+import ast
import asyncio
-from asyncio.base_events import BaseEventLoop
from concurrent.futures import Executor, ProcessPoolExecutor
+from contextlib import contextmanager
from datetime import datetime
from enum import Enum
from functools import lru_cache, partial, wraps
import os
from pathlib import Path
import pickle
-import re
+import regex as re
import signal
import sys
import tempfile
import tokenize
+import traceback
from typing import (
Any,
Callable,
from attr import dataclass, evolve, Factory
import click
import toml
+from typed_ast import ast3, ast27
# lib2to3 fork
from blib2to3.pytree import Node, Leaf, type_repr
from blib2to3.pgen2.grammar import Grammar
from blib2to3.pgen2.parse import ParseError
+from _version import version as __version__
-__version__ = "19.3b0"
DEFAULT_LINE_LENGTH = 88
DEFAULT_EXCLUDES = (
r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|_build|buck-out|build|dist)/"
Priority = int
Index = int
LN = Union[Leaf, Node]
-SplitFunc = Callable[["Line", bool], Iterator["Line"]]
+SplitFunc = Callable[["Line", Collection["Feature"]], Iterator["Line"]]
Timestamp = float
FileSize = int
CacheInfo = Tuple[Timestamp, FileSize]
UNICODE_LITERALS = 1
F_STRINGS = 2
NUMERIC_UNDERSCORES = 3
- TRAILING_COMMA = 4
+ TRAILING_COMMA_IN_CALL = 4
+ TRAILING_COMMA_IN_DEF = 5
+ # The following two feature-flags are mutually exclusive, and exactly one should be
+ # set for every version of python.
+ ASYNC_IDENTIFIERS = 6
+ ASYNC_KEYWORDS = 7
+ ASSIGNMENT_EXPRESSIONS = 8
+ POS_ONLY_ARGUMENTS = 9
VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = {
- TargetVersion.PY27: set(),
- TargetVersion.PY33: {Feature.UNICODE_LITERALS},
- TargetVersion.PY34: {Feature.UNICODE_LITERALS},
- TargetVersion.PY35: {Feature.UNICODE_LITERALS, Feature.TRAILING_COMMA},
+ TargetVersion.PY27: {Feature.ASYNC_IDENTIFIERS},
+ TargetVersion.PY33: {Feature.UNICODE_LITERALS, Feature.ASYNC_IDENTIFIERS},
+ TargetVersion.PY34: {Feature.UNICODE_LITERALS, Feature.ASYNC_IDENTIFIERS},
+ TargetVersion.PY35: {
+ Feature.UNICODE_LITERALS,
+ Feature.TRAILING_COMMA_IN_CALL,
+ Feature.ASYNC_IDENTIFIERS,
+ },
TargetVersion.PY36: {
Feature.UNICODE_LITERALS,
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
- Feature.TRAILING_COMMA,
+ Feature.TRAILING_COMMA_IN_CALL,
+ Feature.TRAILING_COMMA_IN_DEF,
+ Feature.ASYNC_IDENTIFIERS,
},
TargetVersion.PY37: {
Feature.UNICODE_LITERALS,
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
- Feature.TRAILING_COMMA,
+ Feature.TRAILING_COMMA_IN_CALL,
+ Feature.TRAILING_COMMA_IN_DEF,
+ Feature.ASYNC_KEYWORDS,
},
TargetVersion.PY38: {
Feature.UNICODE_LITERALS,
Feature.F_STRINGS,
Feature.NUMERIC_UNDERSCORES,
- Feature.TRAILING_COMMA,
+ Feature.TRAILING_COMMA_IN_CALL,
+ Feature.TRAILING_COMMA_IN_DEF,
+ Feature.ASYNC_KEYWORDS,
+ Feature.ASSIGNMENT_EXPRESSIONS,
+ Feature.POS_ONLY_ARGUMENTS,
},
}
@click.command(context_settings=dict(help_option_names=["-h", "--help"]))
+@click.option("-c", "--code", type=str, help="Format the code passed in as a string.")
@click.option(
"-l",
"--line-length",
"--quiet",
is_flag=True,
help=(
- "Don't emit non-error messages to stderr. Errors are still emitted, "
+ "Don't emit non-error messages to stderr. Errors are still emitted; "
"silence those with 2>/dev/null."
),
)
@click.pass_context
def main(
ctx: click.Context,
+ code: Optional[str],
line_length: int,
target_version: List[TargetVersion],
check: bool,
)
if config and verbose:
out(f"Using configuration from {config}.", bold=False, fg="blue")
+ if code is not None:
+ print(format_str(code, mode=mode))
+ ctx.exit(0)
try:
include_regex = re_compile_maybe_verbose(include)
except re.error:
report = Report(check=check, quiet=quiet, verbose=verbose)
root = find_project_root(src)
sources: Set[Path] = set()
+ path_empty(src, quiet, verbose, ctx)
for s in src:
p = Path(s)
if p.is_dir():
err(f"invalid path: {s}")
if len(sources) == 0:
if verbose or not quiet:
- out("No paths given. Nothing to do 😴")
+ out("No Python files are present to be formatted. Nothing to do 😴")
ctx.exit(0)
if len(sources) == 1:
report=report,
)
else:
- loop = asyncio.get_event_loop()
- executor = ProcessPoolExecutor(max_workers=os.cpu_count())
- try:
- loop.run_until_complete(
- schedule_formatting(
- sources=sources,
- fast=fast,
- write_back=write_back,
- mode=mode,
- report=report,
- loop=loop,
- executor=executor,
- )
- )
- finally:
- shutdown(loop)
+ reformat_many(
+ sources=sources, fast=fast, write_back=write_back, mode=mode, report=report
+ )
+
if verbose or not quiet:
- bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨"
- out(f"All done! {bang}")
+ out("Oh no! 💥 💔 💥" if report.return_code else "All done! ✨ 🍰 ✨")
click.secho(str(report), err=True)
ctx.exit(report.return_code)
+def path_empty(src: Tuple[str], quiet: bool, verbose: bool, ctx: click.Context) -> None:
+ """
+ Exit if there is no `src` provided for formatting
+ """
+ if not src:
+ if verbose or not quiet:
+ out("No Path provided. Nothing to do 😴")
+ ctx.exit(0)
+
+
def reformat_one(
src: Path, fast: bool, write_back: WriteBack, mode: FileMode, report: "Report"
) -> None:
"""Reformat a single file under `src` without spawning child processes.
- If `quiet` is True, non-error messages are not output. `line_length`,
- `write_back`, `fast` and `pyi` options are passed to
+ `fast`, `write_back`, and `mode` options are passed to
:func:`format_file_in_place` or :func:`format_stdin_to_stdout`.
"""
try:
report.failed(src, str(exc))
+def reformat_many(
+ sources: Set[Path],
+ fast: bool,
+ write_back: WriteBack,
+ mode: FileMode,
+ report: "Report",
+) -> None:
+ """Reformat multiple files using a ProcessPoolExecutor."""
+ loop = asyncio.get_event_loop()
+ worker_count = os.cpu_count()
+ if sys.platform == "win32":
+ # Work around https://bugs.python.org/issue26903
+ worker_count = min(worker_count, 61)
+ executor = ProcessPoolExecutor(max_workers=worker_count)
+ try:
+ loop.run_until_complete(
+ schedule_formatting(
+ sources=sources,
+ fast=fast,
+ write_back=write_back,
+ mode=mode,
+ report=report,
+ loop=loop,
+ executor=executor,
+ )
+ )
+ finally:
+ shutdown(loop)
+ executor.shutdown()
+
+
async def schedule_formatting(
sources: Set[Path],
fast: bool,
write_back: WriteBack,
mode: FileMode,
report: "Report",
- loop: BaseEventLoop,
+ loop: asyncio.AbstractEventLoop,
executor: Executor,
) -> None:
"""Run formatting of `sources` in parallel using the provided `executor`.
(Use ProcessPoolExecutors for actual parallelism.)
- `line_length`, `write_back`, `fast`, and `pyi` options are passed to
+ `write_back`, `fast`, and `mode` options are passed to
:func:`format_file_in_place`.
"""
cache: Cache = {}
manager = Manager()
lock = manager.Lock()
tasks = {
- loop.run_in_executor(
- executor, format_file_in_place, src, fast, mode, write_back, lock
+ asyncio.ensure_future(
+ loop.run_in_executor(
+ executor, format_file_in_place, src, fast, mode, write_back, lock
+ )
): src
for src in sorted(sources)
}
- pending: Iterable[asyncio.Task] = tasks.keys()
+ pending: Iterable[asyncio.Future] = tasks.keys()
try:
loop.add_signal_handler(signal.SIGINT, cancel, pending)
loop.add_signal_handler(signal.SIGTERM, cancel, pending)
If `write_back` is DIFF, write a diff to stdout. If it is YES, write reformatted
code to the file.
- `line_length` and `fast` options are passed to :func:`format_file_contents`.
+ `mode` and `fast` options are passed to :func:`format_file_contents`.
"""
if src.suffix == ".pyi":
mode = evolve(mode, is_pyi=True)
src_name = f"{src}\t{then} +0000"
dst_name = f"{src}\t{now} +0000"
diff_contents = diff(src_contents, dst_contents, src_name, dst_name)
- if lock:
- lock.acquire()
- try:
+
+ with lock or nullcontext():
f = io.TextIOWrapper(
sys.stdout.buffer,
encoding=encoding,
)
f.write(diff_contents)
f.detach()
- finally:
- if lock:
- lock.release()
+
return True
If `fast` is False, additionally confirm that the reformatted code is
valid by calling :func:`assert_equivalent` and :func:`assert_stable` on it.
- `line_length` is passed to :func:`format_str`.
+ `mode` is passed to :func:`format_str`.
"""
if src_contents.strip() == "":
raise NothingChanged
def format_str(src_contents: str, *, mode: FileMode) -> FileContent:
"""Reformat a string and return new contents.
- `line_length` determines how many characters per line are allowed.
+ `mode` determines formatting options, such as how many characters per line are
+ allowed.
"""
src_node = lib2to3_parse(src_contents.lstrip(), mode.target_versions)
- dst_contents = ""
+ dst_contents = []
future_imports = get_future_imports(src_node)
if mode.target_versions:
versions = mode.target_versions
elt = EmptyLineTracker(is_pyi=mode.is_pyi)
empty_line = Line()
after = 0
+ split_line_features = {
+ feature
+ for feature in {Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF}
+ if supports_feature(versions, feature)
+ }
for current_line in lines.visit(src_node):
for _ in range(after):
- dst_contents += str(empty_line)
+ dst_contents.append(str(empty_line))
before, after = elt.maybe_empty_lines(current_line)
for _ in range(before):
- dst_contents += str(empty_line)
+ dst_contents.append(str(empty_line))
for line in split_line(
- current_line,
- line_length=mode.line_length,
- supports_trailing_commas=supports_feature(versions, Feature.TRAILING_COMMA),
+ current_line, line_length=mode.line_length, features=split_line_features
):
- dst_contents += str(line)
- return dst_contents
+ dst_contents.append(str(line))
+ return "".join(dst_contents)
def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]:
return tiow.read(), encoding, newline
-GRAMMARS = [
- pygram.python_grammar_no_print_statement_no_exec_statement,
- pygram.python_grammar_no_print_statement,
- pygram.python_grammar,
-]
-
-
def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]:
if not target_versions:
- return GRAMMARS
- elif all(not version.is_python2() for version in target_versions):
- # Python 2-compatible code, so don't try Python 3 grammar.
+ # No target_version specified, so try all grammars.
return [
+ # Python 3.7+
+ pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords,
+ # Python 3.0-3.6
pygram.python_grammar_no_print_statement_no_exec_statement,
+ # Python 2.7 with future print_function import
pygram.python_grammar_no_print_statement,
+ # Python 2.7
+ pygram.python_grammar,
+ ]
+ elif all(version.is_python2() for version in target_versions):
+ # Python 2-only code, so try Python 2 grammars.
+ return [
+ # Python 2.7 with future print_function import
+ pygram.python_grammar_no_print_statement,
+ # Python 2.7
+ pygram.python_grammar,
]
else:
- return [pygram.python_grammar]
+ # Python 3-compatible code, so only try Python 3 grammar.
+ grammars = []
+ # If we have to parse both, try to parse async as a keyword first
+ if not supports_feature(target_versions, Feature.ASYNC_IDENTIFIERS):
+ # Python 3.7+
+ grammars.append(
+ pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords # noqa: B950
+ )
+ if not supports_feature(target_versions, Feature.ASYNC_KEYWORDS):
+ # Python 3.0-3.6
+ grammars.append(pygram.python_grammar_no_print_statement_no_exec_statement)
+ # At least one of the above branches must have been taken, because every Python
+ # version has exactly one of the two 'ASYNC_*' flags
+ return grammars
def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node:
token.DOUBLESTAR,
}
STARS = {token.STAR, token.DOUBLESTAR}
+VARARGS_SPECIALS = STARS | {token.SLASH}
VARARGS_PARENTS = {
syms.arglist,
syms.argument, # double star in arglist
"""Return True if there is an yet unmatched open bracket on the line."""
return bool(self.bracket_match)
- def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> int:
+ def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> Priority:
"""Return the highest priority of a delimiter found on the line.
Values are consistent with what `is_split_*_delimiter()` return.
"""
return max(v for k, v in self.delimiters.items() if k not in exclude)
- def delimiter_count_with_priority(self, priority: int = 0) -> int:
+ def delimiter_count_with_priority(self, priority: Priority = 0) -> int:
"""Return the number of delimiters with the given `priority`.
If no `priority` is passed, defaults to max priority on the line.
return True
return False
- def contains_inner_type_comments(self) -> bool:
+ def contains_uncollapsable_type_comments(self) -> bool:
ignored_ids = set()
try:
last_leaf = self.leaves[-1]
ignored_ids.add(id(last_leaf))
- if last_leaf.type == token.COMMA:
- # When trailing commas are inserted by Black for consistency, comments
- # after the previous last element are not moved (they don't have to,
- # rendering will still be correct). So we ignore trailing commas.
+ if last_leaf.type == token.COMMA or (
+ last_leaf.type == token.RPAR and not last_leaf.value
+ ):
+ # When trailing commas or optional parens are inserted by Black for
+ # consistency, comments after the previous last element are not moved
+ # (they don't have to, rendering will still be correct). So we ignore
+ # trailing commas and invisible.
last_leaf = self.leaves[-2]
ignored_ids.add(id(last_leaf))
except IndexError:
return False
+ # A type comment is uncollapsable if it is attached to a leaf
+ # that isn't at the end of the line (since that could cause it
+ # to get associated to a different argument) or if there are
+ # comments before it (since that could cause it to get hidden
+ # behind a comment.
+ comment_seen = False
for leaf_id, comments in self.comments.items():
- if leaf_id in ignored_ids:
- continue
-
for comment in comments:
if is_type_comment(comment):
- return True
+ if leaf_id not in ignored_ids or comment_seen:
+ return True
+
+ comment_seen = True
+
+ return False
+
+ def contains_unsplittable_type_ignore(self) -> bool:
+ if not self.leaves:
+ return False
+
+ # If a 'type: ignore' is attached to the end of a line, we
+ # can't split the line, because we can't know which of the
+ # subexpressions the ignore was meant to apply to.
+ #
+ # We only want this to apply to actual physical lines from the
+ # original source, though: we don't want the presence of a
+ # 'type: ignore' at the end of a multiline expression to
+ # justify pushing it all onto one line. Thus we
+ # (unfortunately) need to check the actual source lines and
+ # only report an unsplittable 'type: ignore' if this line was
+ # one line in the original code.
+
+ # Like in the type comment check above, we need to skip a black added
+ # trailing comma or invisible paren, since it will be the original leaf
+ # before it that has the original line number.
+ last_idx = -1
+ last_leaf = self.leaves[-1]
+ if len(self.leaves) > 2 and (
+ last_leaf.type == token.COMMA
+ or (last_leaf.type == token.RPAR and not last_leaf.value)
+ ):
+ last_idx = -2
+
+ if self.leaves[0].lineno == self.leaves[last_idx].lineno:
+ for node in self.leaves[last_idx:]:
+ for comment in self.comments.get(id(node), []):
+ if is_type_comment(comment, " ignore"):
+ return True
return False
bracket_depth = leaf.bracket_depth
if bracket_depth == depth and leaf.type == token.COMMA:
commas += 1
- if leaf.parent and leaf.parent.type == syms.arglist:
+ if leaf.parent and leaf.parent.type in {
+ syms.arglist,
+ syms.typedargslist,
+ }:
commas += 1
break
comment.prefix = ""
return False
- self.comments.setdefault(id(self.leaves[-1]), []).append(comment)
+ last_leaf = self.leaves[-1]
+ if (
+ last_leaf.type == token.RPAR
+ and not last_leaf.value
+ and last_leaf.parent
+ and len(list(last_leaf.parent.leaves())) <= 3
+ and not is_type_comment(comment)
+ ):
+ # Comments on an optional parens wrapping a single leaf should belong to
+ # the wrapped node except if it's a type comment. Pinning the comment like
+ # this avoids unstable formatting caused by comment migration.
+ if len(self.leaves) < 2:
+ comment.type = STANDALONE_COMMENT
+ comment.prefix = ""
+ return False
+ last_leaf = self.leaves[-2]
+ self.comments.setdefault(id(last_leaf), []).append(comment)
return True
def comments_after(self, leaf: Leaf) -> List[Leaf]:
lines (two on module-level).
"""
before, after = self._maybe_empty_lines(current_line)
- before -= self.previous_after
+ before = (
+ # Black should not insert empty lines at the beginning
+ # of the file
+ 0
+ if self.previous_line is None
+ else before - self.previous_after
+ )
self.previous_after = after
self.previous_line = current_line
return before, after
self.current_line.append(node)
yield from super().visit_default(node)
+ def visit_atom(self, node: Node) -> Iterator[Line]:
+ # Always make parentheses invisible around a single node, because it should
+ # not be needed (except in the case of yield, where removing the parentheses
+ # produces a SyntaxError).
+ if (
+ len(node.children) == 3
+ and isinstance(node.children[0], Leaf)
+ and node.children[0].type == token.LPAR
+ and isinstance(node.children[2], Leaf)
+ and node.children[2].type == token.RPAR
+ and isinstance(node.children[1], Leaf)
+ and not (
+ node.children[1].type == token.NAME
+ and node.children[1].value == "yield"
+ )
+ ):
+ node.children[0].value = ""
+ node.children[2].value = ""
+ yield from super().visit_default(node)
+
+ def visit_factor(self, node: Node) -> Iterator[Line]:
+ """Force parentheses between a unary op and a binary power:
+
+ -2 ** 8 -> -(2 ** 8)
+ """
+ child = node.children[1]
+ if child.type == syms.power and len(child.children) == 3:
+ lpar = Leaf(token.LPAR, "(")
+ rpar = Leaf(token.RPAR, ")")
+ index = child.remove() or 0
+ node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
+ yield from self.visit_default(node)
+
def visit_INDENT(self, node: Node) -> Iterator[Line]:
"""Increase indentation level, maybe yield a line."""
# In blib2to3 INDENT never holds comments.
# that, too.
return prevp.prefix
- elif prevp.type in STARS:
+ elif prevp.type in VARARGS_SPECIALS:
if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS):
return NO
if not prevp or prevp.type == token.LPAR:
return NO
- elif prev.type in {token.EQUAL} | STARS:
+ elif prev.type in {token.EQUAL} | VARARGS_SPECIALS:
return NO
elif p.type == syms.decorator:
return container
-def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int:
+def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority:
"""Return the priority of the `leaf` delimiter, given a line break after it.
The delimiter priorities returned here are from those delimiters that would
return 0
-def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int:
+def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority:
"""Return the priority of the `leaf` delimiter, given a line break before it.
The delimiter priorities returned here are from those delimiters that would
consumed = 0
nlines = 0
+ ignored_lines = 0
for index, line in enumerate(prefix.split("\n")):
consumed += len(line) + 1 # adding the length of the split '\n'
line = line.lstrip()
if not line:
nlines += 1
if not line.startswith("#"):
+ # Escaped newlines outside of a comment are not really newlines at
+ # all. We treat a single-line comment following an escaped newline
+ # as a simple trailing comment.
+ if line.endswith("\\"):
+ ignored_lines += 1
continue
- if index == 0 and not is_endmarker:
+ if index == ignored_lines and not is_endmarker:
comment_type = token.COMMENT # simple trailing comment
else:
comment_type = STANDALONE_COMMENT
line: Line,
line_length: int,
inner: bool = False,
- supports_trailing_commas: bool = False,
+ features: Collection[Feature] = (),
) -> Iterator[Line]:
"""Split a `line` into potentially many lines.
current `line`, possibly transitively. This means we can fallback to splitting
by delimiters if the LHS/RHS don't yield any results.
- If `supports_trailing_commas` is True, splitting may use the TRAILING_COMMA feature.
+ `features` are syntactical features that may be used in the output.
"""
if line.is_comment:
yield line
line_str = str(line).strip("\n")
if (
- not line.contains_inner_type_comments()
+ not line.contains_uncollapsable_type_comments()
and not line.should_explode
- and is_line_short_enough(line, line_length=line_length, line_str=line_str)
+ and (
+ is_line_short_enough(line, line_length=line_length, line_str=line_str)
+ or line.contains_unsplittable_type_ignore()
+ )
):
yield line
return
split_funcs = [left_hand_split]
else:
- def rhs(line: Line, supports_trailing_commas: bool = False) -> Iterator[Line]:
+ def rhs(line: Line, features: Collection[Feature]) -> Iterator[Line]:
for omit in generate_trailers_to_omit(line, line_length):
- lines = list(
- right_hand_split(
- line, line_length, supports_trailing_commas, omit=omit
- )
- )
+ lines = list(right_hand_split(line, line_length, features, omit=omit))
if is_line_short_enough(lines[0], line_length=line_length):
yield from lines
return
# All splits failed, best effort split with no omits.
# This mostly happens to multiline strings that are by definition
# reported as not fitting a single line.
- yield from right_hand_split(line, supports_trailing_commas)
+ yield from right_hand_split(line, line_length, features=features)
if line.inside_brackets:
split_funcs = [delimiter_split, standalone_comment_split, rhs]
# split altogether.
result: List[Line] = []
try:
- for l in split_func(line, supports_trailing_commas):
+ for l in split_func(line, features):
if str(l).strip("\n") == line_str:
raise CannotSplit("Split function returned an unchanged result")
result.extend(
split_line(
- l,
- line_length=line_length,
- inner=True,
- supports_trailing_commas=supports_trailing_commas,
+ l, line_length=line_length, inner=True, features=features
)
)
except CannotSplit:
yield line
-def left_hand_split(
- line: Line, supports_trailing_commas: bool = False
-) -> Iterator[Line]:
+def left_hand_split(line: Line, features: Collection[Feature] = ()) -> 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.
def right_hand_split(
line: Line,
line_length: int,
- supports_trailing_commas: bool = False,
+ features: Collection[Feature] = (),
omit: Collection[LeafID] = (),
) -> Iterator[Line]:
"""Split line into many lines, starting with the last matching bracket pair.
):
omit = {id(closing_bracket), *omit}
try:
- yield from right_hand_split(
- line,
- line_length,
- supports_trailing_commas=supports_trailing_commas,
- omit=omit,
- )
+ yield from right_hand_split(line, line_length, features=features, omit=omit)
return
except CannotSplit:
if leaves:
# Since body is a new indent level, remove spurious leading whitespace.
normalize_prefix(leaves[0], inside_brackets=True)
- # Ensure a trailing comma when expected.
- if original.is_import:
- if leaves[-1].type != token.COMMA:
- leaves.append(Leaf(token.COMMA, ","))
+ # Ensure a trailing comma for imports and standalone function arguments, but
+ # be careful not to add one after any comments.
+ no_commas = original.is_def and not any(
+ l.type == token.COMMA for l in leaves
+ )
+
+ if original.is_import or no_commas:
+ for i in range(len(leaves) - 1, -1, -1):
+ if leaves[i].type == STANDALONE_COMMENT:
+ continue
+ elif leaves[i].type == token.COMMA:
+ break
+ else:
+ leaves.insert(i + 1, Leaf(token.COMMA, ","))
+ break
# Populate the line
for leaf in leaves:
result.append(leaf, preformatted=True)
"""
@wraps(split_func)
- def split_wrapper(
- line: Line, supports_trailing_commas: bool = False
- ) -> Iterator[Line]:
- for l in split_func(line, supports_trailing_commas):
+ def split_wrapper(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
+ for l in split_func(line, features):
normalize_prefix(l.leaves[0], inside_brackets=True)
yield l
@dont_increase_indentation
-def delimiter_split(
- line: Line, supports_trailing_commas: bool = False
-) -> Iterator[Line]:
+def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]:
"""Split according to delimiters of the highest priority.
- If `supports_trailing_commas` is True, the split will add trailing commas
- also in function signatures that contain `*` and `**`.
+ If the appropriate Features are given, the split will add trailing commas
+ also in function signatures and calls that contain `*` and `**`.
"""
try:
last_leaf = line.leaves[-1]
yield from append_to_line(comment_after)
lowest_depth = min(lowest_depth, leaf.bracket_depth)
- if leaf.bracket_depth == lowest_depth and is_vararg(
- leaf, within=VARARGS_PARENTS
- ):
- trailing_comma_safe = trailing_comma_safe and supports_trailing_commas
+ if leaf.bracket_depth == lowest_depth:
+ if is_vararg(leaf, within={syms.typedargslist}):
+ trailing_comma_safe = (
+ trailing_comma_safe and Feature.TRAILING_COMMA_IN_DEF in features
+ )
+ elif is_vararg(leaf, within={syms.arglist, syms.argument}):
+ trailing_comma_safe = (
+ trailing_comma_safe and Feature.TRAILING_COMMA_IN_CALL in features
+ )
+
leaf_priority = bt.delimiters.get(id(leaf))
if leaf_priority == delimiter_priority:
yield current_line
@dont_increase_indentation
def standalone_comment_split(
- line: Line, supports_trailing_commas: bool = False
+ line: Line, features: Collection[Feature] = ()
) -> Iterator[Line]:
"""Split standalone comments from the rest of the line."""
if not line.contains_standalone_comments(0):
)
-def is_type_comment(leaf: Leaf) -> bool:
+def is_type_comment(leaf: Leaf, suffix: str = "") -> bool:
"""Return True if the given leaf is a special comment.
Only returns true for type comments for now."""
t = leaf.type
v = leaf.value
- return t in {token.COMMENT, t == STANDALONE_COMMENT} and v.startswith("# type:")
+ return t in {token.COMMENT, t == STANDALONE_COMMENT} and v.startswith(
+ "# type:" + suffix
+ )
def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None:
new_body = sub_twice(escaped_orig_quote, rf"\1\2{orig_quote}", new_body)
new_body = sub_twice(unescaped_new_quote, rf"\1\\{new_quote}", new_body)
if "f" in prefix.casefold():
- matches = re.findall(r"[^{]\{(.*?)\}[^}]", new_body)
+ matches = re.findall(
+ r"""
+ (?:[^{]|^)\{ # start of the string or a non-{ followed by a single {
+ ([^{].*?) # contents of the brackets except if begins with {{
+ \}(?:[^}]|$) # A } followed by end of the string or a non-}
+ """,
+ new_body,
+ re.VERBOSE,
+ )
for m in matches:
if "\\" in str(m):
# Do not introduce backslashes in interpolated expressions
def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
"""Make existing optional parentheses invisible or create new ones.
- `parens_after` is a set of string leaf values immeditely after which parens
+ `parens_after` is a set of string leaf values immediately after which parens
should be put.
Standardizes on visible parentheses for single-element tuples, and keeps
check_lpar = False
for index, child in enumerate(list(node.children)):
+ # Add parentheses around long tuple unpacking in assignments.
+ if (
+ index == 0
+ and isinstance(child, Node)
+ and child.type == syms.testlist_star_expr
+ ):
+ check_lpar = True
+
if check_lpar:
+ if is_walrus_assignment(child):
+ continue
if child.type == syms.atom:
- if maybe_make_parens_invisible_in_atom(child):
+ # Determines if the underlying atom should be surrounded with
+ # invisible params - also makes parens invisible recursively
+ # within the atom and removes repeated invisible parens within
+ # the atom
+ should_surround_with_parens = maybe_make_parens_invisible_in_atom(
+ child, parent=node
+ )
+
+ if should_surround_with_parens:
lpar = Leaf(token.LPAR, "")
rpar = Leaf(token.RPAR, "")
index = child.remove() or 0
lpar = Leaf(token.LPAR, "")
rpar = Leaf(token.RPAR, "")
index = child.remove() or 0
- node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
+ prefix = child.prefix
+ child.prefix = ""
+ new_child = Node(syms.atom, [lpar, child, rpar])
+ new_child.prefix = prefix
+ node.insert_child(index, new_child)
check_lpar = isinstance(child, Leaf) and child.value in parens_after
container = container.next_sibling
-def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
+def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> 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.
node.type != syms.atom
or is_empty_tuple(node)
or is_one_tuple(node)
- or is_yield(node)
+ or (is_yield(node) and parent.type != syms.expr_stmt)
or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
):
return False
first = node.children[0]
last = node.children[-1]
if first.type == token.LPAR and last.type == token.RPAR:
+ middle = node.children[1]
# make parentheses invisible
first.value = "" # type: ignore
last.value = "" # type: ignore
- if len(node.children) > 1:
- maybe_make_parens_invisible_in_atom(node.children[1])
+ maybe_make_parens_invisible_in_atom(middle, parent=parent)
+
+ if is_atom_with_invisible_parens(middle):
+ # Strip the invisible parens from `middle` by replacing
+ # it with the child in-between the invisible parens
+ middle.replace(middle.children[1])
+
return False
return True
+def is_atom_with_invisible_parens(node: LN) -> bool:
+ """Given a `LN`, determines whether it's an atom `node` with invisible
+ parens. Useful in dedupe-ing and normalizing parens.
+ """
+ if isinstance(node, Leaf) or node.type != syms.atom:
+ return False
+
+ first, last = node.children[0], node.children[-1]
+ return (
+ isinstance(first, Leaf)
+ and first.type == token.LPAR
+ and first.value == ""
+ and isinstance(last, Leaf)
+ and last.type == token.RPAR
+ and last.value == ""
+ )
+
+
def is_empty_tuple(node: LN) -> bool:
"""Return True if `node` holds an empty tuple."""
return (
)
+def unwrap_singleton_parenthesis(node: LN) -> Optional[LN]:
+ """Returns `wrapped` if `node` is of the shape ( wrapped ).
+
+ Parenthesis can be optional. Returns None otherwise"""
+ if len(node.children) != 3:
+ return None
+ lpar, wrapped, rpar = node.children
+ if not (lpar.type == token.LPAR and rpar.type == token.RPAR):
+ return None
+
+ return wrapped
+
+
def is_one_tuple(node: LN) -> bool:
"""Return True if `node` holds a tuple with one element, with or without parens."""
if node.type == syms.atom:
- if len(node.children) != 3:
- return False
-
- lpar, gexp, rpar = node.children
- if not (
- lpar.type == token.LPAR
- and gexp.type == syms.testlist_gexp
- and rpar.type == token.RPAR
- ):
+ gexp = unwrap_singleton_parenthesis(node)
+ if gexp is None or gexp.type != syms.testlist_gexp:
return False
return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA
)
+def is_walrus_assignment(node: LN) -> bool:
+ """Return True iff `node` is of the shape ( test := test )"""
+ inner = unwrap_singleton_parenthesis(node)
+ return inner is not None and inner.type == syms.namedexpr_test
+
+
def is_yield(node: LN) -> bool:
"""Return True if `node` holds a `yield` or `yield from` expression."""
if node.type == syms.yield_expr:
extended iterable unpacking (PEP 3132) and additional unpacking
generalizations (PEP 448).
"""
- if leaf.type not in STARS or not leaf.parent:
+ if leaf.type not in VARARGS_SPECIALS or not leaf.parent:
return False
p = leaf.parent
)
-def max_delimiter_priority_in_atom(node: LN) -> int:
+def max_delimiter_priority_in_atom(node: LN) -> Priority:
"""Return maximum delimiter priority inside `node`.
This is specific to atoms with contents contained in a pair of parentheses.
"""Make sure parentheses are visible.
They could be invisible as part of some statements (see
- :func:`normalize_invible_parens` and :func:`visit_import_from`).
+ :func:`normalize_invisible_parens` and :func:`visit_import_from`).
"""
if leaf.type == token.LPAR:
leaf.value = "("
Currently looking for:
- f-strings;
- - underscores in numeric literals; and
- - trailing commas after * or ** in function signatures and calls.
+ - underscores in numeric literals;
+ - trailing commas after * or ** in function signatures and calls;
+ - positional only arguments in function signatures and lambdas;
"""
features: Set[Feature] = set()
for n in node.pre_order():
if "_" in n.value: # type: ignore
features.add(Feature.NUMERIC_UNDERSCORES)
+ elif n.type == token.SLASH:
+ if n.parent and n.parent.type in {syms.typedargslist, syms.arglist}:
+ features.add(Feature.POS_ONLY_ARGUMENTS)
+
+ elif n.type == token.COLONEQUAL:
+ features.add(Feature.ASSIGNMENT_EXPRESSIONS)
+
elif (
n.type in {syms.typedargslist, syms.arglist}
and n.children
and n.children[-1].type == token.COMMA
):
+ if n.type == syms.typedargslist:
+ feature = Feature.TRAILING_COMMA_IN_DEF
+ else:
+ feature = Feature.TRAILING_COMMA_IN_CALL
+
for ch in n.children:
if ch.type in STARS:
- features.add(Feature.TRAILING_COMMA)
+ features.add(feature)
if ch.type == syms.argument:
for argch in ch.children:
if argch.type in STARS:
- features.add(Feature.TRAILING_COMMA)
+ features.add(feature)
return features
elif child.type == syms.import_as_names:
yield from get_imports_from_children(child.children)
else:
- assert False, "Invalid syntax parsing imports"
+ raise AssertionError("Invalid syntax parsing imports")
for child in node.children:
if child.type != syms.simple_stmt:
return ", ".join(report) + "."
+def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]:
+ filename = "<unknown>"
+ if sys.version_info >= (3, 8):
+ # TODO: support Python 4+ ;)
+ for minor_version in range(sys.version_info[1], 4, -1):
+ try:
+ return ast.parse(src, filename, feature_version=(3, minor_version))
+ except SyntaxError:
+ continue
+ else:
+ for feature_version in (7, 6):
+ try:
+ return ast3.parse(src, filename, feature_version=feature_version)
+ except SyntaxError:
+ continue
+
+ return ast27.parse(src)
+
+
+def _fixup_ast_constants(
+ node: Union[ast.AST, ast3.AST, ast27.AST]
+) -> Union[ast.AST, ast3.AST, ast27.AST]:
+ """Map ast nodes deprecated in 3.8 to Constant."""
+ # casts are required until this is released:
+ # https://github.com/python/typeshed/pull/3142
+ if isinstance(node, (ast.Str, ast3.Str, ast27.Str, ast.Bytes, ast3.Bytes)):
+ return cast(ast.AST, ast.Constant(value=node.s))
+ elif isinstance(node, (ast.Num, ast3.Num, ast27.Num)):
+ return cast(ast.AST, ast.Constant(value=node.n))
+ elif isinstance(node, (ast.NameConstant, ast3.NameConstant)):
+ return cast(ast.AST, ast.Constant(value=node.value))
+ return node
+
+
def assert_equivalent(src: str, dst: str) -> None:
"""Raise AssertionError if `src` and `dst` aren't equivalent."""
- import ast
- import traceback
-
- def _v(node: ast.AST, depth: int = 0) -> Iterator[str]:
+ def _v(node: Union[ast.AST, ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]:
"""Simple visitor generating strings to compare ASTs by content."""
+
+ node = _fixup_ast_constants(node)
+
yield f"{' ' * depth}{node.__class__.__name__}("
for field in sorted(node._fields):
+ # TypeIgnore has only one field 'lineno' which breaks this comparison
+ type_ignore_classes = (ast3.TypeIgnore, ast27.TypeIgnore)
+ if sys.version_info >= (3, 8):
+ type_ignore_classes += (ast.TypeIgnore,)
+ if isinstance(node, type_ignore_classes):
+ break
+
try:
value = getattr(node, field)
except AttributeError:
# parentheses and they change the AST.
if (
field == "targets"
- and isinstance(node, ast.Delete)
- and isinstance(item, ast.Tuple)
+ and isinstance(node, (ast.Delete, ast3.Delete, ast27.Delete))
+ and isinstance(item, (ast.Tuple, ast3.Tuple, ast27.Tuple))
):
for item in item.elts:
yield from _v(item, depth + 2)
- elif isinstance(item, ast.AST):
+ elif isinstance(item, (ast.AST, ast3.AST, ast27.AST)):
yield from _v(item, depth + 2)
- elif isinstance(value, ast.AST):
+ elif isinstance(value, (ast.AST, ast3.AST, ast27.AST)):
yield from _v(value, depth + 2)
else:
yield f"{' ' * depth}) # /{node.__class__.__name__}"
try:
- src_ast = ast.parse(src)
+ src_ast = parse_ast(src)
except Exception as exc:
- major, minor = sys.version_info[:2]
raise AssertionError(
- f"cannot use --safe with this file; failed to parse source file "
- f"with Python {major}.{minor}'s builtin AST. Re-run with --fast "
- f"or stop using deprecated Python 2 syntax. AST error message: {exc}"
+ f"cannot use --safe with this file; failed to parse source file. "
+ f"AST error message: {exc}"
)
try:
- dst_ast = ast.parse(dst)
+ dst_ast = parse_ast(dst)
except Exception as exc:
log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
raise AssertionError(
f"INTERNAL ERROR: Black produced invalid code: {exc}. "
- f"Please report a bug on https://github.com/ambv/black/issues. "
+ f"Please report a bug on https://github.com/psf/black/issues. "
f"This invalid output might be helpful: {log}"
) from None
raise AssertionError(
f"INTERNAL ERROR: Black produced code that is not equivalent to "
f"the source. "
- f"Please report a bug on https://github.com/ambv/black/issues. "
+ f"Please report a bug on https://github.com/psf/black/issues. "
f"This diff might be helpful: {log}"
) from None
raise AssertionError(
f"INTERNAL ERROR: Black produced different code on the second pass "
f"of the formatter. "
- f"Please report a bug on https://github.com/ambv/black/issues. "
+ f"Please report a bug on https://github.com/psf/black/issues. "
f"This diff might be helpful: {log}"
) from None
def dump_to_file(*output: str) -> str:
"""Dump `output` to a temporary file. Return path to the file."""
- import tempfile
-
with tempfile.NamedTemporaryFile(
mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8"
) as f:
return f.name
+@contextmanager
+def nullcontext() -> Iterator[None]:
+ """Return context manager that does nothing.
+ Similar to `nullcontext` from python 3.7"""
+ yield
+
+
def diff(a: str, b: str, a_name: str, b_name: str) -> str:
"""Return a unified diff string between strings `a` and `b`."""
import difflib
task.cancel()
-def shutdown(loop: BaseEventLoop) -> None:
+def shutdown(loop: asyncio.AbstractEventLoop) -> None:
"""Cancel all pending tasks on `loop`, wait for them, and close the loop."""
try:
if sys.version_info[:2] >= (3, 7):
"""
if "\n" in regex:
regex = "(?x)" + regex
- return re.compile(regex)
+ compiled: Pattern[str] = re.compile(regex)
+ return compiled
def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]:
if "\n" in leaf.value:
return # Multiline strings, we can't continue.
- comment: Optional[Leaf]
for comment in line.comments_after(leaf):
length += len(comment.value)
projects / etc / vim.git / blobdiff