Core/StringExt: Add Substring family of function (unicode-aware)

include/Nazara/Core/StringExt.hpp

@@ -48,6 +48,9 @@ namespace Nz
 	template<typename T> std::basic_string<T>& ReplaceStr(std::basic_string<T>& str, const T* from, const T* to);
 	template<typename T> std::basic_string<T>& ReplaceStr(std::basic_string<T>& str, std::basic_string_view<T> from, std::basic_string_view<T> to);
 
+	inline std::string_view Substring(std::string_view str, std::size_t index, UnicodeAware);
+	NAZARA_CORE_API std::string_view Substring(std::string_view str, std::size_t index, std::size_t count, UnicodeAware);
+
 	inline bool StartsWith(std::string_view str, std::string_view s);
 	NAZARA_CORE_API bool StartsWith(std::string_view lhs, std::string_view rhs, CaseIndependent);
 	NAZARA_CORE_API bool StartsWith(std::string_view lhs, std::string_view rhs, UnicodeAware);

include/Nazara/Core/StringExt.inl

@@ -6,6 +6,7 @@
 #include <algorithm>
 #include <cctype>
 #include <cstring>
+#include <limits>
 #include <Nazara/Core/Debug.hpp>
 
 namespace Nz
@@ -109,6 +110,11 @@ namespace Nz
 		return str;
 	}
 
+	std::string_view Substring(std::string_view str, std::size_t index, UnicodeAware)
+	{
+		return Substring(str, index, std::numeric_limits<std::size_t>::max(), UnicodeAware{});
+	}
+
 	inline bool StartsWith(std::string_view str, std::string_view s)
 	{
 #if NAZARA_CPP_VER >= NAZARA_CPP20

src/Nazara/Core/StringExt.cpp

@@ -389,6 +389,35 @@ namespace Nz
 		return str;
 	}
 
+	std::string_view Substring(std::string_view str, std::size_t index, std::size_t count, UnicodeAware)
+	{
+		const char* start = str.data();
+		const char* end = start + str.size();
+		try
+		{
+			utf8::advance(start, index, end);
+		}
+		catch (const utf8::not_enough_room&)
+		{
+			return {};
+		}
+
+		if (count == std::numeric_limits<std::size_t>::max())
+			return str.substr(start - str.data());
+
+		const char* to = start;
+		try
+		{
+			utf8::advance(to, count, end);
+		}
+		catch (const utf8::not_enough_room&)
+		{
+			return str.substr(start - str.data());
+		}
+
+		return std::string_view(start, to - start);
+	}
+
 	bool StartsWith(std::string_view lhs, std::string_view rhs, CaseIndependent)
 	{
 		NAZARA_USE_ANONYMOUS_NAMESPACE

tests/UnitTests/Engine/Core/StringExtTest.cpp

@@ -112,6 +112,15 @@ SCENARIO("String", "[CORE][STRING]")
 		REQUIRE(Nz::ReplaceStr(str, "Unreal Reality", "Ungine") == "Ungine");
 	}
 
+	WHEN("Getting substring of unicode strings")
+	{
+		std::string_view str = "\u00E0\u00E9\u00E7\u0153\u00C2\u5B98";
+		CHECK(Nz::Substring(str, 0, 150, Nz::UnicodeAware{}) == "\u00E0\u00E9\u00E7\u0153\u00C2\u5B98");
+		CHECK(Nz::Substring(str, 1, Nz::UnicodeAware{}) == "\u00E9\u00E7\u0153\u00C2\u5B98");
+		CHECK(Nz::Substring(str, 1, 2, Nz::UnicodeAware{}) == "\u00E9\u00E7");
+		CHECK(Nz::Substring(str, 1, 10, Nz::UnicodeAware{}) == "\u00E9\u00E7\u0153\u00C2\u5B98");
+	}
+
 	WHEN("Checking if string starts with")
 	{
 		CHECK(Nz::StartsWith("Nazara Engine", ""));