Empty brackets around the `case` of a `match-case` are incorrectly parsed as a pair of square brackets

[sirosen]

This is an oddity which results in data not roundtripping through libcst accurately. It also makes it impossible to know what the original file contents were.

If a match-case is used and the case is an unparenthesized, unbracketed expression, libcst parses this to have lbracket and rbracket incorrectly populated.

Here's a reproducer:

x = (1, 2)
match x:
    case 1, 2:
        print("x matches (1, 2)")
    case _:
        print("x does not match (1, 2)")

Attempting to simply load and dump this module produces case [1, 2]:.

Inspecting the results of parse_module, I find a parse shaped like...

pattern=MatchList(...,
  lbracket=LeftSquareBracket(
    whitespace_after=SimpleWhitespace(
      value='',
    ),
  ),
  rbracket=RightSquareBracket(
    whitespace_before=SimpleWhitespace(
      value='',
    ),
  ),
  lpar=[],
  rpar=[],
)

[zsol]

Whoa this is pretty bad

[zsol]

Thank you for the report!