Handle brackets inside string arguments in split_args

split_args treated every bracket and quote as structural, so a closing
bracket inside a quoted string argument (e.g. reason="closed ]" or an
emoticon "Hello :)") raised ValueError and crashed Colang flow parsing.
Track string state so that while inside a quoted string only the matching
closing quote is special and brackets, commas and the other quote type are
literal, matching the sibling splitters (word_split, char_split,
params_tokenize).

Signed-off-by: winklemad <winklemad@outlook.com>

Author: winklemad

nemoguardrails/colang/v1_0/lang/utils.py

@@ -50,13 +50,21 @@ def split_args(args_str: str) -> List[str]:
     closing_char = {"[": "]", "(": ")", "{": "}", "'": "'", '"': '"'}
 
     for char in args_str:
+        # While inside a string literal, only the matching closing quote is
+        # special; brackets, commas and the other quote type are literal text.
+        if stack and stack[-1] in "\"'":
+            if char == stack[-1]:
+                stack.pop()
+            current.append(char)
+            continue
+
         if char in "([{":
             stack.append(char)
             current.append(char)
-        elif char in "\"'" and (len(stack) == 0 or stack[-1] != char):
+        elif char in "\"'":
             stack.append(char)
             current.append(char)
-        elif char in ")]}\"'":
+        elif char in ")]}":
             if not stack:
                 raise ValueError(f"Invalid syntax for string: {args_str}; unexpected closing '{char}'")
             if char != closing_char[stack[-1]]:

tests/test_parser_utils.py

@@ -55,3 +55,28 @@ def test_split_args_unbalanced_closing_bracket_raises_value_error(args_str):
     # the ValueError the function uses to report invalid syntax.
     with pytest.raises(ValueError):
         split_args(args_str)
+
+
+@pytest.mark.parametrize(
+    "args_str, expected",
+    [
+        # A closing bracket inside a quoted string is literal text, not structural.
+        ('x="]"', ['x="]"']),
+        ('x=")"', ['x=")"']),
+        ('x="}"', ['x="}"']),
+        # Real-world strings: an emoticon or a citation marker.
+        ('msg="Hello :)"', ['msg="Hello :)"']),
+        ('note="cite [1] and (2)"', ['note="cite [1] and (2)"']),
+        ('reason="closed ]"', ['reason="closed ]"']),
+        # A comma inside a string does not split the arguments.
+        ('x="a, b"', ['x="a, b"']),
+        # The other quote character inside a string is literal too.
+        ('x="it\'s ok)"', ['x="it\'s ok)"']),
+        # A bracket-in-string argument alongside a normal argument still splits.
+        ('n=1, x="]"', ["n=1", 'x="]"']),
+    ],
+)
+def test_split_args_bracket_inside_string(args_str, expected):
+    # Brackets, commas, and the other quote character inside a quoted string
+    # argument are literal text and must not be treated as structural syntax.
+    assert split_args(args_str) == expected