X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/cea13f498418784e22f8fbd78db3f9240a2bad11..14b28c89c22659e1f935bc0ac22ee03d90bcc290:/black.py diff --git a/black.py b/black.py index 0ea9b9a..151c1a6 100644 --- a/black.py +++ b/black.py @@ -1,6 +1,7 @@ +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 @@ -11,11 +12,12 @@ from multiprocessing import Manager, freeze_support 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, @@ -40,6 +42,7 @@ from appdirs import user_cache_dir 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 @@ -48,12 +51,10 @@ from blib2to3.pgen2 import driver, token 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)/" -) +DEFAULT_EXCLUDES = r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|\.svn|_build|buck-out|build|dist)/" # noqa: B950 DEFAULT_INCLUDES = r"\.pyi?$" CACHE_DIR = Path(user_cache_dir("black", version=__version__)) @@ -135,19 +136,30 @@ class Feature(Enum): NUMERIC_UNDERSCORES = 3 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_IN_CALL}, + 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_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_IDENTIFIERS, }, TargetVersion.PY37: { Feature.UNICODE_LITERALS, @@ -155,6 +167,7 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { Feature.NUMERIC_UNDERSCORES, Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, }, TargetVersion.PY38: { Feature.UNICODE_LITERALS, @@ -162,6 +175,9 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { Feature.NUMERIC_UNDERSCORES, Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, + Feature.ASSIGNMENT_EXPRESSIONS, + Feature.POS_ONLY_ARGUMENTS, }, } @@ -231,6 +247,7 @@ def read_pyproject_toml( @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", @@ -323,7 +340,7 @@ def read_pyproject_toml( "--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." ), ) @@ -357,6 +374,7 @@ def read_pyproject_toml( @click.pass_context def main( ctx: click.Context, + code: Optional[str], line_length: int, target_version: List[TargetVersion], check: bool, @@ -397,6 +415,9 @@ def main( ) 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: @@ -410,6 +431,7 @@ def main( 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(): @@ -423,7 +445,7 @@ def main( 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: @@ -435,36 +457,32 @@ def main( 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: @@ -492,20 +510,51 @@ def reformat_one( 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 = {} @@ -526,12 +575,14 @@ async def schedule_formatting( 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) @@ -572,7 +623,7 @@ def format_file_in_place( 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) @@ -593,9 +644,8 @@ def format_file_in_place( 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, @@ -604,9 +654,7 @@ def format_file_in_place( ) f.write(diff_contents) f.detach() - finally: - if lock: - lock.release() + return True @@ -650,7 +698,7 @@ def format_file_contents( 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 @@ -668,10 +716,11 @@ def format_file_contents( 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 @@ -694,15 +743,15 @@ def format_str(src_contents: str, *, mode: FileMode) -> FileContent: } 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, 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]: @@ -726,16 +775,38 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]: if not target_versions: # 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 [pygram.python_grammar_no_print_statement, pygram.python_grammar] + return [ + # Python 2.7 with future print_function import + pygram.python_grammar_no_print_statement, + # Python 2.7 + pygram.python_grammar, + ] else: # Python 3-compatible code, so only try Python 3 grammar. - return [pygram.python_grammar_no_print_statement_no_exec_statement] + 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: @@ -875,6 +946,7 @@ MATH_OPERATORS = { token.DOUBLESTAR, } STARS = {token.STAR, token.DOUBLESTAR} +VARARGS_SPECIALS = STARS | {token.SLASH} VARARGS_PARENTS = { syms.arglist, syms.argument, # double star in arglist @@ -1002,7 +1074,7 @@ class BracketTracker: """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. @@ -1010,7 +1082,7 @@ class BracketTracker: """ 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. @@ -1166,6 +1238,61 @@ class Line: Leaf(token.DOT, ".") for _ in range(3) ] + @property + def is_collection_with_optional_trailing_comma(self) -> bool: + """Is this line a collection literal with a trailing comma that's optional? + + Note that the trailing comma in a 1-tuple is not optional. + """ + if not self.leaves or len(self.leaves) < 4: + return False + # Look for and address a trailing colon. + if self.leaves[-1].type == token.COLON: + closer = self.leaves[-2] + close_index = -2 + else: + closer = self.leaves[-1] + close_index = -1 + if closer.type not in CLOSING_BRACKETS or self.inside_brackets: + return False + if closer.type == token.RPAR: + # Tuples require an extra check, because if there's only + # one element in the tuple removing the comma unmakes the + # tuple. + # + # We also check for parens before looking for the trailing + # comma because in some cases (eg assigning a dict + # literal) the literal gets wrapped in temporary parens + # during parsing. This case is covered by the + # collections.py test data. + opener = closer.opening_bracket + for _open_index, leaf in enumerate(self.leaves): + if leaf is opener: + break + else: + # Couldn't find the matching opening paren, play it safe. + return False + commas = 0 + comma_depth = self.leaves[close_index - 1].bracket_depth + for leaf in self.leaves[_open_index + 1 : close_index]: + if leaf.bracket_depth == comma_depth and leaf.type == token.COMMA: + commas += 1 + if commas > 1: + # We haven't looked yet for the trailing comma because + # we might also have caught noop parens. + return self.leaves[close_index - 1].type == token.COMMA + elif commas == 1: + return False # it's either a one-tuple or didn't have a trailing comma + if self.leaves[close_index - 1].type in CLOSING_BRACKETS: + close_index -= 1 + closer = self.leaves[close_index] + if closer.type == token.RPAR: + # TODO: this is a gut feeling. Will we ever see this? + return False + if self.leaves[close_index - 1].type != token.COMMA: + return False + return True + @property def is_def(self) -> bool: """Is this a function definition? (Also returns True for async defs.)""" @@ -1218,27 +1345,67 @@ class 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. + + # Grab the first and last line numbers, skipping generated leaves + first_line = next((l.lineno for l in self.leaves if l.lineno != 0), 0) + last_line = next((l.lineno for l in reversed(self.leaves) if l.lineno != 0), 0) + + if first_line == last_line: + # We look at the last two leaves since a comma or an + # invisible paren could have been added at the end of the + # line. + for node in self.leaves[-2:]: + for comment in self.comments.get(id(node), []): + if is_type_comment(comment, " ignore"): + return True return False @@ -1251,57 +1418,39 @@ class Line: def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: """Remove trailing comma if there is one and it's safe.""" + if not (self.leaves and self.leaves[-1].type == token.COMMA): + return False + # We remove trailing commas only in the case of importing a + # single name from a module. if not ( self.leaves + and self.is_import + and len(self.leaves) > 4 and self.leaves[-1].type == token.COMMA and closing.type in CLOSING_BRACKETS + and self.leaves[-4].type == token.NAME + and ( + # regular `from foo import bar,` + self.leaves[-4].value == "import" + # `from foo import (bar as baz,) + or ( + len(self.leaves) > 6 + and self.leaves[-6].value == "import" + and self.leaves[-3].value == "as" + ) + # `from foo import bar as baz,` + or ( + len(self.leaves) > 5 + and self.leaves[-5].value == "import" + and self.leaves[-3].value == "as" + ) + ) + and closing.type == token.RPAR ): return False - if closing.type == token.RBRACE: - self.remove_trailing_comma() - return True - - if closing.type == token.RSQB: - comma = self.leaves[-1] - if comma.parent and comma.parent.type == syms.listmaker: - self.remove_trailing_comma() - return True - - # For parens let's check if it's safe to remove the comma. - # Imports are always safe. - if self.is_import: - self.remove_trailing_comma() - return True - - # Otherwise, if the trailing one is the only one, we might mistakenly - # change a tuple into a different type by removing the comma. - depth = closing.bracket_depth + 1 - commas = 0 - opening = closing.opening_bracket - for _opening_index, leaf in enumerate(self.leaves): - if leaf is opening: - break - - else: - return False - - for leaf in self.leaves[_opening_index + 1 :]: - if leaf is closing: - break - - 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: - commas += 1 - break - - if commas > 1: - self.remove_trailing_comma() - return True - - return False + self.remove_trailing_comma() + return True def append_comment(self, comment: Leaf) -> bool: """Add an inline or standalone comment to the line.""" @@ -1320,7 +1469,23 @@ class Line: 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]: @@ -1395,7 +1560,13 @@ class EmptyLineTracker: 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 @@ -1543,6 +1714,19 @@ class LineGenerator(Visitor[Line]): self.current_line.append(node) 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. @@ -1732,7 +1916,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa: C901 # 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 @@ -1822,7 +2006,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa: C901 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: @@ -1956,7 +2140,7 @@ def container_of(leaf: Leaf) -> LN: 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 @@ -1970,7 +2154,7 @@ def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int 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 @@ -2120,15 +2304,21 @@ def list_comments(prefix: str, *, is_endmarker: bool) -> List[ProtoComment]: 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 @@ -2183,9 +2373,13 @@ def split_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 not line.is_collection_with_optional_trailing_comma + and ( + is_line_short_enough(line, line_length=line_length, line_str=line_str) + or line.contains_unsplittable_type_ignore() + ) ): yield line return @@ -2403,10 +2597,21 @@ def bracket_split_build_line( 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) @@ -2547,12 +2752,12 @@ def is_import(leaf: Leaf) -> bool: ) -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, STANDALONE_COMMENT} and v.startswith("# type:" + suffix) def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: @@ -2637,7 +2842,15 @@ def normalize_string_quotes(leaf: Leaf) -> 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 @@ -2704,7 +2917,7 @@ def format_float_or_int_string(text: str) -> str: 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 @@ -2717,9 +2930,27 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: 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 @@ -2748,7 +2979,11 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: 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 @@ -2830,8 +3065,10 @@ def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]: 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. @@ -2840,7 +3077,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: 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 @@ -2848,16 +3085,40 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: 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 ( @@ -2868,18 +3129,24 @@ def is_empty_tuple(node: LN) -> bool: ) +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 @@ -2891,6 +3158,12 @@ def is_one_tuple(node: LN) -> bool: ) +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: @@ -2920,7 +3193,7 @@ def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool: 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 @@ -2970,7 +3243,7 @@ def is_stub_body(node: LN) -> bool: ) -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. @@ -3002,7 +3275,7 @@ def ensure_visible(leaf: Leaf) -> None: """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 = "(" @@ -3035,8 +3308,9 @@ def get_features_used(node: Node) -> Set[Feature]: 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(): @@ -3049,6 +3323,13 @@ def get_features_used(node: Node) -> Set[Feature]: 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 @@ -3146,7 +3427,7 @@ def get_future_imports(node: Node) -> Set[str]: 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: @@ -3330,17 +3611,58 @@ class Report: return ", ".join(report) + "." +def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]: + filename = "" + 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: @@ -3354,15 +3676,15 @@ def assert_equivalent(src: str, dst: str) -> None: # 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: @@ -3371,22 +3693,20 @@ def assert_equivalent(src: str, dst: str) -> None: 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 @@ -3397,7 +3717,7 @@ def assert_equivalent(src: str, dst: str) -> 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 @@ -3413,15 +3733,13 @@ def assert_stable(src: str, dst: str, mode: FileMode) -> 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: @@ -3432,6 +3750,13 @@ def dump_to_file(*output: str) -> str: 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 @@ -3450,7 +3775,7 @@ def cancel(tasks: Iterable[asyncio.Task]) -> None: 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): @@ -3492,7 +3817,8 @@ def re_compile_maybe_verbose(regex: str) -> Pattern[str]: """ 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]]: @@ -3519,7 +3845,6 @@ def enumerate_with_length( 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)