X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/1ec2470da7a040e0d7e3900c97305731826ab101..340d87b47cac2bc599ffda8e19d45fe827295434:/black.py?ds=sidebyside diff --git a/black.py b/black.py index c4b9132..7629d9f 100644 --- a/black.py +++ b/black.py @@ -2,7 +2,7 @@ import asyncio from asyncio.base_events import BaseEventLoop from concurrent.futures import Executor, ProcessPoolExecutor from datetime import datetime -from enum import Enum, Flag +from enum import Enum from functools import lru_cache, partial, wraps import io import itertools @@ -37,18 +37,20 @@ from typing import ( ) from appdirs import user_cache_dir -from attr import dataclass, Factory +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 import pygram, pytree from blib2to3.pgen2 import driver, token +from blib2to3.pgen2.grammar import Grammar from blib2to3.pgen2.parse import ParseError -__version__ = "18.9b0" +__version__ = "19.3b0" DEFAULT_LINE_LENGTH = 88 DEFAULT_EXCLUDES = ( r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|_build|buck-out|build|dist)/" @@ -67,7 +69,7 @@ LeafID = int 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] @@ -111,32 +113,97 @@ class Changed(Enum): YES = 2 -class FileMode(Flag): - AUTO_DETECT = 0 - PYTHON36 = 1 - PYI = 2 - NO_STRING_NORMALIZATION = 4 - NO_NUMERIC_UNDERSCORE_NORMALIZATION = 8 +class TargetVersion(Enum): + PY27 = 2 + PY33 = 3 + PY34 = 4 + PY35 = 5 + PY36 = 6 + PY37 = 7 + PY38 = 8 + + def is_python2(self) -> bool: + return self is TargetVersion.PY27 + + +PY36_VERSIONS = {TargetVersion.PY36, TargetVersion.PY37, TargetVersion.PY38} + + +class Feature(Enum): + # All string literals are unicode + UNICODE_LITERALS = 1 + F_STRINGS = 2 + 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 + + +VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { + 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, + Feature.F_STRINGS, + Feature.NUMERIC_UNDERSCORES, + 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_IN_CALL, + Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, + }, +} - @classmethod - def from_configuration( - cls, - *, - py36: bool, - pyi: bool, - skip_string_normalization: bool, - skip_numeric_underscore_normalization: bool, - ) -> "FileMode": - mode = cls.AUTO_DETECT - if py36: - mode |= cls.PYTHON36 - if pyi: - mode |= cls.PYI - if skip_string_normalization: - mode |= cls.NO_STRING_NORMALIZATION - if skip_numeric_underscore_normalization: - mode |= cls.NO_NUMERIC_UNDERSCORE_NORMALIZATION - return mode + +@dataclass +class FileMode: + target_versions: Set[TargetVersion] = Factory(set) + line_length: int = DEFAULT_LINE_LENGTH + string_normalization: bool = True + is_pyi: bool = False + + def get_cache_key(self) -> str: + if self.target_versions: + version_str = ",".join( + str(version.value) + for version in sorted(self.target_versions, key=lambda v: v.value) + ) + else: + version_str = "-" + parts = [ + version_str, + str(self.line_length), + str(int(self.string_normalization)), + str(int(self.is_pyi)), + ] + return ".".join(parts) + + +def supports_feature(target_versions: Set[TargetVersion], feature: Feature) -> bool: + return all(feature in VERSION_TO_FEATURES[version] for version in target_versions) def read_pyproject_toml( @@ -176,6 +243,7 @@ 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", @@ -184,13 +252,25 @@ def read_pyproject_toml( help="How many characters per line to allow.", show_default=True, ) +@click.option( + "-t", + "--target-version", + type=click.Choice([v.name.lower() for v in TargetVersion]), + callback=lambda c, p, v: [TargetVersion[val.upper()] for val in v], + multiple=True, + help=( + "Python versions that should be supported by Black's output. [default: " + "per-file auto-detection]" + ), +) @click.option( "--py36", is_flag=True, help=( "Allow using Python 3.6-only syntax on all input files. This will put " "trailing commas in function signatures and calls also after *args and " - "**kwargs. [default: per-file auto-detection]" + "**kwargs. Deprecated; use --target-version instead. " + "[default: per-file auto-detection]" ), ) @click.option( @@ -207,12 +287,6 @@ def read_pyproject_toml( is_flag=True, help="Don't normalize string quotes or prefixes.", ) -@click.option( - "-N", - "--skip-numeric-underscore-normalization", - is_flag=True, - help="Don't normalize underscores in numeric literals.", -) @click.option( "--check", is_flag=True, @@ -296,14 +370,15 @@ def read_pyproject_toml( @click.pass_context def main( ctx: click.Context, + code: Optional[str], line_length: int, + target_version: List[TargetVersion], check: bool, diff: bool, fast: bool, pyi: bool, py36: bool, skip_string_normalization: bool, - skip_numeric_underscore_normalization: bool, quiet: bool, verbose: bool, include: str, @@ -313,14 +388,32 @@ def main( ) -> None: """The uncompromising code formatter.""" write_back = WriteBack.from_configuration(check=check, diff=diff) - mode = FileMode.from_configuration( - py36=py36, - pyi=pyi, - skip_string_normalization=skip_string_normalization, - skip_numeric_underscore_normalization=skip_numeric_underscore_normalization, + if target_version: + if py36: + err(f"Cannot use both --target-version and --py36") + ctx.exit(2) + else: + versions = set(target_version) + elif py36: + err( + "--py36 is deprecated and will be removed in a future version. " + "Use --target-version py36 instead." + ) + versions = PY36_VERSIONS + else: + # We'll autodetect later. + versions = set() + mode = FileMode( + target_versions=versions, + line_length=line_length, + is_pyi=pyi, + string_normalization=not skip_string_normalization, ) 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: @@ -353,44 +446,24 @@ def main( if len(sources) == 1: reformat_one( src=sources.pop(), - line_length=line_length, fast=fast, write_back=write_back, mode=mode, report=report, ) else: - loop = asyncio.get_event_loop() - executor = ProcessPoolExecutor(max_workers=os.cpu_count()) - try: - loop.run_until_complete( - schedule_formatting( - sources=sources, - line_length=line_length, - 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 reformat_one( - src: Path, - line_length: int, - fast: bool, - write_back: WriteBack, - mode: FileMode, - report: "Report", + src: Path, fast: bool, write_back: WriteBack, mode: FileMode, report: "Report" ) -> None: """Reformat a single file under `src` without spawning child processes. @@ -401,37 +474,60 @@ def reformat_one( try: changed = Changed.NO if not src.is_file() and str(src) == "-": - if format_stdin_to_stdout( - line_length=line_length, fast=fast, write_back=write_back, mode=mode - ): + if format_stdin_to_stdout(fast=fast, write_back=write_back, mode=mode): changed = Changed.YES else: cache: Cache = {} if write_back != WriteBack.DIFF: - cache = read_cache(line_length, mode) + cache = read_cache(mode) res_src = src.resolve() if res_src in cache and cache[res_src] == get_cache_info(res_src): changed = Changed.CACHED if changed is not Changed.CACHED and format_file_in_place( - src, - line_length=line_length, - fast=fast, - write_back=write_back, - mode=mode, + src, fast=fast, write_back=write_back, mode=mode ): changed = Changed.YES if (write_back is WriteBack.YES and changed is not Changed.CACHED) or ( write_back is WriteBack.CHECK and changed is Changed.NO ): - write_cache(cache, [src], line_length, mode) + write_cache(cache, [src], mode) report.done(src, changed) except Exception as exc: 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) + + async def schedule_formatting( sources: Set[Path], - line_length: int, fast: bool, write_back: WriteBack, mode: FileMode, @@ -448,7 +544,7 @@ async def schedule_formatting( """ cache: Cache = {} if write_back != WriteBack.DIFF: - cache = read_cache(line_length, mode) + cache = read_cache(mode) sources, cached = filter_cached(cache, sources) for src in sorted(cached): report.done(src, Changed.CACHED) @@ -464,19 +560,14 @@ async def schedule_formatting( manager = Manager() lock = manager.Lock() tasks = { - loop.run_in_executor( - executor, - format_file_in_place, - src, - line_length, - fast, - write_back, - mode, - 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) @@ -503,15 +594,14 @@ async def schedule_formatting( if cancelled: await asyncio.gather(*cancelled, loop=loop, return_exceptions=True) if sources_to_cache: - write_cache(cache, sources_to_cache, line_length, mode) + write_cache(cache, sources_to_cache, mode) def format_file_in_place( src: Path, - line_length: int, fast: bool, + mode: FileMode, write_back: WriteBack = WriteBack.NO, - mode: FileMode = FileMode.AUTO_DETECT, lock: Any = None, # multiprocessing.Manager().Lock() is some crazy proxy ) -> bool: """Format file under `src` path. Return True if changed. @@ -521,15 +611,13 @@ def format_file_in_place( `line_length` and `fast` options are passed to :func:`format_file_contents`. """ if src.suffix == ".pyi": - mode |= FileMode.PYI + mode = evolve(mode, is_pyi=True) then = datetime.utcfromtimestamp(src.stat().st_mtime) with open(src, "rb") as buf: src_contents, encoding, newline = decode_bytes(buf.read()) try: - dst_contents = format_file_contents( - src_contents, line_length=line_length, fast=fast, mode=mode - ) + dst_contents = format_file_contents(src_contents, fast=fast, mode=mode) except NothingChanged: return False @@ -559,23 +647,19 @@ def format_file_in_place( def format_stdin_to_stdout( - line_length: int, - fast: bool, - write_back: WriteBack = WriteBack.NO, - mode: FileMode = FileMode.AUTO_DETECT, + fast: bool, *, write_back: WriteBack = WriteBack.NO, mode: FileMode ) -> bool: """Format file on stdin. Return True if changed. If `write_back` is YES, write reformatted code back to stdout. If it is DIFF, - write a diff to stdout. - `line_length`, `fast`, `is_pyi`, and `force_py36` arguments are passed to + write a diff to stdout. The `mode` argument is passed to :func:`format_file_contents`. """ then = datetime.utcnow() src, encoding, newline = decode_bytes(sys.stdin.buffer.read()) dst = src try: - dst = format_file_contents(src, line_length=line_length, fast=fast, mode=mode) + dst = format_file_contents(src, fast=fast, mode=mode) return True except NothingChanged: @@ -596,11 +680,7 @@ def format_stdin_to_stdout( def format_file_contents( - src_contents: str, - *, - line_length: int, - fast: bool, - mode: FileMode = FileMode.AUTO_DETECT, + src_contents: str, *, fast: bool, mode: FileMode ) -> FileContent: """Reformat contents a file and return new contents. @@ -611,49 +691,54 @@ def format_file_contents( if src_contents.strip() == "": raise NothingChanged - dst_contents = format_str(src_contents, line_length=line_length, mode=mode) + dst_contents = format_str(src_contents, mode=mode) if src_contents == dst_contents: raise NothingChanged if not fast: assert_equivalent(src_contents, dst_contents) - assert_stable(src_contents, dst_contents, line_length=line_length, mode=mode) + assert_stable(src_contents, dst_contents, mode=mode) return dst_contents -def format_str( - src_contents: str, line_length: int, *, mode: FileMode = FileMode.AUTO_DETECT -) -> FileContent: +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. """ - src_node = lib2to3_parse(src_contents.lstrip()) - dst_contents = "" + src_node = lib2to3_parse(src_contents.lstrip(), mode.target_versions) + dst_contents = [] future_imports = get_future_imports(src_node) - is_pyi = bool(mode & FileMode.PYI) - py36 = bool(mode & FileMode.PYTHON36) or is_python36(src_node) - normalize_strings = not bool(mode & FileMode.NO_STRING_NORMALIZATION) + if mode.target_versions: + versions = mode.target_versions + else: + versions = detect_target_versions(src_node) normalize_fmt_off(src_node) lines = LineGenerator( - remove_u_prefix=py36 or "unicode_literals" in future_imports, - is_pyi=is_pyi, - normalize_strings=normalize_strings, - allow_underscores=py36 - and not bool(mode & FileMode.NO_NUMERIC_UNDERSCORE_NORMALIZATION), + remove_u_prefix="unicode_literals" in future_imports + or supports_feature(versions, Feature.UNICODE_LITERALS), + is_pyi=mode.is_pyi, + normalize_strings=mode.string_normalization, ) - elt = EmptyLineTracker(is_pyi=is_pyi) + 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) - for line in split_line(current_line, line_length=line_length, py36=py36): - dst_contents += str(line) - return dst_contents + dst_contents.append(str(empty_line)) + for line in split_line( + current_line, line_length=mode.line_length, features=split_line_features + ): + dst_contents.append(str(line)) + return "".join(dst_contents) def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]: @@ -673,18 +758,50 @@ 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: + # 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: + # 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) -> Node: +def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node: """Given a string with source, return the lib2to3 Node.""" if src_txt[-1:] != "\n": src_txt += "\n" - for grammar in GRAMMARS: + + for grammar in get_grammars(set(target_versions)): drv = driver.Driver(grammar, pytree.convert) try: result = drv.parse_string(src_txt, True) @@ -1026,9 +1143,7 @@ class Line: depth: int = 0 leaves: List[Leaf] = Factory(list) - # The LeafID keys of comments must remain ordered by the corresponding leaf's index - # in leaves - comments: Dict[LeafID, List[Leaf]] = Factory(dict) + comments: Dict[LeafID, List[Leaf]] = Factory(dict) # keys ordered like `leaves` bracket_tracker: BracketTracker = Factory(BracketTracker) inside_brackets: bool = False should_explode: bool = False @@ -1159,6 +1274,29 @@ class Line: if leaf.type == STANDALONE_COMMENT: if leaf.bracket_depth <= depth_limit: return True + return False + + def contains_inner_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. + last_leaf = self.leaves[-2] + ignored_ids.add(id(last_leaf)) + except IndexError: + return 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 return False @@ -1240,13 +1378,8 @@ class Line: comment.prefix = "" return False - else: - leaf_id = id(self.leaves[-1]) - if leaf_id not in self.comments: - self.comments[leaf_id] = [comment] - else: - self.comments[leaf_id].append(comment) - return True + self.comments.setdefault(id(self.leaves[-1]), []).append(comment) + return True def comments_after(self, leaf: Leaf) -> List[Leaf]: """Generate comments that should appear directly after `leaf`.""" @@ -1254,17 +1387,11 @@ class Line: def remove_trailing_comma(self) -> None: """Remove the trailing comma and moves the comments attached to it.""" - # Remember, the LeafID keys of self.comments are ordered by the - # corresponding leaf's index in self.leaves - # If id(self.leaves[-2]) is in self.comments, the order doesn't change. - # Otherwise, we insert it into self.comments, and it becomes the last entry. - # However, since we delete id(self.leaves[-1]) from self.comments, the invariant - # is maintained - self.comments.setdefault(id(self.leaves[-2]), []).extend( - self.comments.get(id(self.leaves[-1]), []) + trailing_comma = self.leaves.pop() + trailing_comma_comments = self.comments.pop(id(trailing_comma), []) + self.comments.setdefault(id(self.leaves[-1]), []).extend( + trailing_comma_comments ) - self.comments.pop(id(self.leaves[-1]), None) - self.leaves.pop() def is_complex_subscript(self, leaf: Leaf) -> bool: """Return True iff `leaf` is part of a slice with non-trivial exprs.""" @@ -1427,7 +1554,6 @@ class LineGenerator(Visitor[Line]): normalize_strings: bool = True current_line: Line = Factory(Line) remove_u_prefix: bool = False - allow_underscores: bool = False def line(self, indent: int = 0) -> Iterator[Line]: """Generate a line. @@ -1470,11 +1596,31 @@ class LineGenerator(Visitor[Line]): normalize_string_prefix(node, remove_u_prefix=self.remove_u_prefix) normalize_string_quotes(node) if node.type == token.NUMBER: - normalize_numeric_literal(node, self.allow_underscores) + normalize_numeric_literal(node) if node.type not in WHITESPACE: 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_INDENT(self, node: Node) -> Iterator[Line]: """Increase indentation level, maybe yield a line.""" # In blib2to3 INDENT never holds comments. @@ -1594,6 +1740,7 @@ class LineGenerator(Visitor[Line]): self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS) self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"}) self.visit_import_from = partial(v, keywords=Ø, parens={"import"}) + self.visit_del_stmt = partial(v, keywords=Ø, parens={"del"}) self.visit_async_funcdef = self.visit_async_stmt self.visit_decorated = self.visit_decorators @@ -2051,15 +2198,21 @@ 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 @@ -2093,7 +2246,10 @@ def make_comment(content: str) -> str: def split_line( - line: Line, line_length: int, inner: bool = False, py36: bool = False + line: Line, + line_length: int, + inner: bool = False, + features: Collection[Feature] = (), ) -> Iterator[Line]: """Split a `line` into potentially many lines. @@ -2102,8 +2258,7 @@ def split_line( current `line`, possibly transitively. This means we can fallback to splitting by delimiters if the LHS/RHS don't yield any results. - If `py36` is True, splitting may generate syntax that is only compatible - with Python 3.6 and later. + `features` are syntactical features that may be used in the output. """ if line.is_comment: yield line @@ -2111,16 +2266,8 @@ def split_line( line_str = str(line).strip("\n") - # we don't want to split special comments like type annotations - # https://github.com/python/typing/issues/186 - has_special_comment = False - for leaf in line.leaves: - for comment in line.comments_after(leaf): - if leaf.type == token.COMMA and is_special_comment(comment): - has_special_comment = True - if ( - not has_special_comment + not line.contains_inner_type_comments() and not line.should_explode and is_line_short_enough(line, line_length=line_length, line_str=line_str) ): @@ -2132,9 +2279,9 @@ def split_line( split_funcs = [left_hand_split] else: - def rhs(line: Line, py36: 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, py36, 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 @@ -2142,7 +2289,7 @@ def split_line( # 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, py36) + yield from right_hand_split(line, line_length, features=features) if line.inside_brackets: split_funcs = [delimiter_split, standalone_comment_split, rhs] @@ -2154,12 +2301,14 @@ def split_line( # split altogether. result: List[Line] = [] try: - for l in split_func(line, py36): + 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, py36=py36) + split_line( + l, line_length=line_length, inner=True, features=features + ) ) except CannotSplit: continue @@ -2172,7 +2321,7 @@ def split_line( yield line -def left_hand_split(line: Line, py36: 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. @@ -2209,7 +2358,10 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: def right_hand_split( - line: Line, line_length: int, py36: bool = False, omit: Collection[LeafID] = () + line: Line, + line_length: int, + features: Collection[Feature] = (), + omit: Collection[LeafID] = (), ) -> Iterator[Line]: """Split line into many lines, starting with the last matching bracket pair. @@ -2267,7 +2419,7 @@ def right_hand_split( ): omit = {id(closing_bracket), *omit} try: - yield from right_hand_split(line, line_length, py36=py36, omit=omit) + yield from right_hand_split(line, line_length, features=features, omit=omit) return except CannotSplit: @@ -2335,10 +2487,17 @@ 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. + # Ensure a trailing comma for imports, but be careful not to add one after + # any comments. if original.is_import: - if leaves[-1].type != token.COMMA: - leaves.append(Leaf(token.COMMA, ",")) + 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) @@ -2356,8 +2515,8 @@ def dont_increase_indentation(split_func: SplitFunc) -> SplitFunc: """ @wraps(split_func) - def split_wrapper(line: Line, py36: bool = False) -> Iterator[Line]: - for l in split_func(line, py36): + 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 @@ -2365,11 +2524,11 @@ def dont_increase_indentation(split_func: SplitFunc) -> SplitFunc: @dont_increase_indentation -def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: +def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]: """Split according to delimiters of the highest priority. - If `py36` 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] @@ -2408,10 +2567,16 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: 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 py36 + 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 @@ -2429,7 +2594,9 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: @dont_increase_indentation -def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]: +def standalone_comment_split( + line: Line, features: Collection[Feature] = () +) -> Iterator[Line]: """Split standalone comments from the rest of the line.""" if not line.contains_standalone_comments(0): raise CannotSplit("Line does not have any standalone comments") @@ -2471,14 +2638,12 @@ def is_import(leaf: Leaf) -> bool: ) -def is_special_comment(leaf: Leaf) -> bool: +def is_type_comment(leaf: Leaf) -> 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 bool( - (t == token.COMMENT or t == STANDALONE_COMMENT) and (v.startswith("# type:")) - ) + return t in {token.COMMENT, t == STANDALONE_COMMENT} and v.startswith("# type:") def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: @@ -2563,7 +2728,15 @@ 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 @@ -2582,11 +2755,11 @@ def normalize_string_quotes(leaf: Leaf) -> None: leaf.value = f"{prefix}{new_quote}{new_body}{new_quote}" -def normalize_numeric_literal(leaf: Leaf, allow_underscores: bool) -> None: +def normalize_numeric_literal(leaf: Leaf) -> None: """Normalizes numeric (float, int, and complex) literals. All letters used in the representation are normalized to lowercase (except - in Python 2 long literals), and long number literals are split using underscores. + in Python 2 long literals). """ text = leaf.value.lower() if text.startswith(("0o", "0b")): @@ -2604,8 +2777,7 @@ def normalize_numeric_literal(leaf: Leaf, allow_underscores: bool) -> None: sign = "-" elif after.startswith("+"): after = after[1:] - before = format_float_or_int_string(before, allow_underscores) - after = format_int_string(after, allow_underscores) + before = format_float_or_int_string(before) text = f"{before}e{sign}{after}" elif text.endswith(("j", "l")): number = text[:-1] @@ -2613,50 +2785,19 @@ def normalize_numeric_literal(leaf: Leaf, allow_underscores: bool) -> None: # Capitalize in "2L" because "l" looks too similar to "1". if suffix == "l": suffix = "L" - text = f"{format_float_or_int_string(number, allow_underscores)}{suffix}" + text = f"{format_float_or_int_string(number)}{suffix}" else: - text = format_float_or_int_string(text, allow_underscores) + text = format_float_or_int_string(text) leaf.value = text -def format_float_or_int_string(text: str, allow_underscores: bool) -> str: +def format_float_or_int_string(text: str) -> str: """Formats a float string like "1.0".""" if "." not in text: - return format_int_string(text, allow_underscores) - - before, after = text.split(".") - before = format_int_string(before, allow_underscores) if before else "0" - if after: - after = format_int_string(after, allow_underscores, count_from_end=False) - else: - after = "0" - return f"{before}.{after}" - - -def format_int_string( - text: str, allow_underscores: bool, count_from_end: bool = True -) -> str: - """Normalizes underscores in a string to e.g. 1_000_000. - - Input must be a string of digits and optional underscores. - If count_from_end is False, we add underscores after groups of three digits - counting from the beginning instead of the end of the strings. This is used - for the fractional part of float literals. - """ - if not allow_underscores: - return text - - text = text.replace("_", "") - if len(text) <= 5: - # No underscores for numbers <= 5 digits long. return text - if count_from_end: - # Avoid removing leading zeros, which are important if we're formatting - # part of a number like "0.001". - return format(int("1" + text), "3_")[1:].lstrip("_") - else: - return "_".join(text[i : i + 3] for i in range(0, len(text), 3)) + before, after = text.split(".") + return f"{before or 0}.{after or 0}" def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: @@ -2675,9 +2816,17 @@ 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 child.type == syms.atom: - if maybe_make_parens_invisible_in_atom(child): + if maybe_make_parens_invisible_in_atom(child, parent=node): lpar = Leaf(token.LPAR, "") rpar = Leaf(token.RPAR, "") index = child.remove() or 0 @@ -2706,7 +2855,11 @@ 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 @@ -2788,7 +2941,7 @@ 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. Returns whether the node should itself be wrapped in invisible parentheses. @@ -2798,7 +2951,7 @@ 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 @@ -2810,7 +2963,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: 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(node.children[1], parent=parent) return False return True @@ -2988,39 +3141,53 @@ def should_explode(line: Line, opening_bracket: Leaf) -> bool: return max_priority == COMMA_PRIORITY -def is_python36(node: Node) -> bool: - """Return True if the current file is using Python 3.6+ features. +def get_features_used(node: Node) -> Set[Feature]: + """Return a set of (relatively) new Python features used in this file. Currently looking for: - f-strings; - underscores in numeric literals; and - trailing commas after * or ** in function signatures and calls. """ + features: Set[Feature] = set() for n in node.pre_order(): if n.type == token.STRING: value_head = n.value[:2] # type: ignore if value_head in {'f"', 'F"', "f'", "F'", "rf", "fr", "RF", "FR"}: - return True + features.add(Feature.F_STRINGS) elif n.type == token.NUMBER: if "_" in n.value: # type: ignore - return True + features.add(Feature.NUMERIC_UNDERSCORES) 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: - return True + features.add(feature) if ch.type == syms.argument: for argch in ch.children: if argch.type in STARS: - return True + features.add(feature) - return False + return features + + +def detect_target_versions(node: Node) -> Set[TargetVersion]: + """Detect the version to target based on the nodes used.""" + features = get_features_used(node) + return { + version for version in TargetVersion if features <= VERSION_TO_FEATURES[version] + } def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[LeafID]]: @@ -3090,7 +3257,7 @@ 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: @@ -3274,17 +3441,34 @@ class Report: return ", ".join(report) + "." +def parse_ast(src: str) -> Union[ast3.AST, ast27.AST]: + for feature_version in (7, 6): + try: + return ast3.parse(src, feature_version=feature_version) + except SyntaxError: + continue + + return ast27.parse(src) + + 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[ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]: """Simple visitor generating strings to compare ASTs by content.""" yield f"{' ' * depth}{node.__class__.__name__}(" for field in sorted(node._fields): + # TypeIgnore has only one field 'lineno' which breaks this comparison + if isinstance(node, (ast3.TypeIgnore, ast27.TypeIgnore)): + break + + # Ignore str kind which is case sensitive / and ignores unicode_literals + if isinstance(node, (ast3.Str, ast27.Str, ast3.Bytes)) and field == "kind": + continue + try: value = getattr(node, field) except AttributeError: @@ -3294,10 +3478,19 @@ def assert_equivalent(src: str, dst: str) -> None: if isinstance(value, list): for item in value: - if isinstance(item, ast.AST): + # Ignore nested tuples within del statements, because we may insert + # parentheses and they change the AST. + if ( + field == "targets" + and isinstance(node, (ast3.Delete, ast27.Delete)) + and isinstance(item, (ast3.Tuple, ast27.Tuple)) + ): + for item in item.elts: + yield from _v(item, depth + 2) + elif isinstance(item, (ast3.AST, ast27.AST)): yield from _v(item, depth + 2) - elif isinstance(value, ast.AST): + elif isinstance(value, (ast3.AST, ast27.AST)): yield from _v(value, depth + 2) else: @@ -3306,22 +3499,20 @@ 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/python/black/issues. " f"This invalid output might be helpful: {log}" ) from None @@ -3332,16 +3523,14 @@ 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/python/black/issues. " f"This diff might be helpful: {log}" ) from None -def assert_stable( - src: str, dst: str, line_length: int, mode: FileMode = FileMode.AUTO_DETECT -) -> None: +def assert_stable(src: str, dst: str, mode: FileMode) -> None: """Raise AssertionError if `dst` reformats differently the second time.""" - newdst = format_str(dst, line_length=line_length, mode=mode) + newdst = format_str(dst, mode=mode) if dst != newdst: log = dump_to_file( diff(src, dst, "source", "first pass"), @@ -3350,7 +3539,7 @@ def assert_stable( 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/python/black/issues. " f"This diff might be helpful: {log}" ) from None @@ -3390,8 +3579,12 @@ def cancel(tasks: Iterable[asyncio.Task]) -> None: def shutdown(loop: BaseEventLoop) -> None: """Cancel all pending tasks on `loop`, wait for them, and close the loop.""" try: + if sys.version_info[:2] >= (3, 7): + all_tasks = asyncio.all_tasks + else: + all_tasks = asyncio.Task.all_tasks # This part is borrowed from asyncio/runners.py in Python 3.7b2. - to_cancel = [task for task in asyncio.Task.all_tasks(loop) if not task.done()] + to_cancel = [task for task in all_tasks(loop) if not task.done()] if not to_cancel: return @@ -3598,16 +3791,16 @@ def can_omit_invisible_parens(line: Line, line_length: int) -> bool: return False -def get_cache_file(line_length: int, mode: FileMode) -> Path: - return CACHE_DIR / f"cache.{line_length}.{mode.value}.pickle" +def get_cache_file(mode: FileMode) -> Path: + return CACHE_DIR / f"cache.{mode.get_cache_key()}.pickle" -def read_cache(line_length: int, mode: FileMode) -> Cache: +def read_cache(mode: FileMode) -> Cache: """Read the cache if it exists and is well formed. If it is not well formed, the call to write_cache later should resolve the issue. """ - cache_file = get_cache_file(line_length, mode) + cache_file = get_cache_file(mode) if not cache_file.exists(): return {} @@ -3642,11 +3835,9 @@ def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set return todo, done -def write_cache( - cache: Cache, sources: Iterable[Path], line_length: int, mode: FileMode -) -> None: +def write_cache(cache: Cache, sources: Iterable[Path], mode: FileMode) -> None: """Update the cache file.""" - cache_file = get_cache_file(line_length, mode) + cache_file = get_cache_file(mode) try: CACHE_DIR.mkdir(parents=True, exist_ok=True) new_cache = {**cache, **{src.resolve(): get_cache_info(src) for src in sources}}