]> 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:

Fix CI with latest flake8-bugbear (#3412)
[etc/vim.git] / src / black / linegen.py
index 5d92011da9ab1c5de432cc3a1f68421b1dd68cf1..219495e9a5e12771ce8837ddcae4c60c24018b73 100644 (file)
@@ -1,38 +1,73 @@
 """
 Generating lines of code.
 """
-from functools import partial, wraps
 import sys
-from typing import Collection, Iterator, List, Optional, Set, Union
+from enum import Enum, auto
+from functools import partial, wraps
+from typing import Collection, Iterator, List, Optional, Set, Union, cast
 
-from black.nodes import WHITESPACE, RARROW, STATEMENT, STANDALONE_COMMENT
-from black.nodes import ASSIGNMENTS, OPENING_BRACKETS, CLOSING_BRACKETS
-from black.nodes import Visitor, syms, is_arith_like, ensure_visible
+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,
+    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,
+    CLOSING_BRACKETS,
+    OPENING_BRACKETS,
+    RARROW,
+    STANDALONE_COMMENT,
+    STATEMENT,
+    WHITESPACE,
+    Visitor,
+    ensure_visible,
+    is_arith_like,
+    is_atom_with_invisible_parens,
     is_docstring,
     is_empty_tuple,
-    is_one_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_vararg,
+    is_walrus_assignment,
+    is_yield,
+    syms,
+    wrap_in_parentheses,
 )
-from black.nodes import is_name_token, is_lpar_token, is_rpar_token
-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 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, StringSplitter
-from black.trans import StringParenWrapper, StringParenStripper, hug_power_op
-from black.mode import Mode, Feature, Preview
-
-from blib2to3.pytree import Node, Leaf
+from black.strings import (
+    fix_docstring,
+    get_string_prefix,
+    normalize_string_prefix,
+    normalize_string_quotes,
+)
+from black.trans import (
+    CannotTransform,
+    StringMerger,
+    StringParenStripper,
+    StringParenWrapper,
+    StringSplitter,
+    Transformer,
+    hug_power_op,
+)
 from blib2to3.pgen2 import token
-
+from blib2to3.pytree import Leaf, Node
 
 # types
 LeafID = int
@@ -144,6 +179,33 @@ class LineGenerator(Visitor[Line]):
 
             yield from self.visit(child)
 
+    def visit_funcdef(self, node: Node) -> Iterator[Line]:
+        """Visit function definition."""
+        if Preview.annotation_parens not in self.mode:
+            yield from self.visit_stmt(node, keywords={"def"}, parens=set())
+        else:
+            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(), preview=self.mode.preview)
@@ -193,7 +255,9 @@ class LineGenerator(Visitor[Line]):
         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)
@@ -226,6 +290,9 @@ class LineGenerator(Visitor[Line]):
             ):
                 wrap_in_parentheses(node, leaf)
 
+        if Preview.remove_redundant_parens in self.mode:
+            remove_await_parens(node)
+
         yield from self.visit_default(node)
 
     def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
@@ -263,7 +330,24 @@ class LineGenerator(Visitor[Line]):
         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)
+            if Preview.normalize_docstring_quotes_and_prefixes_properly in self.mode:
+                # There was a bug where --skip-string-normalization wouldn't stop us
+                # from normalizing docstring prefixes. To maintain stability, we can
+                # only address this buggy behaviour while the preview style is enabled.
+                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
+            else:
+                # ... otherwise, we'll keep the buggy behaviour >.<
+                docstring = normalize_string_prefix(leaf.value)
             prefix = get_string_prefix(docstring)
             docstring = docstring[len(prefix) :]  # Remove the prefix
             quote_char = docstring[0]
@@ -275,9 +359,9 @@ 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
+            indent = " " * 4 * self.current_line.depth
 
             if is_multiline_string(leaf):
-                indent = " " * 4 * self.current_line.depth
                 docstring = fix_docstring(docstring, indent)
             else:
                 docstring = docstring.strip()
@@ -299,7 +383,30 @@ class LineGenerator(Visitor[Line]):
 
             # 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 Preview.long_docstring_quotes_on_newline in 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, then we need to add the length
+                # of the indent, prefix, and starting quotes. Ending quotes are
+                # handled later.
+                lines = docstring.splitlines()
+                last_line_length = len(lines[-1]) if docstring else 0
+
+                if len(lines) == 1:
+                    last_line_length += len(indent) + len(prefix) + quote_len
+
+                # 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 last_line_length + quote_len > self.mode.line_length:
+                    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)
 
@@ -318,9 +425,14 @@ class LineGenerator(Visitor[Line]):
         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)
+        if self.mode.preview:
+            self.visit_except_clause = partial(
+                v, keywords={"except"}, parens={"except"}
+            )
+            self.visit_with_stmt = partial(v, keywords={"with"}, parens={"with"})
+        else:
+            self.visit_except_clause = partial(v, keywords={"except"}, parens=Ø)
+            self.visit_with_stmt = partial(v, keywords={"with"}, parens=Ø)
         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"})
@@ -455,6 +567,12 @@ def transform_line(
         yield line
 
 
+class _BracketSplitComponent(Enum):
+    head = auto()
+    body = auto()
+    tail = auto()
+
+
 def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator[Line]:
     """Split line into many lines, starting with the first matching bracket pair.
 
@@ -472,7 +590,10 @@ 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
+            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:
@@ -482,9 +603,15 @@ def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator
     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:
@@ -530,9 +657,15 @@ def right_hand_split(
     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)
     if (
         Feature.FORCE_OPTIONAL_PARENTHESES not in features
@@ -606,15 +739,23 @@ def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None
 
 
 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`.
 
-    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)
-    if is_body:
+    if component is _BracketSplitComponent.body:
         result.inside_brackets = True
         result.depth += 1
         if leaves:
@@ -652,12 +793,24 @@ def bracket_split_build_line(
                         leaves.insert(i + 1, new_comma)
                     break
 
+    leaves_to_track: Set[LeafID] = set()
+    if (
+        Preview.handle_trailing_commas_in_head in original.mode
+        and component is _BracketSplitComponent.head
+    ):
+        leaves_to_track = get_leaves_inside_matching_brackets(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)
-    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
 
@@ -670,9 +823,9 @@ def dont_increase_indentation(split_func: Transformer) -> Transformer:
 
     @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
+        for split_line in split_func(line, features):
+            normalize_prefix(split_line.leaves[0], inside_brackets=True)
+            yield split_line
 
     return split_wrapper
 
@@ -840,8 +993,27 @@ def normalize_invisible_parens(
             check_lpar = True
 
         if check_lpar:
-            if child.type == syms.atom:
-                if maybe_make_parens_invisible_in_atom(child, parent=node):
+            if (
+                preview
+                and 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 preview and 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)
@@ -858,28 +1030,127 @@ def normalize_invisible_parens(
                     node.insert_child(index, Leaf(token.LPAR, ""))
                     node.append_child(Leaf(token.RPAR, ""))
                 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)
 
-        check_lpar = isinstance(child, Leaf) and child.value in parens_after
+        comma_check = child.type == token.COMMA if preview else False
+
+        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 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 = cast(Node, node.children[1].children[1])
+            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 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 (
         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
+        )
     ):
         return False
 
@@ -902,7 +1173,11 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool:
         # make parentheses invisible
         first.value = ""
         last.value = ""
-        maybe_make_parens_invisible_in_atom(middle, parent=parent)
+        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