Merge pull request godotengine#105347 from HolonProduction/lsp-utf8

LSP: Account for unicode identifiers

Author: Repiteo

modules/gdscript/language_server/gdscript_extend_parser.cpp

@@ -712,9 +712,9 @@ String ExtendGDScriptParser::get_identifier_under_position(const LSP::Position &
 	LSP::Position pos = p_position;
 	if (
 			pos.character >= line.length() // Cursor at end of line.
-			|| (!is_ascii_identifier_char(line[pos.character]) // Not on valid identifier char.
+			|| (!is_unicode_identifier_continue(line[pos.character]) // Not on valid identifier char.
 					   && (pos.character > 0 // Not line start -> there is a prev char.
-								  && is_ascii_identifier_char(line[pos.character - 1]) // Prev is valid identifier char.
+								  && is_unicode_identifier_continue(line[pos.character - 1]) // Prev is valid identifier char.
 								  ))) {
 		pos.character--;
 	}
@@ -723,7 +723,7 @@ String ExtendGDScriptParser::get_identifier_under_position(const LSP::Position &
 	for (int c = pos.character; c >= 0; c--) {
 		start_pos = c;
 		char32_t ch = line[c];
-		bool valid_char = is_ascii_identifier_char(ch);
+		bool valid_char = is_unicode_identifier_continue(ch);
 		if (!valid_char) {
 			break;
 		}
@@ -732,13 +732,17 @@ String ExtendGDScriptParser::get_identifier_under_position(const LSP::Position &
 	int end_pos = pos.character;
 	for (int c = pos.character; c < line.length(); c++) {
 		char32_t ch = line[c];
-		bool valid_char = is_ascii_identifier_char(ch);
+		bool valid_char = is_unicode_identifier_continue(ch);
 		if (!valid_char) {
 			break;
 		}
 		end_pos = c;
 	}
 
+	if (!is_unicode_identifier_start(line[start_pos + 1])) {
+		return "";
+	}
+
 	if (start_pos < end_pos) {
 		r_range.start.line = r_range.end.line = pos.line;
 		r_range.start.character = start_pos + 1;

modules/gdscript/tests/scripts/lsp/class.gd

@@ -58,16 +58,19 @@ class Inner2:
 		#      |  |   ^^^^ -> class2:func:arg2
 		#      ^^^^ -> class2:func:arg1
 
-class Inner3 extends Inner2:
+class Inner三 extends Inner2:
 #     |    |         ^^^^^^ -> class2
 #     ^^^^^^ class3 -> class3
 	var whatever = "foo"
 	#   ^^^^^^^^ class3:whatever -> class3:whatever
 
+	var ütf8 = ""
+	#   ^^^^ class3:utf8 -> class3:utf8
+
 	func _init():
 	#    ^^^^^ class3:init
 	# Note: no self-ref check here: resolves to `Object._init`.
-	#       usages of `Inner3.new()` DO resolve to this `_init`
+	#       usages of `Inner三.new()` DO resolve to this `_init`
 		pass
 
 	class NestedInInner3:
@@ -97,14 +100,18 @@ func _ready():
 	#     |    |  ^^^^^^ -> func:class1:value2
 	#     ^^^^^^ -> func:class1:value1
 
-	var inner3 = Inner3.new()
+	var inner3 = Inner三.new()
 	#   |    |   |    | ^^^ -> class3:init
 	#   |    |   ^^^^^^ -> class3
 	#   ^^^^^^ func:class3 -> func:class3
 	print(inner3)
 	#     ^^^^^^ -> func:class3
 
-	var nested1 = Inner3.NestedInInner3.new()
+	print(inner3.ütf8)
+	#     |    | ^^^^ -> class3:utf8
+	#     ^^^^^^ -> func:class3
+
+	var nested1 = Inner三.NestedInInner3.new()
 	#   |     |   |    | ^^^^^^^^^^^^^^ -> class3:nested1
 	#   |     |   ^^^^^^ -> class3
 	#   ^^^^^^^ func:class3:nested1 -> func:class3:nested1
@@ -115,7 +122,7 @@ func _ready():
 	print(value_nested1)
 	#     ^^^^^^^^^^^^^ -> func:class3:nested1:value
 
-	var nested2 = Inner3.AnotherNestedInInner3.new()
+	var nested2 = Inner三.AnotherNestedInInner3.new()
 	#   |     |   |    | ^^^^^^^^^^^^^^^^^^^^^ -> class3:nested2
 	#   |     |   ^^^^^^ -> class3
 	#   ^^^^^^^ func:class3:nested2 -> func:class3:nested2