String[size()] should return a default constructed CharType

As per the C++ standard 21.3.4.1 for std::string:
Returns: If pos < size(), returns data()[pos]. Otherwise, if pos ==
size(), the const version returns charT(). Otherwise, the behavior is
undefined.

Since the behavior is undefined Godot now does the same thing for const
and non-const versions of operator[].

This fixes #21242 and fixes #22221.

core/ustring.cpp

@@ -58,6 +58,9 @@
 #define IS_DIGIT(m_d) ((m_d) >= '0' && (m_d) <= '9')
 #define IS_HEX_DIGIT(m_d) (((m_d) >= '0' && (m_d) <= '9') || ((m_d) >= 'a' && (m_d) <= 'f') || ((m_d) >= 'A' && (m_d) <= 'F'))
 
+const char CharString::_null = 0;
+const CharType String::_null = 0;
+
 bool is_symbol(CharType c) {
 	return c != '_' && ((c >= '!' && c <= '/') || (c >= ':' && c <= '@') || (c >= '[' && c <= '`') || (c >= '{' && c <= '~') || c == '\t' || c == ' ');
 }

core/ustring.h

@@ -47,6 +47,7 @@ class CharProxy {
 
 	const int _index;
 	CowData<T> &_cowdata;
+	static const T _null = 0;
 
 	_FORCE_INLINE_ CharProxy(const int &p_index, CowData<T> &cowdata) :
 			_index(p_index),
@@ -54,6 +55,9 @@ class CharProxy {
 
 public:
 	_FORCE_INLINE_ operator T() const {
+		if (unlikely(_index == _cowdata.size()))
+			return _null;
+
 		return _cowdata.get(_index);
 	}
 
@@ -73,6 +77,7 @@ public:
 class CharString {
 
 	CowData<char> _cowdata;
+	static const char _null;
 
 public:
 	_FORCE_INLINE_ char *ptrw() { return _cowdata.ptrw(); }
@@ -83,7 +88,12 @@ public:
 	_FORCE_INLINE_ char get(int p_index) { return _cowdata.get(p_index); }
 	_FORCE_INLINE_ const char get(int p_index) const { return _cowdata.get(p_index); }
 	_FORCE_INLINE_ void set(int p_index, const char &p_elem) { _cowdata.set(p_index, p_elem); }
-	_FORCE_INLINE_ const char &operator[](int p_index) const { return _cowdata.get(p_index); }
+	_FORCE_INLINE_ const char &operator[](int p_index) const {
+		if (unlikely(p_index == _cowdata.size()))
+			return _null;
+
+		return _cowdata.get(p_index);
+	}
 	_FORCE_INLINE_ CharProxy<char> operator[](int p_index) { return CharProxy<char>(p_index, _cowdata); }
 
 	_FORCE_INLINE_ CharString() {}
@@ -112,6 +122,7 @@ struct StrRange {
 class String {
 
 	CowData<CharType> _cowdata;
+	static const CharType _null;
 
 	void copy_from(const char *p_cstr);
 	void copy_from(const CharType *p_cstr, const int p_clip_to = -1);
@@ -138,7 +149,12 @@ public:
 	_FORCE_INLINE_ int size() const { return _cowdata.size(); }
 	Error resize(int p_size) { return _cowdata.resize(p_size); }
 
-	_FORCE_INLINE_ const CharType &operator[](int p_index) const { return _cowdata.get(p_index); }
+	_FORCE_INLINE_ const CharType &operator[](int p_index) const {
+		if (unlikely(p_index == _cowdata.size()))
+			return _null;
+
+		return _cowdata.get(p_index);
+	}
 	_FORCE_INLINE_ CharProxy<CharType> operator[](int p_index) { return CharProxy<CharType>(p_index, _cowdata); }
 
 	bool operator==(const String &p_str) const;

main/tests/test_string.cpp

@@ -941,6 +941,33 @@ bool test_30() {
 	state = state && success;
 	OS::get_singleton()->print("Capitalize %ls: %ls, %s\n", input.c_str(), output.c_str(), success ? "OK" : "FAIL");
 
+	return state;
+}
+
+bool test_31() {
+	bool state = true;
+	bool success;
+
+	String a = "";
+	success = a[0] == 0;
+	OS::get_singleton()->print("Is 0 String[0]:, %s\n", success ? "OK" : "FAIL");
+	if (!success) state = false;
+
+	String b = "Godot";
+	success = b[b.size()] == 0;
+	OS::get_singleton()->print("Is 0 String[size()]:, %s\n", success ? "OK" : "FAIL");
+	if (!success) state = false;
+
+	const String c = "";
+	success = c[0] == 0;
+	OS::get_singleton()->print("Is 0 const String[0]:, %s\n", success ? "OK" : "FAIL");
+	if (!success) state = false;
+
+	const String d = "Godot";
+	success = d[d.size()] == 0;
+	OS::get_singleton()->print("Is 0 const String[size()]:, %s\n", success ? "OK" : "FAIL");
+	if (!success) state = false;
+
 	return state;
 };
 
@@ -978,6 +1005,7 @@ TestFunc test_funcs[] = {
 	test_28,
 	test_29,
 	test_30,
+	test_31,
 	0
 
 };