X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/788268bc39a87d37a24d203fa5ee7b3953af3446..33d861371533d64a54af419798f1ddfa692f388c:/src/black/__init__.py?ds=inline diff --git a/src/black/__init__.py b/src/black/__init__.py index 8d0c70f..c7c5d72 100644 --- a/src/black/__init__.py +++ b/src/black/__init__.py @@ -68,6 +68,7 @@ DEFAULT_LINE_LENGTH = 88 DEFAULT_EXCLUDES = r"/(\.direnv|\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|\.svn|_build|buck-out|build|dist)/" # noqa: B950 DEFAULT_INCLUDES = r"\.pyi?$" CACHE_DIR = Path(user_cache_dir("black", version=__version__)) +STDIN_PLACEHOLDER = "__BLACK_STDIN_FILENAME__" STRING_PREFIX_CHARS: Final = "furbFURB" # All possible string prefix characters. @@ -112,6 +113,10 @@ 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 +179,12 @@ 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 +198,8 @@ 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 +237,17 @@ 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, + }, } @@ -439,7 +455,16 @@ def target_version_option_callback( type=str, 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( @@ -501,6 +526,7 @@ def main( include: str, exclude: str, force_exclude: Optional[str], + stdin_filename: Optional[str], src: Tuple[str, ...], config: Optional[str], ) -> None: @@ -533,6 +559,7 @@ def main( exclude=exclude, force_exclude=force_exclude, report=report, + stdin_filename=stdin_filename, ) path_empty( @@ -572,6 +599,7 @@ def get_sources( exclude: str, force_exclude: Optional[str], report: "Report", + stdin_filename: Optional[str], ) -> Set[Path]: """Compute the set of files to be formatted.""" try: @@ -598,22 +626,14 @@ def get_sources( gitignore = get_gitignore(root) for s in src: - p = Path(s) - if p.is_dir(): - sources.update( - gen_python_files( - p.iterdir(), - root, - include_regex, - exclude_regex, - force_exclude_regex, - report, - gitignore, - ) - ) - elif s == "-": - sources.add(p) - elif p.is_file(): + 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 @@ -628,6 +648,23 @@ def get_sources( 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_regex, + force_exclude_regex, + report, + gitignore, + ) + ) + elif s == "-": sources.add(p) else: err(f"invalid path: {s}") @@ -640,10 +677,9 @@ 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( @@ -656,12 +692,23 @@ 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 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): @@ -735,7 +782,7 @@ 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): @@ -746,7 +793,7 @@ 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() @@ -845,9 +892,9 @@ 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 @@ -857,30 +904,22 @@ 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( @@ -921,13 +960,13 @@ 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) @@ -947,7 +986,7 @@ 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: ... @@ -1061,7 +1100,7 @@ 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)): @@ -1284,6 +1323,7 @@ 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. @@ -1307,8 +1347,16 @@ 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) @@ -1321,6 +1369,8 @@ 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) @@ -1621,14 +1671,13 @@ class Line: def maybe_should_explode(self, closing: Leaf) -> bool: """Return True if this line should explode (always be split), that is when: - - there's a pre-existing trailing comma here; and + - there's a trailing comma here; and - it's not a one-tuple. """ if not ( closing.type in CLOSING_BRACKETS and self.leaves and self.leaves[-1].type == token.COMMA - and not self.leaves[-1].was_checked # pre-existing ): return False @@ -1820,6 +1869,10 @@ 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 ( @@ -1843,8 +1896,11 @@ 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 @@ -2031,14 +2087,20 @@ 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) + lead_len = len(prefix) + 3 + tail_len = -3 + indent = " " * 4 * self.current_line.depth + docstring = fix_docstring(leaf.value[lead_len:tail_len], indent) + if docstring: + if leaf.value[lead_len - 1] == docstring[0]: + docstring = " " + docstring + if leaf.value[tail_len + 1] == docstring[-1]: + docstring = docstring + " " + leaf.value = leaf.value[0:lead_len] + docstring + leaf.value[tail_len:] yield from self.visit_default(leaf) @@ -2157,6 +2219,9 @@ 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 @@ -2627,29 +2692,40 @@ def transform_line( else: def rhs(line: Line, features: Collection[Feature]) -> Iterator[Line]: + """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) + # 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_split, string_paren_wrap, rhs, ] @@ -2671,17 +2747,8 @@ def transform_line( # 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 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) - ) + result = run_transformer(line, transform, mode, features, line_str=line_str) except CannotTransform: continue else: @@ -2722,6 +2789,7 @@ class StringTransformer(ABC): line_length: int normalize_strings: bool + __name__ = "StringTransformer" @abstractmethod def do_match(self, line: Line) -> TMatchResult: @@ -2876,11 +2944,8 @@ 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. @@ -2968,7 +3033,7 @@ 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] @@ -3129,6 +3194,7 @@ 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. @@ -3137,6 +3203,26 @@ 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 @@ -3185,7 +3271,8 @@ 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 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. @@ -3209,6 +3296,14 @@ 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) @@ -3293,11 +3388,17 @@ 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() @@ -3460,9 +3561,12 @@ 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] @@ -3528,7 +3632,8 @@ 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 @@ -3548,13 +3653,14 @@ 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: @@ -3868,11 +3974,23 @@ 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 @@ -3978,12 +4096,13 @@ 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 @@ -4182,7 +4301,7 @@ 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 @@ -4567,7 +4686,9 @@ 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. @@ -4583,7 +4704,7 @@ def append_leaves(new_line: Line, old_line: Line, leaves: List[Leaf]) -> None: for old_leaf in 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) @@ -4740,8 +4861,9 @@ def right_hand_split( tail = bracket_split_build_line(tail_leaves, line, opening_bracket) bracket_split_succeeded_or_raise(head, body, tail) if ( + Feature.FORCE_OPTIONAL_PARENTHESES not in features # the opening bracket is an optional paren - opening_bracket.type == token.LPAR + and opening_bracket.type == token.LPAR and not opening_bracket.value # the closing bracket is an optional paren and closing_bracket.type == token.RPAR @@ -4752,7 +4874,7 @@ 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: @@ -4839,7 +4961,6 @@ def bracket_split_build_line( if leaves[i].type != token.COMMA: new_comma = Leaf(token.COMMA, ",") - new_comma.was_checked = True leaves.insert(i + 1, new_comma) break @@ -4935,7 +5056,6 @@ def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[ and current_line.leaves[-1].type != STANDALONE_COMMENT ): new_comma = Leaf(token.COMMA, ",") - new_comma.was_checked = True current_line.append(new_comma) yield current_line @@ -5114,31 +5234,52 @@ 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" + + Uses lowercase because of similarity between "B" and "8", which + can cause security issues. + see: https://github.com/psf/black/issues/1692 + """ + + before, after = text[:2], text[2:] + return f"{before}{after.lower()}" + + +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: @@ -5178,9 +5319,9 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: if check_lpar: if is_walrus_assignment(child): - continue + pass - if child.type == syms.atom: + elif child.type == syms.atom: if maybe_make_parens_invisible_in_atom(child, parent=node): wrap_in_parentheses(node, child, visible=False) elif is_one_tuple(child): @@ -5440,6 +5581,49 @@ 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: @@ -5559,7 +5743,7 @@ def ensure_visible(leaf: Leaf) -> None: def should_split_body_explode(line: Line, opening_bracket: Leaf) -> bool: - """Should `line` immediately be split with `delimiter_split()` after RHS?""" + """Should `line` be immediately split with `delimiter_split()` after RHS?""" if not (opening_bracket.parent and opening_bracket.value in "[{("): return False @@ -5568,25 +5752,28 @@ def should_split_body_explode(line: Line, opening_bracket: Leaf) -> bool: # 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() - pre_existing_trailing_comma = False + trailing_comma = False try: last_leaf = line.leaves[-1] if last_leaf.type == token.COMMA: - pre_existing_trailing_comma = not last_leaf.was_checked + 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 and ( + trailing_comma # always explode imports - opening_bracket.parent.type in {syms.atom, syms.import_from} - or pre_existing_trailing_comma + 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: @@ -5622,6 +5809,8 @@ 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(): @@ -5641,6 +5830,12 @@ 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 @@ -5678,11 +5873,13 @@ 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.should_explode: + yield omit length = 4 * line.depth opening_bracket: Optional[Leaf] = None @@ -5701,9 +5898,23 @@ 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 ( + line.should_explode + and 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. @@ -5716,6 +5927,16 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf inner_brackets.clear() yield omit + if ( + line.should_explode + and 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 @@ -5790,7 +6011,8 @@ def normalize_path_maybe_ignore( `report` is where "path ignored" output goes. """ try: - normalized_path = path.resolve().relative_to(root).as_posix() + 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 @@ -6121,6 +6343,7 @@ 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"), ) @@ -6297,7 +6520,11 @@ 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 @@ -6320,37 +6547,27 @@ 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.should_explode: + 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 @@ -6371,21 +6588,120 @@ 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.should_explode 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" @@ -6471,13 +6787,53 @@ def patched_main() -> None: main() +def is_docstring(leaf: Leaf) -> bool: + if not is_multiline_string(leaf): + # For the purposes of docstring re-indentation, we don't need to do anything + # with single-line docstrings. + return False + + 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:]: @@ -6494,7 +6850,6 @@ def fix_docstring(docstring: str, prefix: str) -> str: trimmed.append(prefix + stripped_line) else: trimmed.append("") - # Return a single string: return "\n".join(trimmed)