Pattern Analysis

Adds functionality to analyze the minimum and maximum # of
characters a regex may match.

Author: Andersama

include/ctre/evaluation.hpp

@@ -447,6 +447,249 @@ constexpr CTRE_FORCE_INLINE R evaluate(const Iterator begin, Iterator current, c
 // property matching
 
 
+// pattern analysis - returns the minimum and maximum # of characters in order for a regex to match a string
+// a custom std::pair to overload some handy operations that we'll perform w/ a fold
+struct analysis_results : std::pair<size_t, size_t> {
+	// -1 is considered INF, -2 is finite (but perhaps too large to store), all other values are exact counts
+	static constexpr CTRE_FORCE_INLINE size_t saturate_limit(const size_t& lhs, const size_t& rhs) {
+		const constexpr size_t inf = size_t{ 0 } -1;
+		const constexpr size_t lim = size_t{ 0 } -2;
+		size_t ret = inf;
+		if (lhs == inf || rhs == inf) {
+			return ret;
+		} else {
+			ret = lhs + rhs;
+			ret = ret < lhs ? lim : ret == inf ? lim : ret;
+		}
+		return ret;
+	}
+
+	static constexpr CTRE_FORCE_INLINE size_t mult_saturate_limit(const size_t& lhs, const size_t& rhs) {
+		const constexpr size_t inf = size_t{ 0 } -1;
+		const constexpr size_t lim = size_t{ 0 } -2;
+		size_t ret = inf;
+		if (lhs == inf || rhs == inf) {
+			return ret;
+		} else if (lhs == 0 || rhs == 0) {
+			return ret = 0;
+		} else {
+			if (lhs > (SIZE_MAX / rhs))
+				return ret = lim;
+			ret = lhs * rhs;
+			ret = ret == inf ? lim : ret;
+			return ret;
+		}
+	}
+	
+	constexpr inline CTRE_FORCE_INLINE operator bool() const noexcept {
+		return first;
+	}
+	constexpr auto CTRE_FORCE_INLINE operator+(analysis_results other) const noexcept {
+		return analysis_results{std::make_pair(
+			saturate_limit(first, other.first),
+			saturate_limit(second, other.second)
+		)};
+	}
+	constexpr auto CTRE_FORCE_INLINE operator||(analysis_results other) const noexcept {
+		return analysis_results{std::make_pair(
+			std::min(first, other.first),
+			std::max(second, other.second)
+		)};
+	}
+};
+
+template <typename Pattern>
+static constexpr auto trampoline_analysis(Pattern) noexcept;
+
+template <typename... Patterns>
+static constexpr auto trampoline_analysis(ctll::list<Patterns...>) noexcept;
+
+template<typename T, typename R>
+static constexpr auto trampoline_analysis(T, R captures) noexcept;
+
+//processing for each type
+
+//repeat
+template<size_t A, size_t B, typename R, typename... Content>
+static constexpr auto _analyze(repeat<A,B,Content...>, R captures) noexcept {
+	analysis_results ret{ std::make_pair(0ULL, 0ULL) };
+	if constexpr (sizeof...(Content)) {
+		ret = trampoline_analysis(ctll::list<Content...>(), captures);
+		ret.first = analysis_results::mult_saturate_limit(ret.first, A);
+		ret.second = analysis_results::mult_saturate_limit(ret.second, B);
+	}
+	return ret;
+}
+
+//note: all * ? + operations are specialized variations of repeat {A,B}
+//lazy_repeat
+template<size_t A, size_t B, typename R, typename... Content>
+static constexpr auto _analyze(lazy_repeat<A, B, Content...>, R captures) noexcept {
+	return _analyze(repeat<A, B, Content...>(), captures);
+}
+
+//possessive_repeat
+template<size_t A, size_t B, typename R, typename... Content>
+static constexpr auto _analyze(possessive_repeat<A, B, Content...>, R captures) noexcept {
+	return _analyze(repeat<A, B, Content...>(), captures);
+}
+
+//star
+template<typename R, typename... Content>
+static constexpr auto _analyze(star<Content...>, R captures) noexcept {
+	return _analyze(repeat<0ULL, ~(0ULL), Content...>(), captures);
+}
+
+//lazy_star
+template<typename R, typename... Content>
+static constexpr auto _analyze(lazy_star<Content...>, R captures) noexcept {
+	return _analyze(repeat<0ULL, ~(0ULL), Content...>(), captures);
+}
+
+//possessive_star
+template<typename R, typename... Content>
+static constexpr auto _analyze(possessive_star<Content...>, R captures) noexcept {
+	return _analyze(repeat<0ULL, ~(0ULL), Content...>(), captures);
+}
+
+//plus
+template<typename R, typename... Content>
+static constexpr auto _analyze(plus<Content...>, R captures) noexcept {
+	return _analyze(repeat<1ULL, ~(0ULL), Content...>(), captures);
+}
+
+//lazy_plus
+template<typename R, typename... Content>
+static constexpr auto _analyze(lazy_star<Content...>, R captures) noexcept {
+	return _analyze(repeat<1ULL, ~(0ULL), Content...>(), captures);
+}
+
+//possessive_plus
+template<typename R, typename... Content>
+static constexpr auto _analyze(possessive_star<Content...>, R captures) noexcept {
+	return _analyze(repeat<1ULL, ~(0ULL), Content...>(), captures);
+}
+
+//optional
+template<typename R, typename... Content>
+static constexpr auto _analyze(optional<Content...>, R captures) noexcept {
+	return _analyze(repeat<0ULL, 1ULL, Content...>(), captures);
+}
+
+//lazy_optional
+template<typename R, typename... Content>
+static constexpr auto _analyze(lazy_optional<Content...>, R captures) noexcept {
+	return _analyze(repeat<0ULL, 1ULL, Content...>(), captures);
+}
+
+//back_reference
+template<size_t Id, typename R>
+static constexpr auto _analyze(back_reference<Id>, R captures) noexcept {
+	const auto ref = captures.template get<Id>();
+	analysis_results ret{ std::make_pair(0ULL, 0ULL) };
+	if constexpr (size(ref.get_expression())) {
+		ret = trampoline_analysis(ref.get_expression(), captures);
+	}
+	return ret;
+}
+
+//back_reference_with_name
+template<typename Name, typename R>
+static constexpr auto _analyze(back_reference_with_name<Name>, R captures) noexcept {
+	const auto ref = captures.template get<Name>();
+	analysis_results ret{ std::make_pair(0ULL, 0ULL) };
+	if constexpr (size(ref.get_expression())) {
+		ret = trampoline_analysis(ref.get_expression(), captures);
+	}
+	return ret;
+}
+
+//select, this is specialized, we need to take the minimum of all minimums and maximum of all maximums
+template<typename R, typename... Content>
+static constexpr auto _analyze(select<Content...>, R captures) noexcept {
+	analysis_results ret = trampoline_select_analysis(ctll::list<Content...>(), captures);
+	return ret;
+}
+
+//character, any character contributes exactly one to both counts
+template<auto C, typename R>
+static constexpr auto _analyze(character<C>, R captures) noexcept {
+	analysis_results ret{ std::make_pair(1ULL, 1ULL) };
+	return ret;
+}
+
+//strings, any string contributes the # of characters it contains (if we have an empty string that'll be 0)
+template<auto... Str, typename R>
+static constexpr auto _analyze(string<Str...>, R captures) noexcept {
+	analysis_results ret{ std::make_pair(sizeof...(Str), sizeof...(Str)) };
+	return ret;
+}
+
+//we'll process anything that has contents as a regex
+//ctll::list
+template<typename R, typename... Content>
+static constexpr auto _analyze(ctll::list<Content...>,R captures) noexcept {
+	analysis_results ret = trampoline_analysis(ctll::list<Content...>(), captures);
+	return ret;
+}
+
+//sequence
+template<typename R, typename... Content>
+static constexpr auto _analyze(sequence<Content...>, R captures) noexcept {
+	analysis_results ret = trampoline_analysis(ctll::list<Content...>(), captures);
+	return ret;
+}
+
+//capture
+template<size_t Id, typename R, typename... Content>
+static constexpr auto _analyze(capture<Id, Content...>, R captures) noexcept {
+	analysis_results ret = trampoline_analysis(ctll::list<Content...>(), captures);
+	return ret;
+}
+
+//capture_with_name
+template<size_t Id, typename Name, typename R, typename... Content>
+static constexpr auto _analyze(capture_with_name<Id, Name, Content...>, R captures) noexcept {
+	analysis_results ret = trampoline_analysis(ctll::list<Content...>(), captures);
+	return ret;
+}
+
+//everything else, anything we haven't matched already isn't supported and will contribute 0
+template<typename T, typename R>
+static constexpr auto _analyze(T, R captures) noexcept {
+	analysis_results ret{ std::make_pair(0ULL, 0ULL) };
+	return ret;
+}
+//note: ctll::list wraps patterns just like sequences, we'll treat anything that looks like a regex w/ ctll::list
+template <typename... Patterns, typename R>
+static constexpr auto trampoline_analysis(ctll::list<Patterns...>, R captures) noexcept {
+	//fold, for every argument in a ctll::list, calculate its contribution to the limits
+	auto r = ((_analyze(Patterns(), captures)) + ...);
+	//note any reordering of parameters will result in the same limits
+	return r;
+}
+
+template <typename... Patterns, typename R>
+static constexpr auto trampoline_select_analysis(ctll::list<Patterns...>, R captures) noexcept {
+	//fold, each argument in a selection of regexes we take the minimum and maximum of all values
+	auto r = ((trampoline_analysis(Patterns(), captures)) || ...);
+	//note again, order is unimportant
+	return r;
+}
+
+template <typename... Patterns>
+static constexpr auto pattern_analysis(ctll::list<Patterns...>) noexcept {
+	using return_type = decltype(regex_results(std::declval<std::basic_string_view<char>::iterator>(), find_captures(pattern)));
+	return trampoline_analysis(ctll::list<Patterns...>(), return_type{});
+}
+
+template <typename Pattern = empty>
+static constexpr auto pattern_analysis(Pattern pattern = {}) noexcept {
+	using return_type = decltype(regex_results(std::declval<std::basic_string_view<char>::iterator>(), find_captures(pattern)));
+	return trampoline_analysis(ctll::list<Pattern>(), return_type{});
+}
+
+
 }
 
 #endif

include/ctre/find_captures.hpp

@@ -112,12 +112,12 @@ template <typename... Content, typename... Tail, typename Output> constexpr auto
 
 
 template <size_t Id, typename... Content, typename... Tail, typename... Output> constexpr auto find_captures(ctll::list<capture<Id,Content...>, Tail...>, ctll::list<Output...>) noexcept {
-	return find_captures(ctll::list<Content..., Tail...>(), ctll::list<Output..., captured_content<Id>>());
+	return find_captures(ctll::list<Content..., Tail...>(), ctll::list<Output..., captured_content<Id, void, ctll::list<Content...>>>());
 }
 
 
 template <size_t Id, typename Name, typename... Content, typename... Tail, typename... Output> constexpr auto find_captures(ctll::list<capture_with_name<Id,Name,Content...>, Tail...>, ctll::list<Output...>) noexcept {
-	return find_captures(ctll::list<Content..., Tail...>(), ctll::list<Output..., captured_content<Id, Name>>());
+	return find_captures(ctll::list<Content..., Tail...>(), ctll::list<Output..., captured_content<Id, Name, ctll::list<Content...>>>());
 }
 
 

include/ctre/return_type.hpp

@@ -13,15 +13,15 @@ struct not_matched_tag_t { };
 
 static constexpr inline auto not_matched = not_matched_tag_t{};
 
-template <size_t Id, typename Name = void> struct captured_content {
+template <size_t Id, typename Name = void, typename Content = void> struct captured_content {
 	template <typename Iterator> class storage {
 		Iterator _begin{};
 		Iterator _end{};
 
 		bool _matched{false};
 	public:
 		using char_type = typename std::iterator_traits<Iterator>::value_type;
-		
+		using content_type = Content;
 		using name = Name;
 
 		constexpr CTRE_FORCE_INLINE storage() noexcept {}
@@ -86,6 +86,10 @@ template <size_t Id, typename Name = void> struct captured_content {
 		constexpr CTRE_FORCE_INLINE static size_t get_id() noexcept {
 			return Id;
 		}
+		
+		constexpr CTRE_FORCE_INLINE static content_type get_expression() noexcept {
+			return {};
+		}
 	};
 };