X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/c7da3482c7badf5c6d8c0c9a37495ab4529eaaad..a18c7bc09942951e93cbd142fb7384aa2af18951:/src/black/__init__.py diff --git a/src/black/__init__.py b/src/black/__init__.py index 77dda2b..49d088b 100644 --- a/src/black/__init__.py +++ b/src/black/__init__.py @@ -42,14 +42,26 @@ from typing import ( cast, TYPE_CHECKING, ) -from typing_extensions import Final from mypy_extensions import mypyc_attr from appdirs import user_cache_dir from dataclasses import dataclass, field, replace import click import toml -from typed_ast import ast3, ast27 + +try: + from typed_ast import ast3, ast27 +except ImportError: + if sys.version_info < (3, 8): + print( + "The typed_ast package is not installed.\n" + "You can install it with `python3 -m pip install typed-ast`.", + file=sys.stderr, + ) + sys.exit(1) + else: + ast3 = ast27 = ast + from pathspec import PathSpec # lib2to3 fork @@ -61,13 +73,19 @@ from blib2to3.pgen2.parse import ParseError from _black_version import version as __version__ +if sys.version_info < (3, 8): + from typing_extensions import Final +else: + from typing import Final + if TYPE_CHECKING: import colorama # noqa: F401 DEFAULT_LINE_LENGTH = 88 -DEFAULT_EXCLUDES = r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|\.svn|_build|buck-out|build|dist)/" # noqa: B950 +DEFAULT_EXCLUDES = r"/(\.direnv|\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|venv|\.svn|_build|buck-out|build|dist)/" # noqa: B950 DEFAULT_INCLUDES = r"\.pyi?$" CACHE_DIR = Path(user_cache_dir("black", version=__version__)) +STDIN_PLACEHOLDER = "__BLACK_STDIN_FILENAME__" STRING_PREFIX_CHARS: Final = "furbFURB" # All possible string prefix characters. @@ -88,7 +106,7 @@ Transformer = Callable[["Line", Collection["Feature"]], Iterator["Line"]] Timestamp = float FileSize = int CacheInfo = Tuple[Timestamp, FileSize] -Cache = Dict[Path, CacheInfo] +Cache = Dict[str, CacheInfo] out = partial(click.secho, bold=True, err=True) err = partial(click.secho, fg="red", err=True) @@ -112,6 +130,10 @@ class InvalidInput(ValueError): """Raised when input source code fails all parse attempts.""" +class BracketMatchError(KeyError): + """Raised when an opening bracket is unable to be matched to a closing bracket.""" + + T = TypeVar("T") E = TypeVar("E", bound=Exception) @@ -174,14 +196,12 @@ class TargetVersion(Enum): PY36 = 6 PY37 = 7 PY38 = 8 + PY39 = 9 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 @@ -195,6 +215,8 @@ class Feature(Enum): ASYNC_KEYWORDS = 7 ASSIGNMENT_EXPRESSIONS = 8 POS_ONLY_ARGUMENTS = 9 + RELAXED_DECORATORS = 10 + FORCE_OPTIONAL_PARENTHESES = 50 VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { @@ -232,6 +254,17 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { Feature.ASSIGNMENT_EXPRESSIONS, Feature.POS_ONLY_ARGUMENTS, }, + TargetVersion.PY39: { + Feature.UNICODE_LITERALS, + Feature.F_STRINGS, + Feature.NUMERIC_UNDERSCORES, + Feature.TRAILING_COMMA_IN_CALL, + Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, + Feature.ASSIGNMENT_EXPRESSIONS, + Feature.RELAXED_DECORATORS, + Feature.POS_ONLY_ARGUMENTS, + }, } @@ -241,6 +274,8 @@ class Mode: line_length: int = DEFAULT_LINE_LENGTH string_normalization: bool = True is_pyi: bool = False + magic_trailing_comma: bool = True + experimental_string_processing: bool = False def get_cache_key(self) -> str: if self.target_versions: @@ -255,6 +290,8 @@ class Mode: str(self.line_length), str(int(self.string_normalization)), str(int(self.is_pyi)), + str(int(self.magic_trailing_comma)), + str(int(self.experimental_string_processing)), ] return ".".join(parts) @@ -267,11 +304,24 @@ def supports_feature(target_versions: Set[TargetVersion], feature: Feature) -> b return all(feature in VERSION_TO_FEATURES[version] for version in target_versions) -def find_pyproject_toml(path_search_start: str) -> Optional[str]: +def find_pyproject_toml(path_search_start: Tuple[str, ...]) -> Optional[str]: """Find the absolute filepath to a pyproject.toml if it exists""" path_project_root = find_project_root(path_search_start) path_pyproject_toml = path_project_root / "pyproject.toml" - return str(path_pyproject_toml) if path_pyproject_toml.is_file() else None + if path_pyproject_toml.is_file(): + return str(path_pyproject_toml) + + try: + path_user_pyproject_toml = find_user_pyproject_toml() + return ( + str(path_user_pyproject_toml) + if path_user_pyproject_toml.is_file() + else None + ) + except PermissionError as e: + # We do not have access to the user-level config directory, so ignore it. + err(f"Ignoring user configuration directory due to {e!r}") + return None def parse_pyproject_toml(path_config: str) -> Dict[str, Any]: @@ -306,11 +356,19 @@ def read_pyproject_toml( if not config: return None + else: + # Sanitize the values to be Click friendly. For more information please see: + # https://github.com/psf/black/issues/1458 + # https://github.com/pallets/click/issues/1567 + config = { + k: str(v) if not isinstance(v, (list, dict)) else v + for k, v in config.items() + } target_version = config.get("target_version") if target_version is not None and not isinstance(target_version, list): raise click.BadOptionUsage( - "target-version", f"Config key target-version must be a list" + "target-version", "Config key target-version must be a list" ) default_map: Dict[str, Any] = {} @@ -333,6 +391,17 @@ def target_version_option_callback( return [TargetVersion[val.upper()] for val in v] +def validate_regex( + ctx: click.Context, + param: click.Parameter, + value: Optional[str], +) -> Optional[Pattern]: + try: + return re_compile_maybe_verbose(value) if value is not None else None + except re.error: + raise click.BadParameter("Not a valid regular expression") + + @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( @@ -368,12 +437,27 @@ def target_version_option_callback( is_flag=True, help="Don't normalize string quotes or prefixes.", ) +@click.option( + "-C", + "--skip-magic-trailing-comma", + is_flag=True, + help="Don't use trailing commas as a reason to split lines.", +) +@click.option( + "--experimental-string-processing", + is_flag=True, + hidden=True, + help=( + "Experimental option that performs more normalization on string literals." + " Currently disabled because it leads to some crashes." + ), +) @click.option( "--check", is_flag=True, help=( - "Don't write the files back, just return the status. Return code 0 means" - " nothing would change. Return code 1 means some files would be reformatted." + "Don't write the files back, just return the status. Return code 0 means" + " nothing would change. Return code 1 means some files would be reformatted." " Return code 123 means there was an internal error." ), ) @@ -396,11 +480,12 @@ def target_version_option_callback( "--include", type=str, default=DEFAULT_INCLUDES, + callback=validate_regex, help=( "A regular expression that matches files and directories that should be" - " included on recursive searches. An empty value means all files are included" - " regardless of the name. Use forward slashes for directories on all platforms" - " (Windows, too). Exclusions are calculated first, inclusions later." + " included on recursive searches. An empty value means all files are included" + " regardless of the name. Use forward slashes for directories on all platforms" + " (Windows, too). Exclusions are calculated first, inclusions later." ), show_default=True, ) @@ -408,20 +493,40 @@ def target_version_option_callback( "--exclude", type=str, default=DEFAULT_EXCLUDES, + callback=validate_regex, help=( "A regular expression that matches files and directories that should be" - " excluded on recursive searches. An empty value means no paths are excluded." - " Use forward slashes for directories on all platforms (Windows, too). " + " excluded on recursive searches. An empty value means no paths are excluded." + " Use forward slashes for directories on all platforms (Windows, too)." " Exclusions are calculated first, inclusions later." ), show_default=True, ) +@click.option( + "--extend-exclude", + type=str, + callback=validate_regex, + help=( + "Like --exclude, but adds additional files and directories on top of the" + " excluded ones. (Useful if you simply want to add to the default)" + ), +) @click.option( "--force-exclude", type=str, + callback=validate_regex, help=( "Like --exclude, but files and directories matching this regex will be " - "excluded even when they are passed explicitly as arguments" + "excluded even when they are passed explicitly as arguments." + ), +) +@click.option( + "--stdin-filename", + type=str, + help=( + "The name of the file when passing it through stdin. Useful to make " + "sure Black will respect --force-exclude option on some " + "editors that rely on using stdin." ), ) @click.option( @@ -439,7 +544,7 @@ def target_version_option_callback( is_flag=True, help=( "Also emit messages to stderr about files that were not changed or were ignored" - " due to --exclude=." + " due to exclusion patterns." ), ) @click.version_option(version=__version__) @@ -463,7 +568,7 @@ def target_version_option_callback( ), is_eager=True, callback=read_pyproject_toml, - help="Read configuration from PATH.", + help="Read configuration from FILE path.", ) @click.pass_context def main( @@ -477,11 +582,15 @@ def main( fast: bool, pyi: bool, skip_string_normalization: bool, + skip_magic_trailing_comma: bool, + experimental_string_processing: bool, quiet: bool, verbose: bool, - include: str, - exclude: str, - force_exclude: Optional[str], + include: Pattern, + exclude: Pattern, + extend_exclude: Optional[Pattern], + force_exclude: Optional[Pattern], + stdin_filename: Optional[str], src: Tuple[str, ...], config: Optional[str], ) -> None: @@ -497,6 +606,8 @@ def main( line_length=line_length, is_pyi=pyi, string_normalization=not skip_string_normalization, + magic_trailing_comma=not skip_magic_trailing_comma, + experimental_string_processing=experimental_string_processing, ) if config and verbose: out(f"Using configuration from {config}.", bold=False, fg="blue") @@ -511,8 +622,10 @@ def main( verbose=verbose, include=include, exclude=exclude, + extend_exclude=extend_exclude, force_exclude=force_exclude, report=report, + stdin_filename=stdin_filename, ) path_empty( @@ -548,58 +661,62 @@ def get_sources( src: Tuple[str, ...], quiet: bool, verbose: bool, - include: str, - exclude: str, - force_exclude: Optional[str], + include: Pattern[str], + exclude: Pattern[str], + extend_exclude: Optional[Pattern[str]], + force_exclude: Optional[Pattern[str]], report: "Report", + stdin_filename: Optional[str], ) -> Set[Path]: """Compute the set of files to be formatted.""" - try: - include_regex = re_compile_maybe_verbose(include) - except re.error: - err(f"Invalid regular expression for include given: {include!r}") - ctx.exit(2) - try: - exclude_regex = re_compile_maybe_verbose(exclude) - except re.error: - err(f"Invalid regular expression for exclude given: {exclude!r}") - ctx.exit(2) - try: - force_exclude_regex = ( - re_compile_maybe_verbose(force_exclude) if force_exclude else None - ) - except re.error: - err(f"Invalid regular expression for force_exclude given: {force_exclude!r}") - ctx.exit(2) root = find_project_root(src) sources: Set[Path] = set() path_empty(src, "No Path provided. Nothing to do 😴", quiet, verbose, ctx) - exclude_regexes = [exclude_regex] - if force_exclude_regex is not None: - exclude_regexes.append(force_exclude_regex) + gitignore = get_gitignore(root) for s in src: - p = Path(s) - if p.is_dir(): + if s == "-" and stdin_filename: + p = Path(stdin_filename) + is_stdin = True + else: + p = Path(s) + is_stdin = False + + if is_stdin or p.is_file(): + normalized_path = normalize_path_maybe_ignore(p, root, report) + if normalized_path is None: + continue + + normalized_path = "/" + normalized_path + # Hard-exclude any files that matches the `--force-exclude` regex. + if force_exclude: + force_exclude_match = force_exclude.search(normalized_path) + else: + force_exclude_match = None + if force_exclude_match and force_exclude_match.group(0): + report.path_ignored(p, "matches the --force-exclude regular expression") + continue + + if is_stdin: + p = Path(f"{STDIN_PLACEHOLDER}{str(p)}") + + sources.add(p) + elif p.is_dir(): sources.update( gen_python_files( p.iterdir(), root, - include_regex, - exclude_regexes, + include, + exclude, + extend_exclude, + force_exclude, report, - get_gitignore(root), + gitignore, ) ) elif s == "-": sources.add(p) - elif p.is_file(): - sources.update( - gen_python_files( - [p], root, None, exclude_regexes, report, get_gitignore(root) - ) - ) else: err(f"invalid path: {s}") return sources @@ -611,10 +728,9 @@ def path_empty( """ Exit if there is no `src` provided for formatting """ - if len(src) == 0: - if verbose or not quiet: - out(msg) - ctx.exit(0) + if not src and (verbose or not quiet): + out(msg) + ctx.exit(0) def reformat_one( @@ -627,15 +743,29 @@ def reformat_one( """ try: changed = Changed.NO - if not src.is_file() and str(src) == "-": + + if str(src) == "-": + is_stdin = True + elif str(src).startswith(STDIN_PLACEHOLDER): + is_stdin = True + # Use the original name again in case we want to print something + # to the user + src = Path(str(src)[len(STDIN_PLACEHOLDER) :]) + else: + is_stdin = False + + if is_stdin: + if src.suffix == ".pyi": + mode = replace(mode, is_pyi=True) if format_stdin_to_stdout(fast=fast, write_back=write_back, mode=mode): changed = Changed.YES else: cache: Cache = {} - if write_back != WriteBack.DIFF: + if write_back not in (WriteBack.DIFF, WriteBack.COLOR_DIFF): cache = read_cache(mode) res_src = src.resolve() - if res_src in cache and cache[res_src] == get_cache_info(res_src): + res_src_s = str(res_src) + if res_src_s in cache and cache[res_src_s] == get_cache_info(res_src): changed = Changed.CACHED if changed is not Changed.CACHED and format_file_in_place( src, fast=fast, write_back=write_back, mode=mode @@ -647,6 +777,8 @@ def reformat_one( write_cache(cache, [src], mode) report.done(src, changed) except Exception as exc: + if report.verbose: + traceback.print_exc() report.failed(src, str(exc)) @@ -659,13 +791,13 @@ def reformat_many( worker_count = os.cpu_count() if sys.platform == "win32": # Work around https://bugs.python.org/issue26903 - worker_count = min(worker_count, 61) + worker_count = min(worker_count, 60) try: executor = ProcessPoolExecutor(max_workers=worker_count) except (ImportError, OSError): # we arrive here if the underlying system does not support multi-processing # like in AWS Lambda or Termux, in which case we gracefully fallback to - # a ThreadPollExecutor with just a single worker (more workers would not do us + # a ThreadPoolExecutor with just a single worker (more workers would not do us # any good due to the Global Interpreter Lock) executor = ThreadPoolExecutor(max_workers=1) @@ -704,7 +836,7 @@ async def schedule_formatting( :func:`format_file_in_place`. """ cache: Cache = {} - if write_back != WriteBack.DIFF: + if write_back not in (WriteBack.DIFF, WriteBack.COLOR_DIFF): cache = read_cache(mode) sources, cached = filter_cached(cache, sources) for src in sorted(cached): @@ -715,7 +847,7 @@ async def schedule_formatting( cancelled = [] sources_to_cache = [] lock = None - if write_back == WriteBack.DIFF: + if write_back in (WriteBack.DIFF, WriteBack.COLOR_DIFF): # For diff output, we need locks to ensure we don't interleave output # from different processes. manager = Manager() @@ -728,7 +860,7 @@ async def schedule_formatting( ): src for src in sorted(sources) } - pending: Iterable["asyncio.Future[bool]"] = tasks.keys() + pending = tasks.keys() try: loop.add_signal_handler(signal.SIGINT, cancel, pending) loop.add_signal_handler(signal.SIGTERM, cancel, pending) @@ -791,7 +923,7 @@ def format_file_in_place( dst_name = f"{src}\t{now} +0000" diff_contents = diff(src_contents, dst_contents, src_name, dst_name) - if write_back == write_back.COLOR_DIFF: + if write_back == WriteBack.COLOR_DIFF: diff_contents = color_diff(diff_contents) with lock or nullcontext(): @@ -814,9 +946,9 @@ def color_diff(contents: str) -> str: for i, line in enumerate(lines): if line.startswith("+++") or line.startswith("---"): line = "\033[1;37m" + line + "\033[0m" # bold white, reset - if line.startswith("@@"): + elif line.startswith("@@"): line = "\033[36m" + line + "\033[0m" # cyan, reset - if line.startswith("+"): + elif line.startswith("+"): line = "\033[32m" + line + "\033[0m" # green, reset elif line.startswith("-"): line = "\033[31m" + line + "\033[0m" # red, reset @@ -826,30 +958,22 @@ def color_diff(contents: str) -> str: def wrap_stream_for_windows( f: io.TextIOWrapper, -) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32.AnsiToWin32"]: +) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]: """ - Wrap the stream in colorama's wrap_stream so colors are shown on Windows. + Wrap stream with colorama's wrap_stream so colors are shown on Windows. - If `colorama` is not found, then no change is made. If `colorama` does - exist, then it handles the logic to determine whether or not to change - things. + If `colorama` is unavailable, the original stream is returned unmodified. + Otherwise, the `wrap_stream()` function determines whether the stream needs + to be wrapped for a Windows environment and will accordingly either return + an `AnsiToWin32` wrapper or the original stream. """ try: - from colorama import initialise - - # We set `strip=False` so that we can don't have to modify - # test_express_diff_with_color. - f = initialise.wrap_stream( - f, convert=None, strip=False, autoreset=False, wrap=True - ) - - # wrap_stream returns a `colorama.AnsiToWin32.AnsiToWin32` object - # which does not have a `detach()` method. So we fake one. - f.detach = lambda *args, **kwargs: None # type: ignore + from colorama.initialise import wrap_stream except ImportError: - pass - - return f + return f + else: + # Set `strip=False` to avoid needing to modify test_express_diff_with_color. + return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True) def format_stdin_to_stdout( @@ -890,13 +1014,13 @@ def format_stdin_to_stdout( def format_file_contents(src_contents: str, *, fast: bool, mode: Mode) -> FileContent: - """Reformat contents a file and return new contents. + """Reformat contents of a file and return new contents. If `fast` is False, additionally confirm that the reformatted code is valid by calling :func:`assert_equivalent` and :func:`assert_stable` on it. `mode` is passed to :func:`format_str`. """ - if src_contents.strip() == "": + if not src_contents.strip(): raise NothingChanged dst_contents = format_str(src_contents, mode=mode) @@ -905,7 +1029,17 @@ def format_file_contents(src_contents: str, *, fast: bool, mode: Mode) -> FileCo if not fast: assert_equivalent(src_contents, dst_contents) - assert_stable(src_contents, dst_contents, mode=mode) + + # Forced second pass to work around optional trailing commas (becoming + # forced trailing commas on pass 2) interacting differently with optional + # parentheses. Admittedly ugly. + dst_contents_pass2 = format_str(dst_contents, mode=mode) + if dst_contents != dst_contents_pass2: + dst_contents = dst_contents_pass2 + assert_equivalent(src_contents, dst_contents, pass_num=2) + assert_stable(src_contents, dst_contents, mode=mode) + # Note: no need to explicitly call `assert_stable` if `dst_contents` was + # the same as `dst_contents_pass2`. return dst_contents @@ -916,11 +1050,12 @@ def format_str(src_contents: str, *, mode: Mode) -> FileContent: allowed. Example: >>> import black - >>> print(black.format_str("def f(arg:str='')->None:...", mode=Mode())) + >>> print(black.format_str("def f(arg:str='')->None:...", mode=black.Mode())) def f(arg: str = "") -> None: ... A more complex example: + >>> print( ... black.format_str( ... "def f(arg:str='')->None: hey", @@ -947,13 +1082,12 @@ def format_str(src_contents: str, *, mode: Mode) -> FileContent: versions = detect_target_versions(src_node) normalize_fmt_off(src_node) lines = LineGenerator( + mode=mode, 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=mode.is_pyi) - empty_line = Line() + empty_line = Line(mode=mode) after = 0 split_line_features = { feature @@ -965,10 +1099,7 @@ def format_str(src_contents: str, *, mode: Mode) -> FileContent: before, after = elt.maybe_empty_lines(current_line) dst_contents.append(str(empty_line) * before) for line in transform_line( - current_line, - line_length=mode.line_length, - normalize_strings=mode.string_normalization, - features=split_line_features, + current_line, mode=mode, features=split_line_features ): dst_contents.append(str(line)) return "".join(dst_contents) @@ -1032,7 +1163,7 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]: 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": + if not src_txt.endswith("\n"): src_txt += "\n" for grammar in get_grammars(set(target_versions)): @@ -1255,6 +1386,7 @@ class BracketTracker: previous: Optional[Leaf] = None _for_loop_depths: List[int] = field(default_factory=list) _lambda_argument_depths: List[int] = field(default_factory=list) + invisible: List[Leaf] = field(default_factory=list) def mark(self, leaf: Leaf) -> None: """Mark `leaf` with bracket-related metadata. Keep track of delimiters. @@ -1278,8 +1410,16 @@ class BracketTracker: self.maybe_decrement_after_lambda_arguments(leaf) if leaf.type in CLOSING_BRACKETS: self.depth -= 1 - opening_bracket = self.bracket_match.pop((self.depth, leaf.type)) + try: + opening_bracket = self.bracket_match.pop((self.depth, leaf.type)) + except KeyError as e: + raise BracketMatchError( + "Unable to match a closing bracket to the following opening" + f" bracket: {leaf}" + ) from e leaf.opening_bracket = opening_bracket + if not leaf.value: + self.invisible.append(leaf) leaf.bracket_depth = self.depth if self.depth == 0: delim = is_split_before_delimiter(leaf, self.previous) @@ -1292,6 +1432,8 @@ class BracketTracker: if leaf.type in OPENING_BRACKETS: self.bracket_match[self.depth, BRACKET[leaf.type]] = leaf self.depth += 1 + if not leaf.value: + self.invisible.append(leaf) self.previous = leaf self.maybe_increment_lambda_arguments(leaf) self.maybe_increment_for_loop_variable(leaf) @@ -1381,13 +1523,15 @@ class BracketTracker: class Line: """Holds leaves and comments. Can be printed with `str(line)`.""" + mode: Mode depth: int = 0 leaves: List[Leaf] = field(default_factory=list) # keys ordered like `leaves` comments: Dict[LeafID, List[Leaf]] = field(default_factory=dict) bracket_tracker: BracketTracker = field(default_factory=BracketTracker) inside_brackets: bool = False - should_explode: bool = False + should_split_rhs: bool = False + magic_trailing_comma: Optional[Leaf] = None def append(self, leaf: Leaf, preformatted: bool = False) -> None: """Add a new `leaf` to the end of the line. @@ -1413,7 +1557,11 @@ class Line: ) if self.inside_brackets or not preformatted: self.bracket_tracker.mark(leaf) - self.maybe_remove_trailing_comma(leaf) + if self.mode.magic_trailing_comma: + if self.has_magic_trailing_comma(leaf): + self.magic_trailing_comma = leaf + elif self.has_magic_trailing_comma(leaf, ensure_removable=True): + self.remove_trailing_comma() if not self.append_comment(leaf): self.leaves.append(leaf) @@ -1465,69 +1613,6 @@ 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.)""" @@ -1633,8 +1718,10 @@ class Line: # 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) + first_line = next((leaf.lineno for leaf in self.leaves if leaf.lineno != 0), 0) + last_line = next( + (leaf.lineno for leaf in reversed(self.leaves) if leaf.lineno != 0), 0 + ) if first_line == last_line: # We look at the last two leaves since a comma or an @@ -1650,42 +1737,38 @@ class Line: def contains_multiline_strings(self) -> bool: return any(is_multiline_string(leaf) for leaf in self.leaves) - 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. + def has_magic_trailing_comma( + self, closing: Leaf, ensure_removable: bool = False + ) -> bool: + """Return True if we have a magic trailing comma, that is when: + - there's a trailing comma here + - it's not a one-tuple + Additionally, if ensure_removable: + - it's not from square bracket indexing + """ if not ( - self.leaves - and self.is_import - and len(self.leaves) > 4 + closing.type in CLOSING_BRACKETS + and self.leaves 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 - self.remove_trailing_comma() - return True + if closing.type == token.RBRACE: + return True + + if closing.type == token.RSQB: + if not ensure_removable: + return True + comma = self.leaves[-1] + return bool(comma.parent and comma.parent.type == syms.listmaker) + + if self.is_import: + return True + + if not is_one_tuple_between(closing.opening_bracket, closing, self.leaves): + return True + + return False def append_comment(self, comment: Leaf) -> bool: """Add an inline or standalone comment to the line.""" @@ -1756,9 +1839,11 @@ class Line: def clone(self) -> "Line": return Line( + mode=self.mode, depth=self.depth, inside_brackets=self.inside_brackets, - should_explode=self.should_explode, + should_split_rhs=self.should_split_rhs, + magic_trailing_comma=self.magic_trailing_comma, ) def __str__(self) -> str: @@ -1864,6 +1949,10 @@ class EmptyLineTracker: return 0, 0 if self.previous_line.is_decorator: + if self.is_pyi and current_line.is_stub_class: + # Insert an empty line after a decorated stub class + return 0, 1 + return 0, 0 if self.previous_line.depth < current_line.depth and ( @@ -1887,8 +1976,11 @@ class EmptyLineTracker: newlines = 0 else: newlines = 1 - elif current_line.is_def and not self.previous_line.is_def: - # Blank line between a block of functions and a block of non-functions + elif ( + current_line.is_def or current_line.is_decorator + ) and not self.previous_line.is_def: + # Blank line between a block of functions (maybe with preceding + # decorators) and a block of non-functions newlines = 1 else: newlines = 0 @@ -1907,10 +1999,9 @@ class LineGenerator(Visitor[Line]): in ways that will no longer stringify to valid Python code on the tree. """ - is_pyi: bool = False - normalize_strings: bool = True - current_line: Line = field(default_factory=Line) + mode: Mode remove_u_prefix: bool = False + current_line: Line = field(init=False) def line(self, indent: int = 0) -> Iterator[Line]: """Generate a line. @@ -1925,7 +2016,7 @@ class LineGenerator(Visitor[Line]): return # Line is empty, don't emit. Creating a new one unnecessary. complete_line = self.current_line - self.current_line = Line(depth=complete_line.depth + indent) + self.current_line = Line(mode=self.mode, depth=complete_line.depth + indent) yield complete_line def visit_default(self, node: LN) -> Iterator[Line]: @@ -1949,7 +2040,7 @@ class LineGenerator(Visitor[Line]): yield from self.line() normalize_prefix(node, inside_brackets=any_open_brackets) - if self.normalize_strings and node.type == token.STRING: + if self.mode.string_normalization and node.type == token.STRING: normalize_string_prefix(node, remove_u_prefix=self.remove_u_prefix) normalize_string_quotes(node) if node.type == token.NUMBER: @@ -2001,16 +2092,18 @@ class LineGenerator(Visitor[Line]): def visit_suite(self, node: Node) -> Iterator[Line]: """Visit a suite.""" - if self.is_pyi and is_stub_suite(node): + if self.mode.is_pyi and is_stub_suite(node): yield from self.visit(node.children[2]) else: yield from self.visit_default(node) def visit_simple_stmt(self, node: Node) -> Iterator[Line]: """Visit a statement without nested statements.""" + if first_child_is_arith(node): + wrap_in_parentheses(node, node.children[0], visible=False) is_suite_like = node.parent and node.parent.type in STATEMENT if is_suite_like: - if self.is_pyi and is_stub_body(node): + if self.mode.is_pyi and is_stub_body(node): yield from self.visit_default(node) else: yield from self.line(+1) @@ -2018,7 +2111,11 @@ class LineGenerator(Visitor[Line]): yield from self.line(-1) else: - if not self.is_pyi or not node.parent or not is_stub_suite(node.parent): + if ( + not self.mode.is_pyi + or not node.parent + or not is_stub_suite(node.parent) + ): yield from self.line() yield from self.visit_default(node) @@ -2075,19 +2172,52 @@ class LineGenerator(Visitor[Line]): yield from self.visit_default(node) def visit_STRING(self, leaf: Leaf) -> Iterator[Line]: - # Check if it's a docstring - if prev_siblings_are( - leaf.parent, [None, token.NEWLINE, token.INDENT, syms.simple_stmt] - ) and is_multiline_string(leaf): - prefix = " " * self.current_line.depth - docstring = fix_docstring(leaf.value[3:-3], prefix) - leaf.value = leaf.value[0:3] + docstring + leaf.value[-3:] - normalize_string_quotes(leaf) + if is_docstring(leaf) and "\\\n" not in leaf.value: + # We're ignoring docstrings with backslash newline escapes because changing + # indentation of those changes the AST representation of the code. + prefix = get_string_prefix(leaf.value) + docstring = leaf.value[len(prefix) :] # Remove the prefix + quote_char = docstring[0] + # A natural way to remove the outer quotes is to do: + # docstring = docstring.strip(quote_char) + # but that breaks on """""x""" (which is '""x'). + # So we actually need to remove the first character and the next two + # characters but only if they are the same as the first. + quote_len = 1 if docstring[1] != quote_char else 3 + docstring = docstring[quote_len:-quote_len] + + if is_multiline_string(leaf): + indent = " " * 4 * self.current_line.depth + docstring = fix_docstring(docstring, indent) + else: + docstring = docstring.strip() + + if docstring: + # Add some padding if the docstring starts / ends with a quote mark. + if docstring[0] == quote_char: + docstring = " " + docstring + if docstring[-1] == quote_char: + docstring += " " + if docstring[-1] == "\\": + backslash_count = len(docstring) - len(docstring.rstrip("\\")) + if backslash_count % 2: + # Odd number of tailing backslashes, add some padding to + # avoid escaping the closing string quote. + docstring += " " + else: + # Add some padding if the docstring is empty. + docstring = " " + + # We could enforce triple quotes at this point. + quote = quote_char * quote_len + leaf.value = prefix + quote + docstring + quote yield from self.visit_default(leaf) def __post_init__(self) -> None: """You are in a twisty little maze of passages.""" + self.current_line = Line(mode=self.mode) + v = self.visit_stmt Ø: Set[str] = set() self.visit_assert_stmt = partial(v, keywords={"assert"}, parens={"assert", ","}) @@ -2201,6 +2331,9 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa: C901 ): # Python 2 print chevron return NO + elif prevp.type == token.AT and p.parent and p.parent.type == syms.decorator: + # no space in decorators + return NO elif prev.type in OPENING_BRACKETS: return NO @@ -2527,6 +2660,8 @@ def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Pr FMT_OFF = {"# fmt: off", "# fmt:off", "# yapf: disable"} +FMT_SKIP = {"# fmt: skip", "# fmt:skip"} +FMT_PASS = {*FMT_OFF, *FMT_SKIP} FMT_ON = {"# fmt: on", "# fmt:on", "# yapf: enable"} @@ -2581,7 +2716,7 @@ 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")): + for index, line in enumerate(re.split("\r?\n", prefix)): consumed += len(line) + 1 # adding the length of the split '\n' line = line.lstrip() if not line: @@ -2622,16 +2757,20 @@ def make_comment(content: str) -> str: if content[0] == "#": content = content[1:] + NON_BREAKING_SPACE = " " + if ( + content + and content[0] == NON_BREAKING_SPACE + and not content.lstrip().startswith("type:") + ): + content = " " + content[1:] # Replace NBSP by a simple space if content and content[0] not in " !:#'%": content = " " + content return "#" + content def transform_line( - line: Line, - line_length: int, - normalize_strings: bool, - features: Collection[Feature] = (), + line: Line, mode: Mode, features: Collection[Feature] = () ) -> Iterator[Line]: """Transform a `line`, potentially splitting it into many lines. @@ -2647,7 +2786,7 @@ def transform_line( def init_st(ST: Type[StringTransformer]) -> StringTransformer: """Initialize StringTransformer""" - return ST(line_length, normalize_strings) + return ST(mode.line_length, mode.string_normalization) string_merge = init_st(StringMerger) string_paren_strip = init_st(StringParenStripper) @@ -2657,73 +2796,81 @@ def transform_line( transformers: List[Transformer] if ( not line.contains_uncollapsable_type_comments() - and not line.should_explode - and not line.is_collection_with_optional_trailing_comma + and not line.should_split_rhs + and not line.magic_trailing_comma and ( - is_line_short_enough(line, line_length=line_length, line_str=line_str) + is_line_short_enough(line, line_length=mode.line_length, line_str=line_str) or line.contains_unsplittable_type_ignore() ) - and not (line.contains_standalone_comments() and line.inside_brackets) + and not (line.inside_brackets and line.contains_standalone_comments()) ): # Only apply basic string preprocessing, since lines shouldn't be split here. - transformers = [string_merge, string_paren_strip] + if mode.experimental_string_processing: + transformers = [string_merge, string_paren_strip] + else: + transformers = [] elif line.is_def: transformers = [left_hand_split] else: 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, features, omit=omit)) - if is_line_short_enough(lines[0], line_length=line_length): + """Wraps calls to `right_hand_split`. + + The calls increasingly `omit` right-hand trailers (bracket pairs with + content), meaning the trailers get glued together to split on another + bracket pair instead. + """ + for omit in generate_trailers_to_omit(line, mode.line_length): + lines = list( + right_hand_split(line, mode.line_length, features, omit=omit) + ) + # Note: this check is only able to figure out if the first line of the + # *current* transformation fits in the line length. This is true only + # for simple cases. All others require running more transforms via + # `transform_line()`. This check doesn't know if those would succeed. + if is_line_short_enough(lines[0], line_length=mode.line_length): 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. - # line_length=1 here was historically a bug that somehow became a feature. - # See #762 and #781 for the full story. - yield from right_hand_split(line, line_length=1, features=features) - - if line.inside_brackets: - transformers = [ - string_merge, - string_paren_strip, - delimiter_split, - standalone_comment_split, - string_split, - string_paren_wrap, - rhs, - ] + # reported as not fitting a single line, as well as lines that contain + # trailing commas (those have to be exploded). + yield from right_hand_split( + line, line_length=mode.line_length, features=features + ) + + if mode.experimental_string_processing: + if line.inside_brackets: + transformers = [ + string_merge, + string_paren_strip, + string_split, + delimiter_split, + standalone_comment_split, + string_paren_wrap, + rhs, + ] + else: + transformers = [ + string_merge, + string_paren_strip, + string_split, + string_paren_wrap, + rhs, + ] else: - transformers = [ - string_merge, - string_paren_strip, - string_split, - string_paren_wrap, - rhs, - ] + if line.inside_brackets: + transformers = [delimiter_split, standalone_comment_split, rhs] + else: + transformers = [rhs] for transform in transformers: # We are accumulating lines in `result` because we might want to abort # mission and return the original line in the end, or attempt a different # split altogether. - result: List[Line] = [] try: - for l in transform(line, features): - if str(l).strip("\n") == line_str: - raise CannotTransform( - "Line transformer returned an unchanged result" - ) - - result.extend( - transform_line( - l, - line_length=line_length, - normalize_strings=normalize_strings, - features=features, - ) - ) + result = run_transformer(line, transform, mode, features, line_str=line_str) except CannotTransform: continue else: @@ -2752,7 +2899,7 @@ class StringTransformer(ABC): StringTransformer to be applied? Transformations: - If the given Line meets all of the above requirments, which string + If the given Line meets all of the above requirements, which string transformations can you expect to be applied to it by this StringTransformer? @@ -2764,6 +2911,7 @@ class StringTransformer(ABC): line_length: int normalize_strings: bool + __name__ = "StringTransformer" @abstractmethod def do_match(self, line: Line) -> TMatchResult: @@ -2918,11 +3066,8 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): """StringTransformer that merges strings together. Requirements: - (A) The line contains adjacent strings such that at most one substring - has inline comments AND none of those inline comments are pragmas AND - the set of all substring prefixes is either of length 1 or equal to - {"", "f"} AND none of the substrings are raw strings (i.e. are prefixed - with 'r'). + (A) The line contains adjacent strings such that ALL of the validation checks + listed in StringMerger.__validate_msg(...)'s docstring pass. OR (B) The line contains a string which uses line continuation backslashes. @@ -3010,7 +3155,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): ) new_line = line.clone() - new_line.comments = line.comments + new_line.comments = line.comments.copy() append_leaves(new_line, line, LL) new_string_leaf = new_line.leaves[string_idx] @@ -3171,6 +3316,7 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): * Ok(None), if ALL validation checks (listed below) pass. OR * Err(CannotTransform), if any of the following are true: + - The target string group does not contain ANY stand-alone comments. - The target string is not in a string group (i.e. it has no adjacent strings). - The string group has more than one inline comment. @@ -3179,6 +3325,26 @@ class StringMerger(CustomSplitMapMixin, StringTransformer): length greater than one and is not equal to {"", "f"}. - The string group consists of raw strings. """ + # We first check for "inner" stand-alone comments (i.e. stand-alone + # comments that have a string leaf before them AND after them). + for inc in [1, -1]: + i = string_idx + found_sa_comment = False + is_valid_index = is_valid_index_factory(line.leaves) + while is_valid_index(i) and line.leaves[i].type in [ + token.STRING, + STANDALONE_COMMENT, + ]: + if line.leaves[i].type == STANDALONE_COMMENT: + found_sa_comment = True + elif found_sa_comment: + return TErr( + "StringMerger does NOT merge string groups which contain " + "stand-alone comments." + ) + + i += inc + num_of_inline_string_comments = 0 set_of_prefixes = set() num_of_strings = 0 @@ -3227,8 +3393,11 @@ class StringParenStripper(StringTransformer): Requirements: The line contains a string which is surrounded by parentheses and: - - The target string is NOT the only argument to a function call). - - The RPAR is NOT followed by an attribute access (i.e. a dot). + - The target string is NOT the only argument to a function call. + - The target string is NOT a "pointless" string. + - If the target string contains a PERCENT, the brackets are not + preceeded or followed by an operator with higher precedence than + PERCENT. Transformations: The parentheses mentioned in the 'Requirements' section are stripped. @@ -3249,6 +3418,14 @@ class StringParenStripper(StringTransformer): if leaf.type != token.STRING: continue + # If this is a "pointless" string... + if ( + leaf.parent + and leaf.parent.parent + and leaf.parent.parent.type == syms.simple_stmt + ): + continue + # Should be preceded by a non-empty LPAR... if ( not is_valid_index(idx - 1) @@ -3271,14 +3448,51 @@ class StringParenStripper(StringTransformer): string_parser = StringParser() next_idx = string_parser.parse(LL, string_idx) + # if the leaves in the parsed string include a PERCENT, we need to + # make sure the initial LPAR is NOT preceded by an operator with + # higher or equal precedence to PERCENT + if is_valid_index(idx - 2): + # mypy can't quite follow unless we name this + before_lpar = LL[idx - 2] + if token.PERCENT in {leaf.type for leaf in LL[idx - 1 : next_idx]} and ( + ( + before_lpar.type + in { + token.STAR, + token.AT, + token.SLASH, + token.DOUBLESLASH, + token.PERCENT, + token.TILDE, + token.DOUBLESTAR, + token.AWAIT, + token.LSQB, + token.LPAR, + } + ) + or ( + # only unary PLUS/MINUS + before_lpar.parent + and before_lpar.parent.type == syms.factor + and (before_lpar.type in {token.PLUS, token.MINUS}) + ) + ): + continue + # Should be followed by a non-empty RPAR... if ( is_valid_index(next_idx) and LL[next_idx].type == token.RPAR and not is_empty_rpar(LL[next_idx]) ): - # That RPAR should NOT be followed by a '.' symbol. - if is_valid_index(next_idx + 1) and LL[next_idx + 1].type == token.DOT: + # That RPAR should NOT be followed by anything with higher + # precedence than PERCENT + if is_valid_index(next_idx + 1) and LL[next_idx + 1].type in { + token.DOUBLESTAR, + token.LSQB, + token.LPAR, + token.DOT, + }: continue return Ok(string_idx) @@ -3296,11 +3510,17 @@ class StringParenStripper(StringTransformer): yield TErr( "Will not strip parentheses which have comments attached to them." ) + return new_line = line.clone() new_line.comments = line.comments.copy() - - append_leaves(new_line, line, LL[: string_idx - 1]) + try: + append_leaves(new_line, line, LL[: string_idx - 1]) + except BracketMatchError: + # HACK: I believe there is currently a bug somewhere in + # right_hand_split() that is causing brackets to not be tracked + # properly by a shared BracketTracker. + append_leaves(new_line, line, LL[: string_idx - 1], preformatted=True) string_leaf = Leaf(token.STRING, LL[string_idx].value) LL[string_idx - 1].remove() @@ -3308,7 +3528,7 @@ class StringParenStripper(StringTransformer): new_line.append(string_leaf) append_leaves( - new_line, line, LL[string_idx + 1 : rpar_idx] + LL[rpar_idx + 1 :], + new_line, line, LL[string_idx + 1 : rpar_idx] + LL[rpar_idx + 1 :] ) LL[rpar_idx].remove() @@ -3463,9 +3683,12 @@ class BaseStringSplitter(StringTransformer): # WMA4 a single space. offset += 1 - # WMA4 the lengths of any leaves that came before that space. - for leaf in LL[: p_idx + 1]: + # WMA4 the lengths of any leaves that came before that space, + # but after any closing bracket before that space. + for leaf in reversed(LL[: p_idx + 1]): offset += len(str(leaf)) + if leaf.type in CLOSING_BRACKETS: + break if is_valid_index(string_idx + 1): N = LL[string_idx + 1] @@ -3531,7 +3754,8 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): MIN_SUBSTR_SIZE characters. The string will ONLY be split on spaces (i.e. each new substring should - start with a space). + start with a space). Note that the string will NOT be split on a space + which is escaped with a backslash. If the string is an f-string, it will NOT be split in the middle of an f-expression (e.g. in f"FooBar: {foo() if x else bar()}", {foo() if x @@ -3551,13 +3775,14 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): MIN_SUBSTR_SIZE = 6 # Matches an "f-expression" (e.g. {var}) that might be found in an f-string. RE_FEXPR = r""" - (? TMatchResult: @@ -3611,7 +3836,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): # We MAY choose to drop the 'f' prefix from substrings that don't # contain any f-expressions, but ONLY if the original f-string - # containes at least one f-expression. Otherwise, we will alter the AST + # contains at least one f-expression. Otherwise, we will alter the AST # of the program. drop_pointless_f_prefix = ("f" in prefix) and re.search( self.RE_FEXPR, LL[string_idx].value, re.VERBOSE @@ -3701,7 +3926,7 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): max_bidx = max_break_idx - 2 if line_needs_plus() else max_break_idx maybe_break_idx = self.__get_break_idx(rest_value, max_bidx) if maybe_break_idx is None: - # If we are unable to algorthmically determine a good split + # If we are unable to algorithmically determine a good split # and this string has custom splits registered to it, we # fall back to using them--which means we have to start # over from the beginning. @@ -3871,11 +4096,23 @@ class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): section of this classes' docstring would be be met by returning @i. """ is_space = string[i] == " " + + is_not_escaped = True + j = i - 1 + while is_valid_index(j) and string[j] == "\\": + is_not_escaped = not is_not_escaped + j -= 1 + is_big_enough = ( len(string[i:]) >= self.MIN_SUBSTR_SIZE and len(string[:i]) >= self.MIN_SUBSTR_SIZE ) - return is_space and is_big_enough and not breaks_fstring_expression(i) + return ( + is_space + and is_not_escaped + and is_big_enough + and not breaks_fstring_expression(i) + ) # First, we check all indices BELOW @max_break_idx. break_idx = max_break_idx @@ -3981,12 +4218,13 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): def do_splitter_match(self, line: Line) -> TMatchResult: LL = line.leaves - string_idx = None - string_idx = string_idx or self._return_match(LL) - string_idx = string_idx or self._else_match(LL) - string_idx = string_idx or self._assert_match(LL) - string_idx = string_idx or self._assign_match(LL) - string_idx = string_idx or self._dict_match(LL) + string_idx = ( + self._return_match(LL) + or self._else_match(LL) + or self._assert_match(LL) + or self._assign_match(LL) + or self._dict_match(LL) + ) if string_idx is not None: string_value = line.leaves[string_idx].value @@ -4185,7 +4423,7 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): is_valid_index = is_valid_index_factory(LL) insert_str_child = insert_str_child_factory(LL[string_idx]) - comma_idx = len(LL) - 1 + comma_idx = -1 ends_with_comma = False if LL[comma_idx].type == token.COMMA: ends_with_comma = True @@ -4230,9 +4468,11 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): # `StringSplitter` will break it down further if necessary. string_value = LL[string_idx].value string_line = Line( + mode=line.mode, depth=line.depth + 1, inside_brackets=True, - should_explode=line.should_explode, + should_split_rhs=line.should_split_rhs, + magic_trailing_comma=line.magic_trailing_comma, ) string_leaf = Leaf(token.STRING, string_value) insert_str_child(string_leaf) @@ -4278,7 +4518,7 @@ class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): class StringParser: """ A state machine that aids in parsing a string's "trailer", which can be - either non-existant, an old-style formatting sequence (e.g. `% varX` or `% + either non-existent, an old-style formatting sequence (e.g. `% varX` or `% (varX, varY)`), or a method-call / attribute access (e.g. `.format(varX, varY)`). @@ -4372,7 +4612,7 @@ class StringParser: was directly after the string leaf in question (e.g. if our target string is `line.leaves[i]` then the first call to this method must be `line.leaves[i + 1]`). - * On the next call to this function, the leaf paramater passed in + * On the next call to this function, the leaf parameter passed in MUST be the leaf directly following @leaf. Returns: @@ -4570,25 +4810,25 @@ def line_to_string(line: Line) -> str: return str(line).strip("\n") -def append_leaves(new_line: Line, old_line: Line, leaves: List[Leaf]) -> None: +def append_leaves( + new_line: Line, old_line: Line, leaves: List[Leaf], preformatted: bool = False +) -> None: """ Append leaves (taken from @old_line) to @new_line, making sure to fix the underlying Node structure where appropriate. All of the leaves in @leaves are duplicated. The duplicates are then appended to @new_line and used to replace their originals in the underlying - Node structure. Any comments attatched to the old leaves are reattached to + Node structure. Any comments attached to the old leaves are reattached to the new leaves. Pre-conditions: set(@leaves) is a subset of set(@old_line.leaves). """ for old_leaf in leaves: - assert old_leaf in old_line.leaves - new_leaf = Leaf(old_leaf.type, old_leaf.value) replace_child(old_leaf, new_leaf) - new_line.append(new_leaf) + new_line.append(new_leaf, preformatted=preformatted) for comment_leaf in old_line.comments_after(old_leaf): new_line.append(comment_leaf, preformatted=True) @@ -4745,8 +4985,7 @@ def right_hand_split( tail = bracket_split_build_line(tail_leaves, line, opening_bracket) bracket_split_succeeded_or_raise(head, body, tail) if ( - # the body shouldn't be exploded - not body.should_explode + Feature.FORCE_OPTIONAL_PARENTHESES not in features # the opening bracket is an optional paren and opening_bracket.type == token.LPAR and not opening_bracket.value @@ -4759,7 +4998,7 @@ def right_hand_split( # there are no standalone comments in the body and not body.contains_standalone_comments(0) # and we can actually remove the parens - and can_omit_invisible_parens(body, line_length) + and can_omit_invisible_parens(body, line_length, omit_on_explode=omit) ): omit = {id(closing_bracket), *omit} try: @@ -4824,7 +5063,7 @@ def bracket_split_build_line( If `is_body` is True, the result line is one-indented inside brackets and as such has its first leaf's prefix normalized and a trailing comma added when expected. """ - result = Line(depth=original.depth) + result = Line(mode=original.mode, depth=original.depth) if is_body: result.inside_brackets = True result.depth += 1 @@ -4836,7 +5075,7 @@ def bracket_split_build_line( no_commas = ( original.is_def and opening_bracket.value == "(" - and not any(l.type == token.COMMA for l in leaves) + and not any(leaf.type == token.COMMA for leaf in leaves) ) if original.is_import or no_commas: @@ -4845,7 +5084,8 @@ def bracket_split_build_line( continue if leaves[i].type != token.COMMA: - leaves.insert(i + 1, Leaf(token.COMMA, ",")) + new_comma = Leaf(token.COMMA, ",") + leaves.insert(i + 1, new_comma) break # Populate the line @@ -4853,8 +5093,8 @@ def bracket_split_build_line( result.append(leaf, preformatted=True) for comment_after in original.comments_after(leaf): result.append(comment_after, preformatted=True) - if is_body: - result.should_explode = should_explode(result, opening_bracket) + if is_body and should_split_line(result, opening_bracket): + result.should_split_rhs = True return result @@ -4866,9 +5106,9 @@ def dont_increase_indentation(split_func: Transformer) -> Transformer: @wraps(split_func) 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 + for line in split_func(line, features): + normalize_prefix(line.leaves[0], inside_brackets=True) + yield line return split_wrapper @@ -4895,7 +5135,9 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[ if bt.delimiter_count_with_priority(delimiter_priority) == 1: raise CannotSplit("Splitting a single attribute from its owner looks wrong") - current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + current_line = Line( + mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets + ) lowest_depth = sys.maxsize trailing_comma_safe = True @@ -4907,7 +5149,9 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[ except ValueError: yield current_line - current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + current_line = Line( + mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets + ) current_line.append(leaf) for leaf in line.leaves: @@ -4931,7 +5175,9 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[ if leaf_priority == delimiter_priority: yield current_line - current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + current_line = Line( + mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets + ) if current_line: if ( trailing_comma_safe @@ -4939,7 +5185,8 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[ and current_line.leaves[-1].type != token.COMMA and current_line.leaves[-1].type != STANDALONE_COMMENT ): - current_line.append(Leaf(token.COMMA, ",")) + new_comma = Leaf(token.COMMA, ",") + current_line.append(new_comma) yield current_line @@ -4951,7 +5198,9 @@ def standalone_comment_split( if not line.contains_standalone_comments(0): raise CannotSplit("Line does not have any standalone comments") - current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + current_line = Line( + mode=line.mode, depth=line.depth, inside_brackets=line.inside_brackets + ) def append_to_line(leaf: Leaf) -> Iterator[Line]: """Append `leaf` to current line or to new line if appending impossible.""" @@ -4961,7 +5210,9 @@ def standalone_comment_split( except ValueError: yield current_line - current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + current_line = Line( + line.mode, depth=line.depth, inside_brackets=line.inside_brackets + ) current_line.append(leaf) for leaf in line.leaves: @@ -5117,31 +5368,47 @@ def normalize_numeric_literal(leaf: Leaf) -> None: # Leave octal and binary literals alone. pass elif text.startswith("0x"): - # Change hex literals to upper case. - before, after = text[:2], text[2:] - text = f"{before}{after.upper()}" + text = format_hex(text) elif "e" in text: - before, after = text.split("e") - sign = "" - if after.startswith("-"): - after = after[1:] - sign = "-" - elif after.startswith("+"): - after = after[1:] - before = format_float_or_int_string(before) - text = f"{before}e{sign}{after}" + text = format_scientific_notation(text) elif text.endswith(("j", "l")): - number = text[:-1] - suffix = text[-1] - # Capitalize in "2L" because "l" looks too similar to "1". - if suffix == "l": - suffix = "L" - text = f"{format_float_or_int_string(number)}{suffix}" + text = format_long_or_complex_number(text) else: text = format_float_or_int_string(text) leaf.value = text +def format_hex(text: str) -> str: + """ + Formats a hexadecimal string like "0x12B3" + """ + before, after = text[:2], text[2:] + return f"{before}{after.upper()}" + + +def format_scientific_notation(text: str) -> str: + """Formats a numeric string utilizing scentific notation""" + before, after = text.split("e") + sign = "" + if after.startswith("-"): + after = after[1:] + sign = "-" + elif after.startswith("+"): + after = after[1:] + before = format_float_or_int_string(before) + return f"{before}e{sign}{after}" + + +def format_long_or_complex_number(text: str) -> str: + """Formats a long or complex string like `10L` or `10j`""" + number = text[:-1] + suffix = text[-1] + # Capitalize in "2L" because "l" looks too similar to "1". + if suffix == "l": + suffix = "L" + return f"{format_float_or_int_string(number)}{suffix}" + + def format_float_or_int_string(text: str) -> str: """Formats a float string like "1.0".""" if "." not in text: @@ -5180,9 +5447,6 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: 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, parent=node): wrap_in_parentheses(node, child, visible=False) @@ -5222,58 +5486,80 @@ def convert_one_fmt_off_pair(node: Node) -> bool: for leaf in node.leaves(): previous_consumed = 0 for comment in list_comments(leaf.prefix, is_endmarker=False): - if comment.value in FMT_OFF: - # We only want standalone comments. If there's no previous leaf or - # the previous leaf is indentation, it's a standalone comment in - # disguise. - if comment.type != STANDALONE_COMMENT: - prev = preceding_leaf(leaf) - if prev and prev.type not in WHITESPACE: + if comment.value not in FMT_PASS: + previous_consumed = comment.consumed + continue + # We only want standalone comments. If there's no previous leaf or + # the previous leaf is indentation, it's a standalone comment in + # disguise. + if comment.value in FMT_PASS and comment.type != STANDALONE_COMMENT: + prev = preceding_leaf(leaf) + if prev: + if comment.value in FMT_OFF and prev.type not in WHITESPACE: + continue + if comment.value in FMT_SKIP and prev.type in WHITESPACE: continue - ignored_nodes = list(generate_ignored_nodes(leaf)) - if not ignored_nodes: - continue - - first = ignored_nodes[0] # Can be a container node with the `leaf`. - parent = first.parent - prefix = first.prefix - first.prefix = prefix[comment.consumed :] - hidden_value = ( - comment.value + "\n" + "".join(str(n) for n in ignored_nodes) - ) - if hidden_value.endswith("\n"): - # That happens when one of the `ignored_nodes` ended with a NEWLINE - # leaf (possibly followed by a DEDENT). - hidden_value = hidden_value[:-1] - first_idx: Optional[int] = None - for ignored in ignored_nodes: - index = ignored.remove() - if first_idx is None: - first_idx = index - assert parent is not None, "INTERNAL ERROR: fmt: on/off handling (1)" - assert first_idx is not None, "INTERNAL ERROR: fmt: on/off handling (2)" - parent.insert_child( - first_idx, - Leaf( - STANDALONE_COMMENT, - hidden_value, - prefix=prefix[:previous_consumed] + "\n" * comment.newlines, - ), - ) - return True + ignored_nodes = list(generate_ignored_nodes(leaf, comment)) + if not ignored_nodes: + continue - previous_consumed = comment.consumed + first = ignored_nodes[0] # Can be a container node with the `leaf`. + parent = first.parent + prefix = first.prefix + first.prefix = prefix[comment.consumed :] + hidden_value = "".join(str(n) for n in ignored_nodes) + if comment.value in FMT_OFF: + hidden_value = comment.value + "\n" + hidden_value + if comment.value in FMT_SKIP: + hidden_value += " " + comment.value + if hidden_value.endswith("\n"): + # That happens when one of the `ignored_nodes` ended with a NEWLINE + # leaf (possibly followed by a DEDENT). + hidden_value = hidden_value[:-1] + first_idx: Optional[int] = None + for ignored in ignored_nodes: + index = ignored.remove() + if first_idx is None: + first_idx = index + assert parent is not None, "INTERNAL ERROR: fmt: on/off handling (1)" + assert first_idx is not None, "INTERNAL ERROR: fmt: on/off handling (2)" + parent.insert_child( + first_idx, + Leaf( + STANDALONE_COMMENT, + hidden_value, + prefix=prefix[:previous_consumed] + "\n" * comment.newlines, + ), + ) + return True return False -def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]: +def generate_ignored_nodes(leaf: Leaf, comment: ProtoComment) -> Iterator[LN]: """Starting from the container of `leaf`, generate all leaves until `# fmt: on`. + If comment is skip, returns leaf only. Stops at the end of the block. """ container: Optional[LN] = container_of(leaf) + if comment.value in FMT_SKIP: + prev_sibling = leaf.prev_sibling + if comment.value in leaf.prefix and prev_sibling is not None: + leaf.prefix = leaf.prefix.replace(comment.value, "") + siblings = [prev_sibling] + while ( + "\n" not in prev_sibling.prefix + and prev_sibling.prev_sibling is not None + ): + prev_sibling = prev_sibling.prev_sibling + siblings.insert(0, prev_sibling) + for sibling in siblings: + yield sibling + elif leaf.parent is not None: + yield leaf.parent + return while container is not None and container.type != token.ENDMARKER: if is_fmt_on(container): return @@ -5333,6 +5619,7 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool: Returns whether the node should itself be wrapped in invisible parentheses. """ + if ( node.type != syms.atom or is_empty_tuple(node) @@ -5342,6 +5629,18 @@ def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool: ): return False + if is_walrus_assignment(node): + if parent.type in [ + syms.annassign, + syms.expr_stmt, + syms.assert_stmt, + syms.return_stmt, + # these ones aren't useful to end users, but they do please fuzzers + syms.for_stmt, + syms.del_stmt, + ]: + return False + first = node.children[0] last = node.children[-1] if first.type == token.LPAR and last.type == token.RPAR: @@ -5403,6 +5702,17 @@ def unwrap_singleton_parenthesis(node: LN) -> Optional[LN]: return wrapped +def first_child_is_arith(node: Node) -> bool: + """Whether first child is an arithmetic or a binary arithmetic expression""" + expr_types = { + syms.arith_expr, + syms.shift_expr, + syms.xor_expr, + syms.and_expr, + } + return bool(node.children and node.children[0].type in expr_types) + + def wrap_in_parentheses(parent: Node, child: LN, *, visible: bool = True) -> None: """Wrap `child` in parentheses. @@ -5443,6 +5753,49 @@ def is_walrus_assignment(node: LN) -> bool: return inner is not None and inner.type == syms.namedexpr_test +def is_simple_decorator_trailer(node: LN, last: bool = False) -> bool: + """Return True iff `node` is a trailer valid in a simple decorator""" + return node.type == syms.trailer and ( + ( + len(node.children) == 2 + and node.children[0].type == token.DOT + and node.children[1].type == token.NAME + ) + # last trailer can be arguments + or ( + last + and len(node.children) == 3 + and node.children[0].type == token.LPAR + # and node.children[1].type == syms.argument + and node.children[2].type == token.RPAR + ) + ) + + +def is_simple_decorator_expression(node: LN) -> bool: + """Return True iff `node` could be a 'dotted name' decorator + + This function takes the node of the 'namedexpr_test' of the new decorator + grammar and test if it would be valid under the old decorator grammar. + + The old grammar was: decorator: @ dotted_name [arguments] NEWLINE + The new grammar is : decorator: @ namedexpr_test NEWLINE + """ + if node.type == token.NAME: + return True + if node.type == syms.power: + if node.children: + return ( + node.children[0].type == token.NAME + and all(map(is_simple_decorator_trailer, node.children[1:-1])) + and ( + len(node.children) < 2 + or is_simple_decorator_trailer(node.children[-1], last=True) + ) + ) + return False + + def is_yield(node: LN) -> bool: """Return True if `node` holds a `yield` or `yield from` expression.""" if node.type == syms.yield_expr: @@ -5561,24 +5914,63 @@ def ensure_visible(leaf: Leaf) -> None: leaf.value = ")" -def should_explode(line: Line, opening_bracket: Leaf) -> bool: - """Should `line` immediately be split with `delimiter_split()` after RHS?""" +def should_split_line(line: Line, opening_bracket: Leaf) -> bool: + """Should `line` be immediately split with `delimiter_split()` after RHS?""" - if not ( - opening_bracket.parent - and opening_bracket.parent.type in {syms.atom, syms.import_from} - and opening_bracket.value in "[{(" - ): + if not (opening_bracket.parent and opening_bracket.value in "[{("): return False + # We're essentially checking if the body is delimited by commas and there's more + # than one of them (we're excluding the trailing comma and if the delimiter priority + # is still commas, that means there's more). + exclude = set() + trailing_comma = False try: last_leaf = line.leaves[-1] - exclude = {id(last_leaf)} if last_leaf.type == token.COMMA else set() + if last_leaf.type == token.COMMA: + trailing_comma = True + exclude.add(id(last_leaf)) max_priority = line.bracket_tracker.max_delimiter_priority(exclude=exclude) except (IndexError, ValueError): return False - return max_priority == COMMA_PRIORITY + return max_priority == COMMA_PRIORITY and ( + (line.mode.magic_trailing_comma and trailing_comma) + # always explode imports + or opening_bracket.parent.type in {syms.atom, syms.import_from} + ) + + +def is_one_tuple_between(opening: Leaf, closing: Leaf, leaves: List[Leaf]) -> bool: + """Return True if content between `opening` and `closing` looks like a one-tuple.""" + if opening.type != token.LPAR and closing.type != token.RPAR: + return False + + depth = closing.bracket_depth + 1 + for _opening_index, leaf in enumerate(leaves): + if leaf is opening: + break + + else: + raise LookupError("Opening paren not found in `leaves`") + + commas = 0 + _opening_index += 1 + for leaf in leaves[_opening_index:]: + 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 in { + syms.arglist, + syms.typedargslist, + }: + commas += 1 + break + + return commas < 2 def get_features_used(node: Node) -> Set[Feature]: @@ -5589,6 +5981,8 @@ def get_features_used(node: Node) -> Set[Feature]: - underscores in numeric literals; - trailing commas after * or ** in function signatures and calls; - positional only arguments in function signatures and lambdas; + - assignment expression; + - relaxed decorator syntax; """ features: Set[Feature] = set() for n in node.pre_order(): @@ -5608,6 +6002,12 @@ def get_features_used(node: Node) -> Set[Feature]: elif n.type == token.COLONEQUAL: features.add(Feature.ASSIGNMENT_EXPRESSIONS) + elif n.type == syms.decorator: + if len(n.children) > 1 and not is_simple_decorator_expression( + n.children[1] + ): + features.add(Feature.RELAXED_DECORATORS) + elif ( n.type in {syms.typedargslist, syms.arglist} and n.children @@ -5645,11 +6045,13 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf a preceding closing bracket fits in one line. Yielded sets are cumulative (contain results of previous yields, too). First - set is empty. + set is empty, unless the line should explode, in which case bracket pairs until + the one that needs to explode are omitted. """ omit: Set[LeafID] = set() - yield omit + if not line.magic_trailing_comma: + yield omit length = 4 * line.depth opening_bracket: Optional[Leaf] = None @@ -5668,9 +6070,22 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf if leaf is opening_bracket: opening_bracket = None elif leaf.type in CLOSING_BRACKETS: + prev = line.leaves[index - 1] if index > 0 else None + if ( + prev + and prev.type == token.COMMA + and not is_one_tuple_between( + leaf.opening_bracket, leaf, line.leaves + ) + ): + # Never omit bracket pairs with trailing commas. + # We need to explode on those. + break + inner_brackets.add(id(leaf)) elif leaf.type in CLOSING_BRACKETS: - if index > 0 and line.leaves[index - 1].type in OPENING_BRACKETS: + prev = line.leaves[index - 1] if index > 0 else None + if prev and prev.type in OPENING_BRACKETS: # Empty brackets would fail a split so treat them as "inner" # brackets (e.g. only add them to the `omit` set if another # pair of brackets was good enough. @@ -5683,6 +6098,15 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf inner_brackets.clear() yield omit + if ( + prev + and prev.type == token.COMMA + and not is_one_tuple_between(leaf.opening_bracket, leaf, line.leaves) + ): + # Never omit bracket pairs with trailing commas. + # We need to explode on those. + break + if leaf.value: opening_bracket = leaf.opening_bracket closing_bracket = leaf @@ -5740,7 +6164,7 @@ def get_future_imports(node: Node) -> Set[str]: @lru_cache() def get_gitignore(root: Path) -> PathSpec: - """ Return a PathSpec matching gitignore content if present.""" + """Return a PathSpec matching gitignore content if present.""" gitignore = root / ".gitignore" lines: List[str] = [] if gitignore.is_file(): @@ -5749,16 +6173,51 @@ def get_gitignore(root: Path) -> PathSpec: return PathSpec.from_lines("gitwildmatch", lines) +def normalize_path_maybe_ignore( + path: Path, root: Path, report: "Report" +) -> Optional[str]: + """Normalize `path`. May return `None` if `path` was ignored. + + `report` is where "path ignored" output goes. + """ + try: + abspath = path if path.is_absolute() else Path.cwd() / path + normalized_path = abspath.resolve().relative_to(root).as_posix() + except OSError as e: + report.path_ignored(path, f"cannot be read because {e}") + return None + + except ValueError: + if path.is_symlink(): + report.path_ignored(path, f"is a symbolic link that points outside {root}") + return None + + raise + + return normalized_path + + +def path_is_excluded( + normalized_path: str, + pattern: Optional[Pattern[str]], +) -> bool: + match = pattern.search(normalized_path) if pattern else None + return bool(match and match.group(0)) + + def gen_python_files( paths: Iterable[Path], root: Path, include: Optional[Pattern[str]], - exclude_regexes: Iterable[Pattern[str]], + exclude: Pattern[str], + extend_exclude: Optional[Pattern[str]], + force_exclude: Optional[Pattern[str]], report: "Report", gitignore: PathSpec, ) -> Iterator[Path]: """Generate all files under `path` whose paths are not excluded by the - `exclude` regex, but are included by the `include` regex. + `exclude_regex`, `extend_exclude`, or `force_exclude` regexes, + but are included by the `include` regex. Symbolic links pointing outside of the `root` directory are ignored. @@ -5766,43 +6225,44 @@ def gen_python_files( """ assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}" for child in paths: - # Then ignore with `exclude` option. - try: - normalized_path = child.resolve().relative_to(root).as_posix() - except OSError as e: - report.path_ignored(child, f"cannot be read because {e}") + normalized_path = normalize_path_maybe_ignore(child, root, report) + if normalized_path is None: continue - except ValueError: - if child.is_symlink(): - report.path_ignored( - child, f"is a symbolic link that points outside {root}" - ) - continue - - raise # First ignore files matching .gitignore if gitignore.match_file(normalized_path): report.path_ignored(child, "matches the .gitignore file content") continue + # Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options. normalized_path = "/" + normalized_path if child.is_dir(): normalized_path += "/" - is_excluded = False - for exclude in exclude_regexes: - exclude_match = exclude.search(normalized_path) if exclude else None - if exclude_match and exclude_match.group(0): - report.path_ignored(child, "matches the --exclude regular expression") - is_excluded = True - break - if is_excluded: + if path_is_excluded(normalized_path, exclude): + report.path_ignored(child, "matches the --exclude regular expression") + continue + + if path_is_excluded(normalized_path, extend_exclude): + report.path_ignored( + child, "matches the --extend-exclude regular expression" + ) + continue + + if path_is_excluded(normalized_path, force_exclude): + report.path_ignored(child, "matches the --force-exclude regular expression") continue if child.is_dir(): yield from gen_python_files( - child.iterdir(), root, include, exclude_regexes, report, gitignore + child.iterdir(), + root, + include, + exclude, + extend_exclude, + force_exclude, + report, + gitignore, ) elif child.is_file(): @@ -5812,11 +6272,11 @@ def gen_python_files( @lru_cache() -def find_project_root(srcs: Iterable[str]) -> Path: +def find_project_root(srcs: Tuple[str, ...]) -> Path: """Return a directory containing .git, .hg, or pyproject.toml. - That directory can be one of the directories passed in `srcs` or their - common parent. + That directory will be a common parent of all files and directories + passed in `srcs`. If no directory in the tree contains a marker that would specify it's the project root, the root of the file system is returned. @@ -5824,11 +6284,20 @@ def find_project_root(srcs: Iterable[str]) -> Path: if not srcs: return Path("/").resolve() - common_base = min(Path(src).resolve() for src in srcs) - if common_base.is_dir(): - # Append a fake file so `parents` below returns `common_base_dir`, too. - common_base /= "fake-file" - for directory in common_base.parents: + path_srcs = [Path(Path.cwd(), src).resolve() for src in srcs] + + # A list of lists of parents for each 'src'. 'src' is included as a + # "parent" of itself if it is a directory + src_parents = [ + list(path.parents) + ([path] if path.is_dir() else []) for path in path_srcs + ] + + common_base = max( + set.intersection(*(set(parents) for parents in src_parents)), + key=lambda path: path.parts, + ) + + for directory in (common_base, *common_base.parents): if (directory / ".git").exists(): return directory @@ -5841,6 +6310,22 @@ def find_project_root(srcs: Iterable[str]) -> Path: return directory +@lru_cache() +def find_user_pyproject_toml() -> Path: + r"""Return the path to the top-level user configuration for black. + + This looks for ~\.black on Windows and ~/.config/black on Linux and other + Unix systems. + """ + if sys.platform == "win32": + # Windows + user_config_path = Path.home() / ".black" + else: + config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config") + user_config_path = Path(config_root).expanduser() / "black" + return user_config_path.resolve() + + @dataclass class Report: """Provides a reformatting counter. Can be rendered with `str(report)`.""" @@ -5942,7 +6427,12 @@ def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]: return ast3.parse(src, filename, feature_version=feature_version) except SyntaxError: continue - + if ast27.__name__ == "ast": + raise SyntaxError( + "The requested source code has invalid Python 3 syntax.\n" + "If you are trying to format Python 2 files please reinstall Black" + " with the 'python2' extra: `python3 -m pip install black[python2]`." + ) return ast27.parse(src) @@ -6008,12 +6498,22 @@ def _stringify_ast( # Constant strings may be indented across newlines, if they are # docstrings; fold spaces after newlines when comparing. Similarly, # trailing and leading space may be removed. + # Note that when formatting Python 2 code, at least with Windows + # line-endings, docstrings can end up here as bytes instead of + # str so make sure that we handle both cases. if ( isinstance(node, ast.Constant) and field == "value" - and isinstance(value, str) + and isinstance(value, (str, bytes)) ): - normalized = re.sub(r" *\n[ \t]+", "\n ", value).strip() + lineend = "\n" if isinstance(value, str) else b"\n" + # To normalize, we strip any leading and trailing space from + # each line... + stripped = [line.strip() for line in value.splitlines()] + normalized = lineend.join(stripped) # type: ignore[attr-defined] + # ...and remove any blank lines at the beginning and end of + # the whole string + normalized = normalized.strip() else: normalized = value yield f"{' ' * (depth+2)}{normalized!r}, # {value.__class__.__name__}" @@ -6021,7 +6521,7 @@ def _stringify_ast( yield f"{' ' * depth}) # /{node.__class__.__name__}" -def assert_equivalent(src: str, dst: str) -> None: +def assert_equivalent(src: str, dst: str, *, pass_num: int = 1) -> None: """Raise AssertionError if `src` and `dst` aren't equivalent.""" try: src_ast = parse_ast(src) @@ -6036,9 +6536,9 @@ def assert_equivalent(src: str, dst: str) -> None: 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}. Please report a bug" - " on https://github.com/psf/black/issues. This invalid output might be" - f" helpful: {log}" + f"INTERNAL ERROR: Black produced invalid code on pass {pass_num}: {exc}. " + "Please report a bug on https://github.com/psf/black/issues. " + f"This invalid output might be helpful: {log}" ) from None src_ast_str = "\n".join(_stringify_ast(src_ast)) @@ -6047,8 +6547,8 @@ def assert_equivalent(src: str, dst: str) -> None: log = dump_to_file(diff(src_ast_str, dst_ast_str, "src", "dst")) raise AssertionError( "INTERNAL ERROR: Black produced code that is not equivalent to the" - " source. Please report a bug on https://github.com/psf/black/issues. " - f" This diff might be helpful: {log}" + f" source on pass {pass_num}. Please report a bug on " + f"https://github.com/psf/black/issues. This diff might be helpful: {log}" ) from None @@ -6057,6 +6557,7 @@ def assert_stable(src: str, dst: str, mode: Mode) -> None: newdst = format_str(dst, mode=mode) if dst != newdst: log = dump_to_file( + str(mode), diff(src, dst, "source", "first pass"), diff(dst, newdst, "first pass", "second pass"), ) @@ -6068,14 +6569,14 @@ def assert_stable(src: str, dst: str, mode: Mode) -> None: @mypyc_attr(patchable=True) -def dump_to_file(*output: str) -> str: +def dump_to_file(*output: str, ensure_final_newline: bool = True) -> str: """Dump `output` to a temporary file. Return path to the file.""" with tempfile.NamedTemporaryFile( mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8" ) as f: for lines in output: f.write(lines) - if lines and lines[-1] != "\n": + if ensure_final_newline and lines and lines[-1] != "\n": f.write("\n") return f.name @@ -6093,11 +6594,20 @@ 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 - a_lines = [line + "\n" for line in a.splitlines()] - b_lines = [line + "\n" for line in b.splitlines()] - return "".join( - difflib.unified_diff(a_lines, b_lines, fromfile=a_name, tofile=b_name, n=5) - ) + a_lines = [line for line in a.splitlines(keepends=True)] + b_lines = [line for line in b.splitlines(keepends=True)] + diff_lines = [] + for line in difflib.unified_diff( + a_lines, b_lines, fromfile=a_name, tofile=b_name, n=5 + ): + # Work around https://bugs.python.org/issue2142 + # See https://www.gnu.org/software/diffutils/manual/html_node/Incomplete-Lines.html + if line[-1] == "\n": + diff_lines.append(line) + else: + diff_lines.append(line + "\n") + diff_lines.append("\\ No newline at end of file\n") + return "".join(diff_lines) def cancel(tasks: Iterable["asyncio.Task[Any]"]) -> None: @@ -6233,7 +6743,11 @@ def can_be_split(line: Line) -> bool: return True -def can_omit_invisible_parens(line: Line, line_length: int) -> bool: +def can_omit_invisible_parens( + line: Line, + line_length: int, + omit_on_explode: Collection[LeafID] = (), +) -> bool: """Does `line` have a shape safe to reformat without optional parens around it? Returns True for only a subset of potentially nice looking formattings but @@ -6256,37 +6770,27 @@ def can_omit_invisible_parens(line: Line, line_length: int) -> bool: assert len(line.leaves) >= 2, "Stranded delimiter" - first = line.leaves[0] - second = line.leaves[1] - penultimate = line.leaves[-2] - last = line.leaves[-1] - # With a single delimiter, omit if the expression starts or ends with # a bracket. + first = line.leaves[0] + second = line.leaves[1] if first.type in OPENING_BRACKETS and second.type not in CLOSING_BRACKETS: - remainder = False - length = 4 * line.depth - for _index, leaf, leaf_length in enumerate_with_length(line): - if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first: - remainder = True - if remainder: - length += leaf_length - if length > line_length: - break - - if leaf.type in OPENING_BRACKETS: - # There are brackets we can further split on. - remainder = False - - else: - # checked the entire string and line length wasn't exceeded - if len(line.leaves) == _index + 1: - return True + if _can_omit_opening_paren(line, first=first, line_length=line_length): + return True # Note: we are not returning False here because a line might have *both* # a leading opening bracket and a trailing closing bracket. If the # opening bracket doesn't match our rule, maybe the closing will. + penultimate = line.leaves[-2] + last = line.leaves[-1] + if line.magic_trailing_comma: + try: + penultimate, last = last_two_except(line.leaves, omit=omit_on_explode) + except LookupError: + # Turns out we'd omit everything. We cannot skip the optional parentheses. + return False + if ( last.type == token.RPAR or last.type == token.RBRACE @@ -6307,21 +6811,120 @@ def can_omit_invisible_parens(line: Line, line_length: int) -> bool: # unnecessary. return True - length = 4 * line.depth - seen_other_brackets = False - for _index, leaf, leaf_length in enumerate_with_length(line): + if line.magic_trailing_comma and penultimate.type == token.COMMA: + # The rightmost non-omitted bracket pair is the one we want to explode on. + return True + + if _can_omit_closing_paren(line, last=last, line_length=line_length): + return True + + return False + + +def _can_omit_opening_paren(line: Line, *, first: Leaf, line_length: int) -> bool: + """See `can_omit_invisible_parens`.""" + remainder = False + length = 4 * line.depth + _index = -1 + for _index, leaf, leaf_length in enumerate_with_length(line): + if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first: + remainder = True + if remainder: length += leaf_length - if leaf is last.opening_bracket: - if seen_other_brackets or length <= line_length: - return True + if length > line_length: + break - elif leaf.type in OPENING_BRACKETS: + if leaf.type in OPENING_BRACKETS: # There are brackets we can further split on. - seen_other_brackets = True + remainder = False + + else: + # checked the entire string and line length wasn't exceeded + if len(line.leaves) == _index + 1: + return True + + return False + + +def _can_omit_closing_paren(line: Line, *, last: Leaf, line_length: int) -> bool: + """See `can_omit_invisible_parens`.""" + length = 4 * line.depth + seen_other_brackets = False + for _index, leaf, leaf_length in enumerate_with_length(line): + length += leaf_length + if leaf is last.opening_bracket: + if seen_other_brackets or length <= line_length: + return True + + elif leaf.type in OPENING_BRACKETS: + # There are brackets we can further split on. + seen_other_brackets = True return False +def last_two_except(leaves: List[Leaf], omit: Collection[LeafID]) -> Tuple[Leaf, Leaf]: + """Return (penultimate, last) leaves skipping brackets in `omit` and contents.""" + stop_after = None + last = None + for leaf in reversed(leaves): + if stop_after: + if leaf is stop_after: + stop_after = None + continue + + if last: + return leaf, last + + if id(leaf) in omit: + stop_after = leaf.opening_bracket + else: + last = leaf + else: + raise LookupError("Last two leaves were also skipped") + + +def run_transformer( + line: Line, + transform: Transformer, + mode: Mode, + features: Collection[Feature], + *, + line_str: str = "", +) -> List[Line]: + if not line_str: + line_str = line_to_string(line) + result: List[Line] = [] + for transformed_line in transform(line, features): + if str(transformed_line).strip("\n") == line_str: + raise CannotTransform("Line transformer returned an unchanged result") + + result.extend(transform_line(transformed_line, mode=mode, features=features)) + + if not ( + transform.__name__ == "rhs" + and line.bracket_tracker.invisible + and not any(bracket.value for bracket in line.bracket_tracker.invisible) + and not line.contains_multiline_strings() + and not result[0].contains_uncollapsable_type_comments() + and not result[0].contains_unsplittable_type_ignore() + and not is_line_short_enough(result[0], line_length=mode.line_length) + ): + return result + + line_copy = line.clone() + append_leaves(line_copy, line, line.leaves) + features_fop = set(features) | {Feature.FORCE_OPTIONAL_PARENTHESES} + second_opinion = run_transformer( + line_copy, transform, mode, features_fop, line_str=line_str + ) + if all( + is_line_short_enough(ln, line_length=mode.line_length) for ln in second_opinion + ): + result = second_opinion + return result + + def get_cache_file(mode: Mode) -> Path: return CACHE_DIR / f"cache.{mode.get_cache_key()}.pickle" @@ -6358,8 +6961,8 @@ def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set """ todo, done = set(), set() for src in sources: - src = src.resolve() - if cache.get(src) != get_cache_info(src): + res_src = src.resolve() + if cache.get(str(res_src)) != get_cache_info(res_src): todo.add(src) else: done.add(src) @@ -6371,7 +6974,10 @@ def write_cache(cache: Cache, sources: Iterable[Path], mode: Mode) -> None: 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}} + new_cache = { + **cache, + **{str(src.resolve()): get_cache_info(src) for src in sources}, + } with tempfile.NamedTemporaryFile(dir=str(cache_file.parent), delete=False) as f: pickle.dump(new_cache, f, protocol=4) os.replace(f.name, cache_file) @@ -6407,13 +7013,48 @@ def patched_main() -> None: main() +def is_docstring(leaf: Leaf) -> bool: + if prev_siblings_are( + leaf.parent, [None, token.NEWLINE, token.INDENT, syms.simple_stmt] + ): + return True + + # Multiline docstring on the same line as the `def`. + if prev_siblings_are(leaf.parent, [syms.parameters, token.COLON, syms.simple_stmt]): + # `syms.parameters` is only used in funcdefs and async_funcdefs in the Python + # grammar. We're safe to return True without further checks. + return True + + return False + + +def lines_with_leading_tabs_expanded(s: str) -> List[str]: + """ + Splits string into lines and expands only leading tabs (following the normal + Python rules) + """ + lines = [] + for line in s.splitlines(): + # Find the index of the first non-whitespace character after a string of + # whitespace that includes at least one tab + match = re.match(r"\s*\t+\s*(\S)", line) + if match: + first_non_whitespace_idx = match.start(1) + + lines.append( + line[:first_non_whitespace_idx].expandtabs() + + line[first_non_whitespace_idx:] + ) + else: + lines.append(line) + return lines + + def fix_docstring(docstring: str, prefix: str) -> str: # https://www.python.org/dev/peps/pep-0257/#handling-docstring-indentation if not docstring: return "" - # Convert tabs to spaces (following the normal Python rules) - # and split into a list of lines: - lines = docstring.expandtabs().splitlines() + lines = lines_with_leading_tabs_expanded(docstring) # Determine minimum indentation (first line doesn't count): indent = sys.maxsize for line in lines[1:]: @@ -6430,7 +7071,6 @@ def fix_docstring(docstring: str, prefix: str) -> str: trimmed.append(prefix + stripped_line) else: trimmed.append("") - # Return a single string: return "\n".join(trimmed)