]> git.madduck.net Git - etc/vim.git/blobdiff - src/black/linegen.py

madduck's git repository

Every one of the projects in this repository is available at the canonical URL git://git.madduck.net/madduck/pub/<projectpath> — see each project's metadata for the exact URL.

All patches and comments are welcome. Please squash your changes to logical commits before using git-format-patch and git-send-email to patches@git.madduck.net. If you'd read over the Git project's submission guidelines and adhered to them, I'd be especially grateful.

SSH access, as well as push access can be individually arranged.

If you use my repositories frequently, consider adding the following snippet to ~/.gitconfig and using the third clone URL listed for each project:

[url "git://git.madduck.net/madduck/"]
  insteadOf = madduck:

Bump actions/checkout from 3 to 4 (#3883)
[etc/vim.git] / src / black / linegen.py
index 8cf32c973bb6a4880e543ec6a51ec06ff54dc492..507e860190f0c8a31701f5a316e46fa8099ac1e7 100644 (file)
@@ -1,35 +1,80 @@
 """
 Generating lines of code.
 """
 """
 Generating lines of code.
 """
-from functools import partial, wraps
 import sys
 import sys
-from typing import Collection, Iterator, List, Optional, Set, Union
-
-from dataclasses import dataclass, field
-
-from black.nodes import WHITESPACE, RARROW, STATEMENT, STANDALONE_COMMENT
-from black.nodes import ASSIGNMENTS, OPENING_BRACKETS, CLOSING_BRACKETS
-from black.nodes import Visitor, syms, first_child_is_arith, ensure_visible
-from black.nodes import is_docstring, is_empty_tuple, is_one_tuple, is_one_tuple_between
-from black.nodes import is_walrus_assignment, is_yield, is_vararg, is_multiline_string
-from black.nodes import is_stub_suite, is_stub_body, is_atom_with_invisible_parens
-from black.nodes import wrap_in_parentheses
-from black.brackets import max_delimiter_priority_in_atom
-from black.brackets import DOT_PRIORITY, COMMA_PRIORITY
-from black.lines import Line, line_to_string, is_line_short_enough
-from black.lines import can_omit_invisible_parens, can_be_split, append_leaves
-from black.comments import generate_comments, list_comments, FMT_OFF
+from dataclasses import replace
+from enum import Enum, auto
+from functools import partial, wraps
+from typing import Collection, Iterator, List, Optional, Set, Union, cast
+
+from black.brackets import (
+    COMMA_PRIORITY,
+    DOT_PRIORITY,
+    get_leaves_inside_matching_brackets,
+    max_delimiter_priority_in_atom,
+)
+from black.comments import FMT_OFF, generate_comments, list_comments
+from black.lines import (
+    Line,
+    RHSResult,
+    append_leaves,
+    can_be_split,
+    can_omit_invisible_parens,
+    is_line_short_enough,
+    line_to_string,
+)
+from black.mode import Feature, Mode, Preview
+from black.nodes import (
+    ASSIGNMENTS,
+    BRACKETS,
+    CLOSING_BRACKETS,
+    OPENING_BRACKETS,
+    RARROW,
+    STANDALONE_COMMENT,
+    STATEMENT,
+    WHITESPACE,
+    Visitor,
+    ensure_visible,
+    is_arith_like,
+    is_async_stmt_or_funcdef,
+    is_atom_with_invisible_parens,
+    is_docstring,
+    is_empty_tuple,
+    is_lpar_token,
+    is_multiline_string,
+    is_name_token,
+    is_one_sequence_between,
+    is_one_tuple,
+    is_rpar_token,
+    is_stub_body,
+    is_stub_suite,
+    is_tuple_containing_walrus,
+    is_type_ignore_comment_string,
+    is_vararg,
+    is_walrus_assignment,
+    is_yield,
+    syms,
+    wrap_in_parentheses,
+)
 from black.numerics import normalize_numeric_literal
 from black.numerics import normalize_numeric_literal
-from black.strings import get_string_prefix, fix_docstring
-from black.strings import normalize_string_prefix, normalize_string_quotes
-from black.trans import Transformer, CannotTransform, StringMerger
-from black.trans import StringSplitter, StringParenWrapper, StringParenStripper
-from black.mode import Mode
-from black.mode import Feature
-
-from blib2to3.pytree import Node, Leaf
+from black.strings import (
+    fix_docstring,
+    get_string_prefix,
+    normalize_string_prefix,
+    normalize_string_quotes,
+    normalize_unicode_escape_sequences,
+)
+from black.trans import (
+    CannotTransform,
+    StringMerger,
+    StringParenStripper,
+    StringParenWrapper,
+    StringSplitter,
+    Transformer,
+    hug_power_op,
+)
 from blib2to3.pgen2 import token
 from blib2to3.pgen2 import token
-
+from blib2to3.pytree import Leaf, Node
 
 # types
 LeafID = int
 
 # types
 LeafID = int
@@ -40,7 +85,8 @@ class CannotSplit(CannotTransform):
     """A readable split that fits the allotted line length is impossible."""
 
 
     """A readable split that fits the allotted line length is impossible."""
 
 
-@dataclass
+# This isn't a dataclass because @dataclass + Generic breaks mypyc.
+# See also https://github.com/mypyc/mypyc/issues/827.
 class LineGenerator(Visitor[Line]):
     """Generates reformatted Line objects.  Empty lines are not emitted.
 
 class LineGenerator(Visitor[Line]):
     """Generates reformatted Line objects.  Empty lines are not emitted.
 
@@ -48,9 +94,11 @@ class LineGenerator(Visitor[Line]):
     in ways that will no longer stringify to valid Python code on the tree.
     """
 
     in ways that will no longer stringify to valid Python code on the tree.
     """
 
-    mode: Mode
-    remove_u_prefix: bool = False
-    current_line: Line = field(init=False)
+    def __init__(self, mode: Mode, features: Collection[Feature]) -> None:
+        self.mode = mode
+        self.features = features
+        self.current_line: Line
+        self.__post_init__()
 
     def line(self, indent: int = 0) -> Iterator[Line]:
         """Generate a line.
 
     def line(self, indent: int = 0) -> Iterator[Line]:
         """Generate a line.
@@ -64,6 +112,17 @@ class LineGenerator(Visitor[Line]):
             self.current_line.depth += indent
             return  # Line is empty, don't emit. Creating a new one unnecessary.
 
             self.current_line.depth += indent
             return  # Line is empty, don't emit. Creating a new one unnecessary.
 
+        if (
+            Preview.improved_async_statements_handling in self.mode
+            and len(self.current_line.leaves) == 1
+            and is_async_stmt_or_funcdef(self.current_line.leaves[0])
+        ):
+            # Special case for async def/for/with statements. `visit_async_stmt`
+            # adds an `ASYNC` leaf then visits the child def/for/with statement
+            # nodes. Line yields from those nodes shouldn't treat the former
+            # `ASYNC` leaf as a complete line.
+            return
+
         complete_line = self.current_line
         self.current_line = Line(mode=self.mode, depth=complete_line.depth + indent)
         yield complete_line
         complete_line = self.current_line
         self.current_line = Line(mode=self.mode, depth=complete_line.depth + indent)
         yield complete_line
@@ -90,9 +149,7 @@ class LineGenerator(Visitor[Line]):
 
             normalize_prefix(node, inside_brackets=any_open_brackets)
             if self.mode.string_normalization and node.type == token.STRING:
 
             normalize_prefix(node, inside_brackets=any_open_brackets)
             if self.mode.string_normalization and node.type == token.STRING:
-                node.value = normalize_string_prefix(
-                    node.value, remove_u_prefix=self.remove_u_prefix
-                )
+                node.value = normalize_string_prefix(node.value)
                 node.value = normalize_string_quotes(node.value)
             if node.type == token.NUMBER:
                 normalize_numeric_literal(node)
                 node.value = normalize_string_quotes(node.value)
             if node.type == token.NUMBER:
                 normalize_numeric_literal(node)
@@ -100,6 +157,22 @@ class LineGenerator(Visitor[Line]):
                 self.current_line.append(node)
         yield from super().visit_default(node)
 
                 self.current_line.append(node)
         yield from super().visit_default(node)
 
+    def visit_test(self, node: Node) -> Iterator[Line]:
+        """Visit an `x if y else z` test"""
+
+        if Preview.parenthesize_conditional_expressions in self.mode:
+            already_parenthesized = (
+                node.prev_sibling and node.prev_sibling.type == token.LPAR
+            )
+
+            if not already_parenthesized:
+                lpar = Leaf(token.LPAR, "")
+                rpar = Leaf(token.RPAR, "")
+                node.insert_child(0, lpar)
+                node.append_child(rpar)
+
+        yield from self.visit_default(node)
+
     def visit_INDENT(self, node: Leaf) -> Iterator[Line]:
         """Increase indentation level, maybe yield a line."""
         # In blib2to3 INDENT never holds comments.
     def visit_INDENT(self, node: Leaf) -> Iterator[Line]:
         """Increase indentation level, maybe yield a line."""
         # In blib2to3 INDENT never holds comments.
@@ -126,7 +199,7 @@ class LineGenerator(Visitor[Line]):
         """Visit a statement.
 
         This implementation is shared for `if`, `while`, `for`, `try`, `except`,
         """Visit a statement.
 
         This implementation is shared for `if`, `while`, `for`, `try`, `except`,
-        `def`, `with`, `class`, `assert`, `match`, `case` and assignments.
+        `def`, `with`, `class`, `assert`, and assignments.
 
         The relevant Python language `keywords` for a given statement will be
         NAME leaves within it. This methods puts those on a separate line.
 
         The relevant Python language `keywords` for a given statement will be
         NAME leaves within it. This methods puts those on a separate line.
@@ -134,27 +207,100 @@ class LineGenerator(Visitor[Line]):
         `parens` holds a set of string leaf values immediately after which
         invisible parens should be put.
         """
         `parens` holds a set of string leaf values immediately after which
         invisible parens should be put.
         """
-        normalize_invisible_parens(node, parens_after=parens)
+        normalize_invisible_parens(
+            node, parens_after=parens, mode=self.mode, features=self.features
+        )
         for child in node.children:
         for child in node.children:
-            if child.type == token.NAME and child.value in keywords:  # type: ignore
+            if is_name_token(child) and child.value in keywords:
                 yield from self.line()
 
             yield from self.visit(child)
 
                 yield from self.line()
 
             yield from self.visit(child)
 
+    def visit_typeparams(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[0].prefix = ""
+
+    def visit_typevartuple(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[1].prefix = ""
+
+    def visit_paramspec(self, node: Node) -> Iterator[Line]:
+        yield from self.visit_default(node)
+        node.children[1].prefix = ""
+
+    def visit_dictsetmaker(self, node: Node) -> Iterator[Line]:
+        if Preview.wrap_long_dict_values_in_parens in self.mode:
+            for i, child in enumerate(node.children):
+                if i == 0:
+                    continue
+                if node.children[i - 1].type == token.COLON:
+                    if child.type == syms.atom and child.children[0].type == token.LPAR:
+                        if maybe_make_parens_invisible_in_atom(
+                            child,
+                            parent=node,
+                            remove_brackets_around_comma=False,
+                        ):
+                            wrap_in_parentheses(node, child, visible=False)
+                    else:
+                        wrap_in_parentheses(node, child, visible=False)
+        yield from self.visit_default(node)
+
+    def visit_funcdef(self, node: Node) -> Iterator[Line]:
+        """Visit function definition."""
+        yield from self.line()
+
+        # Remove redundant brackets around return type annotation.
+        is_return_annotation = False
+        for child in node.children:
+            if child.type == token.RARROW:
+                is_return_annotation = True
+            elif is_return_annotation:
+                if child.type == syms.atom and child.children[0].type == token.LPAR:
+                    if maybe_make_parens_invisible_in_atom(
+                        child,
+                        parent=node,
+                        remove_brackets_around_comma=False,
+                    ):
+                        wrap_in_parentheses(node, child, visible=False)
+                else:
+                    wrap_in_parentheses(node, child, visible=False)
+                is_return_annotation = False
+
+        for child in node.children:
+            yield from self.visit(child)
+
+    def visit_match_case(self, node: Node) -> Iterator[Line]:
+        """Visit either a match or case statement."""
+        normalize_invisible_parens(
+            node, parens_after=set(), mode=self.mode, features=self.features
+        )
+
+        yield from self.line()
+        for child in node.children:
+            yield from self.visit(child)
+
     def visit_suite(self, node: Node) -> Iterator[Line]:
         """Visit a suite."""
     def visit_suite(self, node: Node) -> Iterator[Line]:
         """Visit a suite."""
-        if self.mode.is_pyi and is_stub_suite(node):
+        if (
+            self.mode.is_pyi or Preview.dummy_implementations in self.mode
+        ) and is_stub_suite(node):
             yield from self.visit(node.children[2])
         else:
             yield from self.visit_default(node)
 
     def visit_simple_stmt(self, node: Node) -> Iterator[Line]:
         """Visit a statement without nested statements."""
             yield from self.visit(node.children[2])
         else:
             yield from self.visit_default(node)
 
     def visit_simple_stmt(self, node: Node) -> Iterator[Line]:
         """Visit a statement without nested statements."""
-        if first_child_is_arith(node):
-            wrap_in_parentheses(node, node.children[0], visible=False)
+        prev_type: Optional[int] = None
+        for child in node.children:
+            if (prev_type is None or prev_type == token.SEMI) and is_arith_like(child):
+                wrap_in_parentheses(node, child, visible=False)
+            prev_type = child.type
+
         is_suite_like = node.parent and node.parent.type in STATEMENT
         if is_suite_like:
         is_suite_like = node.parent and node.parent.type in STATEMENT
         if is_suite_like:
-            if self.mode.is_pyi and is_stub_body(node):
+            if (
+                self.mode.is_pyi or Preview.dummy_implementations in self.mode
+            ) and is_stub_body(node):
                 yield from self.visit_default(node)
             else:
                 yield from self.line(+1)
                 yield from self.visit_default(node)
             else:
                 yield from self.line(+1)
@@ -163,7 +309,7 @@ class LineGenerator(Visitor[Line]):
 
         else:
             if (
 
         else:
             if (
-                not self.mode.is_pyi
+                not (self.mode.is_pyi or Preview.dummy_implementations in self.mode)
                 or not node.parent
                 or not is_stub_suite(node.parent)
             ):
                 or not node.parent
                 or not is_stub_suite(node.parent)
             ):
@@ -178,12 +324,17 @@ class LineGenerator(Visitor[Line]):
         for child in children:
             yield from self.visit(child)
 
         for child in children:
             yield from self.visit(child)
 
-            if child.type == token.ASYNC:
+            if child.type == token.ASYNC or child.type == STANDALONE_COMMENT:
+                # STANDALONE_COMMENT happens when `# fmt: skip` is applied on the async
+                # line.
                 break
 
         internal_stmt = next(children)
                 break
 
         internal_stmt = next(children)
-        for child in internal_stmt.children:
-            yield from self.visit(child)
+        if Preview.improved_async_statements_handling in self.mode:
+            yield from self.visit(internal_stmt)
+        else:
+            for child in internal_stmt.children:
+                yield from self.visit(child)
 
     def visit_decorators(self, node: Node) -> Iterator[Line]:
         """Visit decorators."""
 
     def visit_decorators(self, node: Node) -> Iterator[Line]:
         """Visit decorators."""
@@ -191,6 +342,30 @@ class LineGenerator(Visitor[Line]):
             yield from self.line()
             yield from self.visit(child)
 
             yield from self.line()
             yield from self.visit(child)
 
+    def visit_power(self, node: Node) -> Iterator[Line]:
+        for idx, leaf in enumerate(node.children[:-1]):
+            next_leaf = node.children[idx + 1]
+
+            if not isinstance(leaf, Leaf):
+                continue
+
+            value = leaf.value.lower()
+            if (
+                leaf.type == token.NUMBER
+                and next_leaf.type == syms.trailer
+                # Ensure that we are in an attribute trailer
+                and next_leaf.children[0].type == token.DOT
+                # It shouldn't wrap hexadecimal, binary and octal literals
+                and not value.startswith(("0x", "0b", "0o"))
+                # It shouldn't wrap complex literals
+                and "j" not in value
+            ):
+                wrap_in_parentheses(node, leaf)
+
+        remove_await_parens(node)
+
+        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()
     def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
         """Remove a semicolon and put the other statement on a separate line."""
         yield from self.line()
@@ -223,10 +398,23 @@ class LineGenerator(Visitor[Line]):
         yield from self.visit_default(node)
 
     def visit_STRING(self, leaf: Leaf) -> Iterator[Line]:
         yield from self.visit_default(node)
 
     def visit_STRING(self, leaf: Leaf) -> Iterator[Line]:
+        if Preview.hex_codes_in_unicode_sequences in self.mode:
+            normalize_unicode_escape_sequences(leaf)
+
         if is_docstring(leaf) and "\\\n" not in leaf.value:
             # We're ignoring docstrings with backslash newline escapes because changing
             # indentation of those changes the AST representation of the code.
         if is_docstring(leaf) and "\\\n" not in leaf.value:
             # We're ignoring docstrings with backslash newline escapes because changing
             # indentation of those changes the AST representation of the code.
-            docstring = normalize_string_prefix(leaf.value, self.remove_u_prefix)
+            if self.mode.string_normalization:
+                docstring = normalize_string_prefix(leaf.value)
+                # visit_default() does handle string normalization for us, but
+                # since this method acts differently depending on quote style (ex.
+                # see padding logic below), there's a possibility for unstable
+                # formatting as visit_default() is called *after*. To avoid a
+                # situation where this function formats a docstring differently on
+                # the second pass, normalize it early.
+                docstring = normalize_string_quotes(docstring)
+            else:
+                docstring = leaf.value
             prefix = get_string_prefix(docstring)
             docstring = docstring[len(prefix) :]  # Remove the prefix
             quote_char = docstring[0]
             prefix = get_string_prefix(docstring)
             docstring = docstring[len(prefix) :]  # Remove the prefix
             quote_char = docstring[0]
@@ -238,13 +426,14 @@ class LineGenerator(Visitor[Line]):
             quote_len = 1 if docstring[1] != quote_char else 3
             docstring = docstring[quote_len:-quote_len]
             docstring_started_empty = not docstring
             quote_len = 1 if docstring[1] != quote_char else 3
             docstring = docstring[quote_len:-quote_len]
             docstring_started_empty = not docstring
+            indent = " " * 4 * self.current_line.depth
 
             if is_multiline_string(leaf):
 
             if is_multiline_string(leaf):
-                indent = " " * 4 * self.current_line.depth
                 docstring = fix_docstring(docstring, indent)
             else:
                 docstring = docstring.strip()
 
                 docstring = fix_docstring(docstring, indent)
             else:
                 docstring = docstring.strip()
 
+            has_trailing_backslash = False
             if docstring:
                 # Add some padding if the docstring starts / ends with a quote mark.
                 if docstring[0] == quote_char:
             if docstring:
                 # Add some padding if the docstring starts / ends with a quote mark.
                 if docstring[0] == quote_char:
@@ -257,12 +446,37 @@ class LineGenerator(Visitor[Line]):
                         # Odd number of tailing backslashes, add some padding to
                         # avoid escaping the closing string quote.
                         docstring += " "
                         # Odd number of tailing backslashes, add some padding to
                         # avoid escaping the closing string quote.
                         docstring += " "
+                        has_trailing_backslash = True
             elif not docstring_started_empty:
                 docstring = " "
 
             # We could enforce triple quotes at this point.
             quote = quote_char * quote_len
             elif not docstring_started_empty:
                 docstring = " "
 
             # We could enforce triple quotes at this point.
             quote = quote_char * quote_len
-            leaf.value = prefix + quote + docstring + quote
+
+            # It's invalid to put closing single-character quotes on a new line.
+            if self.mode and quote_len == 3:
+                # We need to find the length of the last line of the docstring
+                # to find if we can add the closing quotes to the line without
+                # exceeding the maximum line length.
+                # If docstring is one line, we don't put the closing quotes on a
+                # separate line because it looks ugly (#3320).
+                lines = docstring.splitlines()
+                last_line_length = len(lines[-1]) if docstring else 0
+
+                # If adding closing quotes would cause the last line to exceed
+                # the maximum line length then put a line break before the
+                # closing quotes
+                if (
+                    len(lines) > 1
+                    and last_line_length + quote_len > self.mode.line_length
+                    and len(indent) + quote_len <= self.mode.line_length
+                    and not has_trailing_backslash
+                ):
+                    leaf.value = prefix + quote + docstring + "\n" + indent + quote
+                else:
+                    leaf.value = prefix + quote + docstring + quote
+            else:
+                leaf.value = prefix + quote + docstring + quote
 
         yield from self.visit_default(leaf)
 
 
         yield from self.visit_default(leaf)
 
@@ -281,9 +495,8 @@ class LineGenerator(Visitor[Line]):
         self.visit_try_stmt = partial(
             v, keywords={"try", "except", "else", "finally"}, parens=Ø
         )
         self.visit_try_stmt = partial(
             v, keywords={"try", "except", "else", "finally"}, parens=Ø
         )
-        self.visit_except_clause = partial(v, keywords={"except"}, parens=Ø)
-        self.visit_with_stmt = partial(v, keywords={"with"}, parens=Ø)
-        self.visit_funcdef = partial(v, keywords={"def"}, parens=Ø)
+        self.visit_except_clause = partial(v, keywords={"except"}, parens={"except"})
+        self.visit_with_stmt = partial(v, keywords={"with"}, parens={"with"})
         self.visit_classdef = partial(v, keywords={"class"}, parens=Ø)
         self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS)
         self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"})
         self.visit_classdef = partial(v, keywords={"class"}, parens=Ø)
         self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS)
         self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"})
@@ -293,8 +506,8 @@ class LineGenerator(Visitor[Line]):
         self.visit_decorated = self.visit_decorators
 
         # PEP 634
         self.visit_decorated = self.visit_decorators
 
         # PEP 634
-        self.visit_match_stmt = partial(v, keywords={"match"}, parens=Ø)
-        self.visit_case_block = partial(v, keywords={"case"}, parens=Ø)
+        self.visit_match_stmt = self.visit_match_case
+        self.visit_case_block = self.visit_match_case
 
 
 def transform_line(
 
 
 def transform_line(
@@ -325,13 +538,13 @@ def transform_line(
         and not line.should_split_rhs
         and not line.magic_trailing_comma
         and (
         and not line.should_split_rhs
         and not line.magic_trailing_comma
         and (
-            is_line_short_enough(line, line_length=mode.line_length, line_str=line_str)
+            is_line_short_enough(line, mode=mode, line_str=line_str)
             or line.contains_unsplittable_type_ignore()
         )
         and not (line.inside_brackets and line.contains_standalone_comments())
     ):
         # Only apply basic string preprocessing, since lines shouldn't be split here.
             or line.contains_unsplittable_type_ignore()
         )
         and not (line.inside_brackets and line.contains_standalone_comments())
     ):
         # Only apply basic string preprocessing, since lines shouldn't be split here.
-        if mode.experimental_string_processing:
+        if Preview.string_processing in mode:
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
             transformers = [string_merge, string_paren_strip]
         else:
             transformers = []
@@ -339,7 +552,9 @@ def transform_line(
         transformers = [left_hand_split]
     else:
 
         transformers = [left_hand_split]
     else:
 
-        def rhs(line: Line, features: Collection[Feature]) -> Iterator[Line]:
+        def _rhs(
+            self: object, line: Line, features: Collection[Feature], mode: Mode
+        ) -> Iterator[Line]:
             """Wraps calls to `right_hand_split`.
 
             The calls increasingly `omit` right-hand trailers (bracket pairs with
             """Wraps calls to `right_hand_split`.
 
             The calls increasingly `omit` right-hand trailers (bracket pairs with
@@ -347,14 +562,12 @@ def transform_line(
             bracket pair instead.
             """
             for omit in generate_trailers_to_omit(line, mode.line_length):
             bracket pair instead.
             """
             for omit in generate_trailers_to_omit(line, mode.line_length):
-                lines = list(
-                    right_hand_split(line, mode.line_length, features, omit=omit)
-                )
+                lines = list(right_hand_split(line, mode, features, omit=omit))
                 # Note: this check is only able to figure out if the first line of the
                 # *current* transformation fits in the line length.  This is true only
                 # for simple cases.  All others require running more transforms via
                 # `transform_line()`.  This check doesn't know if those would succeed.
                 # Note: this check is only able to figure out if the first line of the
                 # *current* transformation fits in the line length.  This is true only
                 # for simple cases.  All others require running more transforms via
                 # `transform_line()`.  This check doesn't know if those would succeed.
-                if is_line_short_enough(lines[0], line_length=mode.line_length):
+                if is_line_short_enough(lines[0], mode=mode):
                     yield from lines
                     return
 
                     yield from lines
                     return
 
@@ -362,11 +575,15 @@ def transform_line(
             # This mostly happens to multiline strings that are by definition
             # reported as not fitting a single line, as well as lines that contain
             # trailing commas (those have to be exploded).
             # This mostly happens to multiline strings that are by definition
             # reported as not fitting a single line, as well as lines that contain
             # trailing commas (those have to be exploded).
-            yield from right_hand_split(
-                line, line_length=mode.line_length, features=features
-            )
+            yield from right_hand_split(line, mode, features=features)
 
 
-        if mode.experimental_string_processing:
+        # HACK: nested functions (like _rhs) compiled by mypyc don't retain their
+        # __name__ attribute which is needed in `run_transformer` further down.
+        # Unfortunately a nested class breaks mypyc too. So a class must be created
+        # via type ... https://github.com/mypyc/mypyc/issues/884
+        rhs = type("rhs", (), {"__call__": _rhs})()
+
+        if Preview.string_processing in mode:
             if line.inside_brackets:
                 transformers = [
                     string_merge,
             if line.inside_brackets:
                 transformers = [
                     string_merge,
@@ -390,6 +607,9 @@ def transform_line(
                 transformers = [delimiter_split, standalone_comment_split, rhs]
             else:
                 transformers = [rhs]
                 transformers = [delimiter_split, standalone_comment_split, rhs]
             else:
                 transformers = [rhs]
+    # It's always safe to attempt hugging of power operations and pretty much every line
+    # could match.
+    transformers.append(hug_power_op)
 
     for transform in transformers:
         # We are accumulating lines in `result` because we might want to abort
 
     for transform in transformers:
         # We are accumulating lines in `result` because we might want to abort
@@ -407,7 +627,15 @@ def transform_line(
         yield line
 
 
         yield line
 
 
-def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator[Line]:
+class _BracketSplitComponent(Enum):
+    head = auto()
+    body = auto()
+    tail = auto()
+
+
+def left_hand_split(
+    line: Line, _features: Collection[Feature], mode: Mode
+) -> Iterator[Line]:
     """Split line into many lines, starting with the first matching bracket pair.
 
     Note: this usually looks weird, only use this for function definitions.
     """Split line into many lines, starting with the first matching bracket pair.
 
     Note: this usually looks weird, only use this for function definitions.
@@ -424,7 +652,10 @@ def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator
             current_leaves is body_leaves
             and leaf.type in CLOSING_BRACKETS
             and leaf.opening_bracket is matching_bracket
             current_leaves is body_leaves
             and leaf.type in CLOSING_BRACKETS
             and leaf.opening_bracket is matching_bracket
+            and isinstance(matching_bracket, Leaf)
         ):
         ):
+            ensure_visible(leaf)
+            ensure_visible(matching_bracket)
             current_leaves = tail_leaves if body_leaves else head_leaves
         current_leaves.append(leaf)
         if current_leaves is head_leaves:
             current_leaves = tail_leaves if body_leaves else head_leaves
         current_leaves.append(leaf)
         if current_leaves is head_leaves:
@@ -434,9 +665,15 @@ def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator
     if not matching_bracket:
         raise CannotSplit("No brackets found")
 
     if not matching_bracket:
         raise CannotSplit("No brackets found")
 
-    head = bracket_split_build_line(head_leaves, line, matching_bracket)
-    body = bracket_split_build_line(body_leaves, line, matching_bracket, is_body=True)
-    tail = bracket_split_build_line(tail_leaves, line, matching_bracket)
+    head = bracket_split_build_line(
+        head_leaves, line, matching_bracket, component=_BracketSplitComponent.head
+    )
+    body = bracket_split_build_line(
+        body_leaves, line, matching_bracket, component=_BracketSplitComponent.body
+    )
+    tail = bracket_split_build_line(
+        tail_leaves, line, matching_bracket, component=_BracketSplitComponent.tail
+    )
     bracket_split_succeeded_or_raise(head, body, tail)
     for result in (head, body, tail):
         if result:
     bracket_split_succeeded_or_raise(head, body, tail)
     for result in (head, body, tail):
         if result:
@@ -445,7 +682,7 @@ def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator
 
 def right_hand_split(
     line: Line,
 
 def right_hand_split(
     line: Line,
-    line_length: int,
+    mode: Mode,
     features: Collection[Feature] = (),
     omit: Collection[LeafID] = (),
 ) -> Iterator[Line]:
     features: Collection[Feature] = (),
     omit: Collection[LeafID] = (),
 ) -> Iterator[Line]:
@@ -457,6 +694,22 @@ def right_hand_split(
 
     Note: running this function modifies `bracket_depth` on the leaves of `line`.
     """
 
     Note: running this function modifies `bracket_depth` on the leaves of `line`.
     """
+    rhs_result = _first_right_hand_split(line, omit=omit)
+    yield from _maybe_split_omitting_optional_parens(
+        rhs_result, line, mode, features=features, omit=omit
+    )
+
+
+def _first_right_hand_split(
+    line: Line,
+    omit: Collection[LeafID] = (),
+) -> RHSResult:
+    """Split the line into head, body, tail starting with the last bracket pair.
+
+    Note: this function should not have side effects. It's relied upon by
+    _maybe_split_omitting_optional_parens to get an opinion whether to prefer
+    splitting on the right side of an assignment statement.
+    """
     tail_leaves: List[Leaf] = []
     body_leaves: List[Leaf] = []
     head_leaves: List[Leaf] = []
     tail_leaves: List[Leaf] = []
     body_leaves: List[Leaf] = []
     head_leaves: List[Leaf] = []
@@ -482,41 +735,82 @@ def right_hand_split(
     tail_leaves.reverse()
     body_leaves.reverse()
     head_leaves.reverse()
     tail_leaves.reverse()
     body_leaves.reverse()
     head_leaves.reverse()
-    head = bracket_split_build_line(head_leaves, line, opening_bracket)
-    body = bracket_split_build_line(body_leaves, line, opening_bracket, is_body=True)
-    tail = bracket_split_build_line(tail_leaves, line, opening_bracket)
+    head = bracket_split_build_line(
+        head_leaves, line, opening_bracket, component=_BracketSplitComponent.head
+    )
+    body = bracket_split_build_line(
+        body_leaves, line, opening_bracket, component=_BracketSplitComponent.body
+    )
+    tail = bracket_split_build_line(
+        tail_leaves, line, opening_bracket, component=_BracketSplitComponent.tail
+    )
     bracket_split_succeeded_or_raise(head, body, tail)
     bracket_split_succeeded_or_raise(head, body, tail)
+    return RHSResult(head, body, tail, opening_bracket, closing_bracket)
+
+
+def _maybe_split_omitting_optional_parens(
+    rhs: RHSResult,
+    line: Line,
+    mode: Mode,
+    features: Collection[Feature] = (),
+    omit: Collection[LeafID] = (),
+) -> Iterator[Line]:
     if (
         Feature.FORCE_OPTIONAL_PARENTHESES not in features
         # the opening bracket is an optional paren
     if (
         Feature.FORCE_OPTIONAL_PARENTHESES not in features
         # the opening bracket is an optional paren
-        and opening_bracket.type == token.LPAR
-        and not opening_bracket.value
+        and rhs.opening_bracket.type == token.LPAR
+        and not rhs.opening_bracket.value
         # the closing bracket is an optional paren
         # the closing bracket is an optional paren
-        and closing_bracket.type == token.RPAR
-        and not closing_bracket.value
+        and rhs.closing_bracket.type == token.RPAR
+        and not rhs.closing_bracket.value
         # it's not an import (optional parens are the only thing we can split on
         # in this case; attempting a split without them is a waste of time)
         and not line.is_import
         # there are no standalone comments in the body
         # it's not an import (optional parens are the only thing we can split on
         # in this case; attempting a split without them is a waste of time)
         and not line.is_import
         # there are no standalone comments in the body
-        and not body.contains_standalone_comments(0)
+        and not rhs.body.contains_standalone_comments(0)
         # and we can actually remove the parens
         # and we can actually remove the parens
-        and can_omit_invisible_parens(body, line_length, omit_on_explode=omit)
+        and can_omit_invisible_parens(rhs, mode.line_length)
     ):
     ):
-        omit = {id(closing_bracket), *omit}
+        omit = {id(rhs.closing_bracket), *omit}
         try:
         try:
-            yield from right_hand_split(line, line_length, features=features, omit=omit)
-            return
+            # The RHSResult Omitting Optional Parens.
+            rhs_oop = _first_right_hand_split(line, omit=omit)
+            if not (
+                Preview.prefer_splitting_right_hand_side_of_assignments in line.mode
+                # the split is right after `=`
+                and len(rhs.head.leaves) >= 2
+                and rhs.head.leaves[-2].type == token.EQUAL
+                # the left side of assignment contains brackets
+                and any(leaf.type in BRACKETS for leaf in rhs.head.leaves[:-1])
+                # the left side of assignment is short enough (the -1 is for the ending
+                # optional paren)
+                and is_line_short_enough(
+                    rhs.head, mode=replace(mode, line_length=mode.line_length - 1)
+                )
+                # the left side of assignment won't explode further because of magic
+                # trailing comma
+                and rhs.head.magic_trailing_comma is None
+                # the split by omitting optional parens isn't preferred by some other
+                # reason
+                and not _prefer_split_rhs_oop(rhs_oop, mode)
+            ):
+                yield from _maybe_split_omitting_optional_parens(
+                    rhs_oop, line, mode, features=features, omit=omit
+                )
+                return
 
         except CannotSplit as e:
             if not (
 
         except CannotSplit as e:
             if not (
-                can_be_split(body)
-                or is_line_short_enough(body, line_length=line_length)
+                can_be_split(rhs.body) or is_line_short_enough(rhs.body, mode=mode)
             ):
                 raise CannotSplit(
                     "Splitting failed, body is still too long and can't be split."
                 ) from e
 
             ):
                 raise CannotSplit(
                     "Splitting failed, body is still too long and can't be split."
                 ) from e
 
-            elif head.contains_multiline_strings() or tail.contains_multiline_strings():
+            elif (
+                rhs.head.contains_multiline_strings()
+                or rhs.tail.contains_multiline_strings()
+            ):
                 raise CannotSplit(
                     "The current optional pair of parentheses is bound to fail to"
                     " satisfy the splitting algorithm because the head or the tail"
                 raise CannotSplit(
                     "The current optional pair of parentheses is bound to fail to"
                     " satisfy the splitting algorithm because the head or the tail"
@@ -524,13 +818,42 @@ def right_hand_split(
                     " line."
                 ) from e
 
                     " line."
                 ) from e
 
-    ensure_visible(opening_bracket)
-    ensure_visible(closing_bracket)
-    for result in (head, body, tail):
+    ensure_visible(rhs.opening_bracket)
+    ensure_visible(rhs.closing_bracket)
+    for result in (rhs.head, rhs.body, rhs.tail):
         if result:
             yield result
 
 
         if result:
             yield result
 
 
+def _prefer_split_rhs_oop(rhs_oop: RHSResult, mode: Mode) -> bool:
+    """
+    Returns whether we should prefer the result from a split omitting optional parens.
+    """
+    has_closing_bracket_after_assign = False
+    for leaf in reversed(rhs_oop.head.leaves):
+        if leaf.type == token.EQUAL:
+            break
+        if leaf.type in CLOSING_BRACKETS:
+            has_closing_bracket_after_assign = True
+            break
+    return (
+        # contains matching brackets after the `=` (done by checking there is a
+        # closing bracket)
+        has_closing_bracket_after_assign
+        or (
+            # the split is actually from inside the optional parens (done by checking
+            # the first line still contains the `=`)
+            any(leaf.type == token.EQUAL for leaf in rhs_oop.head.leaves)
+            # the first line is short enough
+            and is_line_short_enough(rhs_oop.head, mode=mode)
+        )
+        # contains unsplittable type ignore
+        or rhs_oop.head.contains_unsplittable_type_ignore()
+        or rhs_oop.body.contains_unsplittable_type_ignore()
+        or rhs_oop.tail.contains_unsplittable_type_ignore()
+    )
+
+
 def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None:
     """Raise :exc:`CannotSplit` if the last left- or right-hand split failed.
 
 def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None:
     """Raise :exc:`CannotSplit` if the last left- or right-hand split failed.
 
@@ -558,15 +881,23 @@ def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None
 
 
 def bracket_split_build_line(
 
 
 def bracket_split_build_line(
-    leaves: List[Leaf], original: Line, opening_bracket: Leaf, *, is_body: bool = False
+    leaves: List[Leaf],
+    original: Line,
+    opening_bracket: Leaf,
+    *,
+    component: _BracketSplitComponent,
 ) -> Line:
     """Return a new line with given `leaves` and respective comments from `original`.
 
 ) -> Line:
     """Return a new line with given `leaves` and respective comments from `original`.
 
-    If `is_body` is True, the result line is one-indented inside brackets and as such
-    has its first leaf's prefix normalized and a trailing comma added when expected.
+    If it's the head component, brackets will be tracked so trailing commas are
+    respected.
+
+    If it's the body component, the result line is one-indented inside brackets and as
+    such has its first leaf's prefix normalized and a trailing comma added when
+    expected.
     """
     result = Line(mode=original.mode, depth=original.depth)
     """
     result = Line(mode=original.mode, depth=original.depth)
-    if is_body:
+    if component is _BracketSplitComponent.body:
         result.inside_brackets = True
         result.depth += 1
         if leaves:
         result.inside_brackets = True
         result.depth += 1
         if leaves:
@@ -592,6 +923,13 @@ def bracket_split_build_line(
                     )
                     if isinstance(node, Node) and isinstance(node.prev_sibling, Leaf)
                 )
                     )
                     if isinstance(node, Node) and isinstance(node.prev_sibling, Leaf)
                 )
+                # Except the false negatives above for PEP 604 unions where we
+                # can't add the comma.
+                and not (
+                    leaves[0].parent
+                    and leaves[0].parent.next_sibling
+                    and leaves[0].parent.next_sibling.type == token.VBAR
+                )
             )
 
             if original.is_import or no_commas:
             )
 
             if original.is_import or no_commas:
@@ -604,12 +942,21 @@ def bracket_split_build_line(
                         leaves.insert(i + 1, new_comma)
                     break
 
                         leaves.insert(i + 1, new_comma)
                     break
 
+    leaves_to_track: Set[LeafID] = set()
+    if component is _BracketSplitComponent.head:
+        leaves_to_track = get_leaves_inside_matching_brackets(leaves)
     # Populate the line
     for leaf in leaves:
     # Populate the line
     for leaf in leaves:
-        result.append(leaf, preformatted=True)
+        result.append(
+            leaf,
+            preformatted=True,
+            track_bracket=id(leaf) in leaves_to_track,
+        )
         for comment_after in original.comments_after(leaf):
             result.append(comment_after, preformatted=True)
         for comment_after in original.comments_after(leaf):
             result.append(comment_after, preformatted=True)
-    if is_body and should_split_line(result, opening_bracket):
+    if component is _BracketSplitComponent.body and should_split_line(
+        result, opening_bracket
+    ):
         result.should_split_rhs = True
     return result
 
         result.should_split_rhs = True
     return result
 
@@ -621,16 +968,39 @@ def dont_increase_indentation(split_func: Transformer) -> Transformer:
     """
 
     @wraps(split_func)
     """
 
     @wraps(split_func)
-    def split_wrapper(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
-        for line in split_func(line, features):
-            normalize_prefix(line.leaves[0], inside_brackets=True)
-            yield line
+    def split_wrapper(
+        line: Line, features: Collection[Feature], mode: Mode
+    ) -> Iterator[Line]:
+        for split_line in split_func(line, features, mode):
+            normalize_prefix(split_line.leaves[0], inside_brackets=True)
+            yield split_line
 
     return split_wrapper
 
 
 
     return split_wrapper
 
 
+def _get_last_non_comment_leaf(line: Line) -> Optional[int]:
+    for leaf_idx in range(len(line.leaves) - 1, 0, -1):
+        if line.leaves[leaf_idx].type != STANDALONE_COMMENT:
+            return leaf_idx
+    return None
+
+
+def _safe_add_trailing_comma(safe: bool, delimiter_priority: int, line: Line) -> Line:
+    if (
+        safe
+        and delimiter_priority == COMMA_PRIORITY
+        and line.leaves[-1].type != token.COMMA
+        and line.leaves[-1].type != STANDALONE_COMMENT
+    ):
+        new_comma = Leaf(token.COMMA, ",")
+        line.append(new_comma)
+    return line
+
+
 @dont_increase_indentation
 @dont_increase_indentation
-def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
+def delimiter_split(
+    line: Line, features: Collection[Feature], mode: Mode
+) -> Iterator[Line]:
     """Split according to delimiters of the highest priority.
 
     If the appropriate Features are given, the split will add trailing commas
     """Split according to delimiters of the highest priority.
 
     If the appropriate Features are given, the split will add trailing commas
@@ -670,7 +1040,8 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
             )
             current_line.append(leaf)
 
             )
             current_line.append(leaf)
 
-    for leaf in line.leaves:
+    last_non_comment_leaf = _get_last_non_comment_leaf(line)
+    for leaf_idx, leaf in enumerate(line.leaves):
         yield from append_to_line(leaf)
 
         for comment_after in line.comments_after(leaf):
         yield from append_to_line(leaf)
 
         for comment_after in line.comments_after(leaf):
@@ -687,6 +1058,15 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
                     trailing_comma_safe and Feature.TRAILING_COMMA_IN_CALL in features
                 )
 
                     trailing_comma_safe and Feature.TRAILING_COMMA_IN_CALL in features
                 )
 
+        if (
+            Preview.add_trailing_comma_consistently in mode
+            and last_leaf.type == STANDALONE_COMMENT
+            and leaf_idx == last_non_comment_leaf
+        ):
+            current_line = _safe_add_trailing_comma(
+                trailing_comma_safe, delimiter_priority, current_line
+            )
+
         leaf_priority = bt.delimiters.get(id(leaf))
         if leaf_priority == delimiter_priority:
             yield current_line
         leaf_priority = bt.delimiters.get(id(leaf))
         if leaf_priority == delimiter_priority:
             yield current_line
@@ -695,20 +1075,15 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[
                 mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets
             )
     if current_line:
                 mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets
             )
     if current_line:
-        if (
-            trailing_comma_safe
-            and delimiter_priority == COMMA_PRIORITY
-            and current_line.leaves[-1].type != token.COMMA
-            and current_line.leaves[-1].type != STANDALONE_COMMENT
-        ):
-            new_comma = Leaf(token.COMMA, ",")
-            current_line.append(new_comma)
+        current_line = _safe_add_trailing_comma(
+            trailing_comma_safe, delimiter_priority, current_line
+        )
         yield current_line
 
 
 @dont_increase_indentation
 def standalone_comment_split(
         yield current_line
 
 
 @dont_increase_indentation
 def standalone_comment_split(
-    line: Line, features: Collection[Feature] = ()
+    line: Line, features: Collection[Feature], mode: Mode
 ) -> Iterator[Line]:
     """Split standalone comments from the rest of the line."""
     if not line.contains_standalone_comments(0):
 ) -> Iterator[Line]:
     """Split standalone comments from the rest of the line."""
     if not line.contains_standalone_comments(0):
@@ -759,7 +1134,9 @@ def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None:
     leaf.prefix = ""
 
 
     leaf.prefix = ""
 
 
-def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
+def normalize_invisible_parens(
+    node: Node, parens_after: Set[str], *, mode: Mode, features: Collection[Feature]
+) -> None:
     """Make existing optional parentheses invisible or create new ones.
 
     `parens_after` is a set of string leaf values immediately after which parens
     """Make existing optional parentheses invisible or create new ones.
 
     `parens_after` is a set of string leaf values immediately after which parens
@@ -772,12 +1149,21 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
         if pc.value in FMT_OFF:
             # This `node` has a prefix with `# fmt: off`, don't mess with parens.
             return
         if pc.value in FMT_OFF:
             # This `node` has a prefix with `# fmt: off`, don't mess with parens.
             return
+
+    # The multiple context managers grammar has a different pattern, thus this is
+    # separate from the for-loop below. This possibly wraps them in invisible parens,
+    # and later will be removed in remove_with_parens when needed.
+    if node.type == syms.with_stmt:
+        _maybe_wrap_cms_in_parens(node, mode, features)
+
     check_lpar = False
     for index, child in enumerate(list(node.children)):
         # Fixes a bug where invisible parens are not properly stripped from
         # assignment statements that contain type annotations.
         if isinstance(child, Node) and child.type == syms.annassign:
     check_lpar = False
     for index, child in enumerate(list(node.children)):
         # Fixes a bug where invisible parens are not properly stripped from
         # assignment statements that contain type annotations.
         if isinstance(child, Node) and child.type == syms.annassign:
-            normalize_invisible_parens(child, parens_after=parens_after)
+            normalize_invisible_parens(
+                child, parens_after=parens_after, mode=mode, features=features
+            )
 
         # Add parentheses around long tuple unpacking in assignments.
         if (
 
         # Add parentheses around long tuple unpacking in assignments.
         if (
@@ -788,45 +1174,212 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
             check_lpar = True
 
         if check_lpar:
             check_lpar = True
 
         if check_lpar:
-            if child.type == syms.atom:
-                if maybe_make_parens_invisible_in_atom(child, parent=node):
+            if (
+                child.type == syms.atom
+                and node.type == syms.for_stmt
+                and isinstance(child.prev_sibling, Leaf)
+                and child.prev_sibling.type == token.NAME
+                and child.prev_sibling.value == "for"
+            ):
+                if maybe_make_parens_invisible_in_atom(
+                    child,
+                    parent=node,
+                    remove_brackets_around_comma=True,
+                ):
+                    wrap_in_parentheses(node, child, visible=False)
+            elif isinstance(child, Node) and node.type == syms.with_stmt:
+                remove_with_parens(child, node)
+            elif child.type == syms.atom:
+                if maybe_make_parens_invisible_in_atom(
+                    child,
+                    parent=node,
+                ):
                     wrap_in_parentheses(node, child, visible=False)
             elif is_one_tuple(child):
                 wrap_in_parentheses(node, child, visible=True)
             elif node.type == syms.import_from:
                     wrap_in_parentheses(node, child, visible=False)
             elif is_one_tuple(child):
                 wrap_in_parentheses(node, child, visible=True)
             elif node.type == syms.import_from:
-                # "import from" nodes store parentheses directly as part of
-                # the statement
-                if child.type == token.LPAR:
-                    # make parentheses invisible
-                    child.value = ""  # type: ignore
-                    node.children[-1].value = ""  # type: ignore
-                elif child.type != token.STAR:
-                    # insert invisible parentheses
-                    node.insert_child(index, Leaf(token.LPAR, ""))
-                    node.append_child(Leaf(token.RPAR, ""))
+                _normalize_import_from(node, child, index)
                 break
                 break
+            elif (
+                index == 1
+                and child.type == token.STAR
+                and node.type == syms.except_clause
+            ):
+                # In except* (PEP 654), the star is actually part of
+                # of the keyword. So we need to skip the insertion of
+                # invisible parentheses to work more precisely.
+                continue
 
             elif not (isinstance(child, Leaf) and is_multiline_string(child)):
                 wrap_in_parentheses(node, child, visible=False)
 
 
             elif not (isinstance(child, Leaf) and is_multiline_string(child)):
                 wrap_in_parentheses(node, child, visible=False)
 
-        check_lpar = isinstance(child, Leaf) and child.value in parens_after
+        comma_check = child.type == token.COMMA
 
 
+        check_lpar = isinstance(child, Leaf) and (
+            child.value in parens_after or comma_check
+        )
 
 
-def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:
+
+def _normalize_import_from(parent: Node, child: LN, index: int) -> None:
+    # "import from" nodes store parentheses directly as part of
+    # the statement
+    if is_lpar_token(child):
+        assert is_rpar_token(parent.children[-1])
+        # make parentheses invisible
+        child.value = ""
+        parent.children[-1].value = ""
+    elif child.type != token.STAR:
+        # insert invisible parentheses
+        parent.insert_child(index, Leaf(token.LPAR, ""))
+        parent.append_child(Leaf(token.RPAR, ""))
+
+
+def remove_await_parens(node: Node) -> None:
+    if node.children[0].type == token.AWAIT and len(node.children) > 1:
+        if (
+            node.children[1].type == syms.atom
+            and node.children[1].children[0].type == token.LPAR
+        ):
+            if maybe_make_parens_invisible_in_atom(
+                node.children[1],
+                parent=node,
+                remove_brackets_around_comma=True,
+            ):
+                wrap_in_parentheses(node, node.children[1], visible=False)
+
+            # Since await is an expression we shouldn't remove
+            # brackets in cases where this would change
+            # the AST due to operator precedence.
+            # Therefore we only aim to remove brackets around
+            # power nodes that aren't also await expressions themselves.
+            # https://peps.python.org/pep-0492/#updated-operator-precedence-table
+            # N.B. We've still removed any redundant nested brackets though :)
+            opening_bracket = cast(Leaf, node.children[1].children[0])
+            closing_bracket = cast(Leaf, node.children[1].children[-1])
+            bracket_contents = node.children[1].children[1]
+            if isinstance(bracket_contents, Node):
+                if bracket_contents.type != syms.power:
+                    ensure_visible(opening_bracket)
+                    ensure_visible(closing_bracket)
+                elif (
+                    bracket_contents.type == syms.power
+                    and bracket_contents.children[0].type == token.AWAIT
+                ):
+                    ensure_visible(opening_bracket)
+                    ensure_visible(closing_bracket)
+                    # If we are in a nested await then recurse down.
+                    remove_await_parens(bracket_contents)
+
+
+def _maybe_wrap_cms_in_parens(
+    node: Node, mode: Mode, features: Collection[Feature]
+) -> None:
+    """When enabled and safe, wrap the multiple context managers in invisible parens.
+
+    It is only safe when `features` contain Feature.PARENTHESIZED_CONTEXT_MANAGERS.
+    """
+    if (
+        Feature.PARENTHESIZED_CONTEXT_MANAGERS not in features
+        or Preview.wrap_multiple_context_managers_in_parens not in mode
+        or len(node.children) <= 2
+        # If it's an atom, it's already wrapped in parens.
+        or node.children[1].type == syms.atom
+    ):
+        return
+    colon_index: Optional[int] = None
+    for i in range(2, len(node.children)):
+        if node.children[i].type == token.COLON:
+            colon_index = i
+            break
+    if colon_index is not None:
+        lpar = Leaf(token.LPAR, "")
+        rpar = Leaf(token.RPAR, "")
+        context_managers = node.children[1:colon_index]
+        for child in context_managers:
+            child.remove()
+        # After wrapping, the with_stmt will look like this:
+        #   with_stmt
+        #     NAME 'with'
+        #     atom
+        #       LPAR ''
+        #       testlist_gexp
+        #         ... <-- context_managers
+        #       /testlist_gexp
+        #       RPAR ''
+        #     /atom
+        #     COLON ':'
+        new_child = Node(
+            syms.atom, [lpar, Node(syms.testlist_gexp, context_managers), rpar]
+        )
+        node.insert_child(1, new_child)
+
+
+def remove_with_parens(node: Node, parent: Node) -> None:
+    """Recursively hide optional parens in `with` statements."""
+    # Removing all unnecessary parentheses in with statements in one pass is a tad
+    # complex as different variations of bracketed statements result in pretty
+    # different parse trees:
+    #
+    # with (open("file")) as f:                       # this is an asexpr_test
+    #     ...
+    #
+    # with (open("file") as f):                       # this is an atom containing an
+    #     ...                                         # asexpr_test
+    #
+    # with (open("file")) as f, (open("file")) as f:  # this is asexpr_test, COMMA,
+    #     ...                                         # asexpr_test
+    #
+    # with (open("file") as f, open("file") as f):    # an atom containing a
+    #     ...                                         # testlist_gexp which then
+    #                                                 # contains multiple asexpr_test(s)
+    if node.type == syms.atom:
+        if maybe_make_parens_invisible_in_atom(
+            node,
+            parent=parent,
+            remove_brackets_around_comma=True,
+        ):
+            wrap_in_parentheses(parent, node, visible=False)
+        if isinstance(node.children[1], Node):
+            remove_with_parens(node.children[1], node)
+    elif node.type == syms.testlist_gexp:
+        for child in node.children:
+            if isinstance(child, Node):
+                remove_with_parens(child, node)
+    elif node.type == syms.asexpr_test and not any(
+        leaf.type == token.COLONEQUAL for leaf in node.leaves()
+    ):
+        if maybe_make_parens_invisible_in_atom(
+            node.children[0],
+            parent=node,
+            remove_brackets_around_comma=True,
+        ):
+            wrap_in_parentheses(node, node.children[0], visible=False)
+
+
+def maybe_make_parens_invisible_in_atom(
+    node: LN,
+    parent: LN,
+    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 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 (
         node.type != syms.atom
         or is_empty_tuple(node)
         or is_one_tuple(node)
         or (is_yield(node) and parent.type != syms.expr_stmt)
     if (
         node.type != syms.atom
         or is_empty_tuple(node)
         or is_one_tuple(node)
         or (is_yield(node) and parent.type != syms.expr_stmt)
-        or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
+        or (
+            # This condition tries to prevent removing non-optional brackets
+            # around a tuple, however, can be a bit overzealous so we provide
+            # and option to skip this check for `for` and `with` statements.
+            not remove_brackets_around_comma
+            and max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
+        )
+        or is_tuple_containing_walrus(node)
     ):
         return False
 
     ):
         return False
 
@@ -836,20 +1389,33 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:
             syms.expr_stmt,
             syms.assert_stmt,
             syms.return_stmt,
             syms.expr_stmt,
             syms.assert_stmt,
             syms.return_stmt,
+            syms.except_clause,
+            syms.funcdef,
+            syms.with_stmt,
             # these ones aren't useful to end users, but they do please fuzzers
             syms.for_stmt,
             syms.del_stmt,
             # these ones aren't useful to end users, but they do please fuzzers
             syms.for_stmt,
             syms.del_stmt,
+            syms.for_stmt,
         ]:
             return False
 
     first = node.children[0]
     last = node.children[-1]
         ]:
             return False
 
     first = node.children[0]
     last = node.children[-1]
-    if first.type == token.LPAR and last.type == token.RPAR:
+    if is_lpar_token(first) and is_rpar_token(last):
         middle = node.children[1]
         # make parentheses invisible
         middle = node.children[1]
         # make parentheses invisible
-        first.value = ""  # type: ignore
-        last.value = ""  # type: ignore
-        maybe_make_parens_invisible_in_atom(middle, parent=parent)
+        if (
+            # If the prefix of `middle` includes a type comment with
+            # ignore annotation, then we do not remove the parentheses
+            not is_type_ignore_comment_string(middle.prefix.strip())
+        ):
+            first.value = ""
+            last.value = ""
+        maybe_make_parens_invisible_in_atom(
+            middle,
+            parent=parent,
+            remove_brackets_around_comma=remove_brackets_around_comma,
+        )
 
         if is_atom_with_invisible_parens(middle):
             # Strip the invisible parens from `middle` by replacing
 
         if is_atom_with_invisible_parens(middle):
             # Strip the invisible parens from `middle` by replacing
@@ -924,7 +1490,8 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
                 if (
                     prev
                     and prev.type == token.COMMA
                 if (
                     prev
                     and prev.type == token.COMMA
-                    and not is_one_tuple_between(
+                    and leaf.opening_bracket is not None
+                    and not is_one_sequence_between(
                         leaf.opening_bracket, leaf, line.leaves
                     )
                 ):
                         leaf.opening_bracket, leaf, line.leaves
                     )
                 ):
@@ -951,7 +1518,8 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf
             if (
                 prev
                 and prev.type == token.COMMA
             if (
                 prev
                 and prev.type == token.COMMA
-                and not is_one_tuple_between(leaf.opening_bracket, leaf, line.leaves)
+                and leaf.opening_bracket is not None
+                and not is_one_sequence_between(leaf.opening_bracket, leaf, line.leaves)
             ):
                 # Never omit bracket pairs with trailing commas.
                 # We need to explode on those.
             ):
                 # Never omit bracket pairs with trailing commas.
                 # We need to explode on those.
@@ -973,20 +1541,22 @@ def run_transformer(
     if not line_str:
         line_str = line_to_string(line)
     result: List[Line] = []
     if not line_str:
         line_str = line_to_string(line)
     result: List[Line] = []
-    for transformed_line in transform(line, features):
+    for transformed_line in transform(line, features, mode):
         if str(transformed_line).strip("\n") == line_str:
             raise CannotTransform("Line transformer returned an unchanged result")
 
         result.extend(transform_line(transformed_line, mode=mode, features=features))
 
         if str(transformed_line).strip("\n") == line_str:
             raise CannotTransform("Line transformer returned an unchanged result")
 
         result.extend(transform_line(transformed_line, mode=mode, features=features))
 
+    features_set = set(features)
     if (
     if (
-        transform.__name__ != "rhs"
+        Feature.FORCE_OPTIONAL_PARENTHESES in features_set
+        or transform.__class__.__name__ != "rhs"
         or not line.bracket_tracker.invisible
         or any(bracket.value for bracket in line.bracket_tracker.invisible)
         or line.contains_multiline_strings()
         or result[0].contains_uncollapsable_type_comments()
         or result[0].contains_unsplittable_type_ignore()
         or not line.bracket_tracker.invisible
         or any(bracket.value for bracket in line.bracket_tracker.invisible)
         or line.contains_multiline_strings()
         or result[0].contains_uncollapsable_type_comments()
         or result[0].contains_unsplittable_type_ignore()
-        or is_line_short_enough(result[0], line_length=mode.line_length)
+        or is_line_short_enough(result[0], mode=mode)
         # If any leaves have no parents (which _can_ occur since
         # `transform(line)` potentially destroys the line's underlying node
         # structure), then we can't proceed. Doing so would cause the below
         # If any leaves have no parents (which _can_ occur since
         # `transform(line)` potentially destroys the line's underlying node
         # structure), then we can't proceed. Doing so would cause the below
@@ -997,12 +1567,10 @@ def run_transformer(
 
     line_copy = line.clone()
     append_leaves(line_copy, line, line.leaves)
 
     line_copy = line.clone()
     append_leaves(line_copy, line, line.leaves)
-    features_fop = set(features) | {Feature.FORCE_OPTIONAL_PARENTHESES}
+    features_fop = features_set | {Feature.FORCE_OPTIONAL_PARENTHESES}
     second_opinion = run_transformer(
         line_copy, transform, mode, features_fop, line_str=line_str
     )
     second_opinion = run_transformer(
         line_copy, transform, mode, features_fop, line_str=line_str
     )
-    if all(
-        is_line_short_enough(ln, line_length=mode.line_length) for ln in second_opinion
-    ):
+    if all(is_line_short_enough(ln, mode=mode) for ln in second_opinion):
         result = second_opinion
     return result
         result = second_opinion
     return result