Merge pull request #72975 from vnen/gdscript-no-ascii-spoof-check

GDScript: Be more lenient with identifiers

modules/gdscript/gdscript_parser.cpp

@@ -2145,7 +2145,12 @@ GDScriptParser::ExpressionNode *GDScriptParser::parse_precedence(Precedence p_pr
 	make_completion_context(COMPLETION_IDENTIFIER, nullptr);
 
 	GDScriptTokenizer::Token token = current;
-	ParseFunction prefix_rule = get_rule(token.type)->prefix;
+	GDScriptTokenizer::Token::Type token_type = token.type;
+	if (token.is_identifier()) {
+		// Allow keywords that can be treated as identifiers.
+		token_type = GDScriptTokenizer::Token::IDENTIFIER;
+	}
+	ParseFunction prefix_rule = get_rule(token_type)->prefix;
 
 	if (prefix_rule == nullptr) {
 		// Expected expression. Let the caller give the proper error message.
@@ -3010,7 +3015,14 @@ GDScriptParser::ExpressionNode *GDScriptParser::parse_get_node(ExpressionNode *p
 			path_state = PATH_STATE_NODE_NAME;
 		} else if (current.is_node_name()) {
 			advance();
-			get_node->full_path += previous.get_identifier();
+			String identifier = previous.get_identifier();
+#ifdef DEBUG_ENABLED
+			// Check spoofing.
+			if (TS->has_feature(TextServer::FEATURE_UNICODE_SECURITY) && TS->spoof_check(identifier)) {
+				push_warning(get_node, GDScriptWarning::CONFUSABLE_IDENTIFIER, identifier);
+			}
+#endif
+			get_node->full_path += identifier;
 
 			path_state = PATH_STATE_NODE_NAME;
 		} else if (!check(GDScriptTokenizer::Token::SLASH) && !check(GDScriptTokenizer::Token::PERCENT)) {

modules/gdscript/gdscript_tokenizer.cpp

@@ -168,7 +168,11 @@ bool GDScriptTokenizer::Token::is_identifier() const {
 	switch (type) {
 		case IDENTIFIER:
 		case MATCH: // Used in String.match().
-		case CONST_INF: // Used in Vector{2,3,4}.INF
+		// Allow constants to be treated as regular identifiers.
+		case CONST_PI:
+		case CONST_INF:
+		case CONST_NAN:
+		case CONST_TAU:
 			return true;
 		default:
 			return false;
@@ -188,6 +192,10 @@ bool GDScriptTokenizer::Token::is_node_name() const {
 		case CLASS_NAME:
 		case CLASS:
 		case CONST:
+		case CONST_PI:
+		case CONST_INF:
+		case CONST_NAN:
+		case CONST_TAU:
 		case CONTINUE:
 		case ELIF:
 		case ELSE:
@@ -530,9 +538,12 @@ void GDScriptTokenizer::make_keyword_list() {
 #endif // DEBUG_ENABLED
 
 GDScriptTokenizer::Token GDScriptTokenizer::potential_identifier() {
+	bool only_ascii = _peek(-1) < 128;
+
 	// Consume all identifier characters.
 	while (is_unicode_identifier_continue(_peek())) {
-		_advance();
+		char32_t c = _advance();
+		only_ascii = only_ascii && c < 128;
 	}
 
 	int len = _current - _start;
@@ -587,7 +598,7 @@ GDScriptTokenizer::Token GDScriptTokenizer::potential_identifier() {
 
 #ifdef DEBUG_ENABLED
 	// Additional checks for identifiers but only in debug and if it's available in TextServer.
-	if (TS->has_feature(TextServer::FEATURE_UNICODE_SECURITY)) {
+	if (!only_ascii && TS->has_feature(TextServer::FEATURE_UNICODE_SECURITY)) {
 		int64_t confusable = TS->is_confusable(name, keyword_list);
 		if (confusable >= 0) {
 			push_error(vformat(R"(Identifier "%s" is visually similar to the GDScript keyword "%s" and thus not allowed.)", name, keyword_list[confusable]));

modules/gdscript/tests/scripts/parser/features/allow_id_similar_to_keyword_in_ascii.gd

@@ -0,0 +1,3 @@
+func test():
+	var P1 = "ok" # Technically it is visually similar to keyword "PI" but allowed since it's in ASCII range.
+	print(P1)

modules/gdscript/tests/scripts/parser/features/allow_id_similar_to_keyword_in_ascii.out

@@ -0,0 +1,2 @@
+GDTEST_OK
+ok

modules/gdscript/tests/scripts/parser/features/allowed_keywords_as_identifiers.gd

@@ -0,0 +1,16 @@
+func test():
+	# The following keywords are allowed as identifiers:
+	var match = "match"
+	print(match)
+
+	var PI = "PI"
+	print(PI)
+
+	var INF = "INF"
+	print(INF)
+
+	var NAN = "NAN"
+	print(NAN)
+
+	var TAU = "TAU"
+	print(TAU)

modules/gdscript/tests/scripts/parser/features/allowed_keywords_as_identifiers.out

@@ -0,0 +1,6 @@
+GDTEST_OK
+match
+PI
+INF
+NAN
+TAU

modules/gdscript/tests/scripts/parser/warnings/confusable_identifier.gd

@@ -1,5 +1,12 @@
+extends Node
+
 func test():
 	var port = 0 # Only latin characters.
 	var pοrt = 1 # The "ο" is Greek omicron.
 
 	prints(port, pοrt)
+
+# Do not call this since nodes aren't in the tree. It is just a parser check.
+func nodes():
+	var _node1 = $port # Only latin characters.
+	var _node2 = $pοrt # The "ο" is Greek omicron.

modules/gdscript/tests/scripts/parser/warnings/confusable_identifier.out

@@ -1,6 +1,10 @@
 GDTEST_OK
 >> WARNING
->> Line: 3
+>> Line: 5
+>> CONFUSABLE_IDENTIFIER
+>> The identifier "pοrt" has misleading characters and might be confused with something else.
+>> WARNING
+>> Line: 12
 >> CONFUSABLE_IDENTIFIER
 >> The identifier "pοrt" has misleading characters and might be confused with something else.
 0 1