WIP

gdtoolkit/parser/gdscript_indenter.py

@@ -0,0 +1,61 @@
+from typing import Iterator
+
+from lark import Token, indenter
+
+
+class GDScriptIndenter(indenter.Indenter):
+    NL_type = "_NL"
+    OPEN_PAREN_types = ["LPAR", "LSQB", "LBRACE"]
+    CLOSE_PAREN_types = ["RPAR", "RSQB", "RBRACE"]
+    INDENT_type = "_INDENT"
+    DEDENT_type = "_DEDENT"
+    # TODO: guess tab length
+    tab_len = 4
+
+    def handle_NL(self, token: Token) -> Iterator[Token]:
+        if self.paren_level > 0:
+            return              # TODO: special handling for lambdas
+
+        yield token
+
+        indent_str = token.rsplit("\n", 1)[1]  # Tabs and spaces
+        indent = indent_str.count(" ") + indent_str.count("\t") * self.tab_len
+
+        if indent > self.indent_level[-1]:
+            self.indent_level.append(indent)
+            yield Token.new_borrow_pos(self.INDENT_type, indent_str, token)
+        else:
+            while indent < self.indent_level[-1]:
+                self.indent_level.pop()
+                yield Token.new_borrow_pos(self.DEDENT_type, indent_str, token)
+
+            if indent != self.indent_level[-1]:
+                raise DedentError(
+                    "Unexpected dedent to column %s. Expected dedent to %s"
+                    % (indent, self.indent_level[-1])
+                )
+
+    def _process(self, stream):
+        for token in stream:
+            if token.type == self.NL_type:
+                yield from self.handle_NL(token)
+            else:
+                yield token
+
+            if token.type in self.OPEN_PAREN_types:
+                self.paren_level += 1
+            elif token.type in self.CLOSE_PAREN_types:
+                self.paren_level -= 1
+                assert self.paren_level >= 0
+
+        while len(self.indent_level) > 1:
+            self.indent_level.pop()
+            yield Token(self.DEDENT_type, "")
+
+        assert self.indent_level == [0], self.indent_level
+
+    # def process(self, stream):
+    #     import pdb;pdb.set_trace()
+    #     self.paren_level = 0
+    #     self.indent_level = [0]
+    #     return self._process(stream)

gdtoolkit/parser/parser.py

@@ -9,15 +9,7 @@
 
 from lark import Lark, Tree, indenter
 
-
-class Indenter(indenter.Indenter):
-    NL_type = "_NL"
-    OPEN_PAREN_types = ["LPAR", "LSQB", "LBRACE"]
-    CLOSE_PAREN_types = ["RPAR", "RSQB", "RBRACE"]
-    INDENT_type = "_INDENT"
-    DEDENT_type = "_DEDENT"
-    # TODO: guess tab length
-    tab_len = 4
+from .gdscript_indenter import GDScriptIndenter
 
 
 # TODO: when upgrading to Python 3.8, replace with functools.cached_property
@@ -92,7 +84,7 @@ def _get_parser(
             grammar_filepath,
             parser="lalr",
             start="start",
-            postlex=Indenter(),  # type: ignore
+            postlex=GDScriptIndenter(),  # type: ignore
             propagate_positions=add_metadata,
             maybe_placeholders=False,
             cache=cache_filepath,

tests/valid-gd-scripts/lambdas.gd

@@ -3,19 +3,9 @@ extends Node
 func _ready():
 	var l1 = func(x): pass;print('l1', x)
 	l1.call('foo')
-	# var l2 = func(x):
-	# 	pass
-	# 	print('l2', x)
-	# l2.call('bar')
 	var ls = [func(): print('l3'), func(): print('l4')]
 	ls[1].call()
 	var lss = [
-	# 	func():
-	# 		pass
-	# 		print('l5'),
 	func(): print('l6')]
 	lss[1].call()
 	get_tree().process_frame.connect(func(): pass;print('x'))
-	# get_tree().process_frame.connect(func():
-	# 	pass
-	# 	print('y'))

tests/valid-gd-scripts/multiline_lambdas.gd

@@ -0,0 +1,20 @@
+func foo():
+	pass
+	# var f0 = func bar():
+	# 	pass
+	# var f1 = func():
+	# 	pass
+	# var f2s = [func():
+	# 	pass]
+	# var f3s = [func():
+	# 	pass, func():
+	# 	pass]
+	# var f4s = [func():
+	# 	return [1,2,3], func():
+	# 	pass]
+	# Godot 4.3 failing:
+	# var fx = func():
+	# 	pass if true else func():
+	# 	pass
+	# var fx = func():
+	# 	pass is int