optimizes string matching by allowing memcmp like functionality (even…

… on utf8 sequences)

reference: #147
comparison: https://compiler-explorer.com/z/Tz3KhG

include/ctre/evaluation.hpp

@@ -115,11 +115,52 @@ template <typename CharT, typename Iterator, typename EndIterator> constexpr CTR
 	return false;
 }
 
-template <auto... String, size_t... Idx, typename Iterator, typename EndIterator> constexpr CTRE_FORCE_INLINE string_match_result<Iterator> evaluate_match_string(Iterator current, [[maybe_unused]] const EndIterator end, std::index_sequence<Idx...>) noexcept {
-
-	bool same = (compare_character(String, current, end) && ... && true);
+#if __cpp_char8_t >= 201811
+template <size_t N, size_t... Idx, typename Iterator, typename EndIterator> constexpr CTRE_FORCE_INLINE string_match_result<Iterator> evaluate_match_utf8_string(Iterator current, [[maybe_unused]] const EndIterator end, char8_t (&buffer)[N], std::index_sequence<Idx...>) noexcept {
+	//abuse inside knowledge of how utf8_iterator works
+	if constexpr (!std::is_same_v<::std::remove_const_t<EndIterator>, utf8_iterator::sentinel>) {
+		size_t count = end.ptr - current.ptr; //size_t count = std::distance(current.ptr, end.ptr);
+		size_t bump = ((count < N) ? count : N);
+		return { Iterator{current.ptr + bump, current.end}, (count >= N) && (((current.ptr[Idx] == buffer[Idx]) + ... + 0) == sizeof...(Idx)) };
+	} else {
+		size_t count = current.end - current.ptr; //size_t count = std::distance(current.ptr, current.end);
+		size_t bump = ((count < N) ? count : N);
+		return { Iterator{current.ptr + bump, current.end}, (count >= N) && (((current.ptr[Idx] == buffer[Idx]) + ... + 0) == sizeof...(Idx)) };
+	}
+}
+#endif
 
-	return {current, same};
+template <auto... String, size_t... Idx, typename Iterator, typename EndIterator> constexpr CTRE_FORCE_INLINE string_match_result<Iterator> evaluate_match_string(Iterator current, [[maybe_unused]] const EndIterator end, std::index_sequence<Idx...>) noexcept {
+#if __cpp_char8_t >= 201811
+	if constexpr (std::is_same_v<::std::remove_const_t<Iterator>, utf8_iterator> && (std::is_same_v<std::remove_const_t<Iterator>, std::remove_const_t<EndIterator>> || std::is_same_v<::std::remove_const_t<EndIterator>, utf8_iterator::sentinel>)) {
+		constexpr size_t str_length = (utf8_codepoint_length(String) + ... + 0ULL);
+		//encode our String... into it's utf8 representation
+		char8_t utf8_sequence[str_length];
+		char8_t* ptr = utf8_sequence;
+		((ptr = utf32_codepoint_to_utf8_codepoint(String, ptr)), ...);
+		//run the comparison
+		return evaluate_match_utf8_string(current, end, utf8_sequence, std::make_index_sequence<str_length>());
+	} else if constexpr (is_random_accessible(typename std::iterator_traits<Iterator>::iterator_category{}) && std::is_same_v<std::remove_const_t<Iterator>, std::remove_const_t<EndIterator>>) {
+		//check the remaining bytes*
+		size_t count = end - current;
+		//make sure we only "bump" the iterator a safe distance
+		size_t bump = ((count < sizeof...(String)) ? count : sizeof...(String));
+		//do math against how many characters we match, avoid as many branches as possible
+		return { current + bump, (count >= sizeof...(String)) && (((current[Idx] == String) + ... + 0) == sizeof...(String)) };
+	} else {
+		bool same = (compare_character(String, current, end) && ... && true);
+		return { current, same };
+	}
+#else
+	if constexpr (is_random_accessible(typename std::iterator_traits<Iterator>::iterator_category{}) && std::is_same_v<std::remove_const_t<Iterator>, std::remove_const_t<EndIterator>>) {
+		size_t count = end - current;
+		size_t bump = ((count < sizeof...(String)) ? count : sizeof...(String));
+		return { current + bump, (count >= sizeof...(String)) && (((current[Idx] == String) + ... + 0) == sizeof...(String)) };
+	} else {
+		bool same = (compare_character(String, current, end) && ... && true);
+		return { current, same };
+	}
+#endif
 }
 
 template <typename R, typename Iterator, typename EndIterator, auto... String, typename... Tail> 

include/ctre/utf8.hpp

@@ -8,6 +8,44 @@
 #include <iterator>
 
 namespace ctre {
+constexpr char8_t* utf32_codepoint_to_utf8_codepoint(uint32_t code, char8_t *ptr) {
+	if (code < 0x80) {
+		ptr[0] = code;
+		return ptr + 1;
+	} else if (code < 0x800) {   // 00000yyy yyxxxxxx
+		ptr[0] = (0b11000000 | (code >> 6));
+		ptr[1] = (0b10000000 | (code & 0x3f));
+		return ptr + 2;
+	} else if (code < 0x10000) {  // zzzzyyyy yyxxxxxx
+		ptr[0] = (0b11100000 | (code >> 12));         // 1110zzz
+		ptr[1] = (0b10000000 | ((code >> 6) & 0x3f)); // 10yyyyy
+		ptr[2] = (0b10000000 | (code & 0x3f));        // 10xxxxx
+		return ptr + 3;
+	} else if (code < 0x200000) { // 000uuuuu zzzzyyyy yyxxxxxx
+		ptr[0] = (0b11110000 | (code >> 18));          // 11110uuu
+		ptr[1] = (0b10000000 | ((code >> 12) & 0x3f)); // 10uuzzzz
+		ptr[2] = (0b10000000 | ((code >> 6)  & 0x3f)); // 10yyyyyy
+		ptr[3] = (0b10000000 | (code & 0x3f));         // 10xxxxxx
+		return ptr + 4;
+	} else {
+		ptr[0] = 0xff; //invalid start byte
+		return ptr + 1;
+	}
+}
+
+constexpr uint32_t utf8_codepoint_length(uint32_t code) {
+	if (code < 0x80) {
+		return 1;
+	} else if (code < 0x800) {   // 00000yyy yyxxxxxx
+		return 2;
+	} else if (code < 0x10000) {  // zzzzyyyy yyxxxxxx
+		return 3;
+	} else if (code < 0x200000) { // 000uuuuu zzzzyyyy yyxxxxxx
+		return 4;
+	} else {
+		return 1;
+	}
+}
 
 struct utf8_iterator {
 	using self_type = utf8_iterator;