copy_constructible<V>
+    {
+        return base_;
+    }
+
+    constexpr V base() && { return std::move(base_); }
+
+    constexpr iterator begin() {
+        if (!cached_begin_) {
+            cached_begin_.emplace(find_next(std::ranges::begin(base_)));
+        }
+        return {*this, std::ranges::begin(base_), *cached_begin_};
+    }
+
+    constexpr auto end() {
+        if constexpr (std::ranges::common_range<V>) {
+            return iterator(*this, std::ranges::end(base_), {});
+        } else {
+            return sentinel(*this);
+        }
+    }
+
+  private:
+    using base_iterator = std::ranges::iterator_t<V>;
+    using base_subrange = std::ranges::subrange<base_iterator>;
+
+    constexpr base_subrange find_next(base_iterator it) {
+        auto [b, e] = std::ranges::search(std::ranges::subrange(it, std::ranges::end(base_)), pattern_);
+        if (b != std::ranges::end(base_) && std::ranges::empty(pattern_)) {
+            ++b;
+            ++e;
+        }
+        return {b, e};
+    }
+
+    [[no_unique_address]] V                      base_{};
+    [[no_unique_address]] Pattern                pattern_{};
+    detail::non_propagating_cache<base_subrange> cached_begin_{};
+};
+
+template <class R, class Pattern>
+split_view(R&&, Pattern&&) -> split_view<std::ranges::views::all_t<R>, std::ranges::views::all_t<Pattern>>;
+
+template <class R>
+split_view(R&&, std::ranges::range_value_t<R>)
+    -> split_view<std::ranges::views::all_t<R>, detail::single_value_view_t<R>>;
+
+namespace detail {
+
+template <class R, class Pattern>
+using split_view_t = decltype(split_view(std::declval<R>(), std::declval<Pattern>()));
+
+template <std::move_constructible Pattern>
+class split_adaptor_closure : public std::ranges::range_adaptor_closure<split_adaptor_closure<Pattern>> {
+  public:
+    constexpr explicit split_adaptor_closure(Pattern pattern) noexcept(std::is_nothrow_move_constructible_v<Pattern>)
+        : pattern_(std::move(pattern)) {}
+
+    template <class R>
+    constexpr split_view_t<R, Pattern> operator()(R&& range) const& noexcept(
+        std::is_nothrow_constructible_v<split_view_t<R, Pattern>, R, const Pattern&>) {
+        return split_view(std::forward<R>(range), pattern_);
+    }
+
+    template <class R>
+    constexpr split_view_t<R, Pattern>
+    operator()(R&& range) && noexcept(std::is_nothrow_constructible_v<split_view_t<R, Pattern>, R, Pattern>) {
+        return split_view(std::forward<R>(range), std::move(pattern_));
+    }
+
+  private:
+    Pattern pattern_{};
+};
+
+template <class Pattern>
+using split_adaptor_closure_t = split_adaptor_closure<std::decay_t<Pattern>>;
+
+namespace split {
+
+struct fn {
+    template <class R, class Pattern>
+    static constexpr split_view_t<R, Pattern>
+    operator()(R&&       range,
+               Pattern&& pattern) noexcept(std::is_nothrow_constructible_v<split_view_t<R, Pattern>, R, Pattern>) {
+        return split_view(std::forward<R>(range), std::forward<Pattern>(pattern));
+    }
+
+    template <class Pattern>
+    static constexpr split_adaptor_closure_t<Pattern> operator()(Pattern&& pattern) noexcept(
+        std::is_nothrow_constructible_v<split_adaptor_closure_t<Pattern>, Pattern>) {
+        return split_adaptor_closure(std::forward<Pattern>(pattern));
+    }
+};
+
+} // namespace split
+
+} // namespace detail
+
+namespace views {
+
+inline namespace rao {
+
+inline constexpr auto split = detail::split::fn{};
+
+}
+
+} // namespace views
+
+} // namespace beman::str_split
+
+#endif