(🎁) Remove redundant parenthesis in subscript operations

CHANGES.md

@@ -16,6 +16,7 @@
 
 <!-- Changes that affect Black's preview style -->
 
+- Parentheses are removed from subscript operations (#3680)
 - Long type hints are now wrapped in parentheses and properly indented when split across
   multiple lines (#3899)
 - Magic trailing commas are now respected in return types. (#3916)

src/black/linegen.py

@@ -366,6 +366,30 @@ def visit_power(self, node: Node) -> Iterator[Line]:
 
         yield from self.visit_default(node)
 
+    def visit_trailer(self, node: Node) -> Iterator[Line]:
+        if (
+            Preview.remove_redundant_parens_subscript in self.mode
+            # Ensure that we are in a getitem trailer
+            and node.children[0].type == token.LSQB
+            and (
+                # If there are parens in the subscript
+                (
+                    node.children[1].type == syms.atom
+                    and node.children[1].children[0].type == token.LPAR
+                    and not is_walrus_assignment(node.children[1])
+                )
+                or (
+                    is_walrus_assignment(node.children[1])
+                    # This is only supported in py310 and up
+                    and all(t.value >= 10 for t in self.mode.target_versions)
+                )
+            )
+        ):
+            normalize_invisible_parens(
+                node, parens_after={"["}, mode=self.mode, features=self.features
+            )
+        yield from self.visit_default(node)
+
     def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
         """Remove a semicolon and put the other statement on a separate line."""
         yield from self.line()
@@ -1252,6 +1276,8 @@ def normalize_invisible_parens(
                 if maybe_make_parens_invisible_in_atom(
                     child,
                     parent=node,
+                    force_remove_brackets_around_comma=node.type == syms.trailer
+                    and child.children[0].type == token.LPAR,
                 ):
                     wrap_in_parentheses(node, child, visible=False)
             elif is_one_tuple(child):
@@ -1419,17 +1445,20 @@ def maybe_make_parens_invisible_in_atom(
     node: LN,
     parent: LN,
     remove_brackets_around_comma: bool = False,
+    force_remove_brackets_around_comma: bool = False,
 ) -> bool:
     """If it's safe, make the parens in the atom `node` invisible, recursively.
     Additionally, remove repeated, adjacent invisible parens from the atom `node`
     as they are redundant.
 
     Returns whether the node should itself be wrapped in invisible parentheses.
     """
+    if force_remove_brackets_around_comma:
+        remove_brackets_around_comma = True
     if (
         node.type not in (syms.atom, syms.expr)
         or is_empty_tuple(node)
-        or is_one_tuple(node)
+        or (is_one_tuple(node) and not force_remove_brackets_around_comma)
         or (is_yield(node) and parent.type != syms.expr_stmt)
         or (
             # This condition tries to prevent removing non-optional brackets

src/black/mode.py

@@ -185,6 +185,7 @@ class Preview(Enum):
     skip_magic_trailing_comma_in_subscript = auto()
     wrap_long_dict_values_in_parens = auto()
     wrap_multiple_context_managers_in_parens = auto()
+    remove_redundant_parens_subscript = auto()
     dummy_implementations = auto()
     walrus_subscript = auto()
 

tests/data/miscellaneous/subscript.py

@@ -0,0 +1,52 @@
+x[1]
+x[1,]
+x[(1,)]
+x[1, 2]
+x[1, 2,]
+x[(1, 2)]
+x[1, 2]
+x[(1, 2)]
+x[(1, 2),]
+
+x[1] = 1
+x[1,] = 1
+x[(1,)] = 1
+x[1, 2] = 1
+x[1, 2,] = 1
+x[(1, 2)] = 1
+x[1, 2] = 1
+x[(1, 2)] = 1
+x[(1, 2),] = 1
+
+x[(a := 1)]
+x[(a := 1)] = 1
+
+# output
+x[1]
+x[1,]
+x[1,]
+x[1, 2]
+x[
+    1,
+    2,
+]
+x[1, 2]
+x[1, 2]
+x[1, 2]
+x[(1, 2),]
+
+x[1] = 1
+x[1,] = 1
+x[1,] = 1
+x[1, 2] = 1
+x[
+    1,
+    2,
+] = 1
+x[1, 2] = 1
+x[1, 2] = 1
+x[1, 2] = 1
+x[(1, 2),] = 1
+
+x[(a := 1)]
+x[(a := 1)] = 1

tests/data/miscellaneous/subscript_py310.py

@@ -0,0 +1,6 @@
+x[(a := 1)]
+x[(a := 1)] = 1
+
+# output
+x[a:=1]
+x[a:=1] = 1

tests/data/preview/percent_precedence.py

@@ -37,5 +37,5 @@
 ~("" % a)
 2 ** ("" % a)
 await ("" % a)
-b[("" % a)]
+b["" % a]
 b(("" % a))

tests/test_format.py

@@ -218,3 +218,15 @@ def test_type_comment_syntax_error() -> None:
     source, expected = read_data("type_comments", "type_comment_syntax_error")
     assert_format(source, expected)
     black.assert_equivalent(source, expected)
+
+
+def test_subscript() -> None:
+    source, expected = read_data("miscellaneous", "subscript")
+    mode = black.Mode(preview=True, target_versions={black.TargetVersion.PY39})
+    assert_format(source, expected, mode)
+
+
+def test_subscript_py310() -> None:
+    source, expected = read_data("miscellaneous", "subscript_py310")
+    mode = black.Mode(preview=True, target_versions={black.TargetVersion.PY310})
+    assert_format(source, expected, mode)