Make shlex_split always return a token

Matches behavior of split() so is therefore more intuitive

kitty/utils.py

@@ -1215,14 +1215,22 @@ def key_val_matcher(items: Iterable[tuple[str, str]], key_pat: 're.Pattern[str]'
 
 def shlex_split(text: str, allow_ansi_quoted_strings: bool = False) -> Iterator[str]:
     s = Shlex(text, allow_ansi_quoted_strings)
+    yielded = False
     while (q := s.next_word())[0] > -1:
         yield q[1]
+        yielded = True
+    if not yielded:
+        yield ''
 
 
 def shlex_split_with_positions(text: str, allow_ansi_quoted_strings: bool = False) -> Iterator[tuple[int, str]]:
     s = Shlex(text, allow_ansi_quoted_strings)
+    yielded = False
     while (q := s.next_word())[0] > -1:
         yield q
+        yielded = True
+    if not yielded:
+        yield 0, ''
 
 
 def timed_debug_print(*a: Any, sep: str = ' ', end: str = '\n') -> None:

kitty/window.py

@@ -213,8 +213,7 @@ def compile_match_query(exp: str, is_simple: bool = True) -> MatchPatternType:
 def decode_cmdline(x: str) -> str:
     ctype, sep, val = x.partition('=')
     if ctype == 'cmdline':
-        with suppress(StopIteration):
-            return next(shlex_split(val, True))
+        return next(shlex_split(val, True))
     elif ctype == 'cmdline_url':
         from urllib.parse import unquote
         return unquote(val)

kitty_tests/datatypes.py

@@ -629,7 +629,7 @@ def test_shlex_split(self):
             r'x "ab"y \m': ((0, 'x'), (2, 'aby'), (8, 'm')),
             r'''x'y"\z'1''': ((0, 'xy"\\z1'),),
             r'\abc\ d': ((0, 'abc d'),),
-            '': (), '   ': (), ' \tabc\n\t\r ': ((2, 'abc'),),
+            '': ((0, ''),), '   ': ((0, ''),), ' \tabc\n\t\r ': ((2, 'abc'),),
             "$'ab'": ((0, '$ab'),),
         }.items():
             actual = tuple(shlex_split_with_positions(q))